├── .gitattributes ├── .github └── workflows │ └── main.yml ├── .gitignore ├── Antlr ├── AntlrCommon.cs ├── BuildAntlr.csproj ├── CountryListVisitor.cs ├── OwnTypeListVisitor.cs ├── RealtyTypeAndOwnTypeVisitor.cs ├── RealtyTypeListVisitor.cs ├── SoupVisitor.cs ├── SquareListVisitor.cs ├── StrictVisitor.cs ├── TestAntlr.cs ├── generate_csharp_from_antlr.sh ├── generated │ └── .gitignore ├── readme.txt ├── src │ ├── BaseLexer.g4 │ ├── Common.g4 │ ├── CommonLexer.g4 │ ├── CountryLexer.g4 │ ├── CountryList.g4 │ ├── OwnTypeList.g4 │ ├── OwntypeLexer.g4 │ ├── ParseOtherLexer.g4 │ ├── RealtyTypeAndOwnType.g4 │ ├── RealtyTypeLexer.g4 │ ├── RealtyTypeList.g4 │ ├── SkipOtherLexer.g4 │ ├── Soup.g4 │ ├── SoupLexer.g4 │ ├── SquareList.g4 │ ├── Strict.g4 │ └── StrictLexer.g4 └── test_files │ ├── country_list.txt │ ├── country_list.txt.result │ ├── country_list.txt.result.canon │ ├── soup.txt │ ├── soup.txt.result │ ├── soup.txt.result.canon │ ├── square_and_country.txt │ ├── square_and_country.txt.result │ ├── square_and_country.txt.result.canon │ ├── square_list.txt │ ├── square_list.txt.result │ ├── square_list.txt.result.canon │ ├── strict.txt │ ├── strict.txt.result │ └── strict.txt.result.canon ├── DeclaratorApiClient ├── ApiClient.cs ├── DeclaratorApiClient.csproj ├── DeclaratorApiException.cs ├── app.config ├── auth.config └── auth.txt ├── EP.SdkCore ├── EP.Analyzer.Biblio.Core.dll ├── EP.Analyzer.Biblio.Core.xml ├── EP.Analyzer.Decree.Core.dll ├── EP.Analyzer.Decree.Core.xml ├── EP.Analyzer.Goods.Core.dll ├── EP.Analyzer.Goods.Core.xml ├── EP.Analyzer.Location.Core.dll ├── EP.Analyzer.Location.Core.xml ├── EP.Analyzer.Misc.Core.dll ├── EP.Analyzer.Misc.Core.xml ├── EP.Analyzer.Organization.Core.dll ├── EP.Analyzer.Organization.Core.xml ├── EP.Analyzer.Person.Core.dll ├── EP.Analyzer.Person.Core.xml ├── EP.Analyzer.Semantic.Core.dll ├── EP.Analyzer.Semantic.Core.xml ├── EP.Analyzer.Util.Core.dll ├── EP.Analyzer.Util.Core.xml ├── EP.Core.Core.dll ├── EP.Core.Core.xml ├── EP.Explanatory.Core.dll ├── EP.Explanatory.Core.xml ├── EP.Morphology.Core.dll ├── EP.Morphology.Core.xml ├── EP.Sdk.Core.csproj ├── EP.Semantic.Core.dll └── EP.Semantic.Core.xml ├── JsonSerialization ├── DeclarationSerializer.cs ├── JsonSerialization.csproj ├── MiscSerializer.cs ├── ValidationReport.cs ├── app.config ├── import-schema-dicts.json └── import-schema.json ├── ParserCommon ├── CMDLineParser.cs ├── LevenshteinDistance.cs ├── ParserCommon.csproj ├── StringMeasure.cs └── TextHelpers.cs ├── PythonUtils ├── requirements.txt └── unarchive.py ├── RegressionTesting ├── Readme.txt ├── RegressionTesting.csproj ├── ValidityTests.cs ├── files │ ├── 9037 │ │ ├── rykovodstvo_2013.xlsx │ │ ├── rykovodstvo_2013.xlsx_0.json │ │ └── rykovodstvo_2013.xlsx_1.json │ ├── 10639_32.doc │ ├── 10639_32.doc.json │ ├── 11212_66.doc │ ├── 11212_66.doc.json │ ├── 11480_12.doc │ ├── 11480_12.doc.json │ ├── 12104_33.docx │ ├── 12104_33.docx.json │ ├── 13100_13.doc │ ├── 13100_13.doc.json │ ├── 13133_5.doc │ ├── 13133_5.doc.json │ ├── 13252_58.xls │ ├── 13252_58.xls.json │ ├── 13415_200.doc │ ├── 13415_200.doc.json │ ├── 13931_28.doc │ ├── 13931_28.doc.json │ ├── 14421_139.htm │ ├── 14421_139.htm.json │ ├── 14475_0.doc │ ├── 14475_0.doc.json │ ├── 14712_34.doc │ ├── 14712_34.doc.json │ ├── 14845_84.docx │ ├── 14845_84.docx.json │ ├── 15555_0.html │ ├── 15555_0.html.json │ ├── 15555_1.html │ ├── 15555_1.html.json │ ├── 15584_0.html │ ├── 15584_0.html.json │ ├── 16694.docx │ ├── 16694.docx.json │ ├── 16738_12.html │ ├── 16738_12.html.json │ ├── 17207.xls │ ├── 17207.xls.json │ ├── 17335_3.html │ ├── 17335_3.html_0.json │ ├── 17335_3.html_1.json │ ├── 17335_3.html_2.json │ ├── 17339_24.html │ ├── 17339_24.html_0.json │ ├── 17339_24.html_1.json │ ├── 17339_24.html_2.json │ ├── 17339_24.html_3.json │ ├── 17497.xls │ ├── 17497.xls_0.json │ ├── 17497.xls_1.json │ ├── 17497.xls_10.json │ ├── 17497.xls_11.json │ ├── 17497.xls_2.json │ ├── 17497.xls_3.json │ ├── 17497.xls_4.json │ ├── 17497.xls_5.json │ ├── 17497.xls_6.json │ ├── 17497.xls_7.json │ ├── 17497.xls_8.json │ ├── 17497.xls_9.json │ ├── 18261_22.doc │ ├── 18261_22.doc.json │ ├── 1person_pdf_converted.docx │ ├── 20838.doc │ ├── 20838.doc.json │ ├── 20883_1.docx │ ├── 20883_1.docx.json │ ├── 256_Columns.xlsx │ ├── 256_Columns.xlsx.json │ ├── 26937_42.doc │ ├── 26937_42.doc.json │ ├── 28733.doc │ ├── 28733.doc.json │ ├── 28817.xls │ ├── 28817.xls.json │ ├── 30429.docx │ ├── 30429.docx.json │ ├── 31832.xls │ ├── 31832.xls.json │ ├── 31832_1.xls │ ├── 31832_1.xls.json │ ├── 33793.doc │ ├── 33793.doc.json │ ├── 36922.docx │ ├── 36922.docx.json │ ├── 38527_5.docx │ ├── 38527_5.docx.json │ ├── 4002_0.htm │ ├── 4002_0.htm.json │ ├── 4037_9.htm │ ├── 4037_9.htm.json │ ├── 4067_0.docx │ ├── 4067_0.docx.json │ ├── 4144_28.htm │ ├── 4144_28.htm.json │ ├── 4462_10.doc │ ├── 4462_10.doc.json │ ├── 4479_27.doc │ ├── 4479_27.doc.json │ ├── 4778_0.htm │ ├── 4778_0.htm.json │ ├── 50484_2.xls │ ├── 50484_2.xls.json │ ├── 52668.doc │ ├── 52668.doc.json │ ├── 52991.docx │ ├── 52991.docx.json │ ├── 61662.docx │ ├── 61662.docx.json │ ├── 65098_2.docx │ ├── 65098_2.docx.json │ ├── 65098_5.docx │ ├── 65098_5.docx.json │ ├── 7007_10.html │ ├── 7007_10.html.json │ ├── 7007_8.html │ ├── 7007_8.html.json │ ├── 7022_0.htm │ ├── 7022_0.htm.json │ ├── 74.html │ ├── 74.html.json │ ├── 8562.pdf.docx │ ├── 8562.pdf.docx.json │ ├── 8992_15.doc │ ├── 8992_15.doc.json │ ├── BadColumns.doc │ ├── BadColumns.doc.json │ ├── BadOcr.pdf.docx │ ├── BadOcr.pdf.docx.json │ ├── DepEnergo2010.doc │ ├── DepEnergo2010.doc.json │ ├── DepGosPol2012.doc │ ├── DepGosPol2012.doc.json │ ├── Deputy2015.docx │ ├── Deputy2015.docx.json │ ├── Fsin2011.doc │ ├── Fsin2011.doc.json │ ├── HeaderInsideTable.docx │ ├── HeaderInsideTable.docx.json │ ├── IgnoreTableJunk.docx │ ├── IgnoreTableJunk.docx.json │ ├── IncomeNotFirstLine.docx │ ├── IncomeNotFirstLine.docx.json │ ├── MainWorkPositionIncome.docx │ ├── MainWorkPositionIncome.docx.json │ ├── Mchs2010.doc │ ├── Mchs2010.doc.json │ ├── Mchs2013.doc │ ├── Mchs2013.doc.json │ ├── MinDalVostok2015.xlsx │ ├── MinDalVostok2015.xlsx.json │ ├── MinDalVostok2017.xlsx │ ├── MinDalVostok2017.xlsx.json │ ├── MinEkon2013.docx │ ├── MinEkon2013.docx.json │ ├── MinEkon2015.docx │ ├── MinEkon2015.docx.json │ ├── MinEkonom2014.docx │ ├── MinEkonom2014.docx.json │ ├── MinEkonon2017.docx │ ├── MinEkonon2017.docx.json │ ├── MinKult2011.doc │ ├── MinKult2011.doc.json │ ├── MinKult2012.doc │ ├── MinKult2012.doc.json │ ├── MinKult2012.docx │ ├── MinKult2012.docx.json │ ├── MinKult2015.doc │ ├── MinKult2015.doc.json │ ├── MinKult2015.docx │ ├── MinKult2015.docx.json │ ├── MinObor2012.xls │ ├── MinObor2012.xls.json │ ├── MinObr2012.docx │ ├── MinObr2012.docx.json │ ├── MinObr2016.xlsx │ ├── MinObr2016.xlsx.json │ ├── MinProm2013.docx │ ├── MinProm2013.docx.json │ ├── MinRes2011.doc │ ├── MinRes2011.doc.json │ ├── MinSelhoz2013.xlsx │ ├── MinSelhoz2013.xlsx.json │ ├── MinSelhoz2015.xlsx │ ├── MinSelhoz2015.xlsx.json │ ├── MinSevKavkaz2015.docx │ ├── MinSevKavkaz2015.docx.json │ ├── MinSport2016.docx │ ├── MinSport2016.docx.json │ ├── MinStroy2014.docx │ ├── MinStroy2014.docx.json │ ├── MinStroy2017.xlsx │ ├── MinStroy2017.xlsx.json │ ├── MinStroy2017_1.xlsx │ ├── MinStroy2017_1.xlsx.json │ ├── MinTrans2009.doc │ ├── MinTrans2009.doc.json │ ├── MinTrans2011.docx │ ├── MinTrans2011.docx.json │ ├── MinYust2012.doc │ ├── MinYust2012.doc.json │ ├── MinZdorov2015Full.docx │ ├── MinZdorov2015Full.docx.json │ ├── Nedosekov.docx │ ├── Nedosekov.docx.json │ ├── RealtyNaturalText.docx │ ├── RealtyNaturalText.docx.json │ ├── Spasat2016.docx │ ├── Spasat2016.docx.json │ ├── SpellCheckRealtyType.docx │ ├── SpellCheckRealtyType.docx.json │ ├── Sudia2011.htm │ ├── Sudia2011.htm.json │ ├── Unk2014.xlsx │ ├── Unk2014.xlsx.json │ ├── ZabSud2017.docx │ ├── ZabSud2017.docx.json │ ├── ZagranApp2016.doc │ ├── ZagranApp2016.doc.json │ ├── chab2019.xls │ ├── chab2019.xls.json │ ├── disclosures_format.xls │ ├── disclosures_format.xls.json │ ├── dnko-2014.docx │ ├── dnko-2014.docx.json │ ├── fsin2013.docx │ ├── fsin2013.docx.json │ ├── fsn01.html │ ├── fsn01.html.json │ ├── gorshkova_20190706002511.html │ ├── gorshkova_20190706002511.html_0.json │ ├── gorshkova_20190706002511.html_1.json │ ├── gorshkova_20190706002511.html_2.json │ ├── gorshkova_20190706002511.html_3.json │ ├── gorshkova_20190706002511.html_4.json │ ├── gorshkova_20190706002511.html_5.json │ ├── gorshkova_20190706002511.html_6.json │ ├── gorshkova_20190706002511.html_7.json │ ├── html_title.html │ ├── html_title.html.json │ ├── minfin2016.xlsx │ ├── minfin2016.xlsx.json │ ├── minzdrav2015.docx │ ├── minzdrav2015.docx.json │ ├── niz_kam.docx │ ├── niz_kam.docx.json │ ├── one_line_2017.docx │ ├── one_line_2017.docx.json │ ├── privol.xls │ ├── privol.xls.json │ ├── pudoz_01.docx │ ├── pudoz_01.docx.json │ ├── section_example.toloka_json │ ├── section_example.toloka_json.json │ ├── sov_fed.docx │ ├── sov_fed.docx.json │ ├── sud_2016.doc │ ├── sud_2016.doc.json │ ├── toloka.toloka_json │ ├── toloka.toloka_json.json │ ├── two_tables_2017.docx │ ├── two_tables_2017.docx.json │ ├── year_with_space.docx │ ├── year_with_space.docx.json │ ├── zehova2013.doc │ └── zehova2013.doc.json └── log4net.config ├── Smart.Parser.sln ├── lib ├── Adapters │ ├── AdapterSchemes │ │ ├── IAdapterScheme.cs │ │ └── SovetFederaciiDocxScheme.cs │ ├── AngleHtmlAdapter.cs │ ├── AsposeExcelAdapter.cs │ ├── AsposeLicense.cs │ ├── Azure │ │ ├── AzureDtoCache.cs │ │ └── AzureFormRecognizer.cs │ ├── BigramsHolder.cs │ ├── ConvertToDocxAndFix.cs │ ├── Exceptions │ │ └── AsposeCorruptedFileException.cs │ ├── HtmAdapter.cs │ ├── HtmlSchemes │ │ ├── ArbitrationCourt1.cs │ │ ├── ArbitrationCourt2.cs │ │ └── HtmlScheme.cs │ ├── IAdapter.cs │ ├── IAdapterCell.cs │ ├── IAdapterRow.cs │ ├── MicrosoftExcel.cs │ ├── NpoiExcel.cs │ ├── OpenXmlWord.cs │ └── Section.cs ├── BorderFinder.cs ├── DataHelper.cs ├── Declaration.cs ├── DeclarationField.cs ├── DeclarationProperties.cs ├── JsonWriter.cs ├── Logger.cs ├── Parser.cs ├── ParserBase.cs ├── Person.cs ├── PublicServant.cs ├── RealEstateProperty.cs ├── RealtyParser.cs ├── RelationType.cs ├── Relative.cs ├── Resources │ ├── bigrams.txt │ ├── column_trigrams.txt │ └── log4net.config ├── SecondPassParser.cs ├── Smart.Parser.Lib.csproj ├── SmartParserException.cs ├── TableHeader │ ├── HeaderHelpers.cs │ ├── PredictByDataTrigrams.cs │ ├── ReadHeader.cs │ └── TableHeader.cs └── Vehicle.cs ├── nuget.config ├── src ├── Program.cs ├── Resources │ └── versions.txt ├── Smart.Parser.csproj └── readme.md ├── test ├── ColumnDetectorTest.cs ├── DataHelperTest.cs ├── HeaderHelperTest.cs ├── NpoiExcelAdapterBasicTests.cs ├── Test Samples │ └── 2577 │ │ └── 6.docx ├── Test.csproj ├── TestJsonWriter.cs ├── TestUtil.cs ├── TextMeasure.cs ├── XlsxAdapterTest.cs ├── XlsxTest.cs ├── example.json └── testdata │ ├── 17497.xls │ ├── 18664.docx │ ├── 57715.doc │ ├── 82442.doc │ ├── E - min_sport_2012_Rukovoditeli_gospredpriyatij,_podvedomstvennyih_ministerstvu.doc │ ├── M - min_agr_2012 MinSelXoz_Rukovodstvo - 01.xls │ ├── Test.xlsx │ ├── c_sample.xlsx │ ├── customs-tworow-header.xls │ ├── example.json │ ├── fsin_2016_extract.xlsx │ ├── fsin_2016_extract2.xlsx │ ├── glav_44_2010.doc │ └── rabotniki_podved_organizacii_2013.xlsx ├── toloka ├── assignments │ ├── golden_1.tsv │ └── train.tsv └── pools │ ├── test.sh │ └── test_pool.tsv └── tools ├── AntlrTester ├── AntlrTester.csproj └── Program.cs ├── BuildBigrams ├── __init__.py ├── bigrams.py ├── build_dict.py └── texts │ └── _convert.sh ├── BuildCharWidthTable ├── BuildCharWidthTable.csproj └── Program.cs ├── ConvStorage ├── INSTALL.txt ├── conv_storage_server.py ├── conversion_client.py ├── convert_storage.py ├── copy_to_converter2.sh ├── obsolete │ ├── DisclosuresPdfConvService.py │ └── convert_pdf_service.py ├── prod.hft ├── requirements.txt ├── scripts │ ├── check_snowball.py │ ├── convert_pdf.py │ ├── convert_to_snow_ball.py │ ├── docx_is_from_winword.py │ ├── get_docx.py │ ├── get_stats.py │ ├── recreate_database.py │ └── rename_to_sha256.py ├── start_server.sh ├── tests │ ├── __init__.py │ ├── files │ │ ├── 1501.pdf │ │ ├── 18822_cut.pdf │ │ ├── 4043_0.pdf │ │ ├── 4043_1.pdf │ │ ├── 4043_2.pdf │ │ ├── 4119_3.pdf │ │ ├── 4189_5.pdf │ │ ├── 4189_6.pdf │ │ ├── HotFolder.txt │ │ ├── bad.pdf │ │ ├── broken.pdf │ │ ├── complicated.pdf │ │ ├── drm.pdf │ │ ├── for_ocr.pdf │ │ ├── freeze.pdf │ │ ├── good.pdf │ │ └── winword2019_hangs.pdf │ ├── test.hft │ ├── test_conv_storage.py │ ├── test_magic.py │ └── update_finereader_task.py └── windows_task_sheduler.xml ├── CorpusProcess ├── corpus_process.py ├── main.txt ├── parse.bat ├── run.bat ├── run.sh ├── run_smart_parser.backup.py ├── run_smart_parser.py ├── start_parsing.bat └── test_timeout.py ├── DeclDocRecognizer ├── dlrecognizer.py ├── document_types.py ├── external_convertors.py ├── filter_folder.py ├── filter_folder.sh ├── get_metrics.sh ├── many_plus.metrics.txt.canon ├── metrics.txt.canon ├── readme.txt ├── regression_tests │ ├── 1948.doc │ ├── 1948.doc.verdict │ ├── 3199.docx │ ├── 3199.docx.verdict │ ├── 3210.pdf │ ├── 3210.pdf.verdict │ ├── 3223.doc │ ├── 3223.doc.verdict │ ├── 3343.doc │ ├── 3343.doc.verdict │ ├── 3384_0.xls │ ├── 3384_0.xls.verdict │ ├── 35078_3.xlsx │ ├── 35078_3.xlsx.verdict │ ├── 4003_0.htm │ ├── 4003_0.htm.verdict │ ├── 4061_0.htm │ ├── 4061_0.htm.verdict │ ├── archive.rar │ ├── archive.rar.verdict │ ├── simple_minus.rtf │ └── simple_minus.rtf.verdict ├── run_folder.sh ├── run_regression_tests.sh └── test_all.sh ├── DeclMatch ├── decl_match.py ├── decl_match_metric.py ├── golden_check.py └── maj_vote.py ├── DocxFont ├── App.config ├── DocxFont.csproj ├── DocxFont.sln └── Program.cs ├── INSTALL.txt ├── MicrosoftPdf2Docx ├── App.config ├── MicrosoftPdf2Docx.csproj ├── MicrosoftPdf2Docx.sln ├── Program.cs ├── Properties │ └── AssemblyInfo.cs ├── regression_tests │ ├── 3aas.arbitr.ru.txt │ ├── 3aas.arbitr.ru.txt.result_summary │ ├── admkad.ru.txt │ ├── admkad.ru.txt.result_summary │ ├── akrvo.ru.txt │ ├── akrvo.ru.txt.result_summary │ ├── arshush.ru.txt │ ├── arshush.ru.txt.result_summary │ ├── bagaev.donland.ru.txt │ ├── bagaev.donland.ru.txt.result_summary │ ├── gorsovet-podolsk.ru.txt │ ├── gorsovet-podolsk.ru.txt.result_summary │ ├── ing.txt │ ├── ing.txt.result_summary │ ├── kraevoy.hbr.sudrf.ru.txt │ ├── kraevoy.hbr.sudrf.ru.txt.result_summary │ ├── ksl.spb.sudrf.ru.txt │ ├── ksl.spb.sudrf.ru.txt.result_summary │ ├── mid.txt │ ├── mid.txt.result_summary │ ├── mil.txt │ ├── mil.txt.result_summary │ ├── minprom.txt │ ├── minprom.txt.result_summary │ ├── minvr.txt │ ├── minvr.txt.result_summary │ ├── mkrf.txt │ ├── mkrf.txt.result_summary │ ├── oblsud.tula.sudrf.ru.txt │ ├── oblsud.tula.sudrf.ru.txt.result_summary │ ├── oren-rshn.txt │ ├── oren-rshn.txt.result_summary │ ├── ozerny.ru.txt │ ├── ozerny.ru.txt.result_summary │ ├── rosminzdrav.txt │ ├── rosminzdrav.txt.result_summary │ ├── save │ │ └── tests.sav.2020-29-09 │ │ │ ├── 3aas.arbitr.ru.txt.result_summary │ │ │ ├── admkad.ru.txt.result_summary │ │ │ ├── akrvo.ru.txt.result_summary │ │ │ ├── arshush.ru.txt.result_summary │ │ │ ├── bagaev.donland.ru.txt.result_summary │ │ │ ├── gorsovet-podolsk.ru.txt.result_summary │ │ │ ├── ing.txt.result_summary │ │ │ ├── kraevoy.hbr.sudrf.ru.txt.result_summary │ │ │ ├── ksl.spb.sudrf.ru.txt.result_summary │ │ │ ├── mid.txt.result_summary │ │ │ ├── mil.txt.result_summary │ │ │ ├── minprom.txt.result_summary │ │ │ ├── minvr.txt.result_summary │ │ │ ├── mkrf.txt.result_summary │ │ │ ├── oblsud.tula.sudrf.ru.txt.result_summary │ │ │ ├── oren-rshn.txt.result_summary │ │ │ ├── ozerny.ru.txt.result_summary │ │ │ ├── rosminzdrav.txt.result_summary │ │ │ ├── silino.mos.ru.txt.result_summary │ │ │ ├── spbogdo.ru.txt.result_summary │ │ │ ├── vet.txt.result_summary │ │ │ ├── volgograd.izbirkom.ru.txt.result_summary │ │ │ └── zsro.txt.result_summary │ ├── silino.mos.ru.txt │ ├── silino.mos.ru.txt.result_summary │ ├── spbogdo.ru.txt │ ├── spbogdo.ru.txt.result_summary │ ├── vet.txt │ ├── vet.txt.result_summary │ ├── volgograd.izbirkom.ru.txt │ ├── volgograd.izbirkom.ru.txt.result_summary │ ├── zsro.txt │ └── zsro.txt.result_summary └── tests │ ├── negative │ ├── 0468ae09f27cca738585d185b8b4296dde79eddca3b665c9c4ab356776fe0ac8.pdf │ ├── 4043_0.pdf │ ├── 4043_1.pdf │ ├── 4043_2.pdf │ ├── 4119_3.pdf │ ├── 4189_5.pdf │ ├── 4189_6.pdf │ ├── a.pdf │ ├── bad.pdf │ ├── complicated.pdf │ └── not_russian.pdf │ ├── positive │ ├── 0d398293ad70eb0877dc089b2acdecaac5ee9b6afbaded59bd7c0ef83a911d0e.pdf │ ├── 1501.pdf │ ├── 80a446bcedf8d886813dcc300742a4d321b119250168b7c30a3a77902e2447a2.pdf │ ├── 84c3be94280f22fc5e50dbffa1650fb648ed470b9e0fbe1d5c13dab8e8fd6917.pdf │ └── fc96f2af9fcb921837f0506f772232bc2359cec13821aba1cf81a69188e06d4c.pdf │ └── run_tests.sh ├── Office2Txt ├── Office2Txt.csproj └── Program.cs ├── ParsingDeclarationToloka ├── body.css ├── body.html ├── myscript.js ├── toloka_description.docx └── toloka_environment.js ├── TolokaCorpusGen ├── create_toloka.py ├── hash_golden.py └── run.sh ├── ValidateJson ├── App.config ├── Program.cs ├── Properties │ └── AssemblyInfo.cs └── ValidateJson.csproj ├── __init__.py ├── common ├── __init__.py ├── access_log.py ├── archives.py ├── content_types.py ├── decl_title_parser.py ├── django_base_command_monkey.py ├── download.py ├── export_files.py ├── file_storage.py ├── html_parser.py ├── http_request.py ├── languages.py ├── link_info.py ├── logging_wrapper.py ├── popular_sites.py ├── primitives.py ├── recognizer_enum.py ├── russian_fio.py ├── russian_morph_dict.py ├── selenium_driver.py ├── serp_parser.py ├── snow_ball_file_storage.py ├── urllib_parse_pro.py ├── web_site_status.py └── wiki_bots.py ├── deploy ├── docker │ ├── step1 │ │ ├── Dockerfile │ │ └── lic.bin │ └── step2 │ │ └── Dockerfile └── profile.sh ├── disclosures_site ├── __init__.py ├── data │ ├── car_brands.json │ ├── corrections.json │ ├── districts_from_wikidata.json │ ├── dolbilo_requests.txt │ ├── external_links.json │ ├── median_income.csv │ ├── misspell_bin │ │ ├── gramtab.tab │ │ ├── morph.annot │ │ ├── morph.bases │ │ ├── morph.forms_autom │ │ ├── morph.options │ │ ├── morphs.mrd │ │ └── project.mwz │ └── statistics.json ├── declarations │ ├── __init__.py │ ├── admin.py │ ├── apps.py │ ├── car_brands.py │ ├── corrections.py │ ├── documents.py │ ├── gender_recognize.py │ ├── management │ │ ├── __init__.py │ │ └── commands │ │ │ ├── __init__.py │ │ │ ├── add_disclosures_statistics.py │ │ │ ├── build_elastic_index.py │ │ │ ├── build_genders.py │ │ │ ├── build_office_calculated_params.py │ │ │ ├── build_ratings.py │ │ │ ├── build_surname_rank.py │ │ │ ├── car_brand_report.py │ │ │ ├── clear_database.py │ │ │ ├── clear_dedupe_artefacts.py │ │ │ ├── copy_person_id.py │ │ │ ├── create_database.py │ │ │ ├── create_misspell_fio_db.py │ │ │ ├── create_permalink_storage.py │ │ │ ├── create_sql_sequences.py │ │ │ ├── elastic_manage.py │ │ │ ├── external_link_surname_checker.py │ │ │ ├── files_with_bad_fios.py │ │ │ ├── fsin_fix.py │ │ │ ├── gender_report.py │ │ │ ├── generate_dedupe_pairs.py │ │ │ ├── generate_sitemaps.py │ │ │ ├── import_declarator_toloka_pool.py │ │ │ ├── import_json.py │ │ │ ├── init_section_rubric.py │ │ │ ├── name_report.py │ │ │ ├── new_car_model.py │ │ │ ├── new_toloka_tasks.py │ │ │ ├── random_forest_adapter.py │ │ │ ├── rebuild_office_rubrics.py │ │ │ ├── region_report.py │ │ │ ├── renormalize_section_person_name.py │ │ │ ├── test_ml_on_pool.py │ │ │ ├── test_real_clustering_on_pool.py │ │ │ ├── train_ml_on_pool.py │ │ │ └── update_person_redirects.py │ ├── migrations │ │ ├── 0001_initial.py │ │ ├── 0002_region.py │ │ ├── 0003_offices.py │ │ └── __init__.py │ ├── models.py │ ├── permalinks.py │ ├── ratings.py │ ├── section_passport.py │ ├── serializers.py │ ├── sql_helpers.py │ ├── statistics.py │ ├── templates │ │ ├── base.html │ │ ├── file │ │ │ ├── detail.html │ │ │ └── index.html │ │ ├── morda │ │ │ ├── about.html │ │ │ ├── compare_income_descr.html │ │ │ ├── index.html │ │ │ ├── news.html │ │ │ ├── news_mobile.html │ │ │ ├── permalinks.html │ │ │ ├── second_office.html │ │ │ └── smart_parser_spec.html │ │ ├── office │ │ │ ├── detail.html │ │ │ └── index.html │ │ ├── person │ │ │ ├── detail.html │ │ │ └── index.html │ │ ├── region │ │ │ ├── detail.html │ │ │ └── index.html │ │ ├── reports │ │ │ ├── car-brands │ │ │ │ ├── car-brands-by-years.html │ │ │ │ └── index.html │ │ │ ├── genders │ │ │ │ └── index.html │ │ │ ├── names │ │ │ │ └── index.html │ │ │ ├── new-car │ │ │ │ └── index.html │ │ │ ├── offices │ │ │ │ └── index.html │ │ │ ├── offices2020 │ │ │ │ └── index.html │ │ │ ├── regions │ │ │ │ └── index.html │ │ │ └── regions2020 │ │ │ │ └── index.html │ │ ├── robots.txt │ │ ├── search │ │ │ └── paginator.html │ │ ├── section │ │ │ ├── detail.html │ │ │ ├── index.html │ │ │ └── section_table.html │ │ └── statistics │ │ │ └── statistics.html │ ├── tests │ │ ├── acc_log_squeeze │ │ │ ├── __init__.py │ │ │ ├── logs │ │ │ │ └── access.2021-07-30.gz │ │ │ └── test_acc_log.py │ │ ├── car_brands │ │ │ ├── __init__.py │ │ │ ├── cases_1000.txt │ │ │ └── test_car_brands.py │ │ ├── copy_person_id │ │ │ ├── __init__.py │ │ │ └── test_copy_person_id.py │ │ ├── corrected_person │ │ │ ├── __init__.py │ │ │ └── test_corrected_person.py │ │ ├── dedupe │ │ │ ├── __init__.py │ │ │ └── test_dedupe.py │ │ ├── dedupe_base_for_tests.py │ │ ├── dedupe_divide │ │ │ ├── __init__.py │ │ │ └── test_dedupe.py │ │ ├── dedupe_real │ │ │ ├── __init__.py │ │ │ ├── disclosures.sql.person_id_5295.n │ │ │ └── test_dedupe.py │ │ ├── dedupe_reuse_old_person_id │ │ │ ├── __init__.py │ │ │ └── test_dedupe_reuse_person.py │ │ ├── dedupe_with_dump_objects │ │ │ ├── __init__.py │ │ │ ├── dedupe_objects.dump │ │ │ ├── permalinks_declarations_person.dbm │ │ │ └── test_dedupe_andreev.py │ │ ├── elastic │ │ │ ├── __init__.py │ │ │ └── test_elastic.py │ │ ├── export_human_files │ │ │ ├── __init__.py │ │ │ ├── human_files.json │ │ │ └── test_export_human_files.py │ │ ├── fio_clusters │ │ │ ├── __init__.py │ │ │ └── test_fio_clusters.py │ │ ├── import_complex │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ ├── one_line_2017.docx │ │ │ │ │ ├── one_line_2017_copy.docx │ │ │ │ │ ├── one_line_2018.docx │ │ │ │ │ └── one_line_2019.docx │ │ │ └── test_complex_import.py │ │ ├── import_document_order │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ ├── one_line_2017.docx │ │ │ │ │ └── one_line_2017_copy.docx │ │ │ └── test_doc_order_import.py │ │ ├── import_external_year │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── 0.pdf.docx │ │ │ └── test_import_external_year.py │ │ ├── import_fio_spaces │ │ │ ├── dlrobot_human_1.json │ │ │ ├── dlrobot_human_2.json │ │ │ ├── domains_1 │ │ │ │ └── test1.ru │ │ │ │ │ └── one_line_2017.docx │ │ │ ├── domains_2 │ │ │ │ └── test1.ru │ │ │ │ │ └── one_line_2017.docx │ │ │ └── test_fio_spaces.py │ │ ├── import_fsin │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── fsin.xlsx │ │ │ └── test_import_fsin.py │ │ ├── import_fsin_2 │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── fsin.xlsx │ │ │ └── test_import_fsin.py │ │ ├── import_own_type │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── zehova2013.doc │ │ │ └── test_import_own_type.py │ │ ├── import_relatives │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── fsin.docx │ │ │ └── test_import_relative.py │ │ ├── import_second_passport │ │ │ ├── dlrobot_human_1.json │ │ │ ├── domains_1 │ │ │ │ └── test1.ru │ │ │ │ │ └── fsin.docx │ │ │ └── test_import_passport_2.py │ │ ├── import_simple │ │ │ ├── dlrobot_human.json │ │ │ ├── domains │ │ │ │ └── test1.ru │ │ │ │ │ └── one_line_2017.docx │ │ │ └── test_simple_import.py │ │ ├── join_dlrobot_and_human │ │ │ ├── __init__.py │ │ │ ├── dlrobot_human.json │ │ │ ├── human_files.json │ │ │ ├── human_files │ │ │ │ ├── human.xlsx │ │ │ │ └── uniq_from_human.xlsx │ │ │ ├── old │ │ │ │ ├── dlrobot_human.json │ │ │ │ └── domains │ │ │ │ │ └── 05.fsin.su │ │ │ │ │ ├── human_file_deleted_on_site.xlsx │ │ │ │ │ ├── new.xlsx │ │ │ │ │ └── old_dlrobot.xlsx │ │ │ ├── processed_projects │ │ │ │ ├── 05.fsin.su.1590209953 │ │ │ │ │ ├── 05.fsin.su.txt │ │ │ │ │ ├── 05.fsin.su.txt.visited_pages │ │ │ │ │ └── result │ │ │ │ │ │ └── 05.fsin.su │ │ │ │ │ │ ├── copy_dlrobot.xlsx │ │ │ │ │ │ └── new_dlrobot.xlsx │ │ │ │ └── 05.fsin.su │ │ │ │ │ ├── 05.fsin.su.txt │ │ │ │ │ ├── 05.fsin.su.txt.visited_pages │ │ │ │ │ └── result │ │ │ │ │ └── 05.fsin.su │ │ │ │ │ ├── 1984.xlsx │ │ │ │ │ ├── new_dlrobot.xlsx │ │ │ │ │ └── old_dlrobot.xlsx │ │ │ └── test_join_human_and_dlrobot.py │ │ ├── nominal_income │ │ │ ├── __init__.py │ │ │ └── test_nominal_income.py │ │ ├── office_web_site_redirects │ │ │ ├── __init__.py │ │ │ └── test_redirects.py │ │ ├── parse_decl_title │ │ │ ├── __init__.py │ │ │ ├── examples.json │ │ │ └── test_parse_decl_title.py │ │ ├── ratings │ │ │ ├── __init__.py │ │ │ └── test_ratings.py │ │ ├── regions │ │ │ ├── __init__.py │ │ │ └── test_regions.py │ │ ├── rubrics │ │ │ ├── __init__.py │ │ │ └── test_rubrics.py │ │ ├── russian_fio │ │ │ ├── __init__.py │ │ │ └── test_russian_fio.py │ │ ├── smart_parser_for_testing.py │ │ ├── source_doc_for_testing.py │ │ ├── update_person_redirects │ │ │ ├── __init__.py │ │ │ ├── access_log_squeeze.txt │ │ │ └── test_redirects.py │ │ ├── view_office_search │ │ │ ├── __init__.py │ │ │ └── test_search_for_office.py │ │ ├── view_person_name_misspell │ │ │ ├── __init__.py │ │ │ └── test_search_by_misspelled_fio.py │ │ └── view_section_search │ │ │ ├── __init__.py │ │ │ └── test_search_by_fio.py │ ├── urls.py │ └── views.py ├── deduplicate │ ├── __init__.py │ ├── assignments │ │ ├── assignments_disclosures_golden_set.tsv │ │ ├── assignments_disclosures_golden_set_01.tsv │ │ ├── assignments_disclosures_test_01.tsv │ │ ├── assignments_disclosures_test_02.tsv │ │ ├── assignments_disclosures_test_03.tsv │ │ ├── assignments_disclosures_train_01.tsv │ │ └── assignments_disclosures_train_02.tsv │ ├── config.py │ ├── model │ │ ├── points.txt │ │ ├── random_forest.pickle │ │ └── run.sh │ ├── pools │ │ ├── disclosures_test_m.tsv │ │ ├── disclosures_training_20.tsv │ │ ├── fix_list.txt │ │ └── make_pools.sh │ ├── tasks │ │ ├── disclosures_golden_set.tsv │ │ ├── disclosures_test_01.tsv │ │ ├── disclosures_test_02.tsv │ │ ├── disclosures_test_03.tsv │ │ └── disclosures_train_01.tsv │ ├── toloka.py │ └── toloka_project │ │ ├── descr.txt │ │ ├── emulate_toloka.html │ │ ├── instruction.html │ │ ├── package-lock.json │ │ ├── task.css │ │ ├── task.html │ │ ├── task.js │ │ └── toloka_environment.js ├── disclosures │ ├── __init__.py │ ├── elastic_signal.py │ ├── settings │ │ ├── __init__.py │ │ ├── common.py │ │ ├── dev.py │ │ ├── log_setup.py │ │ └── prod.py │ ├── static │ │ ├── a.html │ │ ├── car-brands │ │ │ ├── make_html.py │ │ │ └── run.sh │ │ ├── dlrobot │ │ │ └── stats.html │ │ ├── er_diagram.svg │ │ ├── example_table.png │ │ ├── fonts │ │ │ ├── Roboto.ttf │ │ │ ├── Roboto.woff │ │ │ ├── Robotolight.ttf │ │ │ └── Robotolight.woff │ │ ├── images │ │ │ ├── carbrands │ │ │ │ ├── .png │ │ │ │ ├── ac.png │ │ │ │ ├── acura.png │ │ │ │ ├── alfa_romeo.png │ │ │ │ ├── alpina.png │ │ │ │ ├── ariel.png │ │ │ │ ├── aro.png │ │ │ │ ├── asia.png │ │ │ │ ├── aston_martin.png │ │ │ │ ├── audi.png │ │ │ │ ├── austin.png │ │ │ │ ├── barkas.png │ │ │ │ ├── baw.png │ │ │ │ ├── bentley.png │ │ │ │ ├── bitter.png │ │ │ │ ├── bmw.png │ │ │ │ ├── bogdan.png │ │ │ │ ├── borgward.png │ │ │ │ ├── brabus.png │ │ │ │ ├── brilliance.png │ │ │ │ ├── bristol.png │ │ │ │ ├── bronto.png │ │ │ │ ├── bufori.png │ │ │ │ ├── bugatti.png │ │ │ │ ├── buick.png │ │ │ │ ├── byd.png │ │ │ │ ├── byvin.png │ │ │ │ ├── cadillac.png │ │ │ │ ├── callaway.png │ │ │ │ ├── caterham.png │ │ │ │ ├── changan.png │ │ │ │ ├── changfeng.png │ │ │ │ ├── chaparral.png │ │ │ │ ├── chery.png │ │ │ │ ├── chevrolet.png │ │ │ │ ├── chrysler.png │ │ │ │ ├── citroen.png │ │ │ │ ├── cizeta.png │ │ │ │ ├── cleardot.gif │ │ │ │ ├── coggiola.png │ │ │ │ ├── cord.png │ │ │ │ ├── dacia.png │ │ │ │ ├── dadi.png │ │ │ │ ├── daewoo.png │ │ │ │ ├── daf.png │ │ │ │ ├── daihatsu.png │ │ │ │ ├── daimler.png │ │ │ │ ├── datsun.png │ │ │ │ ├── delage.png │ │ │ │ ├── delorean.png │ │ │ │ ├── derways.png │ │ │ │ ├── desoto.png │ │ │ │ ├── detomaso.png │ │ │ │ ├── dodge.png │ │ │ │ ├── dongfeng.png │ │ │ │ ├── donkervoort.png │ │ │ │ ├── eagle.png │ │ │ │ ├── faw.png │ │ │ │ ├── ferrari.png │ │ │ │ ├── fiat.png │ │ │ │ ├── fisker.png │ │ │ │ ├── ford.png │ │ │ │ ├── foton.png │ │ │ │ ├── fso.png │ │ │ │ ├── gaz.png │ │ │ │ ├── geely.png │ │ │ │ ├── geo.png │ │ │ │ ├── ginetta.png │ │ │ │ ├── gmc.png │ │ │ │ ├── gonow.png │ │ │ │ ├── gordon.png │ │ │ │ ├── great_wall.png │ │ │ │ ├── gumpert.png │ │ │ │ ├── hafei.png │ │ │ │ ├── haima.png │ │ │ │ ├── haval.png │ │ │ │ ├── hawtai.png │ │ │ │ ├── hindustan.png │ │ │ │ ├── hino.png │ │ │ │ ├── holden.png │ │ │ │ ├── honda.png │ │ │ │ ├── horch.png │ │ │ │ ├── huanghai.png │ │ │ │ ├── hummer.png │ │ │ │ ├── hyundai.png │ │ │ │ ├── ifa.png │ │ │ │ ├── ikarus.png │ │ │ │ ├── infiniti.png │ │ │ │ ├── innocenti.png │ │ │ │ ├── international.png │ │ │ │ ├── invicta.png │ │ │ │ ├── isdera.png │ │ │ │ ├── isuzu.png │ │ │ │ ├── iveco.png │ │ │ │ ├── izh.png │ │ │ │ ├── jac.png │ │ │ │ ├── jaguar.png │ │ │ │ ├── jeep.png │ │ │ │ ├── jensen.png │ │ │ │ ├── jmc.png │ │ │ │ ├── jowett.png │ │ │ │ ├── kaiser.png │ │ │ │ ├── kamaz.png │ │ │ │ ├── kavz.png │ │ │ │ ├── kenworth.png │ │ │ │ ├── kia.png │ │ │ │ ├── koenigsegg.png │ │ │ │ ├── kraz.png │ │ │ │ ├── ktm.png │ │ │ │ ├── lada.png │ │ │ │ ├── lamborghini.png │ │ │ │ ├── lancia.png │ │ │ │ ├── land_rover.png │ │ │ │ ├── landwind.png │ │ │ │ ├── laz.png │ │ │ │ ├── lexus.png │ │ │ │ ├── leyland.png │ │ │ │ ├── liaz.png │ │ │ │ ├── lifan.png │ │ │ │ ├── ligier.png │ │ │ │ ├── lincoln.png │ │ │ │ ├── lotus.png │ │ │ │ ├── lti.png │ │ │ │ ├── luaz.png │ │ │ │ ├── luxgen.png │ │ │ │ ├── mahindra.png │ │ │ │ ├── man.png │ │ │ │ ├── marcos.png │ │ │ │ ├── marussia.png │ │ │ │ ├── maruti.png │ │ │ │ ├── maserati.png │ │ │ │ ├── matra.png │ │ │ │ ├── maybach.png │ │ │ │ ├── maz.png │ │ │ │ ├── mazda.png │ │ │ │ ├── mclaren.png │ │ │ │ ├── mercedes-benz.png │ │ │ │ ├── mercury.png │ │ │ │ ├── mg.png │ │ │ │ ├── microcar.png │ │ │ │ ├── mini.png │ │ │ │ ├── mitsubishi.png │ │ │ │ ├── mitsuoka.png │ │ │ │ ├── morgan.png │ │ │ │ ├── morris.png │ │ │ │ ├── moskvich.png │ │ │ │ ├── neoplan.png │ │ │ │ ├── nissan.png │ │ │ │ ├── noble.png │ │ │ │ ├── oka.png │ │ │ │ ├── oldsmobile.png │ │ │ │ ├── opel.png │ │ │ │ ├── osca.png │ │ │ │ ├── packard.png │ │ │ │ ├── pagani.png │ │ │ │ ├── panhard.png │ │ │ │ ├── panoz.png │ │ │ │ ├── paz.png │ │ │ │ ├── perodua.png │ │ │ │ ├── peterbilt.png │ │ │ │ ├── peugeot.png │ │ │ │ ├── pgo.png │ │ │ │ ├── piaggio.png │ │ │ │ ├── plymouth.png │ │ │ │ ├── pontiac.png │ │ │ │ ├── porsche.png │ │ │ │ ├── premier.png │ │ │ │ ├── proton.png │ │ │ │ ├── puch.png │ │ │ │ ├── qoros.png │ │ │ │ ├── raf.png │ │ │ │ ├── ravon.png │ │ │ │ ├── renault.png │ │ │ │ ├── rezvani.png │ │ │ │ ├── rimac.png │ │ │ │ ├── rolls_royce.png │ │ │ │ ├── ronart.png │ │ │ │ ├── rover.png │ │ │ │ ├── saab.png │ │ │ │ ├── saleen.png │ │ │ │ ├── santana.png │ │ │ │ ├── saturn.png │ │ │ │ ├── scania.png │ │ │ │ ├── scion.png │ │ │ │ ├── seat.png │ │ │ │ ├── seaz.png │ │ │ │ ├── setra.png │ │ │ │ ├── shuanghuan.png │ │ │ │ ├── simca.png │ │ │ │ ├── skoda.png │ │ │ │ ├── smart.png │ │ │ │ ├── smz.png │ │ │ │ ├── soueast.png │ │ │ │ ├── spyker.png │ │ │ │ ├── ssangyong.png │ │ │ │ ├── steyr.png │ │ │ │ ├── studebaker.png │ │ │ │ ├── stutz.png │ │ │ │ ├── subaru.png │ │ │ │ ├── sunbeam.png │ │ │ │ ├── suzuki.png │ │ │ │ ├── tagaz.png │ │ │ │ ├── talbot.png │ │ │ │ ├── tata.png │ │ │ │ ├── tatra.png │ │ │ │ ├── tazzari.png │ │ │ │ ├── tesla.png │ │ │ │ ├── tianma.png │ │ │ │ ├── tofas.png │ │ │ │ ├── toyota.png │ │ │ │ ├── trabant.png │ │ │ │ ├── tramontana.png │ │ │ │ ├── triumph.png │ │ │ │ ├── tvr.png │ │ │ │ ├── uaz.png │ │ │ │ ├── ultima.png │ │ │ │ ├── ural.png │ │ │ │ ├── vauxhall.png │ │ │ │ ├── vaz.png │ │ │ │ ├── venturi.png │ │ │ │ ├── vis.png │ │ │ │ ├── volkswagen.png │ │ │ │ ├── volvo.png │ │ │ │ ├── vortex.png │ │ │ │ ├── wartburg.png │ │ │ │ ├── westfield.png │ │ │ │ ├── wiesmann.png │ │ │ │ ├── willys.png │ │ │ │ ├── xin_kai.png │ │ │ │ ├── yamaha.png │ │ │ │ ├── yamal.png │ │ │ │ ├── yo_mobil.png │ │ │ │ ├── yutong.png │ │ │ │ ├── zaz.png │ │ │ │ ├── zenvo.png │ │ │ │ ├── zil.png │ │ │ │ ├── zis.png │ │ │ │ ├── zotye.png │ │ │ │ └── zx.png │ │ │ ├── declarant_index.png │ │ │ ├── declarant_office_income.png │ │ │ ├── declaration_example.png │ │ │ ├── department_example.png │ │ │ ├── favicon-16x16.png │ │ │ ├── favicon-32x32.png │ │ │ ├── favicon.ico │ │ │ ├── favicon.png │ │ │ ├── lambordgini.png │ │ │ ├── maserati.png │ │ │ ├── pozor_putina.jpg │ │ │ ├── realty_example.png │ │ │ ├── realty_example2.png │ │ │ ├── realty_example3.png │ │ │ ├── realty_example4.png │ │ │ └── spouse_office_income.png │ │ ├── main_menu.js │ │ ├── names │ │ │ ├── names.region.txt.gz │ │ │ └── regional_names.py │ │ ├── officereports │ │ │ ├── office-income-report.csv │ │ │ ├── office-income-report.html │ │ │ ├── rubric-1-income-report.csv │ │ │ ├── rubric-1-income-report.html │ │ │ ├── rubric-10-income-report.csv │ │ │ ├── rubric-10-income-report.html │ │ │ ├── rubric-19-income-report.csv │ │ │ ├── rubric-19-income-report.html │ │ │ ├── rubric-2-income-report.csv │ │ │ ├── rubric-2-income-report.html │ │ │ ├── rubric-20-income-report.csv │ │ │ ├── rubric-20-income-report.html │ │ │ ├── rubric-3-income-report.csv │ │ │ ├── rubric-3-income-report.html │ │ │ ├── rubric-4-income-report.csv │ │ │ ├── rubric-4-income-report.html │ │ │ ├── rubric-5-income-report.csv │ │ │ ├── rubric-5-income-report.html │ │ │ ├── rubric-6-income-report.csv │ │ │ ├── rubric-6-income-report.html │ │ │ ├── rubric-7-income-report.csv │ │ │ ├── rubric-7-income-report.html │ │ │ ├── rubric-8-income-report.csv │ │ │ ├── rubric-8-income-report.html │ │ │ ├── rubric-9-income-report.csv │ │ │ ├── rubric-9-income-report.html │ │ │ ├── rubric-income-report.csv │ │ │ └── rubric-income-report.html │ │ ├── regionreports │ │ │ ├── region-income-report-2011.csv │ │ │ ├── region-income-report-2011.html │ │ │ ├── region-income-report-2013.csv │ │ │ ├── region-income-report-2013.html │ │ │ ├── region-income-report-2015.csv │ │ │ ├── region-income-report-2015.html │ │ │ ├── region-income-report-2017.csv │ │ │ ├── region-income-report-2017.html │ │ │ ├── region-income-report-2019.csv │ │ │ └── region-income-report-2019.html │ │ ├── roc_new_car_prod_s5.png │ │ ├── sorttable.js │ │ └── style.css │ ├── urls.py │ └── wsgi.py ├── manage.py └── scripts │ ├── __init__.py │ ├── access_log_squeeze.py │ ├── adhoc │ ├── fix_sha256_html.py │ ├── scrape_car_brand.py │ └── tcp_forward.py │ ├── check_person_id_permanence.py │ ├── create_example_db.sh │ ├── create_sample.py │ ├── csv │ ├── csv_to_html.py │ └── unqote_tsv.py │ ├── dbm_manage.py │ ├── dedupe │ ├── create_golden.py │ ├── descibe_dedupe.py │ ├── import_pools_from_declarator.py │ ├── make_pool.py │ └── prec_recall_curve.py │ ├── dolbilo.py │ ├── etc │ ├── nginx │ │ └── sites-available │ │ │ └── disclosures.ru │ └── systemd │ │ └── system │ │ ├── check_disclosures_health.service │ │ ├── gunicorn.service │ │ └── gunicorn.socket │ ├── export_human_files.py │ ├── intersect_with_wd.py │ ├── rename_db.sh │ ├── rosstat │ ├── average_income.py │ ├── election_2021.py │ ├── population.py │ ├── rosstat.py │ └── salary.py │ └── setup_head_version.py ├── disclosures_stub ├── about.html ├── compare_income_descr.html ├── favicon-16x16.png ├── favicon-32x32.png ├── favicon.ico ├── favicon.png ├── index.html ├── permalinks.html ├── reports │ ├── car-brands │ │ ├── car-brands-by-years.html │ │ └── index.html │ ├── genders │ │ └── index.html │ ├── names │ │ └── index.html │ ├── new-car │ │ └── index.html │ ├── offices │ │ └── index.html │ ├── offices2020 │ │ └── index.html │ ├── regions │ │ └── index.html │ └── regions2020 │ │ └── index.html ├── robots.txt ├── roc_new_car_prod_s5.png ├── second_office.html ├── sitemap-main.xml ├── sitemap.xml ├── smart_parser_spec.html ├── static │ ├── er_diagram.svg │ ├── fonts │ │ ├── Roboto.ttf │ │ ├── Roboto.woff │ │ ├── Robotolight.ttf │ │ └── Robotolight.woff │ ├── images │ │ ├── declarant_index.png │ │ ├── declarant_office_income.png │ │ ├── declaration_example.png │ │ ├── department_example.png │ │ ├── lambordgini.png │ │ ├── maserati.png │ │ ├── realty_example.png │ │ ├── realty_example2.png │ │ ├── realty_example3.png │ │ ├── realty_example4.png │ │ └── spouse_office_income.png │ ├── main_menu.js │ ├── names │ │ ├── names.region.txt.gz │ │ └── regional_names.py │ ├── officereports │ │ ├── office-income-report.csv │ │ ├── office-income-report.html │ │ ├── rubric-1-income-report.csv │ │ ├── rubric-1-income-report.html │ │ ├── rubric-10-income-report.csv │ │ ├── rubric-10-income-report.html │ │ ├── rubric-19-income-report.csv │ │ ├── rubric-19-income-report.html │ │ ├── rubric-2-income-report.csv │ │ ├── rubric-2-income-report.html │ │ ├── rubric-20-income-report.csv │ │ ├── rubric-20-income-report.html │ │ ├── rubric-3-income-report.csv │ │ ├── rubric-3-income-report.html │ │ ├── rubric-4-income-report.csv │ │ ├── rubric-4-income-report.html │ │ ├── rubric-5-income-report.csv │ │ ├── rubric-5-income-report.html │ │ ├── rubric-6-income-report.csv │ │ ├── rubric-6-income-report.html │ │ ├── rubric-7-income-report.csv │ │ ├── rubric-7-income-report.html │ │ ├── rubric-8-income-report.csv │ │ ├── rubric-8-income-report.html │ │ ├── rubric-9-income-report.csv │ │ ├── rubric-9-income-report.html │ │ ├── rubric-income-report.csv │ │ ├── rubric-income-report.html │ │ └── sitemap.xml │ ├── regionreports │ │ ├── region-income-report-2011.csv │ │ ├── region-income-report-2011.html │ │ ├── region-income-report-2013.csv │ │ ├── region-income-report-2013.html │ │ ├── region-income-report-2015.csv │ │ ├── region-income-report-2015.html │ │ ├── region-income-report-2017.csv │ │ ├── region-income-report-2017.html │ │ ├── region-income-report-2019.csv │ │ ├── region-income-report-2019.html │ │ ├── region-income-report-2020.csv │ │ ├── region-income-report-2020.html │ │ └── sitemap.xml │ ├── sorttable.js │ └── style.css ├── statistics │ └── index.html └── stub.html ├── dlrobot ├── __init__.py ├── central │ ├── __init__.py │ ├── data │ │ ├── dlrobot_remote_calls.1.dat.gz │ │ ├── dlrobot_remote_calls.dat │ │ └── dlrobot_rounds.json │ ├── dlrobot_central.py │ ├── scripts │ │ ├── __init__.py │ │ ├── analyze_remote_calls.py │ │ ├── check_fio.py │ │ ├── copy_nginx_logs.sh │ │ ├── create_manually_dlrobot_project.py │ │ ├── create_slf_for_surnames.py │ │ ├── crontab.txt │ │ ├── find_dlrobot_artefacts.py │ │ ├── fns │ │ │ ├── __init__.py │ │ │ ├── add_fns_json_to_html.py │ │ │ ├── add_fns_json_to_html.sh │ │ │ └── unzip_archive.py │ │ ├── monitoring │ │ │ ├── __init__.py │ │ │ ├── check_disclosures_health.py │ │ │ ├── check_disclosures_heath.json │ │ │ ├── dl_monitoring.py │ │ │ └── dl_monitoring.sh │ │ ├── predict_surname.py │ │ ├── profile.sh │ │ ├── send_dlrobot_projects_to_cloud.py │ │ ├── send_docs_to_servers.py │ │ ├── uniq_tab.py │ │ ├── update_declarator_db.sh │ │ ├── update_disclosures.sh │ │ └── yandex_disk.py │ ├── send_docs.py │ └── tests │ │ ├── __init__.py │ │ ├── page.zip │ │ ├── test_dlrobot_server.py │ │ ├── test_primitives.py │ │ └── web_sites │ │ ├── declaration_year_in_anchor │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.html │ │ └── sved.xlsx │ │ ├── simple │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.docx │ │ └── sved.html │ │ └── site_with_subfolder │ │ └── ru │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.html │ │ └── sved.xlsx ├── common │ ├── __init__.py │ ├── central_protocol.py │ ├── dl_robot_round.py │ ├── remote_call.py │ ├── robot_config.py │ ├── robot_project.py │ ├── robot_step.py │ ├── robot_web_site.py │ ├── url_info.py │ └── yandex_cloud.py ├── robot │ ├── __init__.py │ ├── adhoc │ │ ├── __init__.py │ │ ├── gossov_tatarstan_ru.py │ │ ├── tgl_ru.py │ │ └── tomsk.py │ ├── configs │ │ ├── preliminary.json │ │ ├── prod.json │ │ └── test.json │ ├── declaration_link.py │ ├── dl_robot.py │ ├── process_folder.sh │ ├── readme.txt │ ├── regression_tests │ │ ├── 3aas.arbitr.ru.txt │ │ ├── 3aas.arbitr.ru.txt.result_summary │ │ ├── admkad.ru.txt │ │ ├── admkad.ru.txt.result_summary │ │ ├── akrvo.ru.txt │ │ ├── akrvo.ru.txt.result_summary │ │ ├── arshush.ru.txt │ │ ├── arshush.ru.txt.result_summary │ │ ├── bagaev.donland.ru.txt │ │ ├── biradm.ru.txt │ │ ├── biradm.ru.txt.result_summary │ │ ├── culture.gov.ru.txt │ │ ├── culture.gov.ru.txt.result_summary │ │ ├── duma39.ru.txt │ │ ├── duma39.ru.txt.result_summary │ │ ├── gorsovet-podolsk.ru.txt │ │ ├── gorsovet-podolsk.ru.txt.result_summary │ │ ├── ing.txt │ │ ├── ing.txt.result_summary │ │ ├── kraevoy.hbr.sudrf.ru.txt │ │ ├── kraevoy.hbr.sudrf.ru.txt.result_summary │ │ ├── ksl.spb.sudrf.ru.txt │ │ ├── ksl.spb.sudrf.ru.txt.result_summary │ │ ├── mid.txt │ │ ├── mid.txt.result_summary │ │ ├── mil.txt │ │ ├── mil.txt.result_summary │ │ ├── minprom.txt │ │ ├── minprom.txt.result_summary │ │ ├── minvr.txt │ │ ├── minvr.txt.result_summary │ │ ├── oblsud.tula.sudrf.ru.txt │ │ ├── oblsud.tula.sudrf.ru.txt.result_summary │ │ ├── oren-rshn.txt │ │ ├── oren-rshn.txt.result_summary │ │ ├── ozerny.ru.txt │ │ ├── ozerny.ru.txt.result_summary │ │ ├── rosminzdrav.txt │ │ ├── rosminzdrav.txt.result_summary │ │ ├── silino.mos.ru.txt │ │ ├── silino.mos.ru.txt.result_summary │ │ ├── spbogdo.ru.txt │ │ ├── spbogdo.ru.txt.result_summary │ │ ├── vet.txt │ │ ├── vet.txt.result_summary │ │ ├── volgograd.izbirkom.ru.txt │ │ ├── volgograd.izbirkom.ru.txt.result_summary │ │ ├── zsro.txt │ │ └── zsro.txt.result_summary │ ├── run_regression_tests.sh │ ├── scripts │ │ ├── __init__.py │ │ ├── build_dlrecognizer_toloka_pool.py │ │ ├── check_robot_efficiency.py │ │ ├── check_search_engine.py │ │ ├── check_selenium.py │ │ ├── html_parser.py │ │ ├── max_click_path_len.py │ │ ├── print_disclosures_sha256.py │ │ ├── scrape_sudrf.py │ │ ├── site_mirror.sh │ │ └── sitemap_download.py │ ├── tests.sh │ └── tests │ │ ├── __init__.py │ │ ├── common_env.py │ │ ├── declaration_link_for_tests.py │ │ ├── selenium_for_tests.py │ │ ├── test_admkrsk.py │ │ ├── test_akrvo.py │ │ ├── test_akrvo2.py │ │ ├── test_configs.py │ │ ├── test_content_type.py │ │ ├── test_declaration_link.py │ │ ├── test_deleted_web_site.py │ │ ├── test_dlrobot.py │ │ ├── test_duma39.py │ │ ├── test_file_cache.py │ │ ├── test_file_extension_by_contents.py │ │ ├── test_gorsovet_podolsk.py │ │ ├── test_html_parser.py │ │ ├── test_http_redirect.py │ │ ├── test_kolomnagrad1.py │ │ ├── test_kolomnagrad2.py │ │ ├── test_link_speed_optim.py │ │ ├── test_mid.py │ │ ├── test_minzdrav.py │ │ ├── test_one_page_timeout.py │ │ ├── test_prohibited_links.py │ │ ├── test_random_pdf.py │ │ ├── test_regional.py │ │ ├── test_search_engine.py │ │ ├── test_selenium.py │ │ ├── test_sitemap.py │ │ ├── test_ssl.py │ │ ├── test_timeout.py │ │ ├── test_unrar_cyr.py │ │ ├── test_urlparse.py │ │ ├── test_video.py │ │ ├── web_site_monkey.py │ │ └── web_sites │ │ ├── admkrsk │ │ ├── found_links │ │ └── sved.html │ │ ├── admkrsk2 │ │ ├── clerk │ │ │ └── incomes │ │ │ │ ├── Lists │ │ │ │ └── supreme │ │ │ │ │ └── Attachments │ │ │ │ │ └── 92 │ │ │ │ │ └── Одинцов2020.docx │ │ │ │ └── pages │ │ │ │ ├── default.html │ │ │ │ └── supreme.html │ │ └── index.html │ │ ├── archives │ │ ├── archive_with_broken_files.zip │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.docx.7z │ │ ├── sved.docx.rar │ │ ├── sved.docx.zip │ │ └── sved.html │ │ ├── arkvo │ │ ├── found_links │ │ └── sved.html │ │ ├── arkvo2 │ │ ├── 25023.html │ │ └── parent.html │ │ ├── culture.gov.ru │ │ ├── culture.gov.ru.txt │ │ └── found_links │ │ ├── duma39 │ │ ├── parent.html │ │ └── sved.html │ │ ├── enadm │ │ └── project.txt │ │ ├── gorsovet-podolsk │ │ └── project.txt │ │ ├── khabkrai │ │ ├── found_links │ │ └── sved.html │ │ ├── kolomnagrad1 │ │ ├── found_links │ │ └── project.txt │ │ ├── kolomnagrad2 │ │ ├── found_links │ │ └── project.txt │ │ ├── mid │ │ ├── found_links │ │ └── project.txt │ │ ├── minzdrav │ │ ├── 6_4_2.html │ │ └── found_links │ │ ├── minzdrav1 │ │ ├── found_links │ │ └── project.txt │ │ ├── mkrf2 │ │ ├── index.html │ │ ├── mkrf.docx │ │ ├── mkrf.html │ │ ├── protiv.html │ │ ├── sved.docx │ │ └── sved.html │ │ ├── one_page_timeout │ │ └── sved.html │ │ ├── other_website │ │ ├── found_links │ │ └── sved.html │ │ ├── page_text │ │ ├── found_links │ │ └── sved.html │ │ ├── pdf │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.html │ │ └── sved.pdf │ │ ├── random_pdf │ │ ├── index.html │ │ ├── protiv.html │ │ └── sved.html │ │ ├── ros.oblsud │ │ ├── found_links │ │ └── sved.html │ │ ├── simple │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.docx │ │ └── sved.html │ │ ├── simple_doc │ │ ├── found_links │ │ └── sved.html │ │ ├── sitemap │ │ ├── index.html │ │ ├── sitemap.xml │ │ ├── sved.docx │ │ └── sveden_corrupt.html │ │ ├── unrar │ │ └── file.rar │ │ ├── website_with_js │ │ ├── index.html │ │ ├── protiv.html │ │ ├── sved.docx │ │ └── sved.html │ │ ├── with_regional │ │ ├── index.html │ │ ├── magadan.html │ │ ├── protiv.html │ │ ├── protiv_magadan.html │ │ ├── sved.docx │ │ ├── sved.html │ │ └── sved_magadan.html │ │ └── zsro │ │ ├── found_links │ │ └── sved.html ├── worker │ ├── __init__.py │ ├── dlrobot_worker.py │ ├── dlrobot_worker.service │ └── scripts │ │ ├── __init__.py │ │ └── dl_cloud_manager.py └── workstation_monitoring │ ├── workstation_monitoring.service │ └── workstation_monitoring_service.py ├── dlrobot_human ├── __init__.py ├── dlrobot_human_dbm.py ├── input_document.py └── scripts │ ├── __init__.py │ ├── dlrobot_human_manager.py │ └── join_human_and_dlrobot.py ├── office_db ├── countries.py ├── data │ ├── fgup.txt │ ├── office2020 │ │ ├── office_stat_data.csv │ │ ├── office_stat_data.txt │ │ ├── rubric_stat_data.csv │ │ └── rubric_stat_data.txt │ ├── office_current │ │ ├── office_stat_data.csv │ │ ├── office_stat_data.txt │ │ ├── rubric_stat_data.csv │ │ └── rubric_stat_data.txt │ ├── offices.txt │ ├── region_report_table_2020.json │ ├── regions.txt │ ├── ross_stat.json │ ├── sudrf.txt │ └── web_sites_markup.txt ├── declarant_group_stat_data.py ├── declaration_office_website.py ├── offices_in_memory.py ├── region_data.py ├── region_year_snapshot.py ├── rubrics.py ├── rubrics │ └── build_train.py ├── russia.py ├── russian_declarant_position.py ├── russian_geo_word.py ├── russian_office_word.py ├── russian_regions.py ├── scripts │ ├── calc_capital.py │ ├── calc_region.py │ ├── calc_region_from_wikidata.py │ ├── join_offices_and_sites.py │ ├── office_db_manager.py │ ├── set_rubrics.py │ ├── set_wikidata_id.py │ └── update_offices.py ├── web_site_list.py └── year_income.py ├── predict_office ├── __init__.py ├── base_ml_model.py ├── model │ ├── model │ │ ├── keras_metadata.pb │ │ ├── saved_model.pb │ │ └── variables │ │ │ ├── variables.data-00000-of-00001 │ │ │ └── variables.index │ └── office_ngrams.txt ├── office_index.py ├── office_pool.py ├── pools │ ├── test.source.txt │ ├── test_fixed.txt │ ├── train.declarator.source.txt │ ├── train.sud.source.txt │ ├── train.toloka.source.txt │ └── weak_offices_pool.txt ├── prediction_case.py ├── read_office_from_title.py ├── scripts │ ├── __init__.py │ ├── baseline_office.py │ ├── build_office_index.py │ ├── catboost_office.py │ ├── manage_pool.py │ ├── predict_office_dbm.py │ ├── prepare_train_pool.py │ ├── scp_wrapper.py │ ├── tensorflow_examples │ │ ├── binary.py │ │ ├── category.py │ │ ├── category_functional.py │ │ ├── text_feature.py │ │ └── two_categories.py │ ├── tf_office_test.py │ ├── tf_office_toloka.py │ ├── tf_office_train.py │ └── train.sh ├── tensor_flow_model.py └── tests │ ├── __init__.py │ └── predict_office │ ├── __init__.py │ ├── dlrobot_human.json │ ├── processed_projects │ ├── donm.mos.ru.1590209953 │ │ ├── donm.mos.ru.txt │ │ ├── donm.mos.ru.txt.visited_pages │ │ └── result │ │ │ └── mos.ru │ │ │ └── 01.xlsx │ └── tax │ │ └── tax.html │ └── test_predict_office.py ├── requirements.txt ├── smart_parser_http ├── __init__.py ├── scripts │ ├── purge_old_keys.py │ ├── rebuid_smart_parser_worker.py │ └── rebuild_smart_parser.sh ├── smart_parser_client.py ├── smart_parser_server.py └── tests │ ├── __init__.py │ ├── files │ ├── MainWorkPositionIncome.docx │ └── RealtyNaturalText.docx │ └── test_smart_parser_server.py ├── source_doc_http ├── __init__.py ├── etc │ └── systemd │ │ ├── source_doc_central.service │ │ └── system │ │ └── source_declaration_doc.service ├── scripts │ ├── check_storage.py │ └── print_all_keys.py ├── source_doc_client.py ├── source_doc_server.py └── tests │ ├── __init__.py │ └── test_source_doc_client.py └── wikipedia_bots ├── decl2wiki.py ├── delete_spam_links_in_wikidata.py ├── readme.txt └── site_classifier.py /.gitattributes: -------------------------------------------------------------------------------- 1 | text eol=lf -------------------------------------------------------------------------------- /.github/workflows/main.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/.github/workflows/main.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/.gitignore -------------------------------------------------------------------------------- /Antlr/AntlrCommon.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/AntlrCommon.cs -------------------------------------------------------------------------------- /Antlr/BuildAntlr.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/BuildAntlr.csproj -------------------------------------------------------------------------------- /Antlr/CountryListVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/CountryListVisitor.cs -------------------------------------------------------------------------------- /Antlr/OwnTypeListVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/OwnTypeListVisitor.cs -------------------------------------------------------------------------------- /Antlr/RealtyTypeAndOwnTypeVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/RealtyTypeAndOwnTypeVisitor.cs -------------------------------------------------------------------------------- /Antlr/RealtyTypeListVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/RealtyTypeListVisitor.cs -------------------------------------------------------------------------------- /Antlr/SoupVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/SoupVisitor.cs -------------------------------------------------------------------------------- /Antlr/SquareListVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/SquareListVisitor.cs -------------------------------------------------------------------------------- /Antlr/StrictVisitor.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/StrictVisitor.cs -------------------------------------------------------------------------------- /Antlr/TestAntlr.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/TestAntlr.cs -------------------------------------------------------------------------------- /Antlr/generate_csharp_from_antlr.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/generate_csharp_from_antlr.sh -------------------------------------------------------------------------------- /Antlr/generated/.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/generated/.gitignore -------------------------------------------------------------------------------- /Antlr/readme.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/readme.txt -------------------------------------------------------------------------------- /Antlr/src/BaseLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/BaseLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/Common.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/Common.g4 -------------------------------------------------------------------------------- /Antlr/src/CommonLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/CommonLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/CountryLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/CountryLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/CountryList.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/CountryList.g4 -------------------------------------------------------------------------------- /Antlr/src/OwnTypeList.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/OwnTypeList.g4 -------------------------------------------------------------------------------- /Antlr/src/OwntypeLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/OwntypeLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/ParseOtherLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/ParseOtherLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/RealtyTypeAndOwnType.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/RealtyTypeAndOwnType.g4 -------------------------------------------------------------------------------- /Antlr/src/RealtyTypeLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/RealtyTypeLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/RealtyTypeList.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/RealtyTypeList.g4 -------------------------------------------------------------------------------- /Antlr/src/SkipOtherLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/SkipOtherLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/Soup.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/Soup.g4 -------------------------------------------------------------------------------- /Antlr/src/SoupLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/SoupLexer.g4 -------------------------------------------------------------------------------- /Antlr/src/SquareList.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/SquareList.g4 -------------------------------------------------------------------------------- /Antlr/src/Strict.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/Strict.g4 -------------------------------------------------------------------------------- /Antlr/src/StrictLexer.g4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/src/StrictLexer.g4 -------------------------------------------------------------------------------- /Antlr/test_files/country_list.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/country_list.txt -------------------------------------------------------------------------------- /Antlr/test_files/country_list.txt.result: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/country_list.txt.result -------------------------------------------------------------------------------- /Antlr/test_files/country_list.txt.result.canon: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/country_list.txt.result.canon -------------------------------------------------------------------------------- /Antlr/test_files/soup.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/soup.txt -------------------------------------------------------------------------------- /Antlr/test_files/soup.txt.result: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/soup.txt.result -------------------------------------------------------------------------------- /Antlr/test_files/soup.txt.result.canon: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/soup.txt.result.canon -------------------------------------------------------------------------------- /Antlr/test_files/square_and_country.txt: -------------------------------------------------------------------------------- 1 | 475 РФ 2 | 3 | 1. город Москва, 64,7 кв.м. -------------------------------------------------------------------------------- /Antlr/test_files/square_and_country.txt.result: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/square_and_country.txt.result -------------------------------------------------------------------------------- /Antlr/test_files/square_list.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/square_list.txt -------------------------------------------------------------------------------- /Antlr/test_files/square_list.txt.result: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/square_list.txt.result -------------------------------------------------------------------------------- /Antlr/test_files/square_list.txt.result.canon: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/square_list.txt.result.canon -------------------------------------------------------------------------------- /Antlr/test_files/strict.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/strict.txt -------------------------------------------------------------------------------- /Antlr/test_files/strict.txt.result: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/strict.txt.result -------------------------------------------------------------------------------- /Antlr/test_files/strict.txt.result.canon: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Antlr/test_files/strict.txt.result.canon -------------------------------------------------------------------------------- /DeclaratorApiClient/ApiClient.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/DeclaratorApiClient/ApiClient.cs -------------------------------------------------------------------------------- /DeclaratorApiClient/DeclaratorApiClient.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/DeclaratorApiClient/DeclaratorApiClient.csproj -------------------------------------------------------------------------------- /DeclaratorApiClient/DeclaratorApiException.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/DeclaratorApiClient/DeclaratorApiException.cs -------------------------------------------------------------------------------- /DeclaratorApiClient/app.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/DeclaratorApiClient/app.config -------------------------------------------------------------------------------- /DeclaratorApiClient/auth.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/DeclaratorApiClient/auth.config -------------------------------------------------------------------------------- /DeclaratorApiClient/auth.txt: -------------------------------------------------------------------------------- 1 | username 2 | password -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Biblio.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Biblio.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Biblio.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Biblio.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Decree.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Decree.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Decree.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Decree.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Goods.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Goods.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Goods.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Goods.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Location.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Location.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Location.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Location.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Misc.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Misc.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Misc.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Misc.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Organization.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Organization.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Organization.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Organization.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Person.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Person.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Person.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Person.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Semantic.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Semantic.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Semantic.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Semantic.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Util.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Util.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Analyzer.Util.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Analyzer.Util.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Core.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Core.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Core.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Core.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Explanatory.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Explanatory.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Explanatory.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Explanatory.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Morphology.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Morphology.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Morphology.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Morphology.Core.xml -------------------------------------------------------------------------------- /EP.SdkCore/EP.Sdk.Core.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Sdk.Core.csproj -------------------------------------------------------------------------------- /EP.SdkCore/EP.Semantic.Core.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Semantic.Core.dll -------------------------------------------------------------------------------- /EP.SdkCore/EP.Semantic.Core.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/EP.SdkCore/EP.Semantic.Core.xml -------------------------------------------------------------------------------- /JsonSerialization/DeclarationSerializer.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/DeclarationSerializer.cs -------------------------------------------------------------------------------- /JsonSerialization/JsonSerialization.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/JsonSerialization.csproj -------------------------------------------------------------------------------- /JsonSerialization/MiscSerializer.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/MiscSerializer.cs -------------------------------------------------------------------------------- /JsonSerialization/ValidationReport.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/ValidationReport.cs -------------------------------------------------------------------------------- /JsonSerialization/app.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/app.config -------------------------------------------------------------------------------- /JsonSerialization/import-schema-dicts.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/import-schema-dicts.json -------------------------------------------------------------------------------- /JsonSerialization/import-schema.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/JsonSerialization/import-schema.json -------------------------------------------------------------------------------- /ParserCommon/CMDLineParser.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/ParserCommon/CMDLineParser.cs -------------------------------------------------------------------------------- /ParserCommon/LevenshteinDistance.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/ParserCommon/LevenshteinDistance.cs -------------------------------------------------------------------------------- /ParserCommon/ParserCommon.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/ParserCommon/ParserCommon.csproj -------------------------------------------------------------------------------- /ParserCommon/StringMeasure.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/ParserCommon/StringMeasure.cs -------------------------------------------------------------------------------- /ParserCommon/TextHelpers.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/ParserCommon/TextHelpers.cs -------------------------------------------------------------------------------- /PythonUtils/requirements.txt: -------------------------------------------------------------------------------- 1 | chardet==3.0.4 -------------------------------------------------------------------------------- /PythonUtils/unarchive.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/PythonUtils/unarchive.py -------------------------------------------------------------------------------- /RegressionTesting/Readme.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/Readme.txt -------------------------------------------------------------------------------- /RegressionTesting/RegressionTesting.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/RegressionTesting.csproj -------------------------------------------------------------------------------- /RegressionTesting/ValidityTests.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/ValidityTests.cs -------------------------------------------------------------------------------- /RegressionTesting/files/10639_32.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/10639_32.doc -------------------------------------------------------------------------------- /RegressionTesting/files/10639_32.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/10639_32.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/11212_66.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/11212_66.doc -------------------------------------------------------------------------------- /RegressionTesting/files/11212_66.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/11212_66.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/11480_12.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/11480_12.doc -------------------------------------------------------------------------------- /RegressionTesting/files/11480_12.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/11480_12.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/12104_33.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/12104_33.docx -------------------------------------------------------------------------------- /RegressionTesting/files/12104_33.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/12104_33.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/13100_13.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13100_13.doc -------------------------------------------------------------------------------- /RegressionTesting/files/13100_13.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13100_13.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/13133_5.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13133_5.doc -------------------------------------------------------------------------------- /RegressionTesting/files/13133_5.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13133_5.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/13252_58.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13252_58.xls -------------------------------------------------------------------------------- /RegressionTesting/files/13252_58.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13252_58.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/13415_200.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13415_200.doc -------------------------------------------------------------------------------- /RegressionTesting/files/13415_200.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13415_200.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/13931_28.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13931_28.doc -------------------------------------------------------------------------------- /RegressionTesting/files/13931_28.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/13931_28.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/14421_139.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14421_139.htm -------------------------------------------------------------------------------- /RegressionTesting/files/14421_139.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14421_139.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/14475_0.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14475_0.doc -------------------------------------------------------------------------------- /RegressionTesting/files/14475_0.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14475_0.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/14712_34.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14712_34.doc -------------------------------------------------------------------------------- /RegressionTesting/files/14712_34.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14712_34.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/14845_84.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14845_84.docx -------------------------------------------------------------------------------- /RegressionTesting/files/14845_84.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/14845_84.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/15555_0.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15555_0.html -------------------------------------------------------------------------------- /RegressionTesting/files/15555_0.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15555_0.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/15555_1.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15555_1.html -------------------------------------------------------------------------------- /RegressionTesting/files/15555_1.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15555_1.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/15584_0.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15584_0.html -------------------------------------------------------------------------------- /RegressionTesting/files/15584_0.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/15584_0.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/16694.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/16694.docx -------------------------------------------------------------------------------- /RegressionTesting/files/16694.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/16694.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/16738_12.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/16738_12.html -------------------------------------------------------------------------------- /RegressionTesting/files/16738_12.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/16738_12.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/17207.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17207.xls -------------------------------------------------------------------------------- /RegressionTesting/files/17207.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17207.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/17335_3.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17335_3.html -------------------------------------------------------------------------------- /RegressionTesting/files/17335_3.html_0.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17335_3.html_0.json -------------------------------------------------------------------------------- /RegressionTesting/files/17335_3.html_1.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17335_3.html_1.json -------------------------------------------------------------------------------- /RegressionTesting/files/17335_3.html_2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17335_3.html_2.json -------------------------------------------------------------------------------- /RegressionTesting/files/17339_24.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17339_24.html -------------------------------------------------------------------------------- /RegressionTesting/files/17339_24.html_0.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17339_24.html_0.json -------------------------------------------------------------------------------- /RegressionTesting/files/17339_24.html_1.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17339_24.html_1.json -------------------------------------------------------------------------------- /RegressionTesting/files/17339_24.html_2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17339_24.html_2.json -------------------------------------------------------------------------------- /RegressionTesting/files/17339_24.html_3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17339_24.html_3.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_0.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_0.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_1.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_1.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_10.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_10.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_11.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_11.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_2.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_2.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_3.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_3.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_4.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_4.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_5.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_5.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_6.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_6.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_7.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_7.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_8.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_8.json -------------------------------------------------------------------------------- /RegressionTesting/files/17497.xls_9.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/17497.xls_9.json -------------------------------------------------------------------------------- /RegressionTesting/files/18261_22.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/18261_22.doc -------------------------------------------------------------------------------- /RegressionTesting/files/18261_22.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/18261_22.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/1person_pdf_converted.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/1person_pdf_converted.docx -------------------------------------------------------------------------------- /RegressionTesting/files/20838.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/20838.doc -------------------------------------------------------------------------------- /RegressionTesting/files/20838.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/20838.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/20883_1.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/20883_1.docx -------------------------------------------------------------------------------- /RegressionTesting/files/20883_1.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/20883_1.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/256_Columns.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/256_Columns.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/256_Columns.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/256_Columns.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/26937_42.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/26937_42.doc -------------------------------------------------------------------------------- /RegressionTesting/files/26937_42.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/26937_42.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/28733.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/28733.doc -------------------------------------------------------------------------------- /RegressionTesting/files/28733.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/28733.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/28817.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/28817.xls -------------------------------------------------------------------------------- /RegressionTesting/files/28817.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/28817.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/30429.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/30429.docx -------------------------------------------------------------------------------- /RegressionTesting/files/30429.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/30429.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/31832.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/31832.xls -------------------------------------------------------------------------------- /RegressionTesting/files/31832.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/31832.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/31832_1.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/31832_1.xls -------------------------------------------------------------------------------- /RegressionTesting/files/31832_1.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/31832_1.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/33793.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/33793.doc -------------------------------------------------------------------------------- /RegressionTesting/files/33793.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/33793.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/36922.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/36922.docx -------------------------------------------------------------------------------- /RegressionTesting/files/36922.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/36922.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/38527_5.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/38527_5.docx -------------------------------------------------------------------------------- /RegressionTesting/files/38527_5.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/38527_5.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/4002_0.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4002_0.htm -------------------------------------------------------------------------------- /RegressionTesting/files/4002_0.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4002_0.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/4037_9.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4037_9.htm -------------------------------------------------------------------------------- /RegressionTesting/files/4037_9.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4037_9.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/4067_0.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4067_0.docx -------------------------------------------------------------------------------- /RegressionTesting/files/4067_0.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4067_0.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/4144_28.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4144_28.htm -------------------------------------------------------------------------------- /RegressionTesting/files/4144_28.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4144_28.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/4462_10.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4462_10.doc -------------------------------------------------------------------------------- /RegressionTesting/files/4462_10.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4462_10.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/4479_27.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4479_27.doc -------------------------------------------------------------------------------- /RegressionTesting/files/4479_27.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4479_27.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/4778_0.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4778_0.htm -------------------------------------------------------------------------------- /RegressionTesting/files/4778_0.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/4778_0.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/50484_2.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/50484_2.xls -------------------------------------------------------------------------------- /RegressionTesting/files/50484_2.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/50484_2.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/52668.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/52668.doc -------------------------------------------------------------------------------- /RegressionTesting/files/52668.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/52668.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/52991.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/52991.docx -------------------------------------------------------------------------------- /RegressionTesting/files/52991.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/52991.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/61662.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/61662.docx -------------------------------------------------------------------------------- /RegressionTesting/files/61662.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/61662.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/65098_2.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/65098_2.docx -------------------------------------------------------------------------------- /RegressionTesting/files/65098_2.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/65098_2.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/65098_5.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/65098_5.docx -------------------------------------------------------------------------------- /RegressionTesting/files/65098_5.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/65098_5.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/7007_10.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7007_10.html -------------------------------------------------------------------------------- /RegressionTesting/files/7007_10.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7007_10.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/7007_8.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7007_8.html -------------------------------------------------------------------------------- /RegressionTesting/files/7007_8.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7007_8.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/7022_0.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7022_0.htm -------------------------------------------------------------------------------- /RegressionTesting/files/7022_0.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/7022_0.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/74.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/74.html -------------------------------------------------------------------------------- /RegressionTesting/files/74.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/74.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/8562.pdf.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/8562.pdf.docx -------------------------------------------------------------------------------- /RegressionTesting/files/8562.pdf.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/8562.pdf.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/8992_15.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/8992_15.doc -------------------------------------------------------------------------------- /RegressionTesting/files/8992_15.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/8992_15.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/9037/rykovodstvo_2013.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/9037/rykovodstvo_2013.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/BadColumns.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/BadColumns.doc -------------------------------------------------------------------------------- /RegressionTesting/files/BadColumns.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/BadColumns.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/BadOcr.pdf.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/BadOcr.pdf.docx -------------------------------------------------------------------------------- /RegressionTesting/files/BadOcr.pdf.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/BadOcr.pdf.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/DepEnergo2010.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/DepEnergo2010.doc -------------------------------------------------------------------------------- /RegressionTesting/files/DepEnergo2010.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/DepEnergo2010.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/DepGosPol2012.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/DepGosPol2012.doc -------------------------------------------------------------------------------- /RegressionTesting/files/DepGosPol2012.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/DepGosPol2012.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/Deputy2015.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Deputy2015.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Deputy2015.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Deputy2015.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/Fsin2011.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Fsin2011.doc -------------------------------------------------------------------------------- /RegressionTesting/files/Fsin2011.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Fsin2011.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/HeaderInsideTable.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/HeaderInsideTable.docx -------------------------------------------------------------------------------- /RegressionTesting/files/IgnoreTableJunk.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/IgnoreTableJunk.docx -------------------------------------------------------------------------------- /RegressionTesting/files/IgnoreTableJunk.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/IgnoreTableJunk.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/IncomeNotFirstLine.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/IncomeNotFirstLine.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Mchs2010.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Mchs2010.doc -------------------------------------------------------------------------------- /RegressionTesting/files/Mchs2010.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Mchs2010.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/Mchs2013.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Mchs2013.doc -------------------------------------------------------------------------------- /RegressionTesting/files/Mchs2013.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Mchs2013.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinDalVostok2015.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinDalVostok2015.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinDalVostok2015.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinDalVostok2015.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinDalVostok2017.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinDalVostok2017.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinDalVostok2017.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinDalVostok2017.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkon2013.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkon2013.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkon2013.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkon2013.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkon2015.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkon2015.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkon2015.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkon2015.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkonom2014.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkonom2014.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkonom2014.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkonom2014.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkonon2017.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkonon2017.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinEkonon2017.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinEkonon2017.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2011.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2011.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2011.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2011.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2012.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2012.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2012.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2012.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2012.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2012.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2012.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2012.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2015.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2015.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2015.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2015.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2015.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2015.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinKult2015.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinKult2015.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinObor2012.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObor2012.xls -------------------------------------------------------------------------------- /RegressionTesting/files/MinObor2012.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObor2012.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinObr2012.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObr2012.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinObr2012.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObr2012.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinObr2016.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObr2016.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinObr2016.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinObr2016.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinProm2013.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinProm2013.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinProm2013.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinProm2013.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinRes2011.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinRes2011.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinRes2011.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinRes2011.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinSelhoz2013.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSelhoz2013.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinSelhoz2013.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSelhoz2013.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinSelhoz2015.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSelhoz2015.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinSelhoz2015.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSelhoz2015.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinSevKavkaz2015.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSevKavkaz2015.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinSevKavkaz2015.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSevKavkaz2015.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinSport2016.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSport2016.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinSport2016.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinSport2016.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2014.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2014.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2014.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2014.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2017.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2017.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2017.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2017.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2017_1.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2017_1.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/MinStroy2017_1.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinStroy2017_1.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinTrans2009.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinTrans2009.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinTrans2009.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinTrans2009.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinTrans2011.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinTrans2011.docx -------------------------------------------------------------------------------- /RegressionTesting/files/MinTrans2011.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinTrans2011.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinYust2012.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinYust2012.doc -------------------------------------------------------------------------------- /RegressionTesting/files/MinYust2012.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinYust2012.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/MinZdorov2015Full.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/MinZdorov2015Full.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Nedosekov.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Nedosekov.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Nedosekov.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Nedosekov.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/RealtyNaturalText.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/RealtyNaturalText.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Spasat2016.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Spasat2016.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Spasat2016.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Spasat2016.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/SpellCheckRealtyType.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/SpellCheckRealtyType.docx -------------------------------------------------------------------------------- /RegressionTesting/files/Sudia2011.htm: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Sudia2011.htm -------------------------------------------------------------------------------- /RegressionTesting/files/Sudia2011.htm.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Sudia2011.htm.json -------------------------------------------------------------------------------- /RegressionTesting/files/Unk2014.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Unk2014.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/Unk2014.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/Unk2014.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/ZabSud2017.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/ZabSud2017.docx -------------------------------------------------------------------------------- /RegressionTesting/files/ZabSud2017.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/ZabSud2017.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/ZagranApp2016.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/ZagranApp2016.doc -------------------------------------------------------------------------------- /RegressionTesting/files/ZagranApp2016.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/ZagranApp2016.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/chab2019.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/chab2019.xls -------------------------------------------------------------------------------- /RegressionTesting/files/chab2019.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/chab2019.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/disclosures_format.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/disclosures_format.xls -------------------------------------------------------------------------------- /RegressionTesting/files/dnko-2014.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/dnko-2014.docx -------------------------------------------------------------------------------- /RegressionTesting/files/dnko-2014.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/dnko-2014.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/fsin2013.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/fsin2013.docx -------------------------------------------------------------------------------- /RegressionTesting/files/fsin2013.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/fsin2013.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/fsn01.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/fsn01.html -------------------------------------------------------------------------------- /RegressionTesting/files/fsn01.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/fsn01.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/html_title.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/html_title.html -------------------------------------------------------------------------------- /RegressionTesting/files/html_title.html.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/html_title.html.json -------------------------------------------------------------------------------- /RegressionTesting/files/minfin2016.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/minfin2016.xlsx -------------------------------------------------------------------------------- /RegressionTesting/files/minfin2016.xlsx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/minfin2016.xlsx.json -------------------------------------------------------------------------------- /RegressionTesting/files/minzdrav2015.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/minzdrav2015.docx -------------------------------------------------------------------------------- /RegressionTesting/files/minzdrav2015.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/minzdrav2015.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/niz_kam.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/niz_kam.docx -------------------------------------------------------------------------------- /RegressionTesting/files/niz_kam.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/niz_kam.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/one_line_2017.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/one_line_2017.docx -------------------------------------------------------------------------------- /RegressionTesting/files/one_line_2017.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/one_line_2017.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/privol.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/privol.xls -------------------------------------------------------------------------------- /RegressionTesting/files/privol.xls.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/privol.xls.json -------------------------------------------------------------------------------- /RegressionTesting/files/pudoz_01.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/pudoz_01.docx -------------------------------------------------------------------------------- /RegressionTesting/files/pudoz_01.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/pudoz_01.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/sov_fed.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/sov_fed.docx -------------------------------------------------------------------------------- /RegressionTesting/files/sov_fed.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/sov_fed.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/sud_2016.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/sud_2016.doc -------------------------------------------------------------------------------- /RegressionTesting/files/sud_2016.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/sud_2016.doc.json -------------------------------------------------------------------------------- /RegressionTesting/files/toloka.toloka_json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/toloka.toloka_json -------------------------------------------------------------------------------- /RegressionTesting/files/toloka.toloka_json.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/toloka.toloka_json.json -------------------------------------------------------------------------------- /RegressionTesting/files/two_tables_2017.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/two_tables_2017.docx -------------------------------------------------------------------------------- /RegressionTesting/files/two_tables_2017.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/two_tables_2017.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/year_with_space.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/year_with_space.docx -------------------------------------------------------------------------------- /RegressionTesting/files/year_with_space.docx.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/year_with_space.docx.json -------------------------------------------------------------------------------- /RegressionTesting/files/zehova2013.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/zehova2013.doc -------------------------------------------------------------------------------- /RegressionTesting/files/zehova2013.doc.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/files/zehova2013.doc.json -------------------------------------------------------------------------------- /RegressionTesting/log4net.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/RegressionTesting/log4net.config -------------------------------------------------------------------------------- /Smart.Parser.sln: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/Smart.Parser.sln -------------------------------------------------------------------------------- /lib/Adapters/AdapterSchemes/IAdapterScheme.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/AdapterSchemes/IAdapterScheme.cs -------------------------------------------------------------------------------- /lib/Adapters/AngleHtmlAdapter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/AngleHtmlAdapter.cs -------------------------------------------------------------------------------- /lib/Adapters/AsposeExcelAdapter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/AsposeExcelAdapter.cs -------------------------------------------------------------------------------- /lib/Adapters/AsposeLicense.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/AsposeLicense.cs -------------------------------------------------------------------------------- /lib/Adapters/Azure/AzureDtoCache.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/Azure/AzureDtoCache.cs -------------------------------------------------------------------------------- /lib/Adapters/Azure/AzureFormRecognizer.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/Azure/AzureFormRecognizer.cs -------------------------------------------------------------------------------- /lib/Adapters/BigramsHolder.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/BigramsHolder.cs -------------------------------------------------------------------------------- /lib/Adapters/ConvertToDocxAndFix.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/ConvertToDocxAndFix.cs -------------------------------------------------------------------------------- /lib/Adapters/HtmAdapter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/HtmAdapter.cs -------------------------------------------------------------------------------- /lib/Adapters/HtmlSchemes/ArbitrationCourt1.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/HtmlSchemes/ArbitrationCourt1.cs -------------------------------------------------------------------------------- /lib/Adapters/HtmlSchemes/ArbitrationCourt2.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/HtmlSchemes/ArbitrationCourt2.cs -------------------------------------------------------------------------------- /lib/Adapters/HtmlSchemes/HtmlScheme.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/HtmlSchemes/HtmlScheme.cs -------------------------------------------------------------------------------- /lib/Adapters/IAdapter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/IAdapter.cs -------------------------------------------------------------------------------- /lib/Adapters/IAdapterCell.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/IAdapterCell.cs -------------------------------------------------------------------------------- /lib/Adapters/IAdapterRow.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/IAdapterRow.cs -------------------------------------------------------------------------------- /lib/Adapters/MicrosoftExcel.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/MicrosoftExcel.cs -------------------------------------------------------------------------------- /lib/Adapters/NpoiExcel.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/NpoiExcel.cs -------------------------------------------------------------------------------- /lib/Adapters/OpenXmlWord.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/OpenXmlWord.cs -------------------------------------------------------------------------------- /lib/Adapters/Section.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Adapters/Section.cs -------------------------------------------------------------------------------- /lib/BorderFinder.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/BorderFinder.cs -------------------------------------------------------------------------------- /lib/DataHelper.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/DataHelper.cs -------------------------------------------------------------------------------- /lib/Declaration.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Declaration.cs -------------------------------------------------------------------------------- /lib/DeclarationField.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/DeclarationField.cs -------------------------------------------------------------------------------- /lib/DeclarationProperties.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/DeclarationProperties.cs -------------------------------------------------------------------------------- /lib/JsonWriter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/JsonWriter.cs -------------------------------------------------------------------------------- /lib/Logger.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Logger.cs -------------------------------------------------------------------------------- /lib/Parser.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Parser.cs -------------------------------------------------------------------------------- /lib/ParserBase.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/ParserBase.cs -------------------------------------------------------------------------------- /lib/Person.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Person.cs -------------------------------------------------------------------------------- /lib/PublicServant.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/PublicServant.cs -------------------------------------------------------------------------------- /lib/RealEstateProperty.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/RealEstateProperty.cs -------------------------------------------------------------------------------- /lib/RealtyParser.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/RealtyParser.cs -------------------------------------------------------------------------------- /lib/RelationType.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/RelationType.cs -------------------------------------------------------------------------------- /lib/Relative.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Relative.cs -------------------------------------------------------------------------------- /lib/Resources/bigrams.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Resources/bigrams.txt -------------------------------------------------------------------------------- /lib/Resources/column_trigrams.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Resources/column_trigrams.txt -------------------------------------------------------------------------------- /lib/Resources/log4net.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Resources/log4net.config -------------------------------------------------------------------------------- /lib/SecondPassParser.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/SecondPassParser.cs -------------------------------------------------------------------------------- /lib/Smart.Parser.Lib.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Smart.Parser.Lib.csproj -------------------------------------------------------------------------------- /lib/SmartParserException.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/SmartParserException.cs -------------------------------------------------------------------------------- /lib/TableHeader/HeaderHelpers.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/TableHeader/HeaderHelpers.cs -------------------------------------------------------------------------------- /lib/TableHeader/PredictByDataTrigrams.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/TableHeader/PredictByDataTrigrams.cs -------------------------------------------------------------------------------- /lib/TableHeader/ReadHeader.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/TableHeader/ReadHeader.cs -------------------------------------------------------------------------------- /lib/TableHeader/TableHeader.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/TableHeader/TableHeader.cs -------------------------------------------------------------------------------- /lib/Vehicle.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/lib/Vehicle.cs -------------------------------------------------------------------------------- /nuget.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/nuget.config -------------------------------------------------------------------------------- /src/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/src/Program.cs -------------------------------------------------------------------------------- /src/Resources/versions.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/src/Resources/versions.txt -------------------------------------------------------------------------------- /src/Smart.Parser.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/src/Smart.Parser.csproj -------------------------------------------------------------------------------- /src/readme.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/src/readme.md -------------------------------------------------------------------------------- /test/ColumnDetectorTest.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/ColumnDetectorTest.cs -------------------------------------------------------------------------------- /test/DataHelperTest.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/DataHelperTest.cs -------------------------------------------------------------------------------- /test/HeaderHelperTest.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/HeaderHelperTest.cs -------------------------------------------------------------------------------- /test/NpoiExcelAdapterBasicTests.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/NpoiExcelAdapterBasicTests.cs -------------------------------------------------------------------------------- /test/Test Samples/2577/6.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/Test Samples/2577/6.docx -------------------------------------------------------------------------------- /test/Test.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/Test.csproj -------------------------------------------------------------------------------- /test/TestJsonWriter.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/TestJsonWriter.cs -------------------------------------------------------------------------------- /test/TestUtil.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/TestUtil.cs -------------------------------------------------------------------------------- /test/TextMeasure.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/TextMeasure.cs -------------------------------------------------------------------------------- /test/XlsxAdapterTest.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/XlsxAdapterTest.cs -------------------------------------------------------------------------------- /test/XlsxTest.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/XlsxTest.cs -------------------------------------------------------------------------------- /test/example.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/example.json -------------------------------------------------------------------------------- /test/testdata/17497.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/17497.xls -------------------------------------------------------------------------------- /test/testdata/18664.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/18664.docx -------------------------------------------------------------------------------- /test/testdata/57715.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/57715.doc -------------------------------------------------------------------------------- /test/testdata/82442.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/82442.doc -------------------------------------------------------------------------------- /test/testdata/Test.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/Test.xlsx -------------------------------------------------------------------------------- /test/testdata/c_sample.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/c_sample.xlsx -------------------------------------------------------------------------------- /test/testdata/customs-tworow-header.xls: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/customs-tworow-header.xls -------------------------------------------------------------------------------- /test/testdata/example.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/example.json -------------------------------------------------------------------------------- /test/testdata/fsin_2016_extract.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/fsin_2016_extract.xlsx -------------------------------------------------------------------------------- /test/testdata/fsin_2016_extract2.xlsx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/fsin_2016_extract2.xlsx -------------------------------------------------------------------------------- /test/testdata/glav_44_2010.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/test/testdata/glav_44_2010.doc -------------------------------------------------------------------------------- /toloka/assignments/golden_1.tsv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/toloka/assignments/golden_1.tsv -------------------------------------------------------------------------------- /toloka/assignments/train.tsv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/toloka/assignments/train.tsv -------------------------------------------------------------------------------- /toloka/pools/test.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/toloka/pools/test.sh -------------------------------------------------------------------------------- /toloka/pools/test_pool.tsv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/toloka/pools/test_pool.tsv -------------------------------------------------------------------------------- /tools/AntlrTester/AntlrTester.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/AntlrTester/AntlrTester.csproj -------------------------------------------------------------------------------- /tools/AntlrTester/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/AntlrTester/Program.cs -------------------------------------------------------------------------------- /tools/BuildBigrams/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/BuildBigrams/bigrams.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/BuildBigrams/bigrams.py -------------------------------------------------------------------------------- /tools/BuildBigrams/build_dict.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/BuildBigrams/build_dict.py -------------------------------------------------------------------------------- /tools/BuildBigrams/texts/_convert.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/BuildBigrams/texts/_convert.sh -------------------------------------------------------------------------------- /tools/BuildCharWidthTable/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/BuildCharWidthTable/Program.cs -------------------------------------------------------------------------------- /tools/ConvStorage/INSTALL.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/INSTALL.txt -------------------------------------------------------------------------------- /tools/ConvStorage/conv_storage_server.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/conv_storage_server.py -------------------------------------------------------------------------------- /tools/ConvStorage/conversion_client.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/conversion_client.py -------------------------------------------------------------------------------- /tools/ConvStorage/convert_storage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/convert_storage.py -------------------------------------------------------------------------------- /tools/ConvStorage/copy_to_converter2.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/copy_to_converter2.sh -------------------------------------------------------------------------------- /tools/ConvStorage/obsolete/convert_pdf_service.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/obsolete/convert_pdf_service.py -------------------------------------------------------------------------------- /tools/ConvStorage/prod.hft: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/prod.hft -------------------------------------------------------------------------------- /tools/ConvStorage/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/requirements.txt -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/check_snowball.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/check_snowball.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/convert_pdf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/convert_pdf.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/convert_to_snow_ball.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/convert_to_snow_ball.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/docx_is_from_winword.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/docx_is_from_winword.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/get_docx.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/get_docx.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/get_stats.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/get_stats.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/recreate_database.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/recreate_database.py -------------------------------------------------------------------------------- /tools/ConvStorage/scripts/rename_to_sha256.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/scripts/rename_to_sha256.py -------------------------------------------------------------------------------- /tools/ConvStorage/start_server.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/start_server.sh -------------------------------------------------------------------------------- /tools/ConvStorage/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/1501.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/1501.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/18822_cut.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/18822_cut.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4043_0.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4043_0.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4043_1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4043_1.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4043_2.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4043_2.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4119_3.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4119_3.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4189_5.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4189_5.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/4189_6.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/4189_6.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/HotFolder.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/HotFolder.txt -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/bad.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/bad.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/broken.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/broken.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/complicated.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/complicated.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/drm.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/drm.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/for_ocr.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/for_ocr.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/freeze.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/freeze.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/files/good.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/files/good.pdf -------------------------------------------------------------------------------- /tools/ConvStorage/tests/test.hft: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/test.hft -------------------------------------------------------------------------------- /tools/ConvStorage/tests/test_conv_storage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/test_conv_storage.py -------------------------------------------------------------------------------- /tools/ConvStorage/tests/test_magic.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/test_magic.py -------------------------------------------------------------------------------- /tools/ConvStorage/tests/update_finereader_task.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/tests/update_finereader_task.py -------------------------------------------------------------------------------- /tools/ConvStorage/windows_task_sheduler.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ConvStorage/windows_task_sheduler.xml -------------------------------------------------------------------------------- /tools/CorpusProcess/corpus_process.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/corpus_process.py -------------------------------------------------------------------------------- /tools/CorpusProcess/main.txt: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/CorpusProcess/parse.bat: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/parse.bat -------------------------------------------------------------------------------- /tools/CorpusProcess/run.bat: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/run.bat -------------------------------------------------------------------------------- /tools/CorpusProcess/run.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/run.sh -------------------------------------------------------------------------------- /tools/CorpusProcess/run_smart_parser.backup.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/run_smart_parser.backup.py -------------------------------------------------------------------------------- /tools/CorpusProcess/run_smart_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/run_smart_parser.py -------------------------------------------------------------------------------- /tools/CorpusProcess/start_parsing.bat: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/start_parsing.bat -------------------------------------------------------------------------------- /tools/CorpusProcess/test_timeout.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/CorpusProcess/test_timeout.py -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/dlrecognizer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/dlrecognizer.py -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/document_types.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/document_types.py -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/external_convertors.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/external_convertors.py -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/filter_folder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/filter_folder.py -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/filter_folder.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/filter_folder.sh -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/get_metrics.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/get_metrics.sh -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/metrics.txt.canon: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/metrics.txt.canon -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/readme.txt: -------------------------------------------------------------------------------- 1 | follow ../INSTALL.txt 2 | bash test_all.sh 3 | -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/1948.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/regression_tests/1948.doc -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/3199.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/regression_tests/3199.docx -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/3210.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/regression_tests/3210.pdf -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/3223.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/regression_tests/3223.doc -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/3343.doc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/regression_tests/3343.doc -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/regression_tests/archive.rar: -------------------------------------------------------------------------------- 1 | do not read me -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/run_folder.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/run_folder.sh -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/run_regression_tests.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/run_regression_tests.sh -------------------------------------------------------------------------------- /tools/DeclDocRecognizer/test_all.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclDocRecognizer/test_all.sh -------------------------------------------------------------------------------- /tools/DeclMatch/decl_match.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclMatch/decl_match.py -------------------------------------------------------------------------------- /tools/DeclMatch/decl_match_metric.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclMatch/decl_match_metric.py -------------------------------------------------------------------------------- /tools/DeclMatch/golden_check.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclMatch/golden_check.py -------------------------------------------------------------------------------- /tools/DeclMatch/maj_vote.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DeclMatch/maj_vote.py -------------------------------------------------------------------------------- /tools/DocxFont/App.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DocxFont/App.config -------------------------------------------------------------------------------- /tools/DocxFont/DocxFont.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DocxFont/DocxFont.csproj -------------------------------------------------------------------------------- /tools/DocxFont/DocxFont.sln: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DocxFont/DocxFont.sln -------------------------------------------------------------------------------- /tools/DocxFont/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/DocxFont/Program.cs -------------------------------------------------------------------------------- /tools/INSTALL.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/INSTALL.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/App.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/App.config -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/MicrosoftPdf2Docx.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/MicrosoftPdf2Docx.csproj -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/MicrosoftPdf2Docx.sln: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/MicrosoftPdf2Docx.sln -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/Program.cs -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/Properties/AssemblyInfo.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/Properties/AssemblyInfo.cs -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/ing.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/ing.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/mid.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/mid.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/mil.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/mil.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/minvr.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/minvr.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/mkrf.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/mkrf.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/vet.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/vet.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/regression_tests/zsro.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/regression_tests/zsro.txt -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4043_0.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4043_0.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4043_1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4043_1.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4043_2.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4043_2.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4119_3.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4119_3.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4189_5.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4189_5.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/4189_6.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/4189_6.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/a.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/a.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/negative/bad.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/negative/bad.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/positive/1501.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/positive/1501.pdf -------------------------------------------------------------------------------- /tools/MicrosoftPdf2Docx/tests/run_tests.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/MicrosoftPdf2Docx/tests/run_tests.sh -------------------------------------------------------------------------------- /tools/Office2Txt/Office2Txt.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/Office2Txt/Office2Txt.csproj -------------------------------------------------------------------------------- /tools/Office2Txt/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/Office2Txt/Program.cs -------------------------------------------------------------------------------- /tools/ParsingDeclarationToloka/body.css: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ParsingDeclarationToloka/body.css -------------------------------------------------------------------------------- /tools/ParsingDeclarationToloka/body.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ParsingDeclarationToloka/body.html -------------------------------------------------------------------------------- /tools/ParsingDeclarationToloka/myscript.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ParsingDeclarationToloka/myscript.js -------------------------------------------------------------------------------- /tools/TolokaCorpusGen/create_toloka.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/TolokaCorpusGen/create_toloka.py -------------------------------------------------------------------------------- /tools/TolokaCorpusGen/hash_golden.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/TolokaCorpusGen/hash_golden.py -------------------------------------------------------------------------------- /tools/TolokaCorpusGen/run.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/TolokaCorpusGen/run.sh -------------------------------------------------------------------------------- /tools/ValidateJson/App.config: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ValidateJson/App.config -------------------------------------------------------------------------------- /tools/ValidateJson/Program.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ValidateJson/Program.cs -------------------------------------------------------------------------------- /tools/ValidateJson/Properties/AssemblyInfo.cs: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ValidateJson/Properties/AssemblyInfo.cs -------------------------------------------------------------------------------- /tools/ValidateJson/ValidateJson.csproj: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/ValidateJson/ValidateJson.csproj -------------------------------------------------------------------------------- /tools/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/common/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/common/access_log.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/access_log.py -------------------------------------------------------------------------------- /tools/common/archives.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/archives.py -------------------------------------------------------------------------------- /tools/common/content_types.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/content_types.py -------------------------------------------------------------------------------- /tools/common/decl_title_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/decl_title_parser.py -------------------------------------------------------------------------------- /tools/common/django_base_command_monkey.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/django_base_command_monkey.py -------------------------------------------------------------------------------- /tools/common/download.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/download.py -------------------------------------------------------------------------------- /tools/common/export_files.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/export_files.py -------------------------------------------------------------------------------- /tools/common/file_storage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/file_storage.py -------------------------------------------------------------------------------- /tools/common/html_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/html_parser.py -------------------------------------------------------------------------------- /tools/common/http_request.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/http_request.py -------------------------------------------------------------------------------- /tools/common/languages.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/languages.py -------------------------------------------------------------------------------- /tools/common/link_info.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/link_info.py -------------------------------------------------------------------------------- /tools/common/logging_wrapper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/logging_wrapper.py -------------------------------------------------------------------------------- /tools/common/popular_sites.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/popular_sites.py -------------------------------------------------------------------------------- /tools/common/primitives.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/primitives.py -------------------------------------------------------------------------------- /tools/common/recognizer_enum.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/recognizer_enum.py -------------------------------------------------------------------------------- /tools/common/russian_fio.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/russian_fio.py -------------------------------------------------------------------------------- /tools/common/russian_morph_dict.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/russian_morph_dict.py -------------------------------------------------------------------------------- /tools/common/selenium_driver.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/selenium_driver.py -------------------------------------------------------------------------------- /tools/common/serp_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/serp_parser.py -------------------------------------------------------------------------------- /tools/common/snow_ball_file_storage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/snow_ball_file_storage.py -------------------------------------------------------------------------------- /tools/common/urllib_parse_pro.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/urllib_parse_pro.py -------------------------------------------------------------------------------- /tools/common/web_site_status.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/web_site_status.py -------------------------------------------------------------------------------- /tools/common/wiki_bots.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/common/wiki_bots.py -------------------------------------------------------------------------------- /tools/deploy/docker/step1/Dockerfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/deploy/docker/step1/Dockerfile -------------------------------------------------------------------------------- /tools/deploy/docker/step1/lic.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/deploy/docker/step1/lic.bin -------------------------------------------------------------------------------- /tools/deploy/docker/step2/Dockerfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/deploy/docker/step2/Dockerfile -------------------------------------------------------------------------------- /tools/deploy/profile.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/deploy/profile.sh -------------------------------------------------------------------------------- /tools/disclosures_site/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/data/car_brands.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/car_brands.json -------------------------------------------------------------------------------- /tools/disclosures_site/data/corrections.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/corrections.json -------------------------------------------------------------------------------- /tools/disclosures_site/data/dolbilo_requests.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/dolbilo_requests.txt -------------------------------------------------------------------------------- /tools/disclosures_site/data/external_links.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/external_links.json -------------------------------------------------------------------------------- /tools/disclosures_site/data/median_income.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/median_income.csv -------------------------------------------------------------------------------- /tools/disclosures_site/data/misspell_bin/morph.options: -------------------------------------------------------------------------------- 1 | SkipPredictBase 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/data/statistics.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/data/statistics.json -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/admin.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/admin.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/apps.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/apps.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/car_brands.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/car_brands.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/corrections.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/corrections.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/documents.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/documents.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/management/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/management/commands/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/migrations/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/models.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/models.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/permalinks.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/permalinks.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/ratings.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/ratings.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/serializers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/serializers.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/sql_helpers.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/sql_helpers.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/statistics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/statistics.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/acc_log_squeeze/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/car_brands/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/copy_person_id/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/corrected_person/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/dedupe/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/dedupe_divide/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/dedupe_real/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/dedupe_reuse_old_person_id/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/dedupe_with_dump_objects/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/elastic/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/export_human_files/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/fio_clusters/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_complex/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_document_order/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_external_year/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_fsin/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_fsin_2/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_own_type/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/import_relatives/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/human_files/human.xlsx: -------------------------------------------------------------------------------- 1 | 1 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/human_files/uniq_from_human.xlsx: -------------------------------------------------------------------------------- 1 | 4 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/old/domains/05.fsin.su/human_file_deleted_on_site.xlsx: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/old/domains/05.fsin.su/new.xlsx: -------------------------------------------------------------------------------- 1 | 5 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/old/domains/05.fsin.su/old_dlrobot.xlsx: -------------------------------------------------------------------------------- 1 | 3 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/processed_projects/05.fsin.su.1590209953/result/05.fsin.su/copy_dlrobot.xlsx: -------------------------------------------------------------------------------- 1 | 2 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/processed_projects/05.fsin.su.1590209953/result/05.fsin.su/new_dlrobot.xlsx: -------------------------------------------------------------------------------- 1 | 37. -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/processed_projects/05.fsin.su/result/05.fsin.su/1984.xlsx: -------------------------------------------------------------------------------- 1 | 1 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/processed_projects/05.fsin.su/result/05.fsin.su/new_dlrobot.xlsx: -------------------------------------------------------------------------------- 1 | 2 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/join_dlrobot_and_human/processed_projects/05.fsin.su/result/05.fsin.su/old_dlrobot.xlsx: -------------------------------------------------------------------------------- 1 | 3 2 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/nominal_income/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/office_web_site_redirects/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/parse_decl_title/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/ratings/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/regions/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/rubrics/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/russian_fio/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/update_person_redirects/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/view_office_search/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/view_person_name_misspell/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/tests/view_section_search/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/urls.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/urls.py -------------------------------------------------------------------------------- /tools/disclosures_site/declarations/views.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/declarations/views.py -------------------------------------------------------------------------------- /tools/disclosures_site/deduplicate/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/deduplicate/__init__.py -------------------------------------------------------------------------------- /tools/disclosures_site/deduplicate/config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/deduplicate/config.py -------------------------------------------------------------------------------- /tools/disclosures_site/deduplicate/model/run.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/deduplicate/model/run.sh -------------------------------------------------------------------------------- /tools/disclosures_site/deduplicate/toloka.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/deduplicate/toloka.py -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/settings/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/settings/dev.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/disclosures/settings/dev.py -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/static/a.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/disclosures/static/a.html -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/urls.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/disclosures/urls.py -------------------------------------------------------------------------------- /tools/disclosures_site/disclosures/wsgi.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/disclosures/wsgi.py -------------------------------------------------------------------------------- /tools/disclosures_site/manage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/manage.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/create_sample.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/create_sample.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/csv/csv_to_html.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/csv/csv_to_html.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/csv/unqote_tsv.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/csv/unqote_tsv.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/dbm_manage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/dbm_manage.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/dedupe/make_pool.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/dedupe/make_pool.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/dolbilo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/dolbilo.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/rename_db.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/rename_db.sh -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/rosstat/rosstat.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/rosstat/rosstat.py -------------------------------------------------------------------------------- /tools/disclosures_site/scripts/rosstat/salary.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_site/scripts/rosstat/salary.py -------------------------------------------------------------------------------- /tools/disclosures_stub/about.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/about.html -------------------------------------------------------------------------------- /tools/disclosures_stub/compare_income_descr.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/compare_income_descr.html -------------------------------------------------------------------------------- /tools/disclosures_stub/favicon-16x16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/favicon-16x16.png -------------------------------------------------------------------------------- /tools/disclosures_stub/favicon-32x32.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/favicon-32x32.png -------------------------------------------------------------------------------- /tools/disclosures_stub/favicon.ico: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/favicon.ico -------------------------------------------------------------------------------- /tools/disclosures_stub/favicon.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/favicon.png -------------------------------------------------------------------------------- /tools/disclosures_stub/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/permalinks.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/permalinks.html -------------------------------------------------------------------------------- /tools/disclosures_stub/reports/genders/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/reports/genders/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/reports/names/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/reports/names/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/reports/new-car/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/reports/new-car/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/reports/offices/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/reports/offices/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/robots.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/robots.txt -------------------------------------------------------------------------------- /tools/disclosures_stub/roc_new_car_prod_s5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/roc_new_car_prod_s5.png -------------------------------------------------------------------------------- /tools/disclosures_stub/second_office.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/second_office.html -------------------------------------------------------------------------------- /tools/disclosures_stub/sitemap-main.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/sitemap-main.xml -------------------------------------------------------------------------------- /tools/disclosures_stub/sitemap.xml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/sitemap.xml -------------------------------------------------------------------------------- /tools/disclosures_stub/smart_parser_spec.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/smart_parser_spec.html -------------------------------------------------------------------------------- /tools/disclosures_stub/static/er_diagram.svg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/er_diagram.svg -------------------------------------------------------------------------------- /tools/disclosures_stub/static/fonts/Roboto.ttf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/fonts/Roboto.ttf -------------------------------------------------------------------------------- /tools/disclosures_stub/static/fonts/Roboto.woff: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/fonts/Roboto.woff -------------------------------------------------------------------------------- /tools/disclosures_stub/static/main_menu.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/main_menu.js -------------------------------------------------------------------------------- /tools/disclosures_stub/static/sorttable.js: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/sorttable.js -------------------------------------------------------------------------------- /tools/disclosures_stub/static/style.css: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/static/style.css -------------------------------------------------------------------------------- /tools/disclosures_stub/statistics/index.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/statistics/index.html -------------------------------------------------------------------------------- /tools/disclosures_stub/stub.html: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/disclosures_stub/stub.html -------------------------------------------------------------------------------- /tools/dlrobot/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/data/dlrobot_rounds.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/data/dlrobot_rounds.json -------------------------------------------------------------------------------- /tools/dlrobot/central/dlrobot_central.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/dlrobot_central.py -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/check_fio.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/scripts/check_fio.py -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/crontab.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/scripts/crontab.txt -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/fns/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/monitoring/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/profile.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/scripts/profile.sh -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/uniq_tab.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/scripts/uniq_tab.py -------------------------------------------------------------------------------- /tools/dlrobot/central/scripts/yandex_disk.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/scripts/yandex_disk.py -------------------------------------------------------------------------------- /tools/dlrobot/central/send_docs.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/send_docs.py -------------------------------------------------------------------------------- /tools/dlrobot/central/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/central/tests/page.zip: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/tests/page.zip -------------------------------------------------------------------------------- /tools/dlrobot/central/tests/test_primitives.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/central/tests/test_primitives.py -------------------------------------------------------------------------------- /tools/dlrobot/common/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/common/central_protocol.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/central_protocol.py -------------------------------------------------------------------------------- /tools/dlrobot/common/dl_robot_round.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/dl_robot_round.py -------------------------------------------------------------------------------- /tools/dlrobot/common/remote_call.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/remote_call.py -------------------------------------------------------------------------------- /tools/dlrobot/common/robot_config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/robot_config.py -------------------------------------------------------------------------------- /tools/dlrobot/common/robot_project.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/robot_project.py -------------------------------------------------------------------------------- /tools/dlrobot/common/robot_step.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/robot_step.py -------------------------------------------------------------------------------- /tools/dlrobot/common/robot_web_site.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/robot_web_site.py -------------------------------------------------------------------------------- /tools/dlrobot/common/url_info.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/url_info.py -------------------------------------------------------------------------------- /tools/dlrobot/common/yandex_cloud.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/common/yandex_cloud.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/robot/adhoc/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/adhoc/__init__.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/adhoc/tgl_ru.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/adhoc/tgl_ru.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/adhoc/tomsk.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/adhoc/tomsk.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/configs/preliminary.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/configs/preliminary.json -------------------------------------------------------------------------------- /tools/dlrobot/robot/configs/prod.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/configs/prod.json -------------------------------------------------------------------------------- /tools/dlrobot/robot/configs/test.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/configs/test.json -------------------------------------------------------------------------------- /tools/dlrobot/robot/declaration_link.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/declaration_link.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/dl_robot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/dl_robot.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/process_folder.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/process_folder.sh -------------------------------------------------------------------------------- /tools/dlrobot/robot/readme.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/readme.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/ing.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/ing.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/mid.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/mid.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/mil.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/mil.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/minvr.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/minvr.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/vet.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/vet.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/regression_tests/zsro.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/regression_tests/zsro.txt -------------------------------------------------------------------------------- /tools/dlrobot/robot/run_regression_tests.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/run_regression_tests.sh -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/check_selenium.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/scripts/check_selenium.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/html_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/scripts/html_parser.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/scrape_sudrf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/scripts/scrape_sudrf.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/site_mirror.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/scripts/site_mirror.sh -------------------------------------------------------------------------------- /tools/dlrobot/robot/scripts/sitemap_download.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/scripts/sitemap_download.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests.sh -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/common_env.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/common_env.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/selenium_for_tests.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/selenium_for_tests.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_admkrsk.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_admkrsk.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_akrvo.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_akrvo.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_akrvo2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_akrvo2.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_configs.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_configs.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_content_type.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_content_type.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_dlrobot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_dlrobot.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_duma39.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_duma39.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_file_cache.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_file_cache.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_html_parser.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_html_parser.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_http_redirect.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_http_redirect.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_kolomnagrad1.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_kolomnagrad1.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_kolomnagrad2.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_kolomnagrad2.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_mid.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_mid.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_minzdrav.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_minzdrav.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_random_pdf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_random_pdf.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_regional.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_regional.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_search_engine.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_search_engine.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_selenium.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_selenium.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_sitemap.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_sitemap.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_ssl.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_ssl.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_timeout.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_timeout.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_unrar_cyr.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_unrar_cyr.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_urlparse.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_urlparse.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/test_video.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/test_video.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/web_site_monkey.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/robot/tests/web_site_monkey.py -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/web_sites/minzdrav1/found_links: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/robot/tests/web_sites/other_website/found_links: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/worker/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot/worker/dlrobot_worker.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/worker/dlrobot_worker.py -------------------------------------------------------------------------------- /tools/dlrobot/worker/dlrobot_worker.service: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot/worker/dlrobot_worker.service -------------------------------------------------------------------------------- /tools/dlrobot/worker/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot_human/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/dlrobot_human/dlrobot_human_dbm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot_human/dlrobot_human_dbm.py -------------------------------------------------------------------------------- /tools/dlrobot_human/input_document.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/dlrobot_human/input_document.py -------------------------------------------------------------------------------- /tools/dlrobot_human/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/office_db/countries.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/countries.py -------------------------------------------------------------------------------- /tools/office_db/data/fgup.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/fgup.txt -------------------------------------------------------------------------------- /tools/office_db/data/offices.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/offices.txt -------------------------------------------------------------------------------- /tools/office_db/data/regions.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/regions.txt -------------------------------------------------------------------------------- /tools/office_db/data/ross_stat.json: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/ross_stat.json -------------------------------------------------------------------------------- /tools/office_db/data/sudrf.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/sudrf.txt -------------------------------------------------------------------------------- /tools/office_db/data/web_sites_markup.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/data/web_sites_markup.txt -------------------------------------------------------------------------------- /tools/office_db/declarant_group_stat_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/declarant_group_stat_data.py -------------------------------------------------------------------------------- /tools/office_db/declaration_office_website.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/declaration_office_website.py -------------------------------------------------------------------------------- /tools/office_db/offices_in_memory.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/offices_in_memory.py -------------------------------------------------------------------------------- /tools/office_db/region_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/region_data.py -------------------------------------------------------------------------------- /tools/office_db/region_year_snapshot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/region_year_snapshot.py -------------------------------------------------------------------------------- /tools/office_db/rubrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/rubrics.py -------------------------------------------------------------------------------- /tools/office_db/rubrics/build_train.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/rubrics/build_train.py -------------------------------------------------------------------------------- /tools/office_db/russia.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/russia.py -------------------------------------------------------------------------------- /tools/office_db/russian_declarant_position.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/russian_declarant_position.py -------------------------------------------------------------------------------- /tools/office_db/russian_geo_word.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/russian_geo_word.py -------------------------------------------------------------------------------- /tools/office_db/russian_office_word.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/russian_office_word.py -------------------------------------------------------------------------------- /tools/office_db/russian_regions.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/russian_regions.py -------------------------------------------------------------------------------- /tools/office_db/scripts/calc_capital.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/calc_capital.py -------------------------------------------------------------------------------- /tools/office_db/scripts/calc_region.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/calc_region.py -------------------------------------------------------------------------------- /tools/office_db/scripts/office_db_manager.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/office_db_manager.py -------------------------------------------------------------------------------- /tools/office_db/scripts/set_rubrics.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/set_rubrics.py -------------------------------------------------------------------------------- /tools/office_db/scripts/set_wikidata_id.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/set_wikidata_id.py -------------------------------------------------------------------------------- /tools/office_db/scripts/update_offices.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/scripts/update_offices.py -------------------------------------------------------------------------------- /tools/office_db/web_site_list.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/web_site_list.py -------------------------------------------------------------------------------- /tools/office_db/year_income.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/office_db/year_income.py -------------------------------------------------------------------------------- /tools/predict_office/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/predict_office/base_ml_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/base_ml_model.py -------------------------------------------------------------------------------- /tools/predict_office/model/model/saved_model.pb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/model/model/saved_model.pb -------------------------------------------------------------------------------- /tools/predict_office/model/office_ngrams.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/model/office_ngrams.txt -------------------------------------------------------------------------------- /tools/predict_office/office_index.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/office_index.py -------------------------------------------------------------------------------- /tools/predict_office/office_pool.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/office_pool.py -------------------------------------------------------------------------------- /tools/predict_office/pools/test.source.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/pools/test.source.txt -------------------------------------------------------------------------------- /tools/predict_office/pools/test_fixed.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/pools/test_fixed.txt -------------------------------------------------------------------------------- /tools/predict_office/pools/train.sud.source.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/pools/train.sud.source.txt -------------------------------------------------------------------------------- /tools/predict_office/prediction_case.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/prediction_case.py -------------------------------------------------------------------------------- /tools/predict_office/read_office_from_title.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/read_office_from_title.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/predict_office/scripts/baseline_office.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/baseline_office.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/catboost_office.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/catboost_office.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/manage_pool.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/manage_pool.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/scp_wrapper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/scp_wrapper.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/tf_office_test.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/tf_office_test.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/tf_office_train.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/tf_office_train.py -------------------------------------------------------------------------------- /tools/predict_office/scripts/train.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/scripts/train.sh -------------------------------------------------------------------------------- /tools/predict_office/tensor_flow_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/predict_office/tensor_flow_model.py -------------------------------------------------------------------------------- /tools/predict_office/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/predict_office/tests/predict_office/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/requirements.txt -------------------------------------------------------------------------------- /tools/smart_parser_http/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/smart_parser_http/smart_parser_client.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/smart_parser_http/smart_parser_client.py -------------------------------------------------------------------------------- /tools/smart_parser_http/smart_parser_server.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/smart_parser_http/smart_parser_server.py -------------------------------------------------------------------------------- /tools/smart_parser_http/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/source_doc_http/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/source_doc_http/scripts/check_storage.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/source_doc_http/scripts/check_storage.py -------------------------------------------------------------------------------- /tools/source_doc_http/scripts/print_all_keys.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/source_doc_http/scripts/print_all_keys.py -------------------------------------------------------------------------------- /tools/source_doc_http/source_doc_client.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/source_doc_http/source_doc_client.py -------------------------------------------------------------------------------- /tools/source_doc_http/source_doc_server.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/source_doc_http/source_doc_server.py -------------------------------------------------------------------------------- /tools/source_doc_http/tests/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /tools/wikipedia_bots/decl2wiki.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/wikipedia_bots/decl2wiki.py -------------------------------------------------------------------------------- /tools/wikipedia_bots/readme.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/wikipedia_bots/readme.txt -------------------------------------------------------------------------------- /tools/wikipedia_bots/site_classifier.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/TI-Russia/smart_parser/HEAD/tools/wikipedia_bots/site_classifier.py --------------------------------------------------------------------------------