├── .gitignore ├── README.md ├── jseg ├── Emodet │ ├── __init__.py │ ├── emo2.json │ ├── emo3.json │ ├── emo4.json │ ├── emoall.txt │ ├── emodata.json │ ├── emodet.py │ └── emodet2.py ├── __init__.py ├── brill_tagger.pkl ├── dict.txt ├── emoall.txt ├── emodet.py ├── jieba.py ├── prob │ ├── emisP.json │ ├── initP.json │ └── tranP.json ├── ptt_encyc.txt └── test.py ├── pip.txt └── setup.py /.gitignore: -------------------------------------------------------------------------------- 1 | 2 | # Created by https://www.gitignore.io/api/vim,osx,linux,python 3 | 4 | ### Linux ### 5 | *~ 6 | 7 | # temporary files which can be created if a process still has a handle open of a deleted file 8 | .fuse_hidden* 9 | 10 | # KDE directory preferences 11 | .directory 12 | 13 | # Linux trash folder which might appear on any partition or disk 14 | .Trash-* 15 | 16 | # .nfs files are created when an open file is removed but is still being accessed 17 | .nfs* 18 | 19 | ### OSX ### 20 | *.DS_Store 21 | .AppleDouble 22 | .LSOverride 23 | 24 | # Icon must end with two \r 25 | Icon 26 | 27 | # Thumbnails 28 | ._* 29 | 30 | # Files that might appear in the root of a volume 31 | .DocumentRevisions-V100 32 | .fseventsd 33 | .Spotlight-V100 34 | .TemporaryItems 35 | .Trashes 36 | .VolumeIcon.icns 37 | .com.apple.timemachine.donotpresent 38 | 39 | # Directories potentially created on remote AFP share 40 | .AppleDB 41 | .AppleDesktop 42 | Network Trash Folder 43 | Temporary Items 44 | .apdisk 45 | 46 | ### Python ### 47 | # Byte-compiled / optimized / DLL files 48 | __pycache__/ 49 | *.py[cod] 50 | *$py.class 51 | 52 | # C extensions 53 | *.so 54 | 55 | # Distribution / packaging 56 | .Python 57 | build/ 58 | develop-eggs/ 59 | dist/ 60 | downloads/ 61 | eggs/ 62 | .eggs/ 63 | lib/ 64 | lib64/ 65 | parts/ 66 | sdist/ 67 | var/ 68 | wheels/ 69 | *.egg-info/ 70 | .installed.cfg 71 | *.egg 72 | 73 | # PyInstaller 74 | # Usually these files are written by a python script from a template 75 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 76 | *.manifest 77 | *.spec 78 | 79 | # Installer logs 80 | pip-log.txt 81 | pip-delete-this-directory.txt 82 | 83 | # Unit test / coverage reports 84 | htmlcov/ 85 | .tox/ 86 | .coverage 87 | .coverage.* 88 | .cache 89 | nosetests.xml 90 | coverage.xml 91 | *.cover 92 | .hypothesis/ 93 | 94 | # Translations 95 | *.mo 96 | *.pot 97 | 98 | # Django stuff: 99 | *.log 100 | local_settings.py 101 | 102 | # Flask stuff: 103 | instance/ 104 | .webassets-cache 105 | 106 | # Scrapy stuff: 107 | .scrapy 108 | 109 | # Sphinx documentation 110 | docs/_build/ 111 | 112 | # PyBuilder 113 | target/ 114 | 115 | # Jupyter Notebook 116 | .ipynb_checkpoints 117 | 118 | # pyenv 119 | .python-version 120 | 121 | # celery beat schedule file 122 | celerybeat-schedule 123 | 124 | # SageMath parsed files 125 | *.sage.py 126 | 127 | # Environments 128 | .env 129 | .venv 130 | env/ 131 | venv/ 132 | ENV/ 133 | env.bak/ 134 | venv.bak/ 135 | 136 | # Spyder project settings 137 | .spyderproject 138 | .spyproject 139 | 140 | # Rope project settings 141 | .ropeproject 142 | 143 | # mkdocs documentation 144 | /site 145 | 146 | # mypy 147 | .mypy_cache/ 148 | 149 | ### Vim ### 150 | # swap 151 | [._]*.s[a-v][a-z] 152 | [._]*.sw[a-p] 153 | [._]s[a-v][a-z] 154 | [._]sw[a-p] 155 | # session 156 | Session.vim 157 | # temporary 158 | .netrwhist 159 | # auto-generated tag files 160 | tags 161 | 162 | # End of https://www.gitignore.io/api/vim,osx,linux,python 163 | 164 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Jseg 2 | ## A modified version of Jieba 3 | 4 | All credit goes to [fxsjy/jieba](https://github.com/fxsjy/jieba"). 5 | Find more on: https://github.com/fxsjy/jieba 6 | 7 | 8 | ### Synopsis 9 | 10 | 1. **Equipped with Emoticon detection** 11 | Emoticons will not be segmented as sequences of meaningless punctuations. 12 | 13 | 2. **Data are trained with Sinica Corpus** 14 | Results are more accurate when dealing with Traditional Chinese (F1-score = 0.91). 15 | 16 | 3. **Using Brill Tagger** 17 | Training data are trained with Sinica Treebank, which raises the accuracy of POS tagging. 18 | 19 | ### Environment 20 | + Python2.7+ 21 | + Python3.3+ 22 | 23 | 24 | ### Installation 25 | 26 | pip install -U jseg 27 | 28 | 29 | ### Usage 30 | 31 | from jseg import Jieba 32 | j = Jieba() 33 | 34 | ### Add user defined dictionary 35 | 36 | j.add_guaranteed_wordlist(lst) 37 | 38 | 39 | Here's a sample text: 40 | 41 | 42 | sample = '期末要爆炸啦！ ◢▆▅▄▃崩╰(〒皿〒)╯潰▃▄▅▇◣' 43 | 44 | Segmentation *with* POS (part-of-speech) 45 | 46 | j.seg(sample, pos=True) 47 | -------------------------------------------------------------------------------- /jseg/Emodet/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/amigcamel/Jseg/1b8f6aa5e65cfa770cb7e81df0a4fbc15a7bcd35/jseg/Emodet/__init__.py -------------------------------------------------------------------------------- /jseg/Emodet/emo2.json: -------------------------------------------------------------------------------- 1 | ["(\u0c25\u0e3a\u02c7\u0c6a\u02c7\u0c25)", "( \u00b4\u0c25,_\u2025\u0c25\uff40)b", "( \u00b4\u0c25\u0c6a\u0c25\uff09\u03c3", "(\u11c2\u0434\u11c2 )", "(\u229c\u203f\u229c\u2734\uff09", "( \u0384\u25de\u0e34 .\u032b.\u032b \u25df\u0e34\u2035)y", "(;\u0326\u0326\u0326\u0326\u0326\u0326\u0326\u0326\u2182\u20dd\u20d3\u20d9\u20da\u20d8\u0434\u2182\u20dd\u20d3\u20d9\u20da\u20d8;\u0326\u0326\u0326\u0326\u0326\u200b\u0326\u0326\u0326)", "\u2037\u0317\u2182\u51f8\u2182\u2034\u0316", "( \u0c20\u0d60\u0c20 )\uff89", "\u10f0\u10f0\u10f0\u275b\u203f\u275b\u10f4\u10f4\u10f4", "(\u0e51\u0c25\u0e34\u76ca\u0c25\u0e34)", "(\uff0c_\u309d\u06de)", "\u3020_\u3020", "\u2022\u271e_\u271e\u2022", "\u2299\u2583\u2299", "\u0ca4\u0c8e\u0ca4", "\u0c20_\u0c20", "\u0cb8_\u0cb8", "\u2181_\u2181", "\u21ce_\u21ce", "\ud76b_\ud76b", "\u0ca0_\u0ca0", "(\uff65\u0e34\u03c9\uff65\u0e34)", "(\u0ca5_\u0ca5)", "\u25c9\u0e37\u76ca\u25c9\u0e37", "\u0ca0\u76ca\u0ca0", "(\u00b4=\u0e34\u76ca=\u0e34\uff40)", "(\u00b4^\u0e34\u76ca^\u0e34\uff40 )", "(\u25de\u227c\u25c9\u0e37\u227d\u25df;\u76ca;\u25de\u227c\u25c9\u0e37 \u227d\u25df)", "\u261c(:\u265b\u0e3a;\u76ca;\u265b\u0e3a;)\u261e", "(\u256c\u24ea\u76ca\u24ea)", "(\u256c\uff9f\u25e5\u76ca\u25e4\uff9f)", "( \u256c\u25e3 \u76ca\u25e2)y", "(\u256c\u0ca0\u0e34\u76ca\u0ca0\u0e34)", "(\u0e51\u2256\u0e34\u0e34\u76ca\u2256\u0e34\u0e51)", "(\u256c\u265b. \u76ca\u265b. )", "(\u256c\u0ca0\u76ca\u0ca0)\uff7a\uff9e\uff99\u30a1!!", "\u723b((\u256c\u0ca0\u0e34\u76ca\u0ca0\u0e34))\uff72\uff76\uff98\uff7d\uff9e\uff91\uff6f!!", "\u7676(\u7676;:\u309c;\u76ca;\u309c;)\u7676", "(\u0e51\uff65\u0e34\u22d6\u22d7\uff65\u0e34\u0e51)", "(\u0301\u25c9\u25de\u0c6a\u25df\u25c9\u2035)", "( ^\u0e34\u0c6a^\u0e34)", "(\u25de\u227c\u25cf\u227d\u25df\u25de\u0c6a\u25df\u25de\u227c\u25cf\u227d\u25df)", "\uff08 \u00b4\u2623///_\u309d///\u2623\uff40\uff09", "(\u0301\u25d5///\u25de\u0c6a\u25df///\u25d5\u2035)/", "\u30fd(\u0301\u25d5\u25de\u0c6a\u25df\u25d5\u2035)\uff89", "(\u0301=\u25de\u0c6a\u25df=\u2035)", "(*\u2609\u0c6a\u2299*)", "(\u0301^\u25de\u0c6a\u25df<\u2035)", "\uff08\u00b4\u225d\u25de\u2296\u25df\u225d\uff40)", "\u4e41( \u25d4 \u0c6a\u25d4)\u300c", "(\u0301\u2256\u25de\u0c6a\u25df\u2256\u2035)/", "(\u0301>\u25de\u0c6a\u25df<\u2035)\uff89\uff7c", "(\u0301\u261c\u25de\u0c6a\u25df\u261e`)/", "(\u0301\u3002\u25de\u0c6a\u25df\u3002\u2035)", "(\u0301\u3064\u25de\u0c6a\u25df\u2282\u2035)", "(\u0301\u4e00\u25de\u0c6a\u25df\u4e00\u2035)", "(V)(\u0301\u25c9\u25de\u0c6a\u25df\u25c9\u2035)(V)", "(\u0301\u0b87\u25de\u0c6a\u25df\u0b87\u2035)", "(\u0384\u0ca2\u25de\u0c6a\u25df\u0ca2\u2035)\u25c9\u25de\u0c6a\u25df\u25c9)", "(\u0384\u25c9\u25de\u0c6a\u25df\u25c9\u2035)", "(\u0301\u25c9\u25de\u0c6a\u25c9*\u25c9\u25c9\u25de\u25d5\u0c6a\u0c6a\u25df\u25d5\u25c9\u25c9\u03c6\u25c9\u0c6a\u25de\u25c9*\u2035)", "(\u0301\u3020\u25de\u0c6a\u25df\u3020\u2035)", "(\u0301\u25a3\u25de\u0c6a\u25df\u25a3\u2035)", "(\u0301\u273a\u25de\u0c6a\u25df\u273a\u2035)", "(\u0301\u0c25\u25de..\u25df\u0c25\u2035)", "(\u0301\u2299\u25de\u0c6a\u25df\u2299\u2035)", "(\u0301\u2738\u25de\u2649\u25df\u2738\u2035)", "(\u0301\u06de\u25de\u0c6a\u25df\u06de\u2035)", "(\u0301\u0ca4\u25de\u0c6a\u25df\u0ca4\u2035)", "(\u0301\u0b87\u25de\u0c6a\u25df\u0b87\u2035)", "(\u0301\u58a8\u25de\u0c6a\u25df\u93e1\u2035)", "\u0295\u2022\u032b\u0361\u2022\u0294", "\u10da(\u0301\u25c9\u25de\u0c6a\u25df\u25c9\u2035\u10da)", "\uff08\u10da ^\u0e34\u0c6a^\u0e34\uff09\u10da", "\u10da(\u0301\u0ca2.\u25de\u0c6a\u25df\u0ca2\u2035\u10da)", "\u10da\u2182\u203f\u203f\u2182\u10da", "\uff08\u10da ^\u0e34\u0c6a^\u0e34\uff09\u10da", "\u10da(\u2579\u25e1\u2579\u10da)", "\u10da(\u2579\u03b5\u2579\u10da)", "\uff08\u10da ^\u0e34\u0c6a^\u0e34\uff09\u10da", "\u10da(\uff9f\u0434\uff9f\u10da)", "\u10da\uff08\u0b87 e \u0b87\uff40\u3002\u10da\uff09", "\u10da\uff08\u00b4\u2200`\u10da\uff09", "\u10da(\u2070\u2296\u2070\u10da)", "\u10da(\uff65 \u0e34\u03c9\uff65\u0e34\u10da)", "\u10da(^o^\u10da)", "\u10da(\u30fb\u2200\u30fb )\u10da", "\u10da(\uff40\u2200\u00b4\u10da)", "\u10da(\uff3e\u03c9\uff3e\u10da)", "\u10da(\u275b\u25e1\u275b\u273f)\u10da", "\u2501( \u0e34\u25d5\u3268\u25d5 \u0e34)\u2501", "( \u0e34\u25d5\u3268\u25d5 \u0e34)", "(\uff65\u0e34\u3268\uff65\u0e34)", "( \u0e34\u0398\u3268\u0398)\u0e34", "(\u00d2\u3268\u00d3\u05f1)", "(\u00b4\u25c9\u3268\u25c9)", "(\u25d5\u072b\u25d5)", "(\u00d2\u072b\u00d3\u05f1)", "(\u272a\u03c9\u272a)", "(\u273a\u03c9\u273a)", "(\u00d2\u03c9\u00d3\u05f1)", "(\u0301\u25d5\u25de\u3180\u25df\u25d5`)", "(;\u00b4\u0f0e\u0eb6\u0414\u0f0e\u0eb6`)", "\u262e\u2581\u2582\u2583\u2584\u263e \u265b \u25e1 \u265b \u263d\u2584\u2583\u2582\u2581\u262e", "\u25e2\u2586\u2585\u2584\u2583\u5d29\u2570(\u3012\u76bf\u3012)\u256f\u6f70\u2583\u2584\u2585\u2587\u25e3", "\u03a3\uff08\uff9f\u0434\uff9flll\uff09", "(*\u02d8\ufe36\u02d8*).\uff61.:*\u2661", "\u256e(\u256f3\u2570)\u256d\u3000\u256e(\u256f_\u2570)\u256d\u3000\u256e(\u256f\u25bd\u2570)\u256d \u3000\u256e(\uffe3\u25bd\uffe3)\u256d", "(\u256f\u25d5\u25de\u0c6a\u25df\u25d5\u2035)\u256f\ufe35 \u2534\u2500\u2534", "\uff08\u256f\uff0d\uff3f\uff0d\uff09\u256f\u2567\u2567", "\uff08\u256f\u2035\u25a1\u2032\uff09\u256f\ufe35\u2534\u2500\u2534", "\u263b", "\uff08\u30fb\u2200\u30fb\uff09", "(\u309d\u2200\uff65)", "\uff08\uff3e\u03c9\uff3e\uff09", "(o\uff9f\u03c9\uff9fo)", "(*\uff3e\u30fc\uff3e)", "(\u2267\u2200\u2266)", "( \u00b4\u2200`)", "(\u2312o\u2312)", "(^-^*)", "\u30fd( ^\u2200^)\uff89", "(\uff9f\u2200\uff9f )", "(\uff89\u2200\uff40 )", "(\uff40\u2200\u00b4)", "(\uffe3\u30fc\uffe3)", "\u30fd( \u00b4\u30fc)\u30ce", "(-\u03b9_- )", "(\u00b0c\uff3f,\u00b0`)", "\u2727*\uff61\u0669(\u02ca\u15dc\u02cb*)\u0648\u2727*\uff61", "(#\uff9f\u0414\uff9f)", "(\uff40_\u309d\u00b4)", "(`\u76bf\u00b4)", "(\uff89\uff40\u22bf\u00b4)\uff89", "\u30fd(\uff40\u0434\u00b4)\uff89", "( \u00b4\u2022\u0325\u0325\u0325\u03c9\u2022\u0325\u0325\u0325` )", "(;_;)", "(\uff1b\u0434\uff1b)", "(\uff89\u0414`)", "(\u3064\u0414`)", "\uff61\uff9f\uff65(\uff89\u0414`)\u4eba(\u00b4\u0414`)\u4eba(\u0414` )\uff65\uff9f\uff61", "(\u30ce\u0414\uff34)", "(\uff70\uff40 )", "(/\u0414`)~\uff9f\uff61", "\u03a3(\u00b0\u0414\u00b0;", "(\u00b0\u0414\u00b0)", "\u2211(\uffe3\u25a1\uffe3;)", "\u03a3(\u30fb\u03c9\u30fb\u30ce)\u30ce", "(\uff9f\uff9b \uff9f\uff89)\uff89", "\u03a3\u2517(\uff20\u30ed\uff20;)\u251b", "\u03a3\u30fd(\uff9f\u0414 \uff9f; )\uff89", "\u30fe(>\u0414<;)))).....", "(((\u250f(;\uffe3\u25bd\uffe3)\u251b", "\u30fe(\u3002\uffe3\u25a1\uffe3)\u30c4", "\u2514(\uffe3^\uffe3 )\u2510", "(\u309co\u309c(\u2606\u25cb=(-_- )", "(\uff4f\uffe3\u2207\uffe3)\uff1d\u25ef)`\u03bd\u309c)\uff65\uff1b'", "( \u00b4\u2200\uff40)\u03c3", "(\u03c3\u00b4\u25a1\uff40)\u03c3", "( \uff9f\u0414\uff9f)\u03c3", "(\u03c3\uff9f\u2200\uff9f)\u03c3\uff9f\u2200\uff9f)\u03c3\uff9f\u2200\uff9f)\u03c3", "\uffe3\ufe43\uffe3", "( \uff40\u76bf\u00b4)\uff89\"\u9e75=\u5ddb\u5ddb\u5ddb\u5ddb.:;:", "( \u30fbO\u30fb)\u30ce\u2312\u25cf", "( \uff9f\u0414\uff9f)\uff89\u2312\u2464", "(\uff65\u2200\uff65)\u3063/\u51f5", "(o`\uff65\u03c9\uff65)\uff89(\uff89\u0414`)", "(\u00b4\u30fc\uff40)y\u2500\u251b", "\u30fe(*\u03a6\u03c9\u03a6)\u30c4", "(\u00b4\u2200\uff40\u2229)\u2191\u2191", "(\uff9f\u0414\uff9f\u2261\uff9f\u0414\uff9f)?", "( \uff9f\u00c5\uff9f)", "\u2044(\u2044 \u2044\u2022\u2044\u03c9\u2044\u2022\u2044 \u2044)\u2044", "(\u2256\u203f\u309d\u2256)\u2727", "\u51f8(= _=)\u51f8", "( \u0361\u00b0 \u035c\u0296 \u0361\u00b0)", "\u2727(\uc21c_\uc21c)\"", "(\u256f\u1561\u1f64\u1561)\u256f", "\u2db8(\u159b \u0394 \u159c)\u2db8", "\u10da(\u2022\u0300 _ \u2022\u0301 \u10da)", "\u25d4\"L__\u25d4"] -------------------------------------------------------------------------------- /jseg/Emodet/emo3.json: -------------------------------------------------------------------------------- 1 | ["(\uffe3\u25bd\uffe3)", "<(\uffe3\ufe36\uffe3)>", "\u2570(\uffe3\u25bd\uffe3)\u256f", "(\uffe3\ufe4f\uffe3)", "\u256e(\u256f\u25bd\u2570\")\u256d\u2606", "\u2570(\uffe3\ufe36\uffe3)\u256f", "(*\u03c9*)", "(#\uffe3\u25bd\uffe3)o\u256d\u256f \u2606\u2570\u256eo(\uffe3\u25bd\uffe3///)<", "\ufe3f(@\uffe3\ufe36\uffe3@)\ufe3f", "o(\u2035\u25bd\u2032)o", "(\u25cb^\uff5e^\u25cb)", "^(\uff0a)^", "\uff08*^\ufe4f^*\uff09", "(*^\uff20^*)", "\uff5e(\u256f\u25bd\u2570)\uff5e", "\uff5e(\uffe30\uffe3)\uff5e", "~(\uffe3\u25bd\uffe3)~", "(\ufe4b\ufe4f\ufe4c)", "\\(^\u03c9^)/", "\\(\u2267\u03c9\u2266)/", "\u2267\u25bd\u2266", "<\u03a6\ufe4f\u03a6>", ">(\u02d9\u03c9\u02d9)<", "(=\u02d9~\u02d9=)", "(\u03c3 \u2500.\u2500)\u03c3", "q(\uffe3\u03b5\uffe3 )p", "<(\uffe3><\uffe3)>", "(\u2032\u25a1\u2035)\u256d\uff5e", "(\u2032\uff31\u2035)\u256d\uff5e", "\\(\u256f\ufe4f\u2570)", "\u02cb(\u2032\uff5e\u2035\")\u02ca", "\u02cb( \u00b0 \u25bd\ufe51\u00b0|||| )", "^(\u02d9oo\u02d9)^", "^(\uffe3(oo)\uffe3)^", "=(\u2500(oo)\u2500)=", "\u222a(\uffe3\ufe41\uffe3)\u222a", "q(\u2032\u25a1\u2035)p", "(\uffe3y\u25bd\uffe3)\u256d", "\u03b5=\u03b5=\u03b5=\u03b5=\u03b5=\u03b5=\u256d(\uffe3\u25bd\uffe3)\u256f", "\uff5e(\uffe3\u25bd\uffe3)\uff5e(\uff3f\u25b3\uff3f)\uff5e(\uffe3\u25bd\uffe3)\uff5e(\uff3f\u25b3\uff3f)\uff5e(\uffe3\u25bd\uffe3)\uff5e", "( \u2032-`)y-\uff5e", "( \uffe3 c\uffe3)y\u2582\u03be", "(\u2032\u03c9\u2035)y-\uff5e", "\uff08\uffe3(\u221e)\uffe3\uff09y-\uff5e", "(*-*)y~", "<(\u2035\ufe35\u2032)>", "<(\u2035\u25a1\u2032)>", "\u2570(\u2035\u25a1\u2032)\u256f", "<(\u2035\u25b3\u2032)>", "\u2570(\u2035\u25b3\u2032)\u256f", "\u256e(\uffe3\u25a1\uffe3)\u256d", "(#\u2035\u25a1\u2032)\u256f", "\u2570(\u2035\u25a1\u2032#)", "\u51f8(\u2594\ufe35\u2594#)", "(\u256f \u2035\u25a1\u2032)\u256f", "\u51f8(\uffe3\u76bf\uffe3*)", "\u25cb(#\u2035\ufe3f\u2032\u3128)\u25cb", "(\u2035\u25a1\u2032)--\u25cf(#)\uffe3(00)\uffe3)", "( *\u2035\u0414\u2032)--\u25cf(#)\u2035\u22bf\u2032)\u256e", "(#\u2594\ufe35\u2594)=====\u25cf(\uffe3 \u22bf\"\uffe3)\u256d", "(#\u2035\u25bd\u2032)\uff0f~~~~~~~~~(\u2606___\u2606)", "(\u25cb\u2035\u25b3\u2032)=\u25cb\u25cb\u25cb\u25cb(\uffe3#)3\uffe3)", "(o \u2500_\u2500)==O)\u3002O\u3002)", "(#\u2594\u76bf\u2594)\u2500<(*\u2014_____\u2013)", "(*\u2500\ufe35\u2500)====Co( \u4e00//////\u4e00)o", "(#\u256f\uffe3\u25a1\uffe3)\u256f~\u2534\u2500\u2534", "(\u256f \u2035\u0414\u2032)\u256f\u2567\u2500\u2500\u2500\u2567", "\u2569\u2550\u2550\u2569 \uff5e \u2570(\uffe3\u25b3\uffe3\u2570)", "\u2534\u2534~\u2570(\u2035\u25a1\u2032#)", "(\u256f#\u2594__\u2594)\u256f\u2534\u2534\\(\"\uffe3\u25a1\uffe3)", "(\u256f\u2267\u25bd\u2266)\u256f ~\u2569\u2550\u2550\u2550\u2569", "(\u3012__\u3012)", "(\u3012\u25b3\u3012)", "(\u3012\u25c7\u3012)o\uff5e", "(\u3012\ufe35\u3012)", "(\u3012\ufe4f\u3012)", "q(\u3012\u25a1\u3012)p", "\u256e(\u3012\ufe4f\u3012)\u256d", "\uff08 \u2594\ufe3f\u2594\uff09\u256d\uff5e", "(\u03a0\u25b3\u03a0)", "\u03c3 >\ufe3f<\u03bf", "\u03c3(\u2025)", "(\u03a0___\u03a0)", "\u256e(\u256f_\u2570)\u256d", "\u256e\uff08\u256f\ufe4f\u2570\uff09\u256d", "\u256e\uff08\u256f\u2574\u2570\u2225\uff09\u256d", "\u256e(\u256f\uff5e\u2570\" )\u256d", "\u256e(\u2032\u03c9\u2035)\u256d", "\u256e(\ufe40\u03c9\ufe40\")\u256d", "\u256e(\u256f\u25bd\u2570)\u256d", "\u256e(\u2594\u25bd\u2594\")\u256d", "\u256e(\ufe4b\ufe4f\ufe4c)\u256d", "\u256e(\u2594\u03b5\u2594)\u256d", "/(\u256f\u25bd\u2570\")\\", "\u2196\u256d\u2570('(\u03a9)')\u256f", "\u02cb(\u2032\uff5e\u2035\")\u02ca", "\ufe3f(\u2032\u25a1\u2035)\ufe3f", "\u256e(\uffe3(oo)\uffe3)\u256d", "o(.\u301d.)o", "o(\u2027\"\"\u2027)o", "o(\u2027/////\u2027)o", "o(*/////*)o", "o(\uff20//////\uff20)o", "o(\u2299/////\u2299)o", "o(\u2500//////\u2500)o", "\\(\u2299O\u2299)/", "\\\u2299\u25bd\u2299/", "/(\u2032\u02d9\u2035)\\", "/(\u2032\u25a1\u2035)\\", "(\u00af \u25a1 \u00af)", "\\(\"\uffe3\u25a1\uffe3)/", "/(*\uffe3\u25b3\uffe3)\\", "(((m -_-)m", "( \u00af\ufe41\u00af )", "(-_____-x)", "(\uffe3\u25a1\uffe3||)", "\ufe4f\ufe4f(\u310f\uffe3\u25bd\uffe3)\u310f", "\\(\u2032\u25a1\u2035)/", "\u311f(\uffe3\u25bd\uffe3\u311f)\ufe4f\ufe4f", "(#\uffe3\u25bd\uffe3)o\u256d\u256f \u2605", "\\(\uffe3\u25a1\uffe3)/", "\u311f(\uffe3\u25bd\uffe3\u311f)\ufe4f", "\\(\uffe3\u25a1\uffe3*)/", "\u311f(\uffe3\u25bd\uffe3\u311f)\ufe4f\ufe4f", "\ufe4f\ufe4f(\u310f\uffe3\u25b3\uffe3)\u310f", "\u2570(\u3012\u25cb\u3012)\u256f", "\ufe4f\ufe4f(\u310f\uffe3\u25bd\uffe3)\u310f (\u4e00(oo)\u4e00)\uff0a \u311f(\uffe3\u25bd\uffe3\u311f)\ufe4f", "\u256d(\u256d(\u256d (\u310f\uffe3\u25b3\uffe3)\u310f", "\\(\"\uffe3\u25bd\uffe3)/", "\u256d(\u256d(\u256d\uff08\u310f\uffe3\u25a1\uffe3\uff09\u310f", "(\u4e00(oo)\u4e00)\uff0a", "(\u3003^\u25c7^\u3003)", "(*^ \u25c7 ^*)", "(@^\u25c7^@)", "(\u256f^\u25c7^)\u256f", "( \u2267\u25c7\u2266 )", "(\uff34\u25c7\uff34)", "(=\uffe3\u25c7\uffe3=)\uff36", "\\(\uffe3\u25c7\uffe3)/", "(*\uffe3\u25c7\uffe3)\u256f", "\u256e(\uffe3\u25c7\uffe3)\u256d", "(\u3128\u2035\u25c7\u2032)", "(\uff20\u25c7\uff20)", "(*\u2035\u25c7\u2032*)", "(\u3002\u25c7\u3002||)", "\\(\u03a0\u25c7\u03a0)", "(\uffe3\u25c7\uffe3)\u51f8", "\u03a8(\u2267\u25c7\u2266)\u03a8", "((((<||\uff3f\u25c7\uff3f>", "|||||\u256d(\uff3f\u25c7\uff3f)\u256e|||||", "(-\u25ce\u25c7\u25ce-)o", "-{}@{}@{}-\u03c5(\uffe3\u25c7\uffe3*)", "(\uffe3\u2554\u25c7\u2557\uffe3)", "\u5ddd\u2027\u25c7\u2027\u5ddd", "\u5dde\u2027\u25c7\u2027\u5dde", "| \u67f1 |\u2510(\uffe3\u25c7\uffe3*)", "\u2514|\u2027\u25c7\u2027|\u2510", "\u2225\u2225<\u2225\u2032 \u25c7\u2035 || >\u2225\u2225", "|\u03a6|<|\u2032|\u25c7|\u2035|>|\u03a6|", "\u2560\u2566\u2569\u2566\u2566\u2569\u2563\u25c7\u2027)", "\u252c\u2534\u252c\u2534\u2524(\u2027\u25c7\u251c\u252c\u2534\u252c", "\uff3c(\uffe3\u3000\u3000\u3000\u25c7,\u3000\u3000\u3000\uffe3)\uff0f", "\u250c|\u2267\u25c7\u2266|\u2518\u505a\u2514|\u2267\u25c7\u2266|\u2510\u9ad4\u250c|\u2267\u25c7\u2266|\u2518\u64cd", "|||\u256d(\uff3f\u25c7\uff3f)\u256e|||", "(\u3128\u2035\u25c7\u2032)\u51f8", "(\u3112\u25c7\u3112)o\uff5e", "\u5ddd\u2027\u25c7\u2027\u5ddd", "-@{}@{}-\u03c5(\uffe3\u25bd\uffe3*)"] -------------------------------------------------------------------------------- /jseg/Emodet/emo4.json: -------------------------------------------------------------------------------- 1 | ["((((((\u25cb\u309c\u03b5\u309c\u25cb)\u30ce~~", "~~~\u30fe(\u25cb\u309c\u25bd\u309c\u25cb)", "(\uff20\u309c\u25bd\u309c@)\u30ce~~~", "((\u25cf\u309cc_\u309c\u25cf))", "\u30fe(\u25cf\u00b4\u25bd\uff40\u25cf)\u30ce", "(\uff4f\u00b4\u8278\uff40\uff4f)", "\u30fe(\uff20\u309c\u25bd\u309c@)\u30ce", "...\"8-( \u25cf\uff40\u03b5\u00b4\u25cf)\u723b(\u25cf\uff40\u03b5\u00b4\u25cf )-8\"", "(\u2032\u222a`\u8278)\u2312*\u3002", ".\u3002.(\u3002\u00b4\u30fb\u03c9\u30fb\u3002)\u30ce\u309b .\u2661\u3002:", "(\u30ce\u03b5\u309c\u25cf)", "(\u2022\uff56'\u221e\u4eba)", "\u3000\u3000", "\uff08d\u2022\u309cv`\u25cf)+\u309c\u221e", "(\u30ce)'\u0434\uff40(\u30fe)", "\u30fd\u2022\u00b4\u0437`\u2022\u30ce\u266c", "(\u25cf'3`)\u30ce\u2312\u2665", "\u30fe(\u2665'3`\u2660))\u30ce", "\u2191\u2191.+\u309c*\u30fd(\u32a4\u00b4\u2200`*\u32a4)\u30ce\u3002:\u309c+", "(\u309d\u03c9\u30fb)", "\uff08\u3003\u30fb\u222a\u30fb\uff40\uff09", "(**\u25d1\u2200\u25d0)", "(o\u2661\u212b\u2661o)", "(\u30fbc_,\u30fb$)", "(\u3003\u00b4\u2022\u222a\u2022\uff40)", "d(\u309d\uff43_,\u30fb*\uff09b", "(\u25cf*\u00b4\u0414`*)\u30ce??", "(,,\u309cc_,\u309c*)\u30ce", "(\u25cf*>_,\uff1cp\u559cq)*", "\u30fe(o\u309c_,\u309co)\u30ce", "*:.\u3002.\u3002.:*\u30fe(\u25cf\u00b4\u30fb\u03c9\u30fb)(\u30fb\u03c9\u30fb`\u25cb)\u30ce*:.\u3002.\u3002.:*", "\u30ce(\u2606\u00b4\u03b5\uff40\u2261)\u30ce", "\u3001(\u30fbc_,\u30fb\u3002)", "(*\u00b4\u03c9\uff40)\u3063", "(\u30ce\u30fb\u03c9\u30fb)\u30ce\u30fe(\u30fb\u03c9\u30fb\u30fe)", "(\u03c3\u30fb\u0414\u30fb)\u03c3\u2605", "(\uff40\u30fb\u03c9\u00b4\u30fb+)", "(*\u309d\u03c9\u30fb*)\u30ce", "\u30fe(\u25cf\u00b4\uff35\uff40\u25cf)\u30ce", "\u2661\u30fd(\u25cf\u00b4\u03b5\uff40\u25cf)\u30ce\u2661", "\u30fd(\u00b4\u2200\uff40\u25cf)\u30ce", "\u30fe(*\u00b4\u2200\uff40*)\u30ce", "\u309c\u30fb*+:\u309c((((\u00b4*\u30fb\u0434\u30fb*`))))\u309c\u30fb*+:\u3002", "\u3002\uff4f\u25cbd(-\u03c9\u30fb\u3002)", "(\u2605'A`!)", "(\u266f'`\u2605!) ", "\uff08\u3064'A`\u309c\uff09\u309c+\u309c", "\u30fb:*:\u30fb\u30a3\u30a3!!(*\u00b4\u25bc\uff40*\u4eba)\u30fb:*:\u30fb", "(*'v`*)\u309c+ ", "\u30fe(`\u30fb\u03c9\u30fb\u00b4) ", "(\uff42'`*\uff44))", "(\u25cb'\u8278`\u25ce)", "\u25bc*\u309cv\u309c*\u25bc", "\u2514(\u309c\u2200\u309c\u2514)(\u2518\u309c\u2200\u309c)\u2518", "......\u250c\u2524\u00b4\u0434`\u251c\u2510", "(\u309c\u2200\u309c\u30ce\u30ce\u309b\u2606' ", "\u30fd( \u30fb\u2200\u30fb)\u30ce\u250c\u251b\u03a3(\u30ce `\u0414\u00b4)\u30ce", "o\uff1c\uff0d\u2200\uff0d\uff1eo", "(q\uff400\u00b4p)", "\u03a3 \u309c\u30ed\u309c\u2261( \u3000 \u30ce)\u30ce ", "\u3002+\u309c\u2605(\u4eba'v\uff40*)\u2606\u309c+\u3002", "\u3002+\u309c\u30fe(o\u309c\u2200\u309co)\u30ce\u3002+\u309c ", "\u2606.\u3002.:*(\u5b09\u00b4\u0414\uff40\u5b09).\u3002.:*\u2606", "\u30fd(\u25cf\u309d\u2200\u30fb\u25cf)\u30ce\u30fb.\u3002*", "\u30fe@(o\u30fb\u30a7\u30fbo)@\u30d0\u30a4\u30d0\u30a4\u266a", "~~~\u30fe(\u25cb\u309c\u25bd\u309c\u25cb) ", "((((((\u25cb\u309c\u03b5\u309c\u25cb)\u30ce~~", "\u30fb\u309c\u30fb\u30fb:*:\u30fb\u30fb\u309c\u30fb(\u25cf\u00b4\u2200\uff40 \u25cf)\u30fb\u309c\u30fb\u30fb:*:\u30fb\u30fb\u309c\u30fb", "\u03b5=\u03b5=(\u25cf\u30fb\u2207\u30fb\u4eba\u25ce\u2267\u76bf\u2266 \uff09", "\u30fe(\u25cf\u00b4\u25bd\uff40\u25cf)\u30ce", "\u30fe(\uff20\u309c\u25bd\u309c@)\u30ce", "((\u25cf\u309cc_\u309c\u25cf))b", "(\u2605\u309d\u2200\u30fb\u2606)\u30ce", "(*\u00b4\u2200\uff40*)", "\u30fe(o\u309c\u30a7\u309co)\u30ce", "(\u00b4\u30fb\u03c9\u30fb`)(\u00b4\u30fb\u03c9:;.:..(\u00b4\u30fb:;...::;.:.:: ", "\u2605\u309c+.\u30fd(*\u3002>\u0414<)o\u309c+.\u309c\u2606", "(*\u2609\u0c6a\u2299*)", "\u30f5\u309b\u2500\u2501\u2500\u2501(*\u309c\u0414\u2609*)\u2500\u2501\u2500\u2501\u30f3!!", "\u03a3(\u2609A\u2299\uff4c|\uff4c)", "(\u25cf\u00b4\u2200\uff40\u25cf)\u30ce\u3002.:*\u30fb\u309c*:.\u3002.", "(\u25cf\u00b4\u30fb\u03c9\u30fb)\u30ce", "\u30fe(o\u00b4\u03c9\uff40o)\u2663\u2667\uff27\uff4f\uff27\uff4f\uff27\uff4f\uff01\u2667\u2663(o\u00b4\u03c9\uff40o)\u30b7", "(\uff40\u30fb\u03c9\u30fb)\u03c3\u03c3\u3010\u309c+.\u309c\u043d\u0451\u03b9\u03b9\u03bf\u309c+.\u309c\u3011(\u30fb\u03c9\u30fb\u00b4)\u30b7", "(\u25ce\u00b4\u8278\uff40))\u3010\u2666\u2666\u0397\u0451\u03b9\u03c3\u03c3\u2666\u2666\u3011((\u00b4\u8278\uff40\u25ce)", "\u30fe(\u25cf\u00b4\u212b`)\u3010\u266c\u043d\u0451\u03b9\u03b9\u03bf\u266c\u3011(\u00b4\u212b\uff40\u25cf)\u30b7", "(\u30fb\u03c9\u30fb\u3002)", "(\uff0e\uff0e\u3002)", "(\u3002\u30fb\u03c9\u30fb)", "\u30fd\uff08\u309c\u2200\u309c\uff09", "(\u25cf\u309c\uff56\uff40)\u30ce", "\u30fd(\u309c\uff56\uff40\u25cf)", ".+:\u3002(\u25cf\u309d\u03c9\u30fb)\uff5e\u2606NE\uff57!!\u309c.+:", "(\u3002\uff0e\uff0e) ", "\u30fe(\u2312\u2207\u2312*)See You\u266a", "(\u25cf\u2312\u2200\u2312\u25cf)", "\u30fe(\uffe3\u2207\uffe3)Bye\u30fe(\uffe3\u2207\uffe3)Bye"] -------------------------------------------------------------------------------- /jseg/Emodet/emoall.txt: -------------------------------------------------------------------------------- 1 | (ﾒﾟДﾟ)ﾒ 2 | (〃^◇^〃) 3 | Zz(´-ω-`*) 4 | (꒪ͦᴗ̵̍꒪ͦ ) 5 | (`へ´≠) 6 | (♡˙︶˙♡) 7 | |Φ|<|′|◇|‵|>|Φ| 8 | (( へ(へ´∀`)へ 9 | ╮(﹋﹏﹌)╭ 10 | (ó㉨ò) 11 | (́^◞౪◟<‵) 12 | ( ´∀`);y=ｰ(ﾟдﾟ)･∵. ﾀｰﾝ 13 | (*˘︶˘*).｡.:*♡ 14 | (-◎◇◎-)o 15 | ﾟÅﾟ) 16 | 爻((╬ಠิ益ಠิ))ｲｶﾘｽﾞﾑｯ!! 17 | ヽ( ^∀^)ﾉ 18 | (́提◞౪◟供‵) 19 | ╮(′ω‵)╭ 20 | -@{}@{}-υ(￣▽￣*) 21 | (((o(*ﾟ▽ﾟ*)o))) 22 | ╮(╯∀╰)╭ 23 | 。･ﾟ･(つд`ﾟ)･ﾟ･ 24 | (^_っ^) 25 | σ >︿<ο 26 | (　◞≼☸≽◟ ._ゝ◞≼☸≽◟) 27 | ( 3ω3) 28 | (｢･ω･)｢ 29 | ( *´◒`*) 30 | (^-^*) 31 | ( • ̀ω•́ ) 32 | (́✺◞౪◟✺‵) 33 | d(`･∀･)b 34 | (*ﾟ∀ﾟ*) 35 | o(‵▽′)o 36 | ⶸ(ᖛ Δ ᖜ)ⶸ 37 | （ ´ﾟ,_」ﾟ）ﾊﾞｶｼﾞｬﾈｰﾉ 38 | (｡･㉨･｡) 39 | ╠╦╩╦╦╩╣◇‧) 40 | ☻ 41 | (´・ω・)つ旦 42 | ╮(▔▽▔")╭ 43 | ┐(´д`)┌ 44 | (ﾟд⊙) 45 | _(´ཀ`」 ∠)_ 46 | /(*￣△￣)\ 47 | (╬☉д⊙) 48 | ヾ(。￣□￣)ツ 49 | Σ(;ﾟдﾟ) 50 | (￣(エ)￣) 51 | ^(＊)^ 52 | (´・ω・`) 53 | (=˙~˙=) 54 | ⊙▃⊙ 55 | ｡ﾟヽ(ﾟ´Д`)ﾉﾟ｡ 56 | ∑(ι´Дン)ノ 57 | ♡(*´∀｀*)人(*´∀｀*)♡ 58 | <(￣︶￣)> 59 | -`д´- 60 | (◞≼◉ื≽◟;益;◞≼◉ื ≽◟) 61 | (Ｔ◇Ｔ) 62 | (ゝ∀･) 63 | (#╯￣□￣)╯~┴─┴ 64 | ε=ε=ε=ε=ε=ε=╭(￣▽￣)╯ 65 | (゜皿。) 66 | (〒︿〒) 67 | 凸(= _=)凸 68 | ლ(⁰⊖⁰ლ) 69 | (-_____-x) 70 | (ᇂдᇂ ) 71 | (*≥▽≤)ツ┏━┓ 72 | (́ತ◞౪◟ತ‵) 73 | ლ｜＾Д＾ლ｜ 74 | (#‵□′)╯ 75 | ︿(′□‵)︿ 76 | ( ˘•ω•˘ ) 77 | (́墨◞౪◟鏡‵) 78 | (శωశ) 79 | ┌|◎o◎|┘ 80 | ・゜・(PД`q｡)・゜・ 81 | (╬ﾟдﾟ) 82 | ╮（╯╴╰∥）╭ 83 | (눈‸눈) 84 | ◑ω◐ 85 | =.= 86 | ☜(:♛ฺ;益;♛ฺ;)☞ 87 | (*‵◇′*) 88 | (<ゝω・) 綺羅星☆ 89 | (°Д°) 90 | ヾ(´︶`*)ﾉ♬ 91 | ┏( .-. ┏ ) ┓ 92 | (´c_`) 93 | ⊂彡☆))д`) 94 | (￣▽￣) 95 | (┛`д´)┛ 96 | q(′□‵)p 97 | ／/( ◕‿‿◕ )＼ 98 | (๑･ิ⋖⋗･ิ๑) 99 | ლ(╹ε╹ლ) 100 | (o`･ω･)ﾉ(ﾉД`) 101 | ᕕ ( ᐛ ) ᕗ 102 | ჰჰჰ❛‿❛ჴჴჴ 103 | ρ(・ω・、) 104 | (｡-_-｡) 105 | (´･_･`) 106 | ლ(́◕◞౪◟◕‵ლ) 107 | = = 108 | (๑•́ ₃ •̀๑) 109 | ＼＼\\٩( 'ω' )و //／／ 110 | ( ´ﾟДﾟ`) 111 | ( ‘д‘⊂彡☆))Д´) 112 | (;_;) 113 | ლ(´•д• ̀ლ 114 | Σ(ﾟдﾟ) 115 | <*)) >>=< 116 | (•ㅂ•)/ 117 | ヾ(*ΦωΦ)ツ 118 | (　ﾟ∀ﾟ)つ≡≡≡♡♡♡)`ν゜)ｸﾞｼｬ 119 | \(′□‵)/ 120 | ε=ε=ヾ(;ﾟдﾟ)/ 121 | _(┐「﹃ﾟ｡)_ 122 | ㄟ(￣▽￣ㄟ)﹏ 123 | ～(╯▽╰)～ 124 | o(☆Ф∇Ф☆)o 125 | (*´ω`)人(´ω`*) 126 | ┳━┳ノ( OωOノ) 127 | ~(～o￣▽￣)～o.....o～(＿△＿o～)~.. 128 | (ﾟ⊿ﾟ) 129 | (〒︵〒) 130 | (:◎)≡ 131 | ｡ﾟ･(ﾉД`)人(´Д`)人(Д` )･ﾟ｡ 132 | ┳━┳ノ( ' - 'ノ) 133 | (#ﾟДﾟ) 134 | ლ(・∀・ )ლ 135 | (́≖◞౪◟≖‵)/ 136 | (´～`) 137 | (¦3[▓▓] 138 | (╯≧▽≦)╯ ~╩═══╩ 139 | (❛◡❛✿) 140 | 乁( ◔ ౪◔)「 141 | (ｰ｀ ) 142 | ( ′-`)y-～ 143 | (´-ω-｀) 144 | (σ´□｀)σ 145 | ᕦ(ò_óˇ)ᕤ 146 | ₍₍◝(･'ω'･)◟⁾⁾ 147 | ◢▆▅▄▃ 溫╰(〞︶〝) ╯馨 ▃▄▅▆◣ 148 | ( ºΔº ) 149 | 흫_흫 150 | (°c＿,°`) 151 | (╬♛. 益♛. ) 152 | (́✸◞♉◟✸‵) 153 | (=´ᴥ`) 154 | ఠ_ఠ 155 | : ♡｡ﾟ.(*♡´◡` 人´◡` ♡*)ﾟ♡ °・ 156 | (╬ಠ益ಠ)ｺﾞﾙァ!! 157 | ━(ﾟ∀ﾟ)━( ﾟ∀)━( ﾟ)━( )━( )━(ﾟ)━(∀ﾟ)━(ﾟ∀ﾟ)━ 158 | (◔⊖◔)つ 159 | (,,ﾟДﾟ) 160 | ⁄(⁄ ⁄•⁄ω⁄•⁄ ⁄)⁄ 161 | ಸ_ಸ 162 | （＾ω＾） 163 | ლ(・´ｪ`・ლ) 164 | (⁰▿⁰) 165 | (ﾒ3[____] 166 | ﹏﹏(ㄏ￣▽￣)ㄏ 167 | ლ(́ಢ.◞౪◟ಢ‵ლ) 168 | ヽ(゜▽゜　)－C<(/;◇;)/~[拖走] 169 | (´◓Д◔`) 170 | (≧∀≦)ゞ 171 | (´д((☆ミPia!⊂▼(ｏ ‵－′ｏ)▼つPia!彡★))∀`) 172 | ˋ(′～‵")ˊ 173 | (ｏ￣∇￣)＝◯)`ν゜)･；' 174 | ✧◝(⁰▿⁰)◜✧ 175 | <(_ _)> 176 | (｡A｡) 177 | (′Ｑ‵)╭～ 178 | (๑•̀ㅂ•́)و✧ 179 | (☞ﾟ∀ﾟ)ﾟ∀ﾟ)☞ 180 | （´≝◞⊖◟≝｀) 181 | (●б人<●) 182 | ((└(:3」┌)┘)) 183 | ( •́ _ •̀)？ 184 | (́◕///◞౪◟///◕‵)/ 185 | (๑¯∀¯๑) 186 | (◞‸◟) 187 | (*￣◇￣)╯ 188 | ヽ(㊤V㊤*)ﾉ 189 | \("￣□￣)/ 190 | (′･_･`) 191 | Σ(°Д°; 192 | <(￣><￣)> 193 | (ㄨ‵◇′)凸 194 | ◥(ฅº￦ºฅ)◤ 195 | (‘⊙д-) 196 | ლ( • ̀ω•́ )っ 197 | (￣╔◇╗￣) 198 | ㄟ(￣▽￣ㄟ)﹏﹏ 199 | (́۞◞౪◟۞‵) 200 | (＠◇＠) 201 | （￣(∞)￣）y-～ 202 | ( ￣ c￣)y▂ξ 203 | (ÒωÓױ) 204 | (@^◇^@) 205 | (」・ω・)」うー！(／・ω・)／にゃー！ 206 | (﹋﹏﹌) 207 | (|||ﾟдﾟ) 208 | (Ò㉨Óױ) 209 | (￣﹏￣) 210 | (▰˘◡˘▰) 211 | (╯ᕡὤᕡ)╯ 212 | \⊙▽⊙/ 213 | ƪ(•̃͡ε•̃͡)∫ 214 | ( ´∀｀)σ 215 | （’へ’） 216 | (╯°Д°)╯ ┻━┻ 217 | (((┏(;￣▽￣)┛ 218 | ԅ(¯﹃¯ԅ) 219 | (ﾟ∀。) 220 | 州‧◇‧州 221 | ヽ(∀ﾟ )人(ﾟ∀ﾟ)人( ﾟ∀)人(∀ﾟ )人(ﾟ∀ﾟ)人( ﾟ∀)ﾉ 222 | ( ﾟДﾟ)σ 223 | ( ﾟДﾟ)ﾉ⌒⑤ 224 | ヽ( ° ▽°)ノ 225 | /(′˙‵)\ 226 | (●⁰౪⁰●) 227 | (❍ᴥ❍ʋ) 228 | ‷̗ↂ凸ↂ‴̖ 229 | ╰(￣▽￣)╯ 230 | ..._〆(°▽°*) 231 | (ﾟ∀ﾟ ) 232 | ╮(〒﹏〒)╭ 233 | ︿(@￣︶￣@)︿ 234 | (΄ಢ◞౪◟ಢ‵)◉◞౪◟◉) 235 | (ﾒ ﾟ皿ﾟ)ﾒ 236 | (*´з｀*) 237 | ε≡(ノ´＿ゝ｀）ノ 238 | (〃∀〃) 239 | ～(￣▽￣)～(＿△＿)～(￣▽￣)～(＿△＿)～(￣▽￣)～ 240 | ಠ_ಠ 241 | (･ิω･ิ) 242 | ヽ(=^･ω･^=)丿 243 | (｡í _ ì｡) 244 | ˋ( ° ▽﹑°|||| ) 245 | (́థ◞..◟థ‵) 246 | ( ´∀｀)つt[ ] 247 | (￣ε(#￣)☆ 248 | (‵□′)--●(#)￣(00)￣) 249 | (　´・◡・｀) 250 | (:3[___]= 251 | (｀_ゝ´) 252 | (｡・ω・｡) 253 | （´◔​∀◔`) 254 | （ლ ^ิ౪^ิ）ლ 255 | ﹏﹏(ㄏ￣▽￣)ㄏ (一(oo)一)＊ ㄟ(￣▽￣ㄟ)﹏ 256 | (•‾⌣‾•) 257 | ....ˊˋ------｡:.ﾟ_ヽ(_´∀`_)ﾉ_.:｡((浮水 258 | ･*･:≡(　ε:) 259 | 魔貫光殺砲(ﾟДﾟ)σ━00000000000━● 260 | ╰(‵□′#) 261 | (=￣◇￣=)Ｖ 262 | ( ´･･)ﾉ(._.`) 263 | (ﾉ∀｀ ) 264 | ヽ(#`Д´)ﾉ 265 | (ﾟДﾟ≡ﾟДﾟ)? 266 | ┌|≧◇≦|┘做└|≧◇≦|┐體┌|≧◇≦|┘操 267 | (#▔︵▔)=====●(￣ ⊿"￣)╭ 268 | (#‵▽′)／~~~~~~~~~(☆___☆) 269 | ＿ﾉ乙(､ﾝ､)＿ 270 | (´-ωก`) 271 | (*^ ◇ ^*) 272 | (　ᐛ)パァ 273 | ლ(･ ิω･ิლ) 274 | (́>◞౪◟<‵)ﾉｼ 275 | (*‘ v`*) 276 | ( ¯﹁¯ ) 277 | (ヾﾉ･ω･`) 278 | (ﾟω´) 279 | (σ回ω・)σ←↑→↓←↑ 280 | ｷﾀ━━━━(ﾟ∀ﾟ)━━━━!! 281 | _(:3 」∠ )_ 282 | (ﾉ◕ヮ◕)ﾉ*:･ﾟ✧ 283 | o(＠//////＠)o 284 | _(¦3」∠)_ 285 | öㅅö 286 | (ﾟﾛ ﾟﾉ)ﾉ 287 | （╯－＿－）╯╧╧ 288 | (一(oo)一)＊ 289 | (●▼●;) 290 | (╬ﾟдﾟ)▄︻┻┳═一 291 | (^u^) 292 | (`・ω・´) 293 | (　′∀`)σ≡σ☆))Д′) 294 | ( ～'ω')～ 295 | :;(∩´﹏`∩);: 296 | (≖ᴗ≖๑) 297 | (ㄏ￣▽￣)ㄏ ㄟ(￣▽￣ㄟ) 298 | <Φ﹏Φ> 299 | (ﾉ｀⊿´)ﾉ 300 | (ノ・＿・)ノ凹 ┣凹━凹━凹┫ 301 | ╮(╯3╰)╭　╮(╯_╰)╭　╮(╯▽╰)╭ 　╮(￣▽￣)╭ 302 | (･ω´･ ) 303 | ヽ( ´ー)ノ 304 | Φ౪Φ 305 | (*ω*) 306 | ╭(╭(╭（ㄏ￣□￣）ㄏ 307 | ฅ(๑*д*๑)ฅ!! 308 | ╮(╯▽╰")╭☆ 309 | ╭(╭(╭ (ㄏ￣△￣)ㄏ 310 | (́⊙◞౪◟⊙‵) 311 | (灬ºωº灬) 312 | -_-|| 313 | Σ(ﾟДﾟ；≡；ﾟдﾟ) 314 | (΄◉◞౪◟◉‵) 315 | ლ(｀∀´ლ) 316 | (○‵△′)=○○○○(￣#)3￣) 317 | (́。◞౪◟。‵) 318 | (○^～^○) 319 | ٩(｡・ω・｡)و 320 | (“￣▽￣)-o█ █o-(￣▽￣”)/ 321 | ┬┴┬┴┤(‧◇├┬┴┬ 322 | d(・ω・d) 微分！(∫・ω・)∫ 積分！∂(・ω・∂) 偏微分！(∮・ω・)∮ 沿閉曲線的積分！(∬・ω・)∬ 重積分！∇(・ω・∇)梯度！∇・(・ω・∇・)散度！∇×(・ω・∇×)旋度！Δ(・ω・Δ)拉普拉斯！ 323 | ╰(‵△′)╯ 324 | (ಠ益ಠ) 325 | (´,_ゝ`) 326 | (･∀･)っ/凵 327 | (((ﾟДﾟ;))) 328 | (╥﹏╥) 329 | (́〠◞౪◟〠‵) 330 | Ψ(≧◇≦)Ψ 331 | ( ≧◇≦ ) 332 | (/ ↀ3ↀ)/q(﹌ω﹌ ) 333 | ლ(╹◡╹ლ) 334 | └(￣^￣ )┐ 335 | (o´・ω・`)σ)Д`) 336 | (；д；) 337 | _(:3 ⌒ﾞ)_ 338 | (￣◇￣)凸 339 | XD 340 | ╮（╯﹏╰）╭ 341 | (^y^) 342 | (╯ ‵Д′)╯╧───╧ 343 | (σ′▽‵)′▽‵)σ 344 | ಠ益ಠ 345 | (´,,•ω•,,)♡ 346 | (′□‵)╭～ 347 | (/Д`)~ﾟ｡ 348 | (o･e･) 349 | ( ╬◣ 益◢)y 350 | ∥∥<∥′ ◇‵ || >∥∥ 351 | ≧▽≦ 352 | ～(￣0￣)～ 353 | (´;ω;`) 354 | (　っค้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้c ) 355 | ξ( ✿＞◡❛)▄︻▇▇〓▄︻┻┳═一 356 | (´⊙ω⊙`) 357 | (≖＿≖)✧ 358 | ||Φ|(|´|Д|`|)|Φ|| 359 | (∩^o^)⊃━☆ﾟ.*･｡ 360 | o(‧/////‧)o 361 | (=´ω`=) 362 | Σ>―(〃°ω°〃)♡→ 363 | (╯‵□′)╯︵┴─┴ 364 | (*´･д･)? 365 | _(°ω°｣ ∠) 366 | (・∀・)つ⑩ 367 | (-ι_- ) 368 | (́▣◞౪◟▣‵) 369 | ┬─┬ ノ( ' - 'ノ) 370 | ٩(ŏ﹏ŏ、)۶ 371 | (#￣▽￣)o╭╯ ☆╰╮o(￣▽￣///)< 372 | ( శ 3ੜ)～♥ 373 | (づ′▽`)づ 374 | ´-ω-)b 375 | o(*/////*)o 376 | \(≧ω≦)/ 377 | (☉д⊙) 378 | (′ω‵)y-～ 379 | (ΦωΦ) 380 | (́☜◞౪◟☞`)/ 381 | (〒△〒) 382 | ξ( ✿＞◡❛) 383 | （´◔ ₃ ◔`) 384 | (o ─_─)==O)。O。) 385 | (Π△Π) 386 | (=m=) 387 | o(⊙/////⊙)o 388 | (´_ゝ`) 389 | (°ཀ°) 390 | ๛ก(ｰ̀ωｰ́ก) 391 | (￫ܫ￩) 392 | ( ╯' - ')╯ ┻━┻ 393 | (´◉㉨◉) 394 | (=ﾟДﾟ=) ▄︻┻┳━ ·.`.`.`. 395 | (っ・Д・)っ 396 | ｡:.ﾟヽ(*´∀`)ﾉﾟ.:｡ 397 | m(｡≧ｴ≦｡)m 398 | ( ﾒ∀・) 399 | ( ・・)つ―{}@{}@{}- 400 | (◕ܫ◕) 401 | ( ͡° ͜ʖ ͡°) 402 | (〒◇〒)o～ 403 | ヽ(́◕◞౪◟◕‵)ﾉ 404 | (◜◔。◔◝) 405 | ( ¤̴̶̷̤́ ‧̫̮ ¤̴̶̷̤̀ ) 406 | (́◉◞౪◉*◉◉◞◕౪౪◟◕◉◉φ◉౪◞◉*‵) 407 | (*´д`)~♥ 408 | Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰_Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰ 409 | ヽ(｀д´)ﾉ 410 | ლ(╯⊙ε⊙ლ╰) 411 | (੭ु´ ᐜ `)੭ु⁾⁾ 412 | (๑´ڡ`๑) 413 | ฅ●ω●ฅ 414 | (￣ー￣) 415 | ( ￣□￣)/ 敬禮!! <(￣ㄧ￣ ) <(￣ㄧ￣ ) 416 | (`┌_┐´) 417 | ↖╭╰('(Ω)')╯ 418 | (‾◡◝　) 419 | (((m -_-)m 420 | (ㄨ‵◇′) 421 | ( ◕‿‿◕ ) 422 | (*☉౪⊙*) 423 | （ ´☣///_ゝ///☣｀） 424 | Σ┗(＠ロ＠;)┛ 425 | ○(#‵︿′ㄨ)○ 426 | (◞≼●≽◟◞౪◟◞≼●≽◟) 427 | ( *¯ ³¯*)♡ㄘゅ 428 | ╰(￣︶￣)╯ 429 | (*´艸`*) 430 | 凸(▔︵▔#) 431 | ヽ(・×・´)ゞ 432 | Σ(lliдﾟﾉ)ﾉ 433 | (つд⊂) 434 | (//´/◒/`//) 435 | ( ^ิ౪^ิ) 436 | \(Π◇Π) 437 | >///< 438 | ( ´(00)`) 439 | (✧≖‿ゝ≖) 440 | (☍﹏⁰) 441 | (・∀・)ノ三G[__]ｺ 442 | (╬ಠิ益ಠิ) 443 | ⊂(°Д°⊂⌒｀つ≡≡≡(´⌒;;;≡≡≡ 444 | ლ（இ e இ｀。ლ） 445 | (〒﹏〒) 446 | ( ﾟ∀ﾟ)o彡ﾟ 447 | |Д`)ノ⌒●～* 448 | (´>∀)人(´・ω・)ﾉヽ(・ε・*)人(-д-`) 449 | Σ(ﾟωﾟ) 450 | ヾ(●゜▽゜●)♡ 451 | \(￣◇￣)/ 452 | (o´罒`o) 453 | ╮(╯～╰" )╭ 454 | (σﾟ∀ﾟ)σﾟ∀ﾟ)σﾟ∀ﾟ)σ 455 | （´-`）.｡oO 456 | (ﾉ∀`*) 457 | (๑≖ิิ益≖ิ๑) 458 | ☝( ◠‿◠ )☝ 459 | Σ( ° △ °) 460 | (Π___Π) 461 | ＼(●´ϖ`●)／ 462 | ε(*´･∀･｀)зﾞ 463 | (⁰⊖⁰) 464 | (#ﾟ⊿`)凸 465 | (๑ơ ₃ ơ)♥ 466 | (́=◞౪◟=‵) 467 | ╰(‵□′)╯ 468 | (-`ェ´-╬) 469 | どこ━━━━(゜∀゜三゜∀゜)━━━━!!?? 470 | (;´༎ຶД༎ຶ`) 471 | (｡>﹏<)哈啾 472 | ლ(＾ω＾ლ) 473 | 川‧◇‧川 474 | (╯◕◞౪◟◕‵)╯︵ ┴─┴ 475 | ( ﾟ Χ ﾟ) 476 | (っ●ω●)っ 477 | (╬ﾟ◥益◤ﾟ) 478 | ٩(๑•̀ω•́๑)۶ 479 | ↁ_ↁ 480 | il||li _|￣|○ヽ(･ω･｀) 481 | (´Ａ｀。) 482 | =(─(oo)─)= 483 | (♛‿♛) 484 | ʅ（´◔౪◔）ʃ 485 | m9(＾Д＾)ﾌﾟｷﾞｬｰ 486 | (ÒܫÓױ) 487 | ╮(′～‵〞)╭ 488 | (☄◣ω◢)☄ 489 | ヽ(`Д´)ノ 490 | (╬ﾟ ◣ ﾟ) 491 | (σﾟ∀ﾟ)σ..:*☆ 492 | (◔౪◔) 493 | (#`皿´) 494 | (｡◕∀◕｡) 495 | ヾ(◎´・ω・｀)ノ 496 | ತಎತ 497 | (๑´ㅁ`) 498 | (´・Å・`) 499 | (((ﾟдﾟ))) 500 | ( ´థ౪థ）σ 501 | ─=≡Σ((( つ•̀ω•́)つ 502 | (´;ω;`)ヾ(･∀･`) 503 | （つ> _◕）つ︻╦̵̵͇̿̿̿̿╤─── 504 | (´///☁///`) 505 | ε٩(๑> ₃ <)۶з 506 | Σ(*ﾟдﾟﾉ)ﾉ 507 | (ﾟдﾟ≡ﾟдﾟ) 508 | (́இ◞౪◟இ‵) 509 | ⇎_⇎ 510 | (ﾟ3ﾟ)～♪ 511 | |||||╭(＿◇＿)╮||||| 512 | (✘﹏✘ა) 513 | (´=ิ益=ิ｀) 514 | ლ(•̀ _ •́ ლ) 515 | （*^﹏^*） 516 | (❁´ω`❁)*✲ﾟ* 517 | (っ´ω`c) 518 | ☆^(ｏ´Ф∇Ф)o 519 | ﾚ(ﾟ∀ﾟ;)ﾍ=З=З=З 520 | (*＾ー＾) 521 | (⊜‿⊜✴） 522 | (╯^◇^)╯ 523 | ლ(◉◞౪◟◉ )ლ 524 | இдஇ 525 | ( *‵Д′)--●(#)‵⊿′)╮ 526 | ＼(￣　　　◇,　　　￣)／ 527 | (ㆆᴗㆆ) 528 | (*ﾟーﾟ) 529 | ( ｀皿´)ﾉ"鹵=巛巛巛巛.:;: 530 | (〃￣ω￣)人(￣︶￣〃) 531 | σ`∀´)σ 532 | -{}@{}@{}-υ(￣◇￣*) 533 | (・`ω´・) 534 | (//´◜◞⊜◟◝｀////) 535 | (≧∀≦) 536 | (;´д｀).｡ｏO(・・・・) 537 | ∑(￣□￣;) 538 | (ง๑ •̀_•́)ง 539 | (；´ﾟωﾟ｀人) 540 | ╮/(＞▽<)人(>▽＜)╭ 541 | ╩══╩ ～ ╰(￣△￣╰) 542 | (╭￣3￣)╭♡ 543 | ╮(﹀ω﹀")╭ 544 | ( ՞ټ՞) 545 | _(┐「ε:)_ 546 | ((((<||＿◇＿> 547 | ￣﹃￣ 548 | #ﾟÅﾟ）⊂彡☆))ﾟДﾟ)･∵ 549 | （ ▔︿▔）╭～ 550 | ( ¯•ω•¯ ) 551 | |||╭(＿◇＿)╮||| 552 | ( ﾟÅﾟ) 553 | Σ（ﾟдﾟlll） 554 | (´ﾟдﾟ`) 555 | Σ(・ω・ノ)ノ 556 | ლ(•ω •ლ) 557 | ^(˙oo˙)^ 558 | ヽ(✿ﾟ▽ﾟ)ノ 559 | (´∩ω∩｀) 560 | (*^＠^*) 561 | (・ε・) 562 | ლↂ‿‿ↂლ 563 | (´◉‿◉｀) 564 | ◔"L__◔ 565 | ◝(　ﾟ∀ ﾟ )◟ 566 | q(〒□〒)p 567 | ◢▆▅▄▃崩╰(〒皿〒)╯潰▃▄▅▇◣ 568 | ◢▆▅▄▃ 崩╰(〒皿〒)╯潰 ▃▄▅▆◣ 569 | ε≡ﾍ( ´∀`)ﾉ 570 | (;̦̦̦̦̦̦̦̦ↂ⃝⃓⃙⃚⃘дↂ⃝⃓⃙⃚⃘;̦̦̦̦̦​̦̦̦) 571 | (´σ-`) 572 | (ﾟ皿ﾟﾒ) 573 | (^ρ^)/ 574 | ｡･ﾟ･(ﾉД`)ヽ(ﾟДﾟ )秀秀 575 | ( ఠൠఠ )ﾉ 576 | _:(´□`」 ∠):_ 577 | <(‵□′)> 578 | ლ（´∀`ლ） 579 | (✺ω✺) 580 | (๑•ૅω•´๑) 581 | ┌(┌^o^)┐ 582 | (́순◞౪◟순‵) 583 | (#￣▽￣)o╭╯ ★ 584 | Σ(｀L_｀ ) 585 | ^(￣(oo)￣)^ 586 | ლ(^o^ლ) 587 | (／‵Д′)／~ ╧╧ 588 | (*´Д`)つ))´∀`) 589 | ( ☉_☉)≡☞o────★° 590 | ლ(́◉◞౪◟◉‵ლ) 591 | ┏┛學校┗┓λλλλλ 592 | (*´д`) 593 | (V)(́◉◞౪◟◉‵)(V) 594 | Σ Σ Σ (」○ ω○ )／ 595 | ヾ(´ε`ヾ) 596 | 0(:3　)～ ('､3_ヽ)_ 597 | =└(┐卍^o^)卍 598 | (;ﾟдﾟ) 599 | (*ˇωˇ*人) 600 | ( ♥д♥) 601 | ☮▁▂▃▄☾ ♛ ◡ ♛ ☽▄▃▂▁☮ 602 | /(′□‵)\ 603 | ( º﹃º ) 604 | ( ´థ,_‥థ｀)b 605 | ( Φ ω Φ ) 606 | థ౪థ 607 | (*-*)y~ 608 | (ﾟ∀ﾟ) 609 | |柱|ω・`) 610 | (́つ◞౪◟⊂‵) 611 | ( ´-ω ･)▄︻┻┳══━ 612 | ＼(●o○;)ノ 613 | >(˙ω˙)< 614 | d(d＇∀＇) 615 | (´^ิ益^ิ｀ ) 616 | ( ´･ω) 617 | =_= 618 | \(￣□￣)/ 619 | (∂ω∂) 620 | (థฺˇ౪ˇథ) 621 | (✪ω✪) 622 | (σ ─.─)σ 623 | └|‧◇‧|┐ 624 | |////|　( 　)ﾉ　|////|(自動門 625 | 〠_〠 626 | (`皿´) 627 | 。゜+.(人-ω◕)゜+.゜ 628 | ╮(￣◇￣)╭ 629 | _(√ ζ ε:)_ 630 | (´ー｀)y─┛ 631 | ༼ つ/̵͇̿̿/’̿’̿ ̿ ̿̿ ̿̿◕ _◕ ༽つ/̵͇̿̿/’̿’̿ ̿ ̿̿ ̿̿ ̿̿ 632 | (`д´) 633 | (ノДＴ) 634 | (¯ □ ¯) 635 | /(╯▽╰")\ 636 | (ﾉД`) 637 | (　◜◡‾) 638 | o(─//////─)o 639 | (:3っ)へ ヽ(´Д｀●ヽ) 640 | (#‵)3′)▂▂▂▃▄▅～～～嗡嗡嗡嗡嗡 641 | ☆⌒(*^-゜)v 642 | ╮(￣(oo)￣)╭ 643 | (´≖◞౪◟≖) 644 | (´∀｀∩)↑↑ 645 | ( ˘•ω•˘ )◞⁽˙³˙⁾ 646 | ( ×ω× ) 647 | (#`Д´)ﾉ 648 | ლ(❛◡❛✿)ლ 649 | -//(ǒ.ǒ)//- 650 | (́◉◞౪◟◉‵) 651 | ━( ิ◕㉨◕ ิ)━ 652 | ╮(￣□￣)╭ 653 | (́一◞౪◟一‵) 654 | ヾ(；ﾟ(OO)ﾟ)ﾉ 655 | q(￣ε￣ )p 656 | (́◕◞ㆀ◟◕`) 657 | ╮(╯▽╰)╭ 658 | ⊙谷⊙ 659 | (๑థิ益థิ) 660 | <(‵△′)> 661 | ლ(ﾟдﾟლ) 662 | (╬⓪益⓪) 663 | ♥(´∀` )人 664 | ∪(￣﹁￣)∪ 665 | (･ิ㉨･ิ) 666 | ( ￣ 3￣)y▂ξ 667 | ┏((＝￣(ｴ)￣=))┛ 668 | ≧〔゜゜〕≦ 669 | =͟͟͞͞( •̀д•́) 670 | (，_ゝ۞) 671 | (つД`) 672 | \(￣□￣*)/ 673 | (*─︵─)====Co( 一//////一)o 674 | \(⊙O⊙)/ 675 | σ(‥) 676 | ｡ﾟ(ﾟ´ω`ﾟ)ﾟ｡ 677 | ( ´∀`) 678 | (ゝ∀･)⌒☆ 679 | (`3´) 680 | (　˙灬˙　) 681 | (oﾟωﾟo) 682 | (＊゜ー゜)b 683 | ╰(〒○〒)╯ 684 | (っ﹏-) .｡o 685 | ✧(순_순)" 686 | ╯-____-)╯~═╩════╩═ 687 | (●｀･(ｴ)･´●) 688 | (。◇。||) 689 | ╮(╯_╰)╭ 690 | Σヽ(ﾟД ﾟ; )ﾉ 691 | ( ΄◞ิ .̫.̫ ◟ิ‵)y 692 | (#▔皿▔)─<(*—_____–) 693 | (　っธิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิ์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์ლ) 694 | (｡´∀｀)ﾉ―⊂ZZZ⊃ 695 | ( ิΘ㉨Θ)ิ 696 | (´･д･｀)ﾊ(･д･｀*) 697 | (╯°O°)╯┻━┻ 698 | σ ﾟ∀ ﾟ) ﾟ∀ﾟ)σ 699 | (⌒o⌒) 700 | (´-ι_-｀) 701 | ( ・O・)ノ⌒● 702 | *ଘ(੭*ˊᵕˋ)੭* ੈ✩‧₊˚ 703 | (*´∀`)~♥ 704 | (,,・ω・,,) 705 | (゜o゜(☆○=(-_- ) 706 | ( ^ω^) 707 | (〒__〒) 708 | ┏(_д_┏)┓)) 709 | (◓Д◒)✄╰⋃╯ 710 | ヾ(;ﾟ;Д;ﾟ;)ﾉﾞ 711 | (д) ﾟﾟ 712 | | 柱 |┐(￣◇￣*) 713 | o(.〝.)o 714 | (ㄒ◇ㄒ)o～ 715 | ก็ʕ•͡ᴥ•ʔ ก้ 716 | <(‵︵′)> 717 | (╯#▔__▔)╯┴┴\("￣□￣) 718 | ﹏﹏(ㄏ￣△￣)ㄏ 719 | ( ˘･з･) 720 | (╯°▽°)╯ ┻━┻ 721 | \("￣▽￣)/ 722 | (•ө•) 723 | ~(￣▽￣)~ 724 | ( ิ◕㉨◕ ิ) 725 | ┴┴~╰(‵□′#) 726 | (:3[__]4 727 | (›´ω`‹ ) 728 | ( ´Д`)y━･~~ 729 | ◑▂◐ 730 | (°ﾛ°٥) 731 | （・∀・） 732 | (｡ŏ_ŏ) 733 | (ﾉ>ω<)ﾉ 734 | (　ﾟ∀ﾟ) ﾉ♡ 735 | ヽ(●´ε｀●)ノ 736 | •✞_✞• 737 | (ゝ∀･)b 738 | (๑• . •๑) 739 | ◉ื益◉ื 740 | (╯ ‵□′)╯ 741 | (´;;◉;⊖;◉;;｀) 742 | (｡•ㅅ•｡)♡ 743 | (ㅅ˘ㅂ˘) 744 | ٩(๑´3｀๑)۶ 745 | (´ΘωΘ`) 746 | (≖‿ゝ≖)✧ 747 | (￣y▽￣)╭ 748 | ,,Ծ‸Ծ,, 749 | σ(´∀｀*) 750 | (´▽`ʃ♡ƪ) 751 | O-(///￣皿￣)☞ ─═≡☆゜★█▇▆▅▄▃▂＿ 752 | ʕ•̫͡•ʔ 753 | o(‧""‧)o 754 | (｀∀´) 755 | ⁽⁽ ◟(∗ ˊωˋ ∗)◞ ⁾⁾ 756 | (◍•ᴗ•◍)ゝ 757 | (*’ｰ’*) 758 | ( σ՞ਊ ՞)σ 759 | ( ˘•ω•˘ ).oOஇ 760 | (￣□￣||) 761 | (●｀ 艸 ´) 762 | (ಥ_ಥ) 763 | ⋉(● ∸ ●)⋊ 764 | ✧*｡٩(ˊᗜˋ*)و✧*｡ 765 | ( ´•̥̥̥ω•̥̥̥` ) 766 | \(^ω^)/ 767 | 癶(癶;:゜;益;゜;)癶 768 | (●´ω｀●)ゞ 769 | (ｏﾟﾛﾟ)┌┛Σ(ﾉ´*ω*`)ﾉ 770 | \(╯﹏╰) 771 | |柱|ωﾟ)---c<` дﾟ)! 772 | (๑´ㅂ`๑) 773 | （╯‵□′）╯︵┴─┴ 774 | ╮(▔ε▔)╭ 775 | 凸(￣皿￣*) 776 | (ﾟ々｡) 777 | ヾ(>Д<;))))..... 778 | ヾ(:3ﾉｼヾ)ﾉｼ 三[____] 779 | (#￣▽￣#) 780 | <(￣︶￣)/ 781 | <(￣3￣)> 782 | づ(・ω・)づ 783 | い・ω・づ＊ -------------------------------------------------------------------------------- /jseg/Emodet/emodata.json: -------------------------------------------------------------------------------- 1 | ["(\uff92\uff9f\u0414\uff9f)\uff92", "Zz(\u00b4-\u03c9-`*)", "(\ua4aa\u0366\u1d17\u0335\u030d\ua4aa\u0366 )", "( *\u00b4\u25d2`*)", "(\u2661\u02d9\ufe36\u02d9\u2661)", "(( \u3078(\u3078\u00b4\u2200`)\u3078", "(\u00f3\u3268\u00f2)", "( \u00b4\u2200`);y=\uff70(\uff9f\u0434\uff9f)\uff65\u2235. \uff80\uff70\uff9d", "\uff9f\u00c5\uff9f)", "(\u0301\u63d0\u25de\u0c6a\u25df\u4f9b\u2035)", "0", "(((o(*\uff9f\u25bd\uff9f*)o)))", "\u256e(\u256f\u2200\u2570)\u256d", "\u3002\uff65\uff9f\uff65(\u3064\u0434`\uff9f)\uff65\uff9f\uff65", "(^_\u3063^)", "(\u3000\u25de\u227c\u2638\u227d\u25df ._\u309d\u25de\u227c\u2638\u227d\u25df)", "( 3\u03c93)", "(\uff62\uff65\u03c9\uff65)\uff62", "(`\u3078\u00b4\u2260)", "( \u2022 \u0300\u03c9\u2022\u0301 )", "d(`\uff65\u2200\uff65)b", "(*\uff9f\u2200\uff9f*)", "\uff08 \u00b4\uff9f,_\u300d\uff9f\uff09\uff8a\uff9e\uff76\uff7c\uff9e\uff6c\uff88\uff70\uff89", "(\uff61\uff65\u3268\uff65\uff61)", "(\u00b4\u30fb\u03c9\u30fb)\u3064\u65e6", "\u2510(\u00b4\u0434`)\u250c", "(\uff9f\u0434\u2299)", "_(\u00b4\u0f40`\u300d \u2220)_ ", "(\u256c\u2609\u0434\u2299)", "\u03a3(;\uff9f\u0434\uff9f)", "(\uffe3(\u30a8)\uffe3)", "(\u00b4\u30fb\u03c9\u30fb`)", "(\u309d\u2200\uff65)\u2312\u2606", "\uff61\uff9f\u30fd(\uff9f\u00b4\u0414`)\uff89\uff9f\uff61", "\u2211(\u03b9\u00b4\u0414\u30f3)\u30ce", "\u2661(*\u00b4\u2200\uff40*)\u4eba(*\u00b4\u2200\uff40*)\u2661", "-`\u0434\u00b4-", "(\u309d\u2200\uff65)", "\uff08 \u00b4\u2623///_\u309d///\u2623\uff40\uff09", "(\u3012\ufe3f\u3012)", "(*\u2265\u25bd\u2264)\u30c4\u250f\u2501\u2513", "\u10da\uff5c\uff3e\u0414\uff3e\u10da\uff5c", "( \u02d8\u2022\u03c9\u2022\u02d8 )", "(\u0c36\u03c9\u0c36)", "( \u055e\u067c\u055e)", "\u30fb\u309c\u30fb(P\u0414`q\uff61)\u30fb\u309c\u30fb", " (//\u00b4/\u25d2/`//)", " (\u260d\ufe4f\u2070) ", "\u25d1\u03c9\u25d0", "(<\u309d\u03c9\u30fb) \u7dba\u7f85\u661f\u2606", " \u30fe(\u00b4\ufe36`*)\uff89\u266c", "\u250f( .-. \u250f ) \u2513", "(\u00b4c_`)", "\u2282\u5f61\u2606))\u0434`)", "(\u251b`\u0434\u00b4)\u251b", "\uff0f/( \u25d5\u203f\u203f\u25d5 )\uff3c", "\u1555 ( \u141b ) \u1557", "\u03c1(\u30fb\u03c9\u30fb\u3001)", "(\uff61-_-\uff61) ", "\u25dd(\u3000\uff9f\u2200 \uff9f )\u25df", "\u10da(\u0301\u25d5\u25de\u0c6a\u25df\u25d5\u2035\u10da)", "(\u0e51\u2022\u0301 \u2083 \u2022\u0300\u0e51)", "\uff3c\uff3c\\\\\u0669( '\u03c9' )\u0648 //\uff0f\uff0f", " ( \u00b4\uff9f\u0414\uff9f`)", "( \u2018\u0434\u2018\u2282\u5f61\u2606))\u0414\u00b4)", "\u10da(\u00b4\u2022\u0434\u2022 \u0300\u10da", "(\u2022\u04e9\u2022)", "\u03a3(\uff9f\u0434\uff9f)", "<*)) >>=<", "(\u2022\u3142\u2022)/", "\u30fe(*\u03a6\u03c9\u03a6)\u30c4", "(\u3000\uff9f\u2200\uff9f)\u3064\u2261\u2261\u2261\u2661\u2661\u2661)`\u03bd\u309c)\uff78\uff9e\uff7c\uff6c", "\u03b5=\u03b5=\u30fe(;\uff9f\u0434\uff9f)/", "_(\u2510\u300c\ufe43\uff9f\uff61)_", "o(\u2606\u0424\u2207\u0424\u2606)o", "(*\u00b4\u03c9`)\u4eba(\u00b4\u03c9`*)", "\u2533\u2501\u2533\u30ce( O\u03c9O\u30ce)", "~(\uff5eo\uffe3\u25bd\uffe3)\uff5eo.....o\uff5e(\uff3f\u25b3\uff3fo\uff5e)~..", "(\uff9f\u22bf\uff9f)", "(:\u25ce)\u2261", "\u2533\u2501\u2533\u30ce( ' - '\u30ce) ", "(\u00b4\uff5e`)", "(\u00a63[\u2593\u2593]\t", "(\u275b\u25e1\u275b\u273f)", "(\u00b4-\u03c9-\uff40)", "\u1566(\u00f2_\u00f3\u02c7)\u1564", "\u208d\u208d\u25dd(\uff65'\u03c9'\uff65)\u25df\u207e\u207e ", "( \u00ba\u0394\u00ba )", "(=\u00b4\u1d25`)", ": \u2661\uff61\uff9f.(*\u2661\u00b4\u25e1` \u4eba\u00b4\u25e1` \u2661*)\uff9f\u2661 \u00b0\u30fb", "(*\u00b4\u8278`*)", "(\u25d4\u2296\u25d4)\u3064", "(,,\uff9f\u0414\uff9f)", "\u2044(\u2044 \u2044\u2022\u2044\u03c9\u2044\u2022\u2044 \u2044)\u2044", "( \u0e34\u25d5\u3268\u25d5 \u0e34)", "\u10da(\u30fb\u00b4\uff6a`\u30fb\u10da)", "(\u2070\u25bf\u2070)", "(\uff923[____]", "\u30fd(\u309c\u25bd\u309c\u3000)\uff0dC<(/;\u25c7;)/~[\u62d6\u8d70] ", "(\uffeb\u072b\uffe9)", "(\u2267\u2200\u2266)\u309e", "(\u00b4\u0434((\u2606\u30dfPia!\u2282\u25bc(\uff4f \u2035\uff0d\u2032\uff4f)\u25bc\u3064Pia!\u5f61\u2605))\u2200`)", "(\u25cf\uff40\uff65(\uff74)\uff65\u00b4\u25cf)", "\u2727\u25dd(\u2070\u25bf\u2070)\u25dc\u2727", "<(_ _)>", "(\uff61A\uff61)", "(\u0e51\u2022\u0300\u3142\u2022\u0301)\u0648\u2727", "(\u261e\uff9f\u2200\uff9f)\uff9f\u2200\uff9f)\u261e", "(\u25cf\u0431\u4eba<\u25cf)", "((\u2514(:3\u300d\u250c)\u2518))", " ( \u2022\u0301 _ \u2022\u0300)\uff1f", "(\u0e51\u00af\u2200\u00af\u0e51)", "(\u25de\u2038\u25df)", "\u30fd(\u32a4V\u32a4*)\uff89", "\u03a3(\u00b0\u0414\u00b0;", "\u25e5(\u0e05\u00ba\uffe6\u00ba\u0e05)\u25e4", "(\u2018\u2299\u0434-)", " \u10da( \u2022 \u0300\u03c9\u2022\u0301 )\u3063", "(\u300d\u30fb\u03c9\u30fb)\u300d\u3046\u30fc\uff01(\uff0f\u30fb\u03c9\u30fb)\uff0f\u306b\u3083\u30fc\uff01", "(|||\uff9f\u0434\uff9f)", "(\u25b0\u02d8\u25e1\u02d8\u25b0)", "\u01aa(\u2022\u0303\u0361\u03b5\u2022\u0303\u0361)\u222b", "\uff08\u2019\u3078\u2019\uff09", "(\u256f\u00b0\u0414\u00b0)\u256f \u253b\u2501\u253b", "\uff9a(\uff9f\u2200\uff9f;)\uff8d=\u0417=\u0417=\u0417", "(\uff9f\u2200\u3002)", "\u30fd(\u2200\uff9f )\u4eba(\uff9f\u2200\uff9f)\u4eba( \uff9f\u2200)\u4eba(\u2200\uff9f )\u4eba(\uff9f\u2200\uff9f)\u4eba( \uff9f\u2200)\uff89", "(\u00b4\uff65\u0434\uff65\uff40)\uff8a(\uff65\u0434\uff65\uff40*)", "\u30fd( \u00b0 \u25bd\u00b0)\u30ce", "(\u25cf\u2070\u0c6a\u2070\u25cf)", "(\u274d\u1d25\u274d\u028b)", "..._\u3006(\u00b0\u25bd\u00b0*)", "(\u0384\u0ca2\u25de\u0c6a\u25df\u0ca2\u2035)\u25c9\u25de\u0c6a\u25df\u25c9)", "(\uff92 \uff9f\u76bf\uff9f)\uff92", "(*\u00b4\u0437\uff40*)", "\u03b5\u2261(\u30ce\u00b4\uff3f\u309d\uff40\uff09\u30ce", "(\u3003\u2200\u3003)", "\u0ca0_\u0ca0", "\u30fd(=^\uff65\u03c9\uff65^=)\u4e3f", "(\uff61\u00ed _ \u00ec\uff61)", "( \u00b4\u2200\uff40)\u3064t[ ]", "(\u2727\u2256\u203f\u309d\u2256)", " (\u3000\u00b4\u30fb\u25e1\u30fb\uff40)", "(:3[___]=", "(\uff61\u30fb\u03c9\u30fb\uff61)", "\uff08\u00b4\u25d4\u200b\u2200\u25d4`)", " (\u2022\u203e\u2323\u203e\u2022)", "....\u02ca\u02cb------\uff61:.\uff9f_\u30fd(_\u00b4\u2200`_)\uff89_.:\uff61((\u6d6e\u6c34", "\uff65*\uff65:\u2261(\u3000\u03b5:)", "\u9b54\u8cab\u5149\u6bba\u7832(\uff9f\u0414\uff9f)\u03c3\u250100000000000\u2501\u25cf", "( \u00b4\uff65\uff65)\uff89(._.`)", "\u30fd(#`\u0414\u00b4)\uff89", "\uff3f\uff89\u4e59(\uff64\uff9d\uff64)\uff3f", "(\u00b4-\u03c9\u0e01`)", "(\u3000\u141b)\u30d1\u30a1", "(*\u2018 v`*)", "(\u30fe\uff89\uff65\u03c9\uff65`)", "(\uff9f\u03c9\u00b4)", "(\u03c3\u56de\u03c9\u30fb)\u03c3\u2190\u2191\u2192\u2193\u2190\u2191", "(\u00b4\u25d3\u0414\u25d4`)", "\uff77\uff80\u2501\u2501\u2501\u2501(\uff9f\u2200\uff9f)\u2501\u2501\u2501\u2501!!", "(\u00b4>\u2200)\u4eba(\u00b4\u30fb\u03c9\u30fb)\uff89\u30fd(\u30fb\u03b5\u30fb*)\u4eba(-\u0434-`)", "(\uff89\u25d5\u30ee\u25d5)\uff89*:\uff65\uff9f\u2727", "_(\u00a63\u300d\u2220)_", "\u00f6\u3145\u00f6", "(\u25cf\u25bc\u25cf;)", "(\u256c\uff9f\u0434\uff9f)\u2584\ufe3b\u253b\u2533\u2550\u4e00", "(^u^)", "(`\u30fb\u03c9\u30fb\u00b4)", "(\u3000\u2032\u2200`)\u03c3\u2261\u03c3\u2606))\u0414\u2032)", "( \uff5e'\u03c9')\uff5e", ":;(\u2229\u00b4\ufe4f`\u2229);:", "(\u2256\u1d17\u2256\u0e51)", "(\u310f\uffe3\u25bd\uffe3)\u310f \u311f(\uffe3\u25bd\uffe3\u311f)", "(\u30ce\u30fb\uff3f\u30fb)\u30ce\u51f9 \u2523\u51f9\u2501\u51f9\u2501\u51f9\u252b", "(\uff65\u03c9\u00b4\uff65 )", "\u03a6\u0c6a\u03a6", "\u0e05(\u0e51*\u0434*\u0e51)\u0e05!!", "\u03a3(\uff9f\u0414\uff9f\uff1b\u2261\uff1b\uff9f\u0434\uff9f)", "\u0669(\uff61\u30fb\u03c9\u30fb\uff61)\ufeff\u0648", "(\u201c\uffe3\u25bd\uffe3)-o\u2588 \u2588o-(\uffe3\u25bd\uffe3\u201d)/", "d(\u30fb\u03c9\u30fbd) \u5fae\u5206\uff01(\u222b\u30fb\u03c9\u30fb)\u222b \u7a4d\u5206\uff01\u2202(\u30fb\u03c9\u30fb\u2202) \u504f\u5fae\u5206\uff01(\u222e\u30fb\u03c9\u30fb)\u222e \u6cbf\u9589\u66f2\u7dda\u7684\u7a4d\u5206\uff01(\u222c\u30fb\u03c9\u30fb)\u222c \u91cd\u7a4d\u5206\uff01\u2207(\u30fb\u03c9\u30fb\u2207)\u68af\u5ea6\uff01\u2207\u30fb(\u30fb\u03c9\u30fb\u2207\u30fb)\u6563\u5ea6\uff01\u2207\u00d7(\u30fb\u03c9\u30fb\u2207\u00d7)\u65cb\u5ea6\uff01\u0394(\u30fb\u03c9\u30fb\u0394)\u62c9\u666e\u62c9\u65af\uff01", "(\u0ca0\u76ca\u0ca0)", "(\u00b4,_\u309d`)", "(((\uff9f\u0414\uff9f;)))", "(\u2565\ufe4f\u2565)", "(/ \u21803\u2180)/q(\ufe4c\u03c9\ufe4c )", "\u10da(\u2579\u25e1\u2579\u10da)", "(o\u00b4\u30fb\u03c9\u30fb`)\u03c3)\u0414`)", "_(:3 \u2312\uff9e)_", "(^y^)", "(\u03c3\u2032\u25bd\u2035)\u2032\u25bd\u2035)\u03c3", "(\u00b4,,\u2022\u03c9\u2022,,)\u2661", "\u261d( \u25e0\u203f\u25e0 )\u261d", "(\u00b4;\u03c9;`)", "(\u3000\u3063\u0e04\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49\u0e49c )", "\u03be( \u273f\uff1e\u25e1\u275b)\u2584\ufe3b\u2587\u2587\u3013\u2584\ufe3b\u253b\u2533\u2550\u4e00", "(\u00b4\u2299\u03c9\u2299`)", "(\u2256\uff3f\u2256)\u2727", "||\u03a6|(|\u00b4|\u0414|`|)|\u03a6||", "O-(///\uffe3\u76bf\uffe3)\u261e \u2500\u2550\u2261\u2606\u309c\u2605\u2588\u2587\u2586\u2585\u2584\u2583\u2582\uff3f\u3000", "( \u00b4\u0c25\u0c6a\u0c25\uff09\u03c3", "(\u25d3\u0414\u25d2)\u2704\u2570\u22c3\u256f", "(\u256f\u2035\u25a1\u2032)\u256f\ufe35\u2534\u2500\u2534", "(*\u00b4\uff65\u0434\uff65)?", "_(\u00b0\u03c9\u00b0\uff63 \u2220)", "(\u30fb\u2200\u30fb)\u3064\u2469", "\u252c\u2500\u252c \u30ce( ' - '\u30ce) ", "\u0669(\u014f\ufe4f\u014f\u3001)\u06f6", "(\u00b4\uff9f\u0434\uff9f`)", "\u25e2\u2586\u2585\u2584\u2583\u5d29\u2570(\u3012\u76bf\u3012)\u256f\u6f70\u2583\u2584\u2585\u2587\u25e3", "\u00b4-\u03c9-)b", "(\u2609\u0434\u2299)", "(\u309c\u76bf\u3002)", "(\u03a6\u03c9\u03a6)", "\u03b5(*\u00b4\uff65\u2200\uff65\uff40)\u0437\uff9e", "\uff08\u00b4\u25d4 \u2083 \u25d4`)", "(=m=)", "(\u00b4_\u309d`)", "(\u00b0\u0f40\u00b0)", "\u0e5b\u0e01(\uff70\u0300\u03c9\uff70\u0301\u0e01)", "\u2727*\uff61\u0669(\u02ca\u15dc\u02cb*)\u0648\u2727*\uff61", "( \u256f' - ')\u256f \u253b\u2501\u253b ", "\u0e01\u0e47\u0295\u2022\u0361\u1d25\u2022\u0294 \u0e01\u0e49", "(\u3063\u30fb\u0414\u30fb)\u3063", "\uff61:.\uff9f\u30fd(*\u00b4\u2200`)\uff89\uff9f.:\uff61", "m(\uff61\u2267\uff74\u2266\uff61)m", "( \uff92\u2200\u30fb)", " ( \u30fb\u30fb)\u3064\u2015{}@{}@{}-", "(\u25d5\u072b\u25d5)", "( \u0361\u00b0 \u035c\u0296 \u0361\u00b0)", " (\u25dc\u25d4\u3002\u25d4\u25dd)", " ( \u00a4\u0334\u0336\u0337\u0324\u0301 \u2027\u032b\u032e \u00a4\u0334\u0336\u0337\u0324\u0300 )", "(*\u00b4\u0434`)~\u2665", "\u1ece\u0337\u0356\u0348\u031e\u0329\u034e\u033b\u032b\u032b\u031c\u0349\u0320\u032b\u0355\u032d\u032d\u032b\u032b\u0339\u0317\u0339\u0348\u033c\u0320\u0316\u034d\u035a\u0325\u0348\u032e\u033c\u0355\u0320\u0324\u032f\u033b\u0325\u032c\u0317\u033c\u0333\u0324\u0333\u032c\u032a\u0339\u035a\u031e\u033c\u0320\u0355\u033c\u0320\u0326\u035a\u032b\u0354\u032f\u0339\u0349\u0349\u0318\u034e\u0355\u033c\u0323\u031d\u0359\u0331\u031f\u0339\u0329\u031f\u0333\u0326\u032d\u0349\u032e\u0316\u032d\u0323\u0323\u031e\u0319\u0317\u031c\u033a\u032d\u033b\u0325\u035a\u0359\u031d\u0326\u0332\u0331\u0349\u0356\u0349\u0330\u0326\u034e\u032b\u0323\u033c\u034e\u034d\u0320\u032e\u0353\u0339\u0339\u0349\u0324\u0330\u0317\u0319\u0355\u0347\u0354\u0331\u0355\u032d\u0348\u0333\u0317\u032d\u0354\u0318\u0316\u033a\u032e\u031c\u0320\u0356\u0318\u0353\u0333\u0355\u031f\u0320\u0331\u032b\u0324\u0353\u0354\u0318\u0330\u0332\u0359\u034d\u0347\u0319\u034e\u0323\u033c\u0317\u0316\u0359\u032f\u0349\u0320\u031f\u0348\u034d\u0355\u032a\u0353\u031d\u0329\u0326\u0316\u0339\u033c\u0320\u0318\u032e\u035a\u031f\u0349\u033a\u031c\u034d\u0353\u032f\u0333\u0331\u033b\u0355\u0323\u0333\u0349\u033b\u032d\u032d\u0331\u034d\u032a\u0329\u032d\u033a\u0355\u033a\u033c\u0325\u032a\u0356\u0326\u031f\u034e\u033b\u0330_\u1ece\u0337\u0356\u0348\u031e\u0329\u034e\u033b\u032b\u032b\u031c\u0349\u0320\u032b\u0355\u032d\u032d\u032b\u032b\u0339\u0317\u0339\u0348\u033c\u0320\u0316\u034d\u035a\u0325\u0348\u032e\u033c\u0355\u0320\u0324\u032f\u033b\u0325\u032c\u0317\u033c\u0333\u0324\u0333\u032c\u032a\u0339\u035a\u031e\u033c\u0320\u0355\u033c\u0320\u0326\u035a\u032b\u0354\u032f\u0339\u0349\u0349\u0318\u034e\u0355\u033c\u0323\u031d\u0359\u0331\u031f\u0339\u0329\u031f\u0333\u0326\u032d\u0349\u032e\u0316\u032d\u0323\u0323\u031e\u0319\u0317\u031c\u033a\u032d\u033b\u0325\u035a\u0359\u031d\u0326\u0332\u0331\u0349\u0356\u0349\u0330\u0326\u034e\u032b\u0323\u033c\u034e\u034d\u0320\u032e\u0353\u0339\u0339\u0349\u0324\u0330\u0317\u0319\u0355\u0347\u0354\u0331\u0355\u032d\u0348\u0333\u0317\u032d\u0354\u0318\u0316\u033a\u032e\u031c\u0320\u0356\u0318\u0353\u0333\u0355\u031f\u0320\u0331\u032b\u0324\u0353\u0354\u0318\u0330\u0332\u0359\u034d\u0347\u0319\u034e\u0323\u033c\u0317\u0316\u0359\u032f\u0349\u0320\u031f\u0348\u034d\u0355\u032a\u0353\u031d\u0329\u0326\u0316\u0339\u033c\u0320\u0318\u032e\u035a\u031f\u0349\u033a\u031c\u034d\u0353\u032f\u0333\u0331\u033b\u0355\u0323\u0333\u0349\u033b\u032d\u032d\u0331\u034d\u032a\u0329\u032d\u033a\u0355\u033a\u033c\u0325\u032a\u0356\u0326\u031f\u034e\u033b\u0330", "\u10da(\u256f\u2299\u03b5\u2299\u10da\u2570)", " (\u0a6d\u0941\u00b4 \u141c `)\u0a6d\u0941\u207e\u207e", "(\u0e51\u00b4\u06a1`\u0e51)", "\u0e05\u25cf\u03c9\u25cf\u0e05", "( \uffe3\u25a1\uffe3)/ \u656c\u79ae!! <(\uffe3\u3127\uffe3 ) <(\uffe3\u3127\uffe3 )", "(`\u250c_\u2510\u00b4)", "(\u203e\u25e1\u25dd\u3000)", "( \u25d5\u203f\u203f\u25d5 )", "\u250c|\u25ceo\u25ce|\u2518", "( *\u00af \u00b3\u00af*)\u2661\u3118\u3085", "\u2501(\uff9f\u2200\uff9f)\u2501( \uff9f\u2200)\u2501( \uff9f)\u2501( )\u2501( )\u2501(\uff9f)\u2501(\u2200\uff9f)\u2501(\uff9f\u2200\uff9f)\u2501", "\u30fd(\u30fb\u00d7\u30fb\u00b4)\u309e", "\u03a3(lli\u0434\uff9f\uff89)\uff89", "(\u3064\u0434\u2282)", "(\u256c\uff9f\u0434\uff9f)", "( \u00b4(00)`)", "(\uffe3\u03b5(#\uffe3)\u2606", "(\ub208\u2038\ub208)", "(\u30fb\u2200\u30fb)\u30ce\u4e09G[__]\uff7a", "\u2282(\u00b0\u0414\u00b0\u2282\u2312\uff40\u3064\u2261\u2261\u2261(\u00b4\u2312;;;\u2261\u2261\u2261", "( \uff9f\u2200\uff9f)o\u5f61\uff9f", " |\u0414`)\u30ce\u2312\u25cf\uff5e*", "_(:3 \u300d\u2220 )_", "\u03a3(\uff9f\u03c9\uff9f)", "\u30fe(\u25cf\u309c\u25bd\u309c\u25cf)\u2661 ", "(o\u00b4\u7f52`o)", "\uff08\u00b4-`\uff09.\uff61oO", "(\uff89\u2200`*)", "\u25e2\u2586\u2585\u2584\u2583 \u6eab\u2570(\u301e\ufe36\u301d) \u256f\u99a8 \u2583\u2584\u2585\u2586\u25e3", "\u03a3( \u00b0 \u25b3 \u00b0)", "\u03be( \u273f\uff1e\u25e1\u275b)", "(\u2070\u2296\u2070)", "(#\uff9f\u22bf`)\u51f8", "(\u0e51\u01a1 \u2083 \u01a1)\u2665", "(\u0301=\u25de\u0c6a\u25df=\u2035)", "(-`\u30a7\u00b4-\u256c)", "\u3069\u3053\u2501\u2501\u2501\u2501(\u309c\u2200\u309c\u4e09\u309c\u2200\u309c)\u2501\u2501\u2501\u2501!!??", "(;\u00b4\u0f0e\u0eb6\u0414\u0f0e\u0eb6`)", "(\uff61>\ufe4f<)\u54c8\u557e", "( \uff9f \u03a7 \uff9f)", "(\u3063\u25cf\u03c9\u25cf)\u3063", "\u0669(\u0e51\u2022\u0300\u03c9\u2022\u0301\u0e51)\u06f6", "il||li _|\uffe3|\u25cb\u30fd(\uff65\u03c9\uff65\uff40)", "(\u00b4\uff21\uff40\u3002)", "(o\uff65e\uff65)", "(\u265b\u203f\u265b)", "\u0285\uff08\u00b4\u25d4\u0c6a\u25d4\uff09\u0283", "m9(\uff3e\u0414\uff3e)\uff8c\uff9f\uff77\uff9e\uff6c\uff70", "\u256e(\u2032\uff5e\u2035\u301e)\u256d", "(\u2604\u25e3\u03c9\u25e2)\u2604", "\u30fd(`\u0414\u00b4)\u30ce", "(\u256c\uff9f \u25e3 \uff9f)", "(\u03c3\uff9f\u2200\uff9f)\u03c3..:*\u2606", "(\u25d4\u0c6a\u25d4)", "( \u00b4\u0414`)y\u2501\uff65~~", "(\uff61\u25d5\u2200\u25d5\uff61)", "\u30fe(\u25ce\u00b4\u30fb\u03c9\u30fb\uff40)\u30ce ", "(\u0e51\u00b4\u3141`)", "(\u00b4\u30fb\u00c5\u30fb`)", "(((\uff9f\u0434\uff9f)))", "(\u2229^o^)\u2283\u2501\u2606\uff9f.*\uff65\uff61", "\u2500=\u2261\u03a3((( \u3064\u2022\u0300\u03c9\u2022\u0301)\u3064", "(\u00b4;\u03c9;`)\u30fe(\uff65\u2200\uff65`)", "\u250f((\uff1d\uffe3(\uff74)\uffe3=))\u251b", "\uff08\u3064> _\u25d5\uff09\u3064\ufe3b\u2566\u0335\u0335\u0347\u033f\u033f\u033f\u033f\u2564\u2500\u2500\u2500", "(\u00b0\uff9b\u00b0\u0665)", "\u03b5\u0669(\u0e51> \u2083 <)\u06f6\u0437", "\u03a3(*\uff9f\u0434\uff9f\uff89)\uff89", "(\uff9f\u0434\uff9f\u2261\uff9f\u0434\uff9f)", "(\uff9f3\uff9f)\uff5e\u266a", "(\u2741\u00b4\u03c9`\u2741)*\u2732\uff9f*", "(\u3063\u00b4\u03c9`c)", "\u2606^(\uff4f\u00b4\u0424\u2207\u0424)o", "\u0505(\u00af\ufe43\u00af\u0505)", "\u10da(\u25c9\u25de\u0c6a\u25df\u25c9 )\u10da", "\u0b87\u0434\u0b87", "(\u3186\u1d17\u3186)", "(*\uff9f\u30fc\uff9f)", "(\u3003\uffe3\u03c9\uffe3)\u4eba(\uffe3\ufe36\uffe3\u3003) ", "\u03c3`\u2200\u00b4)\u03c3", "(\u30fb`\u03c9\u00b4\u30fb)", " (//\u00b4\u25dc\u25de\u229c\u25df\u25dd\uff40////) ", "(;\u00b4\u0434\uff40).\uff61\uff4fO(\u30fb\u30fb\u30fb\u30fb)", "(\u0e07\u0e51 \u2022\u0300_\u2022\u0301)\u0e07", "(\uff1b\u00b4\uff9f\u03c9\uff9f\uff40\u4eba)", "\u256e/(\uff1e\u25bd<)\u4eba(>\u25bd\uff1c)\u256d", "(\u256d\uffe33\uffe3)\u256d\u2661 ", "( \uff9f\u0414\uff9f)\u03c3", "_(\u2510\u300c\u03b5:)_", "#\uff9f\u00c5\uff9f\uff09\u2282\u5f61\u2606))\uff9f\u0414\uff9f)\uff65\u2235", "( \u00af\u2022\u03c9\u2022\u00af )", "\u30fe(:3\uff89\uff7c\u30fe)\uff89\uff7c \u4e09[____]", "( \u0c36 3\u0a5c)\uff5e\u2665", " \u10da(\u2022\u03c9 \u2022\u10da)", "\u30fd(\u273f\uff9f\u25bd\uff9f)\u30ce", " (\u00b4\u2229\u03c9\u2229\uff40) ", "(\u30fb\u03b5\u30fb)", "(\u00b4\u25c9\u203f\u25c9\uff40)", "(\u00b4\uff65_\uff65`)", "(\u3065\u2032\u25bd`)\u3065", "\u03b5\u2261\uff8d( \u00b4\u2200`)\uff89", "(\u00b4\u03c3-`)", "(\uff9f\u76bf\uff9f\uff92)", "(^\u03c1^)/", "\uff61\uff65\uff9f\uff65(\uff89\u0414`)\u30fd(\uff9f\u0414\uff9f )\u79c0\u79c0", "( \u0c20\u0d60\u0c20 )\uff89", "_:(\u00b4\u25a1`\u300d \u2220):_", "(\u273a\u03c9\u273a)", "(\u0e51\u2022\u0ac5\u03c9\u2022\u00b4\u0e51)", "(\u0301\uc21c\u25de\u0c6a\u25df\uc21c\u2035) ", "\u03a3(\uff40L_\uff40 )", "(\uff0f\u2035\u0414\u2032)\uff0f~ \u2567\u2567", "(:3[__]4", "(*\u00b4\u0414`)\u3064))\u00b4\u2200`)", "( \u2609_\u2609)\u2261\u261eo\u2500\u2500\u2500\u2500\u2605\u00b0", "\u250f\u251b\u5b78\u6821\u2517\u2513\u03bb\u03bb\u03bb\u03bb\u03bb", "(*\u00b4\u0434`)", "\u03a3 \u03a3 \u03a3 (\u300d\u25cb \u03c9\u25cb )\uff0f", "\u30fe(\u00b4\u03b5`\u30fe)", "0(:3\u3000)\uff5e ('\uff643_\u30fd)_", "=\u2514(\u2510\u534d^o^)\u534d", "(;\uff9f\u0434\uff9f)", "(*\u02c7\u03c9\u02c7*\u4eba)", "( \u2665\u0434\u2665)", "( \u00ba\ufe43\u00ba )", "( \u03a6 \u03c9 \u03a6 )", "\u0c25\u0c6a\u0c25", "(\uff9f\u2200\uff9f)", "|\u67f1|\u03c9\u30fb`)", "( \u00b4-\u03c9 \uff65)\u2584\ufe3b\u253b\u2533\u2550\u2550\u2501", "\uff3c(\u25cfo\u25cb;)\u30ce", "d(d\uff07\u2200\uff07)", "( \u00b4\uff65\u03c9)", "(\u2202\u03c9\u2202)", "|////|\u3000( \u3000)\uff89\u3000|////|(\u81ea\u52d5\u9580", "\u3002\u309c+.(\u4eba-\u03c9\u25d5)\u309c+.\u309c", "_(\u221a \u03b6 \u03b5:)_", "\u0f3c \u3064/\u0335\u0347\u033f\u033f/\u2019\u033f\u2019\u033f \u033f \u033f\u033f \u033f\u033f\u25d5 _\u25d5 \u0f3d\u3064/\u0335\u0347\u033f\u033f/\u2019\u033f\u2019\u033f \u033f \u033f\u033f \u033f\u033f \u033f\u033f", "(`\u0434\u00b4)", "(\u3000\u25dc\u25e1\u203e)", " (:3\u3063)\u3078 \u30fd(\u00b4\u0414\uff40\u25cf\u30fd)", "(#\u2035)3\u2032)\u2582\u2582\u2582\u2583\u2584\u2585\uff5e\uff5e\uff5e\u55e1\u55e1\u55e1\u55e1\u55e1", "\u2606\u2312(*^-\u309c)v", "(\u00b4\u2256\u25de\u0c6a\u25df\u2256)", "( \u02d8\u2022\u03c9\u2022\u02d8 )\u25de\u207d\u02d9\u00b3\u02d9\u207e", "( \u00d7\u03c9\u00d7 )", "(#`\u0414\u00b4)\uff89", "-//(\u01d2.\u01d2)//-", "(\u0301\u25c9\u25de\u0c6a\u25df\u25c9\u2035)", "\u30fe(\uff1b\uff9f(OO)\uff9f)\uff89", "\u2299\u8c37\u2299", "\u10da(\uff9f\u0434\uff9f\u10da)", "\u2665(\u00b4\u2200` )\u4eba", "( \uffe3 3\uffe3)y\u2582\u03be", "\uff61\uff9f(\uff9f\u00b4\u03c9`\uff9f)\uff9f\uff61", "\u2267\u3014\u309c\u309c\u3015\u2266", "=\u035f\u035f\u035e\u035e( \u2022\u0300\u0434\u2022\u0301)", "(\u2718\ufe4f\u2718\u10d0)", "(`3\u00b4)", "(\u3000\u02d9\u706c\u02d9\u3000)", "(\uff0a\u309c\u30fc\u309c)b", "(\u3063\ufe4f-) .\uff61o", "\u256f-____-)\u256f~\u2550\u2569\u2550\u2550\u2550\u2550\u2569\u2550", "(\u706c\u00ba\u03c9\u00ba\u706c)", "\u256e(\u256f_\u2570)\u256d", "\u03a3\u30fd(\uff9f\u0414 \uff9f; )\uff89", "(\u3000\u3063\u0e18\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e34\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u0e4c\u10da)", "(\uff61\u00b4\u2200\uff40)\uff89\u2015\u2282ZZZ\u2283", "\u03a3>\u2015(\u3003\u00b0\u03c9\u00b0\u3003)\u2661\u2192", "(\u256f\u00b0O\u00b0)\u256f\u253b\u2501\u253b", "\u03c3 \uff9f\u2200 \uff9f) \uff9f\u2200\uff9f)\u03c3 ", "(\u00b4-\u03b9_-\uff40)", "*\u0b18(\u0a6d*\u02ca\u1d55\u02cb)\u0a6d* \u0a48\u2729\u2027\u208a\u02da", "(*\u00b4\u2200`)~\u2665", "(,,\u30fb\u03c9\u30fb,,)", "( ^\u03c9^)", "\u250f(_\u0434_\u250f)\u2513))", "(=\u00b4\u03c9`=)", "\u30fe(;\uff9f;\u0414;\uff9f;)\uff89\uff9e", "(\u0434) \uff9f\uff9f", "(=\uff9f\u0414\uff9f=) \u2584\ufe3b\u253b\u2533\u2501 \u00b7.`.`.`.", "( \u02d8\uff65\u0437\uff65)", "(\u256f\u00b0\u25bd\u00b0)\u256f \u253b\u2501\u253b", "(\u272a\u03c9\u272a)", "\uff3c(\u25cf\u00b4\u03d6`\u25cf)\uff0f", "\u30fd(\u25cf\u00b4\u03b5\uff40\u25cf)\u30ce", "(\u203a\u00b4\u03c9`\u2039 )", "(#`\u76bf\u00b4)", "\u25d1\u2582\u25d0", "(\u00b4///\u2601///`)", "(\uff61\u014f_\u014f)", "(\uff89>\u03c9<)\uff89", "(\u3000\uff9f\u2200\uff9f) \uff89\u2661", "\u2211(\uffe3\u25a1\uffe3;)", "(\u309d\u2200\uff65)b", "(\u0e51\u2022 . \u2022\u0e51)", "(\u00b4;;\u25c9;\u2296;\u25c9;;\uff40)", "(\uff61\u2022\u3145\u2022\uff61)\u2661", "(\u3145\u02d8\u3142\u02d8)", "\u0669(\u0e51\u00b43\uff40\u0e51)\u06f6", "(\u00b4\u0398\u03c9\u0398`)", ",,\u053e\u2038\u053e,,", "\u03c3(\u00b4\u2200\uff40*)", "(\u00b4\u25bd`\u0283\u2661\u01aa)", "\u250c(\u250c^o^)\u2510", "\u207d\u207d \u25df(\u2217 \u02ca\u03c9\u02cb \u2217)\u25de \u207e\u207e", "(\u25cd\u2022\u1d17\u2022\u25cd)\u309d", "(*\u2019\uff70\u2019*)", "( \u03c3\u055e\u0a0a \u055e)\u03c3 ", "( \u02d8\u2022\u03c9\u2022\u02d8 ).oO\u0b87", "(\u25cf\uff40 \u8278 \u00b4)", "\u22c9(\u25cf \u2238 \u25cf)\u22ca", "( \u00b4\u2022\u0325\u0325\u0325\u03c9\u2022\u0325\u0325\u0325` )", "(\u25cf\u00b4\u03c9\uff40\u25cf)\u309e", "(\uff4f\uff9f\uff9b\uff9f)\u250c\u251b\u03a3(\uff89\u00b4*\u03c9*`)\uff89", "|\u67f1|\u03c9\uff9f)---c<` \u0434\uff9f)!", "(\u0e51\u00b4\u3142`\u0e51)", "(\uff9f\u3005\uff61)", "\u30fd(\u0301\u25d5\u25de\u0c6a\u25df\u25d5\u2035)\uff89"] -------------------------------------------------------------------------------- /jseg/Emodet/emodet.py: -------------------------------------------------------------------------------- 1 | # -*-coding:utf-8-*- 2 | from os.path import abspath, dirname, join 3 | import urllib2 4 | import re 5 | import time 6 | import json 7 | 8 | CUR_PATH = dirname(abspath(__file__)) 9 | 10 | 11 | def craw_emo(save=True): 12 | base = 'http://facemood.grtimed.com/index.php?view=facemood&tid=%d' 13 | pat = re.compile(u'''flashvars="str=(.*?)"''') 14 | emos = [] 15 | 16 | cnt1 = 1 17 | while cnt1 < 200: 18 | print cnt1 19 | url = base % cnt1 20 | raw = urllib2.urlopen(url).read().decode('utf-8') 21 | res = pat.findall(raw) 22 | if res == []: 23 | cnt1 += 1 24 | else: 25 | emos += res 26 | cnt2 = 2 27 | while True: 28 | url2 = url + '&page=%d' % cnt2 29 | print url2 30 | raw2 = urllib2.urlopen(url2).read().decode('utf-8') 31 | res2 = pat.findall(raw2) 32 | if res2 == []: 33 | break 34 | else: 35 | emos += res2 36 | cnt2 += 1 37 | time.sleep(0.5) 38 | cnt1 += 1 39 | time.sleep(0.5) 40 | print '%d emos collected' % len(emos) 41 | emos = list(set(emos)) 42 | if save: 43 | with open('emodata.json', 'w') as jf: 44 | json.dump(emos, jf) 45 | print 'emoticons saved!' 46 | else: 47 | return emos 48 | 49 | 50 | def load_emo(): 51 | with open(join(CUR_PATH, 'emoall.txt'), encoding="utf-8") as f: 52 | "") as f: 53 | emos = f.read().decode('utf-8') 54 | emos = emos.split('\n') 55 | emos = [i for i in emos if i != '\n'] 56 | # with open(join(CUR_PATH, 'emodata.json')) as jf1, \ 57 | # open(join(CUR_PATH, 'emo2.json')) as jf2, \ 58 | # open(join(CUR_PATH, 'emo3.json')) as jf3, \ 59 | # open(join(CUR_PATH, 'emo4.json')) as jf4: 60 | # emos1 = json.load(jf1) 61 | # emos2 = json.load(jf2) 62 | # emos3 = json.load(jf3) 63 | # emos4 = json.load(jf4) 64 | # emos = emos1 + emos2 + emos3 65 | emo_ext = [u'= =', u'=.=', u'=_=', u'>///<', u'> <', u'orz', 66 | u'^ ^', u'XD', u'(′･_･`)', u'-_-||', u'>"<', u'T.T'] 67 | emos += emo_ext 68 | emos = list(set(emos)) 69 | emos = [i.strip() for i in emos] 70 | return emos 71 | 72 | 73 | def ensure_unicode(string): 74 | if not isinstance(string, unicode): 75 | try: 76 | string = string.decode('utf-8') 77 | except: 78 | raise UnicodeError('Input should be UTF8 or UNICODE') 79 | return string 80 | 81 | emos = load_emo() 82 | 83 | 84 | def find_emo(txt, source=False): 85 | global emos 86 | txt = ensure_unicode(txt) 87 | res = [i for i in emos if i in txt] 88 | if source is True: 89 | for x in res: 90 | txt = txt.replace(x, '' + x + '') 91 | return (res, txt) 92 | return res 93 | 94 | # source 95 | # http://facemood.grtimed.com/ (顏文字卡) --> emodata.json 96 | # http://news.gamme.com.tw/55689 (宅宅新聞) --> emo2.json 97 | # http://www.ptt.cc/bbs/cksh77th17/M.1158075878.A.9D2.html (PTT) --> emo3.json 98 | # http://j831124g2009.pixnet.net/blog/post/17973045-素材│日本顏文字（可愛）(*☉౪⊙*) 99 | # (痞客邦) --> emo4.json 100 | 101 | # problems 102 | # 兩行以上的emoticons怎麼辦？ 103 | #　　　　 ▲ ▲ ▲ 104 | # ﹏﹏(ㄏ￣▽￣)ㄏ q(〒□〒)p ㄟ(￣▽￣ㄟ)﹏﹏ 通通變成阿飄 105 | -------------------------------------------------------------------------------- /jseg/Emodet/emodet2.py: -------------------------------------------------------------------------------- 1 | #-*-coding:utf-8-*- 2 | import re 3 | 4 | metachr = '^$*+?{}[]\|()' 5 | 6 | heads = u'''>ζↁ━இㅇ（┌／∑┐ψ〒ʕ└⊙＜√＝☜ㄟ〠Σ•ತ✧╰(*-/☮Oε┴‷〈ಸ爻<︿乁♡ｂˋ⁄ˇఠ◉ಠ○vｍ#◎⇎Ψmb◔♀ⶸლ\｡￣◢づ‧≧흫…╭o╮ჰ≡癶凸~ヽヾ''' 7 | tails = u'''ↁﾉ「↑̖〠ิ▽♀⁄ˇ◣～⑤╮凵︿・|━）dʔಠ√ತ'+/°3ืಸ;?ξツఠｍ﹏◎"⇎｡≡╩╭oჴஇ☆┌з┐〒↗⊙☮☞•…*.┴:＼︴♡σ♂ｄ○●Z￣b≦♪╬╰癶>ｼ~＃〉３↘┛”；!✧)Ψ-ⶸ=＿ㅇψˊノ◔ლaづ╧흫m╯y凸''' 8 | for i in metachr: 9 | heads = heads.replace(i, '\\'+i) 10 | tails = tails.replace(i, '\\'+i) 11 | 12 | 13 | def ensure_unicode(string): 14 | if not isinstance(string, unicode): 15 | try: 16 | string = string.decode('utf-8') 17 | except: 18 | raise UnicodeError('Input should be UTF8 or UNICODE') 19 | return string 20 | 21 | def clear_num(txt): 22 | num_pat = re.compile(u'''([0-9\-.:/]{2,})''') 23 | res = num_pat.sub('', txt) 24 | return res 25 | 26 | def post_check(candlst): 27 | con = [] 28 | for i in candlst: 29 | while True: 30 | i_mod = i.strip() 31 | i_mod = re.sub(u'\([一-龥]{1,}[\)]{0,1}', '', i_mod) 32 | i_mod = re.sub(u'（[一-龥]{1,}）', '', i_mod) 33 | if len(set(i_mod)) < 3: 34 | break 35 | else: 36 | if re.search(u'^m:[0-9a-zA-Z一-龥]', i_mod): # m:大大 37 | break 38 | elif re.search(u'^~[0-9a-zA-Z一-龥]', i_mod): #~4次 39 | break 40 | elif i_mod == i: 41 | con.append(i_mod) 42 | break 43 | else: 44 | i = i_mod 45 | continue 46 | return con 47 | 48 | def exclude_cht(txt): 49 | pat = re.compile(u'''(?!.*[.a-zA-Z][.a-zA-Z])([一-龥][一-龥]{2,})''') 50 | while True: 51 | res = pat.sub('', txt) 52 | if res == txt: 53 | return res 54 | else: 55 | txt = res 56 | exclude_cht(txt) 57 | 58 | def find_emo(txt, source=False): 59 | txt = ensure_unicode(txt) 60 | txt_mod = clear_num(txt) 61 | txt_mod = exclude_cht(txt_mod) 62 | emo_pat = re.compile(u'''(?!.*[.a-zA-Z][.a-zA-Z])([%s].{3,})''' % heads) 63 | res = emo_pat.findall(txt_mod) 64 | res = post_check(res) 65 | res = [i for i in res if i in txt] 66 | if source == True: 67 | for x in res: 68 | txt = txt.replace(x, ''+x+'') 69 | return (res,txt) 70 | return res 71 | 72 | -------------------------------------------------------------------------------- /jseg/__init__.py: -------------------------------------------------------------------------------- 1 | from .jieba import Jieba # noqa 2 | __version__ = '0.0.4' 3 | -------------------------------------------------------------------------------- /jseg/brill_tagger.pkl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/amigcamel/Jseg/1b8f6aa5e65cfa770cb7e81df0a4fbc15a7bcd35/jseg/brill_tagger.pkl -------------------------------------------------------------------------------- /jseg/emoall.txt: -------------------------------------------------------------------------------- 1 | (ﾒﾟДﾟ)ﾒ 2 | (〃^◇^〃) 3 | Zz(´-ω-`*) 4 | (꒪ͦᴗ̵̍꒪ͦ ) 5 | (`へ´≠) 6 | (♡˙︶˙♡) 7 | |Φ|<|′|◇|‵|>|Φ| 8 | (( へ(へ´∀`)へ 9 | ╮(﹋﹏﹌)╭ 10 | (ó㉨ò) 11 | (́^◞౪◟<‵) 12 | ( ´∀`);y=ｰ(ﾟдﾟ)･∵. ﾀｰﾝ 13 | (*˘︶˘*).｡.:*♡ 14 | (-◎◇◎-)o 15 | ﾟÅﾟ) 16 | 爻((╬ಠิ益ಠิ))ｲｶﾘｽﾞﾑｯ!! 17 | ヽ( ^∀^)ﾉ 18 | (́提◞౪◟供‵) 19 | ╮(′ω‵)╭ 20 | -@{}@{}-υ(￣▽￣*) 21 | (((o(*ﾟ▽ﾟ*)o))) 22 | ╮(╯∀╰)╭ 23 | 。･ﾟ･(つд`ﾟ)･ﾟ･ 24 | (^_っ^) 25 | σ >︿<ο 26 | (　◞≼☸≽◟ ._ゝ◞≼☸≽◟) 27 | ( 3ω3) 28 | (｢･ω･)｢ 29 | ( *´◒`*) 30 | (^-^*) 31 | ( • ̀ω•́ ) 32 | (́✺◞౪◟✺‵) 33 | d(`･∀･)b 34 | (*ﾟ∀ﾟ*) 35 | o(‵▽′)o 36 | ⶸ(ᖛ Δ ᖜ)ⶸ 37 | （ ´ﾟ,_」ﾟ）ﾊﾞｶｼﾞｬﾈｰﾉ 38 | (｡･㉨･｡) 39 | ╠╦╩╦╦╩╣◇‧) 40 | ☻ 41 | (´・ω・)つ旦 42 | ╮(▔▽▔")╭ 43 | ┐(´д`)┌ 44 | (ﾟд⊙) 45 | _(´ཀ`」 ∠)_ 46 | /(*￣△￣)\ 47 | (╬☉д⊙) 48 | ヾ(。￣□￣)ツ 49 | Σ(;ﾟдﾟ) 50 | (￣(エ)￣) 51 | ^(＊)^ 52 | (´・ω・`) 53 | (=˙~˙=) 54 | ⊙▃⊙ 55 | ｡ﾟヽ(ﾟ´Д`)ﾉﾟ｡ 56 | ∑(ι´Дン)ノ 57 | ♡(*´∀｀*)人(*´∀｀*)♡ 58 | <(￣︶￣)> 59 | -`д´- 60 | (◞≼◉ื≽◟;益;◞≼◉ื ≽◟) 61 | (Ｔ◇Ｔ) 62 | (ゝ∀･) 63 | (#╯￣□￣)╯~┴─┴ 64 | ε=ε=ε=ε=ε=ε=╭(￣▽￣)╯ 65 | (゜皿。) 66 | (〒︿〒) 67 | 凸(= _=)凸 68 | ლ(⁰⊖⁰ლ) 69 | (-_____-x) 70 | (ᇂдᇂ ) 71 | (*≥▽≤)ツ┏━┓ 72 | (́ತ◞౪◟ತ‵) 73 | ლ｜＾Д＾ლ｜ 74 | (#‵□′)╯ 75 | ︿(′□‵)︿ 76 | ( ˘•ω•˘ ) 77 | (́墨◞౪◟鏡‵) 78 | (శωశ) 79 | ┌|◎o◎|┘ 80 | ・゜・(PД`q｡)・゜・ 81 | (╬ﾟдﾟ) 82 | ╮（╯╴╰∥）╭ 83 | (눈‸눈) 84 | ◑ω◐ 85 | =.= 86 | ☜(:♛ฺ;益;♛ฺ;)☞ 87 | (*‵◇′*) 88 | (<ゝω・) 綺羅星☆ 89 | (°Д°) 90 | ヾ(´︶`*)ﾉ♬ 91 | ┏( .-. ┏ ) ┓ 92 | (´c_`) 93 | ⊂彡☆))д`) 94 | (￣▽￣) 95 | (┛`д´)┛ 96 | q(′□‵)p 97 | ／/( ◕‿‿◕ )＼ 98 | (๑･ิ⋖⋗･ิ๑) 99 | ლ(╹ε╹ლ) 100 | (o`･ω･)ﾉ(ﾉД`) 101 | ᕕ ( ᐛ ) ᕗ 102 | ჰჰჰ❛‿❛ჴჴჴ 103 | ρ(・ω・、) 104 | (｡-_-｡) 105 | (´･_･`) 106 | ლ(́◕◞౪◟◕‵ლ) 107 | = = 108 | (๑•́ ₃ •̀๑) 109 | ＼＼\\٩( 'ω' )و //／／ 110 | ( ´ﾟДﾟ`) 111 | ( ‘д‘⊂彡☆))Д´) 112 | (;_;) 113 | ლ(´•д• ̀ლ 114 | Σ(ﾟдﾟ) 115 | <*)) >>=< 116 | (•ㅂ•)/ 117 | ヾ(*ΦωΦ)ツ 118 | (　ﾟ∀ﾟ)つ≡≡≡♡♡♡)`ν゜)ｸﾞｼｬ 119 | \(′□‵)/ 120 | ε=ε=ヾ(;ﾟдﾟ)/ 121 | _(┐「﹃ﾟ｡)_ 122 | ㄟ(￣▽￣ㄟ)﹏ 123 | ～(╯▽╰)～ 124 | o(☆Ф∇Ф☆)o 125 | (*´ω`)人(´ω`*) 126 | ┳━┳ノ( OωOノ) 127 | ~(～o￣▽￣)～o.....o～(＿△＿o～)~.. 128 | (ﾟ⊿ﾟ) 129 | (〒︵〒) 130 | (:◎)≡ 131 | ｡ﾟ･(ﾉД`)人(´Д`)人(Д` )･ﾟ｡ 132 | ┳━┳ノ( ' - 'ノ) 133 | (#ﾟДﾟ) 134 | ლ(・∀・ )ლ 135 | (́≖◞౪◟≖‵)/ 136 | (´～`) 137 | (¦3[▓▓] 138 | (╯≧▽≦)╯ ~╩═══╩ 139 | (❛◡❛✿) 140 | 乁( ◔ ౪◔)「 141 | (ｰ｀ ) 142 | ( ′-`)y-～ 143 | (´-ω-｀) 144 | (σ´□｀)σ 145 | ᕦ(ò_óˇ)ᕤ 146 | ₍₍◝(･'ω'･)◟⁾⁾ 147 | ◢▆▅▄▃ 溫╰(〞︶〝) ╯馨 ▃▄▅▆◣ 148 | ( ºΔº ) 149 | 흫_흫 150 | (°c＿,°`) 151 | (╬♛. 益♛. ) 152 | (́✸◞♉◟✸‵) 153 | (=´ᴥ`) 154 | ఠ_ఠ 155 | : ♡｡ﾟ.(*♡´◡` 人´◡` ♡*)ﾟ♡ °・ 156 | (╬ಠ益ಠ)ｺﾞﾙァ!! 157 | ━(ﾟ∀ﾟ)━( ﾟ∀)━( ﾟ)━( )━( )━(ﾟ)━(∀ﾟ)━(ﾟ∀ﾟ)━ 158 | (◔⊖◔)つ 159 | (,,ﾟДﾟ) 160 | ⁄(⁄ ⁄•⁄ω⁄•⁄ ⁄)⁄ 161 | ಸ_ಸ 162 | （＾ω＾） 163 | ლ(・´ｪ`・ლ) 164 | (⁰▿⁰) 165 | (ﾒ3[____] 166 | ﹏﹏(ㄏ￣▽￣)ㄏ 167 | ლ(́ಢ.◞౪◟ಢ‵ლ) 168 | ヽ(゜▽゜　)－C<(/;◇;)/~[拖走] 169 | (´◓Д◔`) 170 | (≧∀≦)ゞ 171 | (´д((☆ミPia!⊂▼(ｏ ‵－′ｏ)▼つPia!彡★))∀`) 172 | ˋ(′～‵")ˊ 173 | (ｏ￣∇￣)＝◯)`ν゜)･；' 174 | ✧◝(⁰▿⁰)◜✧ 175 | <(_ _)> 176 | (｡A｡) 177 | (′Ｑ‵)╭～ 178 | (๑•̀ㅂ•́)و✧ 179 | (☞ﾟ∀ﾟ)ﾟ∀ﾟ)☞ 180 | （´≝◞⊖◟≝｀) 181 | (●б人<●) 182 | ((└(:3」┌)┘)) 183 | ( •́ _ •̀)？ 184 | (́◕///◞౪◟///◕‵)/ 185 | (๑¯∀¯๑) 186 | (◞‸◟) 187 | (*￣◇￣)╯ 188 | ヽ(㊤V㊤*)ﾉ 189 | \("￣□￣)/ 190 | (′･_･`) 191 | Σ(°Д°; 192 | <(￣><￣)> 193 | (ㄨ‵◇′)凸 194 | ◥(ฅº￦ºฅ)◤ 195 | (‘⊙д-) 196 | ლ( • ̀ω•́ )っ 197 | (￣╔◇╗￣) 198 | ㄟ(￣▽￣ㄟ)﹏﹏ 199 | (́۞◞౪◟۞‵) 200 | (＠◇＠) 201 | （￣(∞)￣）y-～ 202 | ( ￣ c￣)y▂ξ 203 | (ÒωÓױ) 204 | (@^◇^@) 205 | (」・ω・)」うー！(／・ω・)／にゃー！ 206 | (﹋﹏﹌) 207 | (|||ﾟдﾟ) 208 | (Ò㉨Óױ) 209 | (￣﹏￣) 210 | (▰˘◡˘▰) 211 | (╯ᕡὤᕡ)╯ 212 | \⊙▽⊙/ 213 | ƪ(•̃͡ε•̃͡)∫ 214 | ( ´∀｀)σ 215 | （’へ’） 216 | (╯°Д°)╯ ┻━┻ 217 | (((┏(;￣▽￣)┛ 218 | ԅ(¯﹃¯ԅ) 219 | (ﾟ∀。) 220 | 州‧◇‧州 221 | ヽ(∀ﾟ )人(ﾟ∀ﾟ)人( ﾟ∀)人(∀ﾟ )人(ﾟ∀ﾟ)人( ﾟ∀)ﾉ 222 | ( ﾟДﾟ)σ 223 | ( ﾟДﾟ)ﾉ⌒⑤ 224 | ヽ( ° ▽°)ノ 225 | /(′˙‵)\ 226 | (●⁰౪⁰●) 227 | (❍ᴥ❍ʋ) 228 | ‷̗ↂ凸ↂ‴̖ 229 | ╰(￣▽￣)╯ 230 | ..._〆(°▽°*) 231 | (ﾟ∀ﾟ ) 232 | ╮(〒﹏〒)╭ 233 | ︿(@￣︶￣@)︿ 234 | (΄ಢ◞౪◟ಢ‵)◉◞౪◟◉) 235 | (ﾒ ﾟ皿ﾟ)ﾒ 236 | (*´з｀*) 237 | ε≡(ノ´＿ゝ｀）ノ 238 | (〃∀〃) 239 | ～(￣▽￣)～(＿△＿)～(￣▽￣)～(＿△＿)～(￣▽￣)～ 240 | ಠ_ಠ 241 | (･ิω･ิ) 242 | ヽ(=^･ω･^=)丿 243 | (｡í _ ì｡) 244 | ˋ( ° ▽﹑°|||| ) 245 | (́థ◞..◟థ‵) 246 | ( ´∀｀)つt[ ] 247 | (￣ε(#￣)☆ 248 | (‵□′)--●(#)￣(00)￣) 249 | (　´・◡・｀) 250 | (:3[___]= 251 | (｀_ゝ´) 252 | (｡・ω・｡) 253 | （´◔​∀◔`) 254 | （ლ ^ิ౪^ิ）ლ 255 | ﹏﹏(ㄏ￣▽￣)ㄏ (一(oo)一)＊ ㄟ(￣▽￣ㄟ)﹏ 256 | (•‾⌣‾•) 257 | ....ˊˋ------｡:.ﾟ_ヽ(_´∀`_)ﾉ_.:｡((浮水 258 | ･*･:≡(　ε:) 259 | 魔貫光殺砲(ﾟДﾟ)σ━00000000000━● 260 | ╰(‵□′#) 261 | (=￣◇￣=)Ｖ 262 | ( ´･･)ﾉ(._.`) 263 | (ﾉ∀｀ ) 264 | ヽ(#`Д´)ﾉ 265 | (ﾟДﾟ≡ﾟДﾟ)? 266 | ┌|≧◇≦|┘做└|≧◇≦|┐體┌|≧◇≦|┘操 267 | (#▔︵▔)=====●(￣ ⊿"￣)╭ 268 | (#‵▽′)／~~~~~~~~~(☆___☆) 269 | ＿ﾉ乙(､ﾝ､)＿ 270 | (´-ωก`) 271 | (*^ ◇ ^*) 272 | (　ᐛ)パァ 273 | ლ(･ ิω･ิლ) 274 | (́>◞౪◟<‵)ﾉｼ 275 | (*‘ v`*) 276 | ( ¯﹁¯ ) 277 | (ヾﾉ･ω･`) 278 | (ﾟω´) 279 | (σ回ω・)σ←↑→↓←↑ 280 | ｷﾀ━━━━(ﾟ∀ﾟ)━━━━!! 281 | _(:3 」∠ )_ 282 | (ﾉ◕ヮ◕)ﾉ*:･ﾟ✧ 283 | o(＠//////＠)o 284 | _(¦3」∠)_ 285 | öㅅö 286 | (ﾟﾛ ﾟﾉ)ﾉ 287 | （╯－＿－）╯╧╧ 288 | (一(oo)一)＊ 289 | (●▼●;) 290 | (╬ﾟдﾟ)▄︻┻┳═一 291 | (^u^) 292 | (`・ω・´) 293 | (　′∀`)σ≡σ☆))Д′) 294 | ( ～'ω')～ 295 | :;(∩´﹏`∩);: 296 | (≖ᴗ≖๑) 297 | (ㄏ￣▽￣)ㄏ ㄟ(￣▽￣ㄟ) 298 | <Φ﹏Φ> 299 | (ﾉ｀⊿´)ﾉ 300 | (ノ・＿・)ノ凹 ┣凹━凹━凹┫ 301 | ╮(╯3╰)╭　╮(╯_╰)╭　╮(╯▽╰)╭ 　╮(￣▽￣)╭ 302 | (･ω´･ ) 303 | ヽ( ´ー)ノ 304 | Φ౪Φ 305 | (*ω*) 306 | ╭(╭(╭（ㄏ￣□￣）ㄏ 307 | ฅ(๑*д*๑)ฅ!! 308 | ╮(╯▽╰")╭☆ 309 | ╭(╭(╭ (ㄏ￣△￣)ㄏ 310 | (́⊙◞౪◟⊙‵) 311 | (灬ºωº灬) 312 | -_-|| 313 | Σ(ﾟДﾟ；≡；ﾟдﾟ) 314 | (΄◉◞౪◟◉‵) 315 | ლ(｀∀´ლ) 316 | (○‵△′)=○○○○(￣#)3￣) 317 | (́。◞౪◟。‵) 318 | (○^～^○) 319 | ٩(｡・ω・｡)و 320 | (“￣▽￣)-o█ █o-(￣▽￣”)/ 321 | ┬┴┬┴┤(‧◇├┬┴┬ 322 | d(・ω・d) 微分！(∫・ω・)∫ 積分！∂(・ω・∂) 偏微分！(∮・ω・)∮ 沿閉曲線的積分！(∬・ω・)∬ 重積分！∇(・ω・∇)梯度！∇・(・ω・∇・)散度！∇×(・ω・∇×)旋度！Δ(・ω・Δ)拉普拉斯！ 323 | ╰(‵△′)╯ 324 | (ಠ益ಠ) 325 | (´,_ゝ`) 326 | (･∀･)っ/凵 327 | (((ﾟДﾟ;))) 328 | (╥﹏╥) 329 | (́〠◞౪◟〠‵) 330 | Ψ(≧◇≦)Ψ 331 | ( ≧◇≦ ) 332 | (/ ↀ3ↀ)/q(﹌ω﹌ ) 333 | ლ(╹◡╹ლ) 334 | └(￣^￣ )┐ 335 | (o´・ω・`)σ)Д`) 336 | (；д；) 337 | _(:3 ⌒ﾞ)_ 338 | (￣◇￣)凸 339 | XD 340 | ╮（╯﹏╰）╭ 341 | (^y^) 342 | (╯ ‵Д′)╯╧───╧ 343 | (σ′▽‵)′▽‵)σ 344 | ಠ益ಠ 345 | (´,,•ω•,,)♡ 346 | (′□‵)╭～ 347 | (/Д`)~ﾟ｡ 348 | (o･e･) 349 | ( ╬◣ 益◢)y 350 | ∥∥<∥′ ◇‵ || >∥∥ 351 | ≧▽≦ 352 | ～(￣0￣)～ 353 | (´;ω;`) 354 | (　っค้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้้c ) 355 | ξ( ✿＞◡❛)▄︻▇▇〓▄︻┻┳═一 356 | (´⊙ω⊙`) 357 | (≖＿≖)✧ 358 | ||Φ|(|´|Д|`|)|Φ|| 359 | (∩^o^)⊃━☆ﾟ.*･｡ 360 | o(‧/////‧)o 361 | (=´ω`=) 362 | Σ>―(〃°ω°〃)♡→ 363 | (╯‵□′)╯︵┴─┴ 364 | (*´･д･)? 365 | _(°ω°｣ ∠) 366 | (・∀・)つ⑩ 367 | (-ι_- ) 368 | (́▣◞౪◟▣‵) 369 | ┬─┬ ノ( ' - 'ノ) 370 | ٩(ŏ﹏ŏ、)۶ 371 | (#￣▽￣)o╭╯ ☆╰╮o(￣▽￣///)< 372 | ( శ 3ੜ)～♥ 373 | (づ′▽`)づ 374 | ´-ω-)b 375 | o(*/////*)o 376 | \(≧ω≦)/ 377 | (☉д⊙) 378 | (′ω‵)y-～ 379 | (ΦωΦ) 380 | (́☜◞౪◟☞`)/ 381 | (〒△〒) 382 | ξ( ✿＞◡❛) 383 | （´◔ ₃ ◔`) 384 | (o ─_─)==O)。O。) 385 | (Π△Π) 386 | (=m=) 387 | o(⊙/////⊙)o 388 | (´_ゝ`) 389 | (°ཀ°) 390 | ๛ก(ｰ̀ωｰ́ก) 391 | (￫ܫ￩) 392 | ( ╯' - ')╯ ┻━┻ 393 | (´◉㉨◉) 394 | (=ﾟДﾟ=) ▄︻┻┳━ ·.`.`.`. 395 | (っ・Д・)っ 396 | ｡:.ﾟヽ(*´∀`)ﾉﾟ.:｡ 397 | m(｡≧ｴ≦｡)m 398 | ( ﾒ∀・) 399 | ( ・・)つ―{}@{}@{}- 400 | (◕ܫ◕) 401 | ( ͡° ͜ʖ ͡°) 402 | (〒◇〒)o～ 403 | ヽ(́◕◞౪◟◕‵)ﾉ 404 | (◜◔。◔◝) 405 | ( ¤̴̶̷̤́ ‧̫̮ ¤̴̶̷̤̀ ) 406 | (́◉◞౪◉*◉◉◞◕౪౪◟◕◉◉φ◉౪◞◉*‵) 407 | (*´д`)~♥ 408 | Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰_Ỏ̷͖͈̞̩͎̻̫̫̜͉̠̫͕̭̭̫̫̹̗̹͈̼̠̖͍͚̥͈̮̼͕̠̤̯̻̥̬̗̼̳̤̳̬̪̹͚̞̼̠͕̼̠̦͚̫͔̯̹͉͉̘͎͕̼̣̝͙̱̟̹̩̟̳̦̭͉̮̖̭̣̣̞̙̗̜̺̭̻̥͚͙̝̦̲̱͉͖͉̰̦͎̫̣̼͎͍̠̮͓̹̹͉̤̰̗̙͕͇͔̱͕̭͈̳̗̭͔̘̖̺̮̜̠͖̘͓̳͕̟̠̱̫̤͓͔̘̰̲͙͍͇̙͎̣̼̗̖͙̯͉̠̟͈͍͕̪͓̝̩̦̖̹̼̠̘̮͚̟͉̺̜͍͓̯̳̱̻͕̣̳͉̻̭̭̱͍̪̩̭̺͕̺̼̥̪͖̦̟͎̻̰ 409 | ヽ(｀д´)ﾉ 410 | ლ(╯⊙ε⊙ლ╰) 411 | (੭ु´ ᐜ `)੭ु⁾⁾ 412 | (๑´ڡ`๑) 413 | ฅ●ω●ฅ 414 | (￣ー￣) 415 | ( ￣□￣)/ 敬禮!! <(￣ㄧ￣ ) <(￣ㄧ￣ ) 416 | (`┌_┐´) 417 | ↖╭╰('(Ω)')╯ 418 | (‾◡◝　) 419 | (((m -_-)m 420 | (ㄨ‵◇′) 421 | ( ◕‿‿◕ ) 422 | (*☉౪⊙*) 423 | （ ´☣///_ゝ///☣｀） 424 | Σ┗(＠ロ＠;)┛ 425 | ○(#‵︿′ㄨ)○ 426 | (◞≼●≽◟◞౪◟◞≼●≽◟) 427 | ( *¯ ³¯*)♡ㄘゅ 428 | ╰(￣︶￣)╯ 429 | (*´艸`*) 430 | 凸(▔︵▔#) 431 | ヽ(・×・´)ゞ 432 | Σ(lliдﾟﾉ)ﾉ 433 | (つд⊂) 434 | (//´/◒/`//) 435 | ( ^ิ౪^ิ) 436 | \(Π◇Π) 437 | >///< 438 | ( ´(00)`) 439 | (✧≖‿ゝ≖) 440 | (☍﹏⁰) 441 | (・∀・)ノ三G[__]ｺ 442 | (╬ಠิ益ಠิ) 443 | ⊂(°Д°⊂⌒｀つ≡≡≡(´⌒;;;≡≡≡ 444 | ლ（இ e இ｀。ლ） 445 | (〒﹏〒) 446 | ( ﾟ∀ﾟ)o彡ﾟ 447 | |Д`)ノ⌒●～* 448 | (´>∀)人(´・ω・)ﾉヽ(・ε・*)人(-д-`) 449 | Σ(ﾟωﾟ) 450 | ヾ(●゜▽゜●)♡ 451 | \(￣◇￣)/ 452 | (o´罒`o) 453 | ╮(╯～╰" )╭ 454 | (σﾟ∀ﾟ)σﾟ∀ﾟ)σﾟ∀ﾟ)σ 455 | （´-`）.｡oO 456 | (ﾉ∀`*) 457 | (๑≖ิิ益≖ิ๑) 458 | ☝( ◠‿◠ )☝ 459 | Σ( ° △ °) 460 | (Π___Π) 461 | ＼(●´ϖ`●)／ 462 | ε(*´･∀･｀)зﾞ 463 | (⁰⊖⁰) 464 | (#ﾟ⊿`)凸 465 | (๑ơ ₃ ơ)♥ 466 | (́=◞౪◟=‵) 467 | ╰(‵□′)╯ 468 | (-`ェ´-╬) 469 | どこ━━━━(゜∀゜三゜∀゜)━━━━!!?? 470 | (;´༎ຶД༎ຶ`) 471 | (｡>﹏<)哈啾 472 | ლ(＾ω＾ლ) 473 | 川‧◇‧川 474 | (╯◕◞౪◟◕‵)╯︵ ┴─┴ 475 | ( ﾟ Χ ﾟ) 476 | (っ●ω●)っ 477 | (╬ﾟ◥益◤ﾟ) 478 | ٩(๑•̀ω•́๑)۶ 479 | ↁ_ↁ 480 | il||li _|￣|○ヽ(･ω･｀) 481 | (´Ａ｀。) 482 | =(─(oo)─)= 483 | (♛‿♛) 484 | ʅ（´◔౪◔）ʃ 485 | m9(＾Д＾)ﾌﾟｷﾞｬｰ 486 | (ÒܫÓױ) 487 | ╮(′～‵〞)╭ 488 | (☄◣ω◢)☄ 489 | ヽ(`Д´)ノ 490 | (╬ﾟ ◣ ﾟ) 491 | (σﾟ∀ﾟ)σ..:*☆ 492 | (◔౪◔) 493 | (#`皿´) 494 | (｡◕∀◕｡) 495 | ヾ(◎´・ω・｀)ノ 496 | ತಎತ 497 | (๑´ㅁ`) 498 | (´・Å・`) 499 | (((ﾟдﾟ))) 500 | ( ´థ౪థ）σ 501 | ─=≡Σ((( つ•̀ω•́)つ 502 | (´;ω;`)ヾ(･∀･`) 503 | （つ> _◕）つ︻╦̵̵͇̿̿̿̿╤─── 504 | (´///☁///`) 505 | ε٩(๑> ₃ <)۶з 506 | Σ(*ﾟдﾟﾉ)ﾉ 507 | (ﾟдﾟ≡ﾟдﾟ) 508 | (́இ◞౪◟இ‵) 509 | ⇎_⇎ 510 | (ﾟ3ﾟ)～♪ 511 | |||||╭(＿◇＿)╮||||| 512 | (✘﹏✘ა) 513 | (´=ิ益=ิ｀) 514 | ლ(•̀ _ •́ ლ) 515 | （*^﹏^*） 516 | (❁´ω`❁)*✲ﾟ* 517 | (っ´ω`c) 518 | ☆^(ｏ´Ф∇Ф)o 519 | ﾚ(ﾟ∀ﾟ;)ﾍ=З=З=З 520 | (*＾ー＾) 521 | (⊜‿⊜✴） 522 | (╯^◇^)╯ 523 | ლ(◉◞౪◟◉ )ლ 524 | இдஇ 525 | ( *‵Д′)--●(#)‵⊿′)╮ 526 | ＼(￣　　　◇,　　　￣)／ 527 | (ㆆᴗㆆ) 528 | (*ﾟーﾟ) 529 | ( ｀皿´)ﾉ"鹵=巛巛巛巛.:;: 530 | (〃￣ω￣)人(￣︶￣〃) 531 | σ`∀´)σ 532 | -{}@{}@{}-υ(￣◇￣*) 533 | (・`ω´・) 534 | (//´◜◞⊜◟◝｀////) 535 | (≧∀≦) 536 | (;´д｀).｡ｏO(・・・・) 537 | ∑(￣□￣;) 538 | (ง๑ •̀_•́)ง 539 | (；´ﾟωﾟ｀人) 540 | ╮/(＞▽<)人(>▽＜)╭ 541 | ╩══╩ ～ ╰(￣△￣╰) 542 | (╭￣3￣)╭♡ 543 | ╮(﹀ω﹀")╭ 544 | ( ՞ټ՞) 545 | _(┐「ε:)_ 546 | ((((<||＿◇＿> 547 | ￣﹃￣ 548 | #ﾟÅﾟ）⊂彡☆))ﾟДﾟ)･∵ 549 | （ ▔︿▔）╭～ 550 | ( ¯•ω•¯ ) 551 | |||╭(＿◇＿)╮||| 552 | ( ﾟÅﾟ) 553 | Σ（ﾟдﾟlll） 554 | (´ﾟдﾟ`) 555 | Σ(・ω・ノ)ノ 556 | ლ(•ω •ლ) 557 | ^(˙oo˙)^ 558 | ヽ(✿ﾟ▽ﾟ)ノ 559 | (´∩ω∩｀) 560 | (*^＠^*) 561 | (・ε・) 562 | ლↂ‿‿ↂლ 563 | (´◉‿◉｀) 564 | ◔"L__◔ 565 | ◝(　ﾟ∀ ﾟ )◟ 566 | q(〒□〒)p 567 | ◢▆▅▄▃崩╰(〒皿〒)╯潰▃▄▅▇◣ 568 | ◢▆▅▄▃ 崩╰(〒皿〒)╯潰 ▃▄▅▆◣ 569 | ε≡ﾍ( ´∀`)ﾉ 570 | (;̦̦̦̦̦̦̦̦ↂ⃝⃓⃙⃚⃘дↂ⃝⃓⃙⃚⃘;̦̦̦̦̦​̦̦̦) 571 | (´σ-`) 572 | (ﾟ皿ﾟﾒ) 573 | (^ρ^)/ 574 | ｡･ﾟ･(ﾉД`)ヽ(ﾟДﾟ )秀秀 575 | ( ఠൠఠ )ﾉ 576 | _:(´□`」 ∠):_ 577 | <(‵□′)> 578 | ლ（´∀`ლ） 579 | (✺ω✺) 580 | (๑•ૅω•´๑) 581 | ┌(┌^o^)┐ 582 | (́순◞౪◟순‵) 583 | (#￣▽￣)o╭╯ ★ 584 | Σ(｀L_｀ ) 585 | ^(￣(oo)￣)^ 586 | ლ(^o^ლ) 587 | (／‵Д′)／~ ╧╧ 588 | (*´Д`)つ))´∀`) 589 | ( ☉_☉)≡☞o────★° 590 | ლ(́◉◞౪◟◉‵ლ) 591 | ┏┛學校┗┓λλλλλ 592 | (*´д`) 593 | (V)(́◉◞౪◟◉‵)(V) 594 | Σ Σ Σ (」○ ω○ )／ 595 | ヾ(´ε`ヾ) 596 | 0(:3　)～ ('､3_ヽ)_ 597 | =└(┐卍^o^)卍 598 | (;ﾟдﾟ) 599 | (*ˇωˇ*人) 600 | ( ♥д♥) 601 | ☮▁▂▃▄☾ ♛ ◡ ♛ ☽▄▃▂▁☮ 602 | /(′□‵)\ 603 | ( º﹃º ) 604 | ( ´థ,_‥థ｀)b 605 | ( Φ ω Φ ) 606 | థ౪థ 607 | (*-*)y~ 608 | (ﾟ∀ﾟ) 609 | |柱|ω・`) 610 | (́つ◞౪◟⊂‵) 611 | ( ´-ω ･)▄︻┻┳══━ 612 | ＼(●o○;)ノ 613 | >(˙ω˙)< 614 | d(d＇∀＇) 615 | (´^ิ益^ิ｀ ) 616 | ( ´･ω) 617 | =_= 618 | \(￣□￣)/ 619 | (∂ω∂) 620 | (థฺˇ౪ˇథ) 621 | (✪ω✪) 622 | (σ ─.─)σ 623 | └|‧◇‧|┐ 624 | |////|　( 　)ﾉ　|////|(自動門 625 | 〠_〠 626 | (`皿´) 627 | 。゜+.(人-ω◕)゜+.゜ 628 | ╮(￣◇￣)╭ 629 | _(√ ζ ε:)_ 630 | (´ー｀)y─┛ 631 | ༼ つ/̵͇̿̿/’̿’̿ ̿ ̿̿ ̿̿◕ _◕ ༽つ/̵͇̿̿/’̿’̿ ̿ ̿̿ ̿̿ ̿̿ 632 | (`д´) 633 | (ノДＴ) 634 | (¯ □ ¯) 635 | /(╯▽╰")\ 636 | (ﾉД`) 637 | (　◜◡‾) 638 | o(─//////─)o 639 | (:3っ)へ ヽ(´Д｀●ヽ) 640 | (#‵)3′)▂▂▂▃▄▅～～～嗡嗡嗡嗡嗡 641 | ☆⌒(*^-゜)v 642 | ╮(￣(oo)￣)╭ 643 | (´≖◞౪◟≖) 644 | (´∀｀∩)↑↑ 645 | ( ˘•ω•˘ )◞⁽˙³˙⁾ 646 | ( ×ω× ) 647 | (#`Д´)ﾉ 648 | ლ(❛◡❛✿)ლ 649 | -//(ǒ.ǒ)//- 650 | (́◉◞౪◟◉‵) 651 | ━( ิ◕㉨◕ ิ)━ 652 | ╮(￣□￣)╭ 653 | (́一◞౪◟一‵) 654 | ヾ(；ﾟ(OO)ﾟ)ﾉ 655 | q(￣ε￣ )p 656 | (́◕◞ㆀ◟◕`) 657 | ╮(╯▽╰)╭ 658 | ⊙谷⊙ 659 | (๑థิ益థิ) 660 | <(‵△′)> 661 | ლ(ﾟдﾟლ) 662 | (╬⓪益⓪) 663 | ♥(´∀` )人 664 | ∪(￣﹁￣)∪ 665 | (･ิ㉨･ิ) 666 | ( ￣ 3￣)y▂ξ 667 | ┏((＝￣(ｴ)￣=))┛ 668 | ≧〔゜゜〕≦ 669 | =͟͟͞͞( •̀д•́) 670 | (，_ゝ۞) 671 | (つД`) 672 | \(￣□￣*)/ 673 | (*─︵─)====Co( 一//////一)o 674 | \(⊙O⊙)/ 675 | σ(‥) 676 | ｡ﾟ(ﾟ´ω`ﾟ)ﾟ｡ 677 | ( ´∀`) 678 | (ゝ∀･)⌒☆ 679 | (`3´) 680 | (　˙灬˙　) 681 | (oﾟωﾟo) 682 | (＊゜ー゜)b 683 | ╰(〒○〒)╯ 684 | (っ﹏-) .｡o 685 | ✧(순_순)" 686 | ╯-____-)╯~═╩════╩═ 687 | (●｀･(ｴ)･´●) 688 | (。◇。||) 689 | ╮(╯_╰)╭ 690 | Σヽ(ﾟД ﾟ; )ﾉ 691 | ( ΄◞ิ .̫.̫ ◟ิ‵)y 692 | (#▔皿▔)─<(*—_____–) 693 | (　っธิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิิ์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์์ლ) 694 | (｡´∀｀)ﾉ―⊂ZZZ⊃ 695 | ( ิΘ㉨Θ)ิ 696 | (´･д･｀)ﾊ(･д･｀*) 697 | (╯°O°)╯┻━┻ 698 | σ ﾟ∀ ﾟ) ﾟ∀ﾟ)σ 699 | (⌒o⌒) 700 | (´-ι_-｀) 701 | ( ・O・)ノ⌒● 702 | *ଘ(੭*ˊᵕˋ)੭* ੈ✩‧₊˚ 703 | (*´∀`)~♥ 704 | (,,・ω・,,) 705 | (゜o゜(☆○=(-_- ) 706 | ( ^ω^) 707 | (〒__〒) 708 | ┏(_д_┏)┓)) 709 | (◓Д◒)✄╰⋃╯ 710 | ヾ(;ﾟ;Д;ﾟ;)ﾉﾞ 711 | (д) ﾟﾟ 712 | | 柱 |┐(￣◇￣*) 713 | o(.〝.)o 714 | (ㄒ◇ㄒ)o～ 715 | ก็ʕ•͡ᴥ•ʔ ก้ 716 | <(‵︵′)> 717 | (╯#▔__▔)╯┴┴\("￣□￣) 718 | ﹏﹏(ㄏ￣△￣)ㄏ 719 | ( ˘･з･) 720 | (╯°▽°)╯ ┻━┻ 721 | \("￣▽￣)/ 722 | (•ө•) 723 | ~(￣▽￣)~ 724 | ( ิ◕㉨◕ ิ) 725 | ┴┴~╰(‵□′#) 726 | (:3[__]4 727 | (›´ω`‹ ) 728 | ( ´Д`)y━･~~ 729 | ◑▂◐ 730 | (°ﾛ°٥) 731 | （・∀・） 732 | (｡ŏ_ŏ) 733 | (ﾉ>ω<)ﾉ 734 | (　ﾟ∀ﾟ) ﾉ♡ 735 | ヽ(●´ε｀●)ノ 736 | •✞_✞• 737 | (ゝ∀･)b 738 | (๑• . •๑) 739 | ◉ื益◉ื 740 | (╯ ‵□′)╯ 741 | (´;;◉;⊖;◉;;｀) 742 | (｡•ㅅ•｡)♡ 743 | (ㅅ˘ㅂ˘) 744 | ٩(๑´3｀๑)۶ 745 | (´ΘωΘ`) 746 | (≖‿ゝ≖)✧ 747 | (￣y▽￣)╭ 748 | ,,Ծ‸Ծ,, 749 | σ(´∀｀*) 750 | (´▽`ʃ♡ƪ) 751 | O-(///￣皿￣)☞ ─═≡☆゜★█▇▆▅▄▃▂＿ 752 | ʕ•̫͡•ʔ 753 | o(‧""‧)o 754 | (｀∀´) 755 | ⁽⁽ ◟(∗ ˊωˋ ∗)◞ ⁾⁾ 756 | (◍•ᴗ•◍)ゝ 757 | (*’ｰ’*) 758 | ( σ՞ਊ ՞)σ 759 | ( ˘•ω•˘ ).oOஇ 760 | (￣□￣||) 761 | (●｀ 艸 ´) 762 | (ಥ_ಥ) 763 | ⋉(● ∸ ●)⋊ 764 | ✧*｡٩(ˊᗜˋ*)و✧*｡ 765 | ( ´•̥̥̥ω•̥̥̥` ) 766 | \(^ω^)/ 767 | 癶(癶;:゜;益;゜;)癶 768 | (●´ω｀●)ゞ 769 | (ｏﾟﾛﾟ)┌┛Σ(ﾉ´*ω*`)ﾉ 770 | \(╯﹏╰) 771 | |柱|ωﾟ)---c<` дﾟ)! 772 | (๑´ㅂ`๑) 773 | （╯‵□′）╯︵┴─┴ 774 | ╮(▔ε▔)╭ 775 | 凸(￣皿￣*) 776 | (ﾟ々｡) 777 | ヾ(>Д<;))))..... 778 | ヾ(:3ﾉｼヾ)ﾉｼ 三[____] 779 | (#￣▽￣#) 780 | <(￣︶￣)/ 781 | <(￣3￣)> 782 | づ(・ω・)づ 783 | い・ω・づ＊ -------------------------------------------------------------------------------- /jseg/emodet.py: -------------------------------------------------------------------------------- 1 | # -*-coding:utf-8-*- 2 | from os.path import abspath, dirname, join 3 | 4 | CUR_PATH = dirname(abspath(__file__)) 5 | 6 | 7 | def load_emo(): 8 | with open(join(CUR_PATH, 'emoall.txt'), encoding="utf-8") as f: 9 | emos = f.read() 10 | emos = emos.split('\n') 11 | emos = [i for i in emos if i != '\n'] 12 | emo_ext = [u'= =', u'=.=', u'=_=', u'>///<', u'> <', u'orz', u'^ ^', u'XD', u'(′･_･`)', u'-_-||', u'>"<', u'T.T'] 13 | emos += emo_ext 14 | emos = list(set(emos)) 15 | emos = [i.strip() for i in emos] 16 | return emos 17 | 18 | 19 | def find_emo(txt): 20 | emos = load_emo() 21 | res = [i for i in emos if i in txt] 22 | return res 23 | -------------------------------------------------------------------------------- /jseg/jieba.py: -------------------------------------------------------------------------------- 1 | # -*- coding: utf-8 -*- 2 | """This is a modified version of "jieba" Chinese segmentator. 3 | 4 | All credit goes to fxsjy/jieba. 5 | Find more on: https://github.com/fxsjy/jieba 6 | """ 7 | from multiprocessing import Pool, cpu_count 8 | from functools import partial 9 | from os.path import dirname, abspath, join 10 | from string import punctuation 11 | from math import log 12 | import pickle 13 | import json 14 | import re 15 | import logging 16 | 17 | import six 18 | 19 | from .emodet import find_emo 20 | 21 | logging.basicConfig(level=logging.DEBUG) 22 | logger = logging.getLogger(__name__) 23 | 24 | 25 | CUR_PATH = dirname(abspath(__file__)) 26 | with open(join(CUR_PATH, 'brill_tagger.pkl'), 'rb') as f: 27 | BrillTagger = pickle.load(f) 28 | 29 | 30 | class Hmm: 31 | """Hidden Markov Model.""" 32 | 33 | def __init__(self): 34 | """Initialize probabilities.""" 35 | self._min_float = -3.14e100 36 | self._prev_status = { 37 | 'B': ('E', 'S'), 38 | 'M': ('M', 'B'), 39 | 'S': ('S', 'E'), 40 | 'E': ('B', 'M') 41 | } 42 | 43 | self._prob = dict() 44 | probs = ['initP', 'tranP', 'emisP'] 45 | for prob in probs: 46 | with open(join(CUR_PATH, 'prob', prob + '.json')) as jf: 47 | self._prob[prob] = json.load(jf) 48 | 49 | def _viterbi(self, obs, states, start_p, trans_p, emit_p): 50 | _v = [{}] # tabular 51 | path = {} 52 | for y in states: # init 53 | _v[0][y] = start_p[y] + emit_p[y].get(obs[0], self._min_float) 54 | path[y] = [y] 55 | for t in range(1, len(obs)): 56 | _v.append({}) 57 | newpath = {} 58 | for y in states: 59 | em_p = emit_p[y].get(obs[t], self._min_float) 60 | (prob, state) = max( 61 | [ 62 | ( 63 | _v[t - 1][y0] + 64 | trans_p[y0].get(y, self._min_float) + 65 | em_p, y0 66 | ) 67 | for y0 in self._prev_status[y] 68 | ] 69 | ) 70 | _v[t][y] = prob 71 | newpath[y] = path[state] + [y] 72 | path = newpath 73 | 74 | (prob, state) = max([(_v[len(obs) - 1][y], y) for y in ('E', 'S')]) 75 | 76 | return (prob, path[state]) 77 | 78 | def hmm_seg(self, sentence): 79 | """Segmentate with HMM method.""" 80 | con = [] 81 | prob, pos_list = self._viterbi( 82 | sentence, ('B', 'M', 'E', 'S'), 83 | self._prob['initP'], 84 | self._prob['tranP'], 85 | self._prob['emisP'] 86 | ) 87 | begin, next = 0, 0 88 | for i, char in enumerate(sentence): 89 | pos = pos_list[i] 90 | if pos == 'B': 91 | begin = i 92 | elif pos == 'E': 93 | con.append(sentence[begin:i + 1]) 94 | next = i + 1 95 | elif pos == 'S': 96 | con.append(char) 97 | next = i + 1 98 | if next < len(sentence): 99 | con.append(sentence[next:]) 100 | return con 101 | 102 | 103 | class Jieba(Hmm): 104 | """Main class.""" 105 | 106 | def __init__(self, load_ptt_dict=False): 107 | """__init__ method. 108 | 109 | :param: load_ptt_dict: load PTT dictionary (default: False) 110 | """ 111 | Hmm.__init__(self) 112 | self._freq = {} 113 | self._trie = {} 114 | self._min_freq = None 115 | self._gw, self._gwr = {}, {} 116 | self._gw_num = 0 117 | 118 | self._gen_trie() 119 | 120 | if load_ptt_dict: 121 | logger.debug('loading ptt dictionary') 122 | with open(join(CUR_PATH, "ptt_encyc.txt"), encoding="utf-8") as tf: 123 | raw = tf.read() 124 | guarantee_wlst = filter(None, raw.split('\n')) 125 | self.add_guaranteed_wordlist(guarantee_wlst) 126 | 127 | def _load_dic(self): 128 | logger.debug('loading default dictionary') 129 | with open(join(CUR_PATH, "dict.txt"), encoding="utf-8") as tf: 130 | raw = tf.read() 131 | dic = re.split('\r{0,1}\n', raw) 132 | dic = filter(None, dic) 133 | # 字典只得有中文 134 | 135 | return dic 136 | 137 | def add_guaranteed_wordlist(self, lst=None): 138 | """User-defined dictoinary. 139 | 140 | :param: lst: list of words. 141 | """ 142 | gw_num = self._gw_num + 1 143 | for num, word in enumerate(lst, gw_num): 144 | self._gw_num = num 145 | self._gw[word] = '_gw' + str(num) + '@' 146 | self._gwr['_gw' + str(num) + '@'] = word 147 | 148 | def _gen_trie(self): 149 | dic = self._load_dic() 150 | total = 0.0 151 | for item in dic: 152 | word, freq = item.split(' ') 153 | freq = float(freq) 154 | self._freq[word] = freq 155 | total += freq 156 | p = self._trie 157 | for char in word: 158 | if char not in p: 159 | p[char] = {} 160 | p = p[char] 161 | p[''] = '' # ending flag 162 | 163 | self._freq = dict( 164 | [(w, log(float(c) / total)) for w, c in self._freq.items()] 165 | ) 166 | self._min_freq = min(self._freq.values()) 167 | 168 | def _get_dag(self, sentence): 169 | senlen = len(sentence) 170 | i, j = 0, 0 171 | p = self._trie 172 | dag = {} 173 | while i < senlen: 174 | c = sentence[j] 175 | if c in p: 176 | p = p[c] 177 | if '' in p: 178 | if i not in dag: 179 | dag[i] = [] 180 | dag[i].append(j) 181 | j += 1 182 | if j >= senlen: 183 | i += 1 184 | j = i 185 | p = self._trie 186 | else: 187 | p = self._trie 188 | i += 1 189 | j = i 190 | for i in range(len(sentence)): 191 | if i not in dag: 192 | dag[i] = [i] 193 | return dag 194 | 195 | def _calc_route(self, sentence, dag, idx): 196 | route = {} 197 | senlen = len(sentence) 198 | route[senlen] = (0.0, '') 199 | for idx in range(senlen - 1, -1, -1): 200 | candidates = [ 201 | ( 202 | self._freq.get( 203 | sentence[idx:x + 1], self._min_freq 204 | ) + route[x + 1][0], 205 | x 206 | ) 207 | for x 208 | in dag[idx] 209 | ] 210 | route[idx] = max(candidates) 211 | return route 212 | 213 | def _seg(self, sentence): # cut DAG or hmm seg 214 | dag = self._get_dag(sentence) 215 | route = self._calc_route(sentence, dag, 0) 216 | dag_con = [] 217 | x = 0 218 | buf = u'' 219 | _n = len(sentence) 220 | while x < _n: 221 | y = route[x][1] + 1 222 | l_word = sentence[x:y] 223 | if y - x == 1: 224 | buf += l_word 225 | else: 226 | if len(buf) > 0: 227 | if len(buf) == 1: 228 | dag_con.append(buf) 229 | buf = u'' 230 | else: 231 | if buf not in self._freq: 232 | regognized = self.hmm_seg(buf) 233 | for t in regognized: 234 | dag_con.append(t) 235 | else: 236 | for elem in buf: 237 | dag_con.append(elem) 238 | buf = u'' 239 | dag_con.append(l_word) 240 | x = y 241 | 242 | if len(buf) > 0: 243 | if len(buf) == 1: 244 | dag_con.append(buf) 245 | else: 246 | if buf not in self._freq: 247 | regognized = self.hmm_seg(buf) 248 | for t in regognized: 249 | dag_con.append(t) 250 | else: 251 | for elem in buf: 252 | dag_con.append(elem) 253 | return dag_con 254 | 255 | def seg(self, text, pos=False): 256 | """Segmentate words. 257 | 258 | :param: text: text to be segmentated 259 | :param: pos: show POS tagging (default: False) 260 | """ 261 | # find emoticons 262 | emos = find_emo(text) 263 | emocan, emocan_r = {}, {} 264 | 265 | for num, emo in enumerate(emos): 266 | emocan[emo] = '_emo' + str(num) + '@' 267 | emocan_r['_emo' + str(num) + '@'] = emo 268 | 269 | for emo in emocan.keys(): 270 | text = text.replace(emo, emocan[emo]) 271 | 272 | # guarantee wlst 273 | # for gw in self._gw.keys(): 274 | gws = self._gw.keys() 275 | gws = sorted(gws, key=lambda x: len(x), reverse=True) # 長詞優先 276 | for gw in gws: 277 | if gw in text: 278 | text = text.replace(gw, self._gw[gw]) 279 | 280 | re_han = re.compile( 281 | r"(http://.*?\s|_gw\d+?@|_emo\d+?@|[ㄅ-ㄩ一-龥]+)", re.U 282 | ) 283 | # re_ch = re.compile(ur"[ㄅ-ㄩ\u4E00-\u9FA5]+") 284 | # re_alphnumeric = re.compile(ur'([a-zA-Z0-9]+)') 285 | re_eng = re.compile(r'[a-zA-Z]+') 286 | re_neu = re.compile(r'[0-9]+') 287 | re_url = re.compile(r'(http://.*?)[\s\n\r\n]') 288 | re_skip = re.compile(r"(\r{0,1}\n|\s)", re.U) 289 | 290 | puns = punctuation 291 | puns += ''.join([six.unichr(ord(i) + 65248) 292 | for i in puns]) # full-width punctuations 293 | puns += u'、。「」…“”' # 要在加上一些常用標點符號 294 | br = u'\r{0,1}\n|\s' 295 | # metachr = u'^$*+?{}[]\|()' 296 | re_pun = re.compile(u'(%s|[%s])' % (br, puns)) 297 | 298 | blocks = re_han.split(text) 299 | 300 | con = [] 301 | for blk in blocks: 302 | if blk.startswith('_gw'): 303 | word = self._gwr[blk] 304 | logger.debug('find in dict: %s' % word) 305 | con.append((word, None)) 306 | 307 | elif blk.startswith('_emo'): 308 | word = emocan_r[blk] 309 | con.append((word, 'EMOTICON')) 310 | elif re_url.match(blk): 311 | con.append((re_url.match(blk).group(1), 'URL')) 312 | # elif re_eng.match(blk): 313 | # con.append((blk, 'FW')) 314 | elif re_han.match(blk): 315 | for word in self._seg(blk): 316 | con.append((word, None)) 317 | else: 318 | tmp = re_pun.split(blk) 319 | for x in tmp: 320 | if x != ' ' and x != '': 321 | if re_skip.match(x): 322 | con.append(('\n', 'LINEBREAK')) 323 | else: 324 | con.append((x, None)) 325 | 326 | con_w, con_p1 = zip(*con) 327 | if pos: 328 | res = BrillTagger.tag(con_w) 329 | 330 | res_con = [] 331 | for word_fin, pos_fin in res: 332 | if pos_fin == '-None-': 333 | if re_pun.match(word_fin): 334 | pos_fin = 'PUNCTUATION' 335 | elif re_eng.match(word_fin): 336 | pos_fin = 'FW' 337 | elif re_neu.match(word_fin): 338 | pos_fin = 'Neu' 339 | res_con.append((word_fin, pos_fin)) 340 | 341 | word_seg = [i[0] for i in res_con] 342 | con_p2 = [i[1] for i in res_con] 343 | 344 | pos_con = [] 345 | for pos_p1, pos_p2 in zip(con_p1, con_p2): 346 | if pos_p1 is None: 347 | pos_out = pos_p2 348 | else: 349 | pos_out = pos_p1 350 | pos_con.append(pos_out) 351 | 352 | output = zip(word_seg, pos_con) 353 | output_mod = [] 354 | for word, pos in output: 355 | output_mod.append((word, pos)) 356 | return output_mod 357 | else: 358 | return con_w 359 | 360 | def _seg_multi(self, texts, pos=False, pool_size=None): 361 | """Use `multiprocessing` to segmentate. 362 | 363 | Notice that this is a experimental method, which seems 364 | to work better with Python3.3+ 365 | :param: texts: list of text 366 | :param: pos: show POS tags (default: False) 367 | :param: pool_size: pool size of `multiprocessing` 368 | """ 369 | import sys 370 | assert (sys.version_info.major == 3 and sys.version_info.minor >= 3) 371 | with Pool(pool_size or cpu_count()) as pool: 372 | output = pool.map(partial(self.seg, pos=pos), texts) 373 | 374 | return output 375 | 376 | 377 | # TODO 378 | # add construction detection 379 | # 標點符號分割？ 380 | # backslash 可能有問題 381 | -------------------------------------------------------------------------------- /jseg/prob/initP.json: -------------------------------------------------------------------------------- 1 | {"M": -3.14e+100, "S": -1.2051185781089053, "B": -0.35618429962370884, "E": -3.14e+100} -------------------------------------------------------------------------------- /jseg/prob/tranP.json: -------------------------------------------------------------------------------- 1 | {"M": {"M": -1.2603623820268226, "E": -0.33344856811948514}, "S": {"S": -0.6658631448798212, "B": -0.7211965654669841}, "B": {"M": -0.916290731874155, "E": -0.51082562376599}, "E": {"S": -0.8085250474669937, "B": -0.5897149736854513}} -------------------------------------------------------------------------------- /jseg/ptt_encyc.txt: -------------------------------------------------------------------------------- 1 | 幫蓋 2 | 筆記 3 | 崩潰 4 | 北車 5 | 撥接魂 6 | 幫高調 7 | 八方雲集 8 | 笨舨板規定九 9 | 不要問！很恐怖！ 10 | 不要問!很恐怖! 11 | 脾氣不好 12 | 沒空 13 | 墨鏡 14 | 媽！我在這 15 | 媽!我在這 16 | 馬路三寶 17 | 沒圖沒真相 18 | 苗栗小五郎 19 | 富奸 20 | ㄈㄈ尺 21 | 反指標 22 | 反詐騙 23 | 反串文 24 | 費雯 25 | 飛鴿傳書 26 | 放大絕 27 | 廢文大賽 28 | 分手擂台 29 | 發文不附圖，此風不可長 30 | 發文不附圖,此風不可長 31 | 大大 32 | 電梯向下 33 | 低調 34 | 打臉 35 | 釣魚文 36 | 地震文 37 | 道歉文 38 | 丹丹之亂 39 | 大同電鍋 40 | 淡定哥 41 | 淡定紅茶 42 | 淡淡的哀傷 43 | 蛋蛋的哀傷 44 | 大師兄黎明 45 | 頭推 46 | 推齊 47 | 台肯 48 | 臺肯 49 | 天龍人 50 | 天龍國 51 | 她，我可以 52 | 她,我可以 53 | 推文朝聖 54 | 鐵手套與藏獒 55 | 天大地大台科大 56 | 腿了 57 | 捏他 58 | 你誰啊，亂公告 59 | 你誰啊,亂公告 60 | 你我都推了一把 61 | 你是在大聲什麼啦 62 | 你嗎知道你在這裡發廢文嗎？ 63 | 你嗎知道你在這裡發廢文嗎? 64 | 娘子，快和牛魔王出來看上帝 65 | 你好糟糕，請暫時不要和我說話 66 | 樓層 67 | 綠爆 68 | 藍爆 69 | 魯蛇 70 | 懶人包 71 | 聯合報 72 | 聯合重工 73 | 來大都找我吧 74 | 李組長眉頭一皺，發覺案情並不單純 75 | 李組長眉頭一皺,發覺案情並不單純 76 | 關燈 77 | 肛肛 78 | 肛肛好 79 | 關鍵字問卦 80 | 公海到了沒？ 81 | 給開司一罐啤酒 82 | 跟鄉民認真就輸了 83 | 趕快推，不然別人會以為我們看不懂 84 | 趕快推,不然別人會以為我們看不懂 85 | 開燈 86 | 科科 87 | 可魯 88 | 空白文 89 | 卡卡獸 90 | 塊陶啊 91 | 靠么！圖咧！ 92 | 靠么!圖咧! 93 | 恐怖連結 94 | 看笨版長知識 95 | 黃爆 96 | 厚問 97 | 好兇 98 | 華肯 99 | 紅的明顯 100 | 還好我都用pcman 101 | 還子的學習不能等 102 | 鹼民 103 | 頸推 104 | 甲甲 105 | 祭品文 106 | 集氣文 107 | 金排球 108 | 救援成功 109 | 鍵盤柯南 110 | 鍵盤小妹 111 | 接龍推文 112 | 記者快來抄 113 | 借轉可愛版 114 | 雞排珍奶板凳 115 | 欺負長ID 116 | 青藍爆 117 | 蛆蛆 118 | 搶頭香 119 | 強暴了 120 | 強者我朋友 121 | 鄉民 122 | 胸推 123 | 匈奴 124 | 瞎了 125 | 宵夜文 126 | 新警察 127 | 下手不知輕重 128 | 謝謝你9257 129 | 先承認你就是你朋友 130 | 先別說這個了，你聽過安麗嗎？ 131 | 先別說這個了,你聽過安麗嗎? 132 | 喜歡你溫暖的文字，洋溢著滿滿的幸福 133 | 喜歡你溫暖的文字,洋溢著滿滿的幸福 134 | 吱吱 135 | 中肯 136 | 炸裂 137 | 週經文 138 | 肘擊她奶子 139 | 這就是人參啊 140 | 正常能量釋放 141 | 這是什麼巫術 142 | 直到我的膝蓋中了一箭 143 | 踹共 144 | 趁亂告白 145 | 出來面對 146 | 吃鍋貼 147 | 純推不下 148 | 神人 149 | 閃光 150 | 舒服文 151 | 十里坡劍神 152 | 數學老ㄙ常請假 153 | 日經文 154 | 人肉搜索 155 | 讓他生讓他生 156 | 讓他爆 157 | 讓他昇 158 | 紫爆 159 | 自介文 160 | 在我床上 161 | 在一起～在一起～ 162 | 在一起~在一起~ 163 | 左邊是超大型巨人，右邊是鎧之巨人 164 | 左邊是超大型巨人,右邊是鎧之巨人 165 | 藏頭文 166 | 酸民 167 | 四海遊龍 168 | 三重劉德華 169 | 喔～喔喔～喔喔～喔喔～爪爪 170 | 喔~喔喔~喔喔~喔喔~爪爪 171 | 養ID 172 | 一行文 173 | 葉佩雯 174 | 顏色不對 175 | 顏色對了 176 | 有下有推 177 | 雅量背影出師表 178 | 有神快拜 179 | 五樓 180 | 溫馨 181 | 溫拿 182 | 我難過 183 | 為看先猜 184 | 勿忘督割 185 | 萬用三熊圖 186 | 我開始相信你了 187 | 萬人響應，一人到場 188 | 萬人響應,一人到場 189 | 月經文 190 | 雨八令刀 191 | 原po是正妹 192 | 56家族 193 | 在你後面，他非常火 194 | 在你後面,他非常火 195 | +1 196 | 689 197 | 609 198 | 30cm 199 | BJ4 200 | bj4 201 | CD中 202 | END 203 | GG 204 | lag了 205 | LUB 206 | OP 207 | puma -------------------------------------------------------------------------------- /jseg/test.py: -------------------------------------------------------------------------------- 1 | # -*-coding:utf-8-*- 2 | from jieba import Jieba 3 | import multi 4 | import time 5 | import logging 6 | logger = logging.getLogger(__name__) 7 | 8 | 9 | text = '''網友瘋傳高市新興分局發給各單位協助慈濟骨髓捐贈者戶籍查訪「交辦單」，但不是說「警務大瘦身、大砍20項業務嗎？」引發基層員警議論紛紛；新興分局表示，交辦時間是17日，在警政署20日宣布簡化警察協辦業務前，未來對於類似服務需求將會視個案狀況審慎處理。 10 | 警政署20日公布20項簡化協辦業務，讓警察回歸專業，網路即瘋傳一張高市新興分局指派各警勤區員警，到骨髓捐贈者的戶籍地查訪、通告「交辦單」；網友們還在交辦單上打上「署長一直在減別人的爛攤子、高雄卻也是一直撿爛攤子」嘲諷引發群起議論。 11 | 不少網友留言：「這種急難救助，不是應該衛生單位訪查嗎？怎會是基層員警？」，也有人認為這太誇張，說「慈濟是收費的，不是免費幫人看病，這不是為民服務，根本是幫慈濟跑腿」；連慈濟義工都看不下，直言「這是我們慈濟人工作，不應交辦給警察，還用他們的警勤時間，做我們的事。」 12 | 新興分局指出，慈濟骨髓幹細胞中心請各縣警察局代為協尋已配對成功、但失去聯繫的骨髓幹細胞志願捐贈者，屬急難救助業務一環，盼結合警政分布系統，在最短時間內找尋該志願者，轉送通知信函，速與慈濟骨髓幹細胞中心聯絡，以辦理捐贈移植手續。 13 | 新興分局表示，尊重基層同仁想法及建議，對此類服務將儘量婉拒，但如遇到有緊急、非迅速找到配對捐贈者無法救命者，考慮人道精神，仍依個案來協助；絕不給基層員警造成任何工作負荷。''' 14 | 15 | texts = [text] * 1000 16 | 17 | j = Jieba() 18 | 19 | 20 | def timeit(method): 21 | 22 | def timed(*args, **kw): 23 | ts = time.time() 24 | result = method(*args, **kw) 25 | te = time.time() 26 | logger.info('%r %f sec' % (method.__name__, te-ts)) 27 | return result 28 | 29 | return timed 30 | 31 | 32 | @timeit 33 | def single_test(): 34 | con = [] 35 | for text in texts: 36 | con.append(j.seg(text)) 37 | 38 | 39 | @timeit 40 | def multi_test(): 41 | multi.seg(texts, 8) 42 | 43 | 44 | if __name__ == '__main__': 45 | single_test() 46 | multi_test() 47 | -------------------------------------------------------------------------------- /pip.txt: -------------------------------------------------------------------------------- 1 | nltk==3.0.3 2 | six==1.11.0 3 | -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- 1 | """Python setup settings.""" 2 | from setuptools import setup, find_packages 3 | 4 | setup( 5 | name='jseg', 6 | version='0.0.4', 7 | packages=find_packages(), 8 | install_requires=[ 9 | 'nltk==3.0.3', 10 | 'six==1.11.0', 11 | ], 12 | author='Aji Liu', 13 | author_email='amigcamel@gmail.com', 14 | description='''A modified version of Jieba''', 15 | package_data={ 16 | 'jseg': [ 17 | 'prob/*', 18 | 'dict.txt', 19 | 'emoall.txt', 20 | 'brill_tagger.pkl' 21 | ] 22 | } 23 | ) 24 | --------------------------------------------------------------------------------