├── .gitignore ├── .rspec ├── .travis.yml ├── .yardopts ├── Gemfile ├── README.md ├── bin └── corpus-processor ├── corpus-processor.gemspec ├── lib ├── corpus-processor.rb └── corpus-processor │ ├── categories.rb │ ├── categories │ └── default.yml │ ├── cli.rb │ ├── generators.rb │ ├── generators │ └── stanford_ner.rb │ ├── parsers.rb │ ├── parsers │ └── lampada.rb │ ├── processor.rb │ ├── token.rb │ └── version.rb └── spec ├── corpus-processor ├── cli_spec.rb ├── generators │ └── stanford_ner_spec.rb ├── parsers │ └── lampada_spec.rb ├── processor_spec.rb └── token_spec.rb └── spec_helper.rb /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/.gitignore -------------------------------------------------------------------------------- /.rspec: -------------------------------------------------------------------------------- 1 | --color 2 | --format progress 3 | -------------------------------------------------------------------------------- /.travis.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/.travis.yml -------------------------------------------------------------------------------- /.yardopts: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/.yardopts -------------------------------------------------------------------------------- /Gemfile: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/Gemfile -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/README.md -------------------------------------------------------------------------------- /bin/corpus-processor: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/bin/corpus-processor -------------------------------------------------------------------------------- /corpus-processor.gemspec: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/corpus-processor.gemspec -------------------------------------------------------------------------------- /lib/corpus-processor.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor.rb -------------------------------------------------------------------------------- /lib/corpus-processor/categories.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/categories.rb -------------------------------------------------------------------------------- /lib/corpus-processor/categories/default.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/categories/default.yml -------------------------------------------------------------------------------- /lib/corpus-processor/cli.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/cli.rb -------------------------------------------------------------------------------- /lib/corpus-processor/generators.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/generators.rb -------------------------------------------------------------------------------- /lib/corpus-processor/generators/stanford_ner.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/generators/stanford_ner.rb -------------------------------------------------------------------------------- /lib/corpus-processor/parsers.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/parsers.rb -------------------------------------------------------------------------------- /lib/corpus-processor/parsers/lampada.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/parsers/lampada.rb -------------------------------------------------------------------------------- /lib/corpus-processor/processor.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/processor.rb -------------------------------------------------------------------------------- /lib/corpus-processor/token.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/lib/corpus-processor/token.rb -------------------------------------------------------------------------------- /lib/corpus-processor/version.rb: -------------------------------------------------------------------------------- 1 | module CorpusProcessor 2 | VERSION = '0.3.0' 3 | end 4 | -------------------------------------------------------------------------------- /spec/corpus-processor/cli_spec.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/corpus-processor/cli_spec.rb -------------------------------------------------------------------------------- /spec/corpus-processor/generators/stanford_ner_spec.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/corpus-processor/generators/stanford_ner_spec.rb -------------------------------------------------------------------------------- /spec/corpus-processor/parsers/lampada_spec.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/corpus-processor/parsers/lampada_spec.rb -------------------------------------------------------------------------------- /spec/corpus-processor/processor_spec.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/corpus-processor/processor_spec.rb -------------------------------------------------------------------------------- /spec/corpus-processor/token_spec.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/corpus-processor/token_spec.rb -------------------------------------------------------------------------------- /spec/spec_helper.rb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/dasdad/corpus-processor/HEAD/spec/spec_helper.rb --------------------------------------------------------------------------------