nltk 3.2.1__zip → 3.3__zip

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (404) hide show
  1. {nltk-3.2.1 → nltk-3.3}/LICENSE.txt +1 -1
  2. {nltk-3.2.1/nltk.egg-info → nltk-3.3}/PKG-INFO +13 -5
  3. nltk-3.3/README.md +49 -0
  4. nltk-3.3/nltk/VERSION +1 -0
  5. {nltk-3.2.1 → nltk-3.3}/nltk/__init__.py +3 -3
  6. {nltk-3.2.1 → nltk-3.3}/nltk/app/__init__.py +2 -3
  7. {nltk-3.2.1 → nltk-3.3}/nltk/app/chartparser_app.py +92 -92
  8. {nltk-3.2.1 → nltk-3.3}/nltk/app/chunkparser_app.py +11 -13
  9. {nltk-3.2.1 → nltk-3.3}/nltk/app/collocations_app.py +8 -11
  10. {nltk-3.2.1 → nltk-3.3}/nltk/app/concordance_app.py +9 -11
  11. {nltk-3.2.1 → nltk-3.3}/nltk/app/nemo_app.py +18 -17
  12. {nltk-3.2.1 → nltk-3.3}/nltk/app/rdparser_app.py +10 -12
  13. {nltk-3.2.1 → nltk-3.3}/nltk/app/srparser_app.py +16 -16
  14. {nltk-3.2.1 → nltk-3.3}/nltk/app/wordfreq_app.py +1 -1
  15. {nltk-3.2.1 → nltk-3.3}/nltk/app/wordnet_app.py +11 -11
  16. {nltk-3.2.1 → nltk-3.3}/nltk/book.py +2 -1
  17. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/__init__.py +1 -1
  18. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/api.py +67 -47
  19. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/chart.py +7 -5
  20. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/combinator.py +42 -11
  21. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/lexicon.py +1 -1
  22. {nltk-3.2.1 → nltk-3.3}/nltk/ccg/logic.py +1 -1
  23. {nltk-3.2.1 → nltk-3.3}/nltk/chat/__init__.py +1 -1
  24. {nltk-3.2.1 → nltk-3.3}/nltk/chat/eliza.py +1 -1
  25. {nltk-3.2.1 → nltk-3.3}/nltk/chat/iesha.py +1 -1
  26. {nltk-3.2.1 → nltk-3.3}/nltk/chat/rude.py +1 -1
  27. {nltk-3.2.1 → nltk-3.3}/nltk/chat/suntsu.py +1 -1
  28. {nltk-3.2.1 → nltk-3.3}/nltk/chat/util.py +12 -10
  29. {nltk-3.2.1 → nltk-3.3}/nltk/chat/zen.py +1 -1
  30. {nltk-3.2.1 → nltk-3.3}/nltk/chunk/__init__.py +1 -1
  31. {nltk-3.2.1 → nltk-3.3}/nltk/chunk/api.py +1 -1
  32. {nltk-3.2.1 → nltk-3.3}/nltk/chunk/named_entity.py +13 -13
  33. {nltk-3.2.1 → nltk-3.3}/nltk/chunk/regexp.py +9 -3
  34. {nltk-3.2.1 → nltk-3.3}/nltk/chunk/util.py +12 -12
  35. {nltk-3.2.1 → nltk-3.3}/nltk/classify/__init__.py +1 -1
  36. {nltk-3.2.1 → nltk-3.3}/nltk/classify/api.py +1 -1
  37. {nltk-3.2.1 → nltk-3.3}/nltk/classify/decisiontree.py +18 -18
  38. {nltk-3.2.1 → nltk-3.3}/nltk/classify/maxent.py +26 -8
  39. {nltk-3.2.1 → nltk-3.3}/nltk/classify/megam.py +5 -4
  40. {nltk-3.2.1 → nltk-3.3}/nltk/classify/naivebayes.py +29 -26
  41. {nltk-3.2.1 → nltk-3.3}/nltk/classify/rte_classify.py +58 -89
  42. {nltk-3.2.1 → nltk-3.3}/nltk/classify/scikitlearn.py +4 -2
  43. {nltk-3.2.1 → nltk-3.3}/nltk/classify/senna.py +24 -22
  44. {nltk-3.2.1 → nltk-3.3}/nltk/classify/svm.py +1 -1
  45. {nltk-3.2.1 → nltk-3.3}/nltk/classify/tadm.py +4 -3
  46. {nltk-3.2.1 → nltk-3.3}/nltk/classify/textcat.py +1 -1
  47. {nltk-3.2.1 → nltk-3.3}/nltk/classify/util.py +13 -1
  48. {nltk-3.2.1 → nltk-3.3}/nltk/classify/weka.py +7 -6
  49. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/__init__.py +1 -1
  50. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/api.py +9 -5
  51. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/em.py +1 -1
  52. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/gaac.py +3 -3
  53. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/kmeans.py +1 -1
  54. {nltk-3.2.1 → nltk-3.3}/nltk/cluster/util.py +35 -20
  55. nltk-3.3/nltk/collections.py +640 -0
  56. {nltk-3.2.1 → nltk-3.3}/nltk/collocations.py +2 -2
  57. nltk-3.3/nltk/compat.py +365 -0
  58. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/__init__.py +37 -16
  59. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/europarl_raw.py +1 -1
  60. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/__init__.py +4 -2
  61. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/aligned.py +4 -3
  62. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/api.py +12 -10
  63. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/bnc.py +1 -1
  64. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/bracket_parse.py +1 -1
  65. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/categorized_sents.py +4 -3
  66. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/chasen.py +5 -4
  67. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/childes.py +8 -9
  68. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/chunked.py +4 -4
  69. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/cmudict.py +4 -2
  70. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/comparative_sents.py +4 -2
  71. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/conll.py +19 -12
  72. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/crubadan.py +2 -2
  73. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/dependency.py +1 -1
  74. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/framenet.py +1085 -119
  75. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/ieer.py +4 -3
  76. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/indian.py +4 -5
  77. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/ipipan.py +6 -5
  78. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/knbc.py +2 -2
  79. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/lin.py +1 -1
  80. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/mte.py +5 -4
  81. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/nkjp.py +4 -3
  82. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/nombank.py +8 -8
  83. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/nps_chat.py +1 -1
  84. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/opinion_lexicon.py +3 -3
  85. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/panlex_lite.py +1 -1
  86. nltk-3.3/nltk/corpus/reader/pl196x.py +292 -0
  87. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/plaintext.py +3 -4
  88. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/ppattach.py +4 -3
  89. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/propbank.py +8 -10
  90. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/pros_cons.py +5 -3
  91. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/reviews.py +3 -1
  92. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/rte.py +5 -7
  93. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/semcor.py +1 -1
  94. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/senseval.py +4 -2
  95. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/sentiwordnet.py +3 -3
  96. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/sinica_treebank.py +1 -1
  97. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/string_category.py +4 -3
  98. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/switchboard.py +1 -1
  99. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/tagged.py +4 -3
  100. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/timit.py +10 -8
  101. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/toolbox.py +4 -4
  102. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/twitter.py +4 -6
  103. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/util.py +7 -3
  104. nltk-3.3/nltk/corpus/reader/verbnet.py +595 -0
  105. nltk-3.3/nltk/corpus/reader/wordlist.py +134 -0
  106. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/wordnet.py +357 -285
  107. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/xmldocs.py +10 -5
  108. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/ycoe.py +25 -24
  109. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/util.py +22 -5
  110. {nltk-3.2.1 → nltk-3.3}/nltk/data.py +65 -29
  111. {nltk-3.2.1 → nltk-3.3}/nltk/decorators.py +5 -1
  112. {nltk-3.2.1 → nltk-3.3}/nltk/downloader.py +59 -55
  113. {nltk-3.2.1 → nltk-3.3}/nltk/draw/__init__.py +4 -4
  114. {nltk-3.2.1 → nltk-3.3}/nltk/draw/cfg.py +5 -8
  115. {nltk-3.2.1 → nltk-3.3}/nltk/draw/dispersion.py +1 -1
  116. {nltk-3.2.1 → nltk-3.3}/nltk/draw/table.py +4 -4
  117. {nltk-3.2.1 → nltk-3.3}/nltk/draw/tree.py +2 -5
  118. {nltk-3.2.1 → nltk-3.3}/nltk/draw/util.py +17 -17
  119. {nltk-3.2.1 → nltk-3.3}/nltk/featstruct.py +290 -173
  120. {nltk-3.2.1 → nltk-3.3}/nltk/grammar.py +87 -44
  121. {nltk-3.2.1 → nltk-3.3}/nltk/help.py +1 -1
  122. {nltk-3.2.1 → nltk-3.3}/nltk/inference/__init__.py +1 -1
  123. {nltk-3.2.1 → nltk-3.3}/nltk/inference/api.py +27 -19
  124. {nltk-3.2.1 → nltk-3.3}/nltk/inference/discourse.py +20 -16
  125. {nltk-3.2.1 → nltk-3.3}/nltk/inference/nonmonotonic.py +1 -1
  126. {nltk-3.2.1 → nltk-3.3}/nltk/inference/prover9.py +1 -1
  127. {nltk-3.2.1 → nltk-3.3}/nltk/inference/resolution.py +1 -1
  128. {nltk-3.2.1 → nltk-3.3}/nltk/inference/tableau.py +1 -1
  129. {nltk-3.2.1 → nltk-3.3}/nltk/internals.py +44 -44
  130. {nltk-3.2.1 → nltk-3.3}/nltk/jsontags.py +1 -1
  131. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/__init__.py +2 -1
  132. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/agreement.py +29 -9
  133. nltk-3.3/nltk/metrics/aline.py +607 -0
  134. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/association.py +8 -4
  135. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/confusionmatrix.py +1 -1
  136. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/distance.py +12 -7
  137. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/paice.py +1 -1
  138. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/scores.py +7 -7
  139. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/segmentation.py +4 -5
  140. {nltk-3.2.1 → nltk-3.3}/nltk/metrics/spearman.py +1 -1
  141. {nltk-3.2.1 → nltk-3.3}/nltk/misc/__init__.py +1 -1
  142. {nltk-3.2.1 → nltk-3.3}/nltk/misc/chomsky.py +4 -2
  143. {nltk-3.2.1 → nltk-3.3}/nltk/misc/minimalset.py +1 -1
  144. {nltk-3.2.1 → nltk-3.3}/nltk/misc/sort.py +1 -1
  145. {nltk-3.2.1 → nltk-3.3}/nltk/misc/wordfinder.py +1 -1
  146. {nltk-3.2.1 → nltk-3.3}/nltk/parse/__init__.py +2 -1
  147. {nltk-3.2.1 → nltk-3.3}/nltk/parse/api.py +1 -1
  148. {nltk-3.2.1 → nltk-3.3}/nltk/parse/bllip.py +1 -1
  149. {nltk-3.2.1 → nltk-3.3}/nltk/parse/chart.py +8 -7
  150. nltk-3.3/nltk/parse/corenlp.py +780 -0
  151. {nltk-3.2.1 → nltk-3.3}/nltk/parse/dependencygraph.py +4 -2
  152. {nltk-3.2.1 → nltk-3.3}/nltk/parse/earleychart.py +4 -3
  153. {nltk-3.2.1 → nltk-3.3}/nltk/parse/evaluate.py +1 -1
  154. {nltk-3.2.1 → nltk-3.3}/nltk/parse/featurechart.py +6 -4
  155. {nltk-3.2.1 → nltk-3.3}/nltk/parse/generate.py +15 -4
  156. {nltk-3.2.1 → nltk-3.3}/nltk/parse/malt.py +2 -2
  157. {nltk-3.2.1 → nltk-3.3}/nltk/parse/nonprojectivedependencyparser.py +3 -3
  158. {nltk-3.2.1 → nltk-3.3}/nltk/parse/pchart.py +1 -1
  159. {nltk-3.2.1 → nltk-3.3}/nltk/parse/projectivedependencyparser.py +6 -5
  160. {nltk-3.2.1 → nltk-3.3}/nltk/parse/recursivedescent.py +2 -3
  161. {nltk-3.2.1 → nltk-3.3}/nltk/parse/shiftreduce.py +2 -3
  162. {nltk-3.2.1 → nltk-3.3}/nltk/parse/stanford.py +53 -28
  163. {nltk-3.2.1 → nltk-3.3}/nltk/parse/transitionparser.py +11 -9
  164. {nltk-3.2.1 → nltk-3.3}/nltk/parse/util.py +1 -1
  165. {nltk-3.2.1 → nltk-3.3}/nltk/parse/viterbi.py +1 -1
  166. {nltk-3.2.1 → nltk-3.3}/nltk/probability.py +75 -20
  167. {nltk-3.2.1 → nltk-3.3}/nltk/sem/__init__.py +1 -1
  168. {nltk-3.2.1 → nltk-3.3}/nltk/sem/boxer.py +15 -15
  169. {nltk-3.2.1 → nltk-3.3}/nltk/sem/chat80.py +6 -6
  170. {nltk-3.2.1 → nltk-3.3}/nltk/sem/cooper_storage.py +1 -1
  171. {nltk-3.2.1 → nltk-3.3}/nltk/sem/drt.py +9 -8
  172. {nltk-3.2.1 → nltk-3.3}/nltk/sem/drt_glue_demo.py +5 -9
  173. {nltk-3.2.1 → nltk-3.3}/nltk/sem/evaluate.py +9 -3
  174. {nltk-3.2.1 → nltk-3.3}/nltk/sem/glue.py +8 -6
  175. {nltk-3.2.1 → nltk-3.3}/nltk/sem/hole.py +4 -2
  176. {nltk-3.2.1 → nltk-3.3}/nltk/sem/lfg.py +4 -2
  177. {nltk-3.2.1 → nltk-3.3}/nltk/sem/linearlogic.py +4 -2
  178. {nltk-3.2.1 → nltk-3.3}/nltk/sem/logic.py +14 -14
  179. {nltk-3.2.1 → nltk-3.3}/nltk/sem/relextract.py +11 -18
  180. {nltk-3.2.1 → nltk-3.3}/nltk/sem/skolemize.py +1 -1
  181. {nltk-3.2.1 → nltk-3.3}/nltk/sem/util.py +1 -1
  182. {nltk-3.2.1 → nltk-3.3}/nltk/sentiment/__init__.py +1 -1
  183. {nltk-3.2.1 → nltk-3.3}/nltk/sentiment/sentiment_analyzer.py +1 -1
  184. {nltk-3.2.1 → nltk-3.3}/nltk/sentiment/util.py +14 -4
  185. {nltk-3.2.1 → nltk-3.3}/nltk/sentiment/vader.py +41 -42
  186. {nltk-3.2.1 → nltk-3.3}/nltk/stem/__init__.py +1 -3
  187. {nltk-3.2.1 → nltk-3.3}/nltk/stem/api.py +7 -4
  188. nltk-3.3/nltk/stem/arlstem.py +369 -0
  189. {nltk-3.2.1 → nltk-3.3}/nltk/stem/isri.py +1 -1
  190. {nltk-3.2.1 → nltk-3.3}/nltk/stem/lancaster.py +39 -12
  191. nltk-3.3/nltk/stem/porter.py +710 -0
  192. {nltk-3.2.1 → nltk-3.3}/nltk/stem/regexp.py +2 -2
  193. {nltk-3.2.1 → nltk-3.3}/nltk/stem/rslp.py +1 -1
  194. {nltk-3.2.1 → nltk-3.3}/nltk/stem/snowball.py +537 -13
  195. {nltk-3.2.1 → nltk-3.3}/nltk/stem/util.py +11 -1
  196. {nltk-3.2.1 → nltk-3.3}/nltk/stem/wordnet.py +1 -1
  197. {nltk-3.2.1 → nltk-3.3}/nltk/tag/__init__.py +32 -7
  198. {nltk-3.2.1 → nltk-3.3}/nltk/tag/api.py +13 -8
  199. {nltk-3.2.1 → nltk-3.3}/nltk/tag/brill.py +4 -6
  200. {nltk-3.2.1 → nltk-3.3}/nltk/tag/crf.py +1 -1
  201. {nltk-3.2.1 → nltk-3.3}/nltk/tag/hmm.py +11 -11
  202. {nltk-3.2.1 → nltk-3.3}/nltk/tag/hunpos.py +4 -4
  203. {nltk-3.2.1 → nltk-3.3}/nltk/tag/mapping.py +1 -1
  204. {nltk-3.2.1 → nltk-3.3}/nltk/tag/perceptron.py +25 -10
  205. {nltk-3.2.1 → nltk-3.3}/nltk/tag/senna.py +25 -23
  206. {nltk-3.2.1 → nltk-3.3}/nltk/tag/sequential.py +19 -15
  207. {nltk-3.2.1 → nltk-3.3}/nltk/tag/stanford.py +59 -36
  208. {nltk-3.2.1 → nltk-3.3}/nltk/tag/tnt.py +11 -20
  209. {nltk-3.2.1 → nltk-3.3}/nltk/tag/util.py +1 -1
  210. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/__init__.py +1 -1
  211. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/demo.py +1 -1
  212. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/erroranalysis.py +1 -1
  213. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/feature.py +5 -7
  214. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/rule.py +6 -6
  215. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/template.py +18 -23
  216. {nltk-3.2.1 → nltk-3.3}/nltk/test/__init__.py +1 -1
  217. {nltk-3.2.1 → nltk-3.3}/nltk/test/all.py +7 -6
  218. {nltk-3.2.1 → nltk-3.3}/nltk/test/bnc.doctest +1 -1
  219. {nltk-3.2.1 → nltk-3.3}/nltk/test/ccg.doctest +1 -1
  220. {nltk-3.2.1 → nltk-3.3}/nltk/test/ccg_semantics.doctest +1 -1
  221. {nltk-3.2.1 → nltk-3.3}/nltk/test/chat80.doctest +1 -1
  222. {nltk-3.2.1 → nltk-3.3}/nltk/test/childes_fixt.py +1 -0
  223. {nltk-3.2.1 → nltk-3.3}/nltk/test/chunk.doctest +1 -1
  224. {nltk-3.2.1 → nltk-3.3}/nltk/test/classify.doctest +1 -1
  225. {nltk-3.2.1 → nltk-3.3}/nltk/test/classify_fixt.py +2 -1
  226. nltk-3.3/nltk/test/collections.doctest +20 -0
  227. {nltk-3.2.1 → nltk-3.3}/nltk/test/collocations.doctest +1 -1
  228. {nltk-3.2.1 → nltk-3.3}/nltk/test/compat.doctest +0 -4
  229. {nltk-3.2.1 → nltk-3.3}/nltk/test/compat_fixt.py +1 -0
  230. nltk-3.3/nltk/test/concordance.doctest +55 -0
  231. {nltk-3.2.1 → nltk-3.3}/nltk/test/corpus.doctest +38 -49
  232. {nltk-3.2.1 → nltk-3.3}/nltk/test/crubadan.doctest +1 -1
  233. {nltk-3.2.1 → nltk-3.3}/nltk/test/data.doctest +1 -1
  234. {nltk-3.2.1 → nltk-3.3}/nltk/test/dependency.doctest +1 -1
  235. {nltk-3.2.1 → nltk-3.3}/nltk/test/discourse.doctest +1 -1
  236. {nltk-3.2.1 → nltk-3.3}/nltk/test/discourse_fixt.py +1 -0
  237. {nltk-3.2.1 → nltk-3.3}/nltk/test/doctest_nose_plugin.py +8 -5
  238. {nltk-3.2.1 → nltk-3.3}/nltk/test/drt.doctest +1 -1
  239. {nltk-3.2.1 → nltk-3.3}/nltk/test/featgram.doctest +1 -1
  240. {nltk-3.2.1 → nltk-3.3}/nltk/test/featstruct.doctest +2 -3
  241. {nltk-3.2.1 → nltk-3.3}/nltk/test/framenet.doctest +82 -33
  242. {nltk-3.2.1 → nltk-3.3}/nltk/test/generate.doctest +1 -1
  243. {nltk-3.2.1 → nltk-3.3}/nltk/test/gensim.doctest +6 -7
  244. {nltk-3.2.1 → nltk-3.3}/nltk/test/gensim_fixt.py +1 -0
  245. {nltk-3.2.1 → nltk-3.3}/nltk/test/gluesemantics.doctest +13 -15
  246. {nltk-3.2.1 → nltk-3.3}/nltk/test/gluesemantics_malt.doctest +1 -1
  247. {nltk-3.2.1 → nltk-3.3}/nltk/test/gluesemantics_malt_fixt.py +1 -0
  248. {nltk-3.2.1 → nltk-3.3}/nltk/test/grammar.doctest +1 -1
  249. {nltk-3.2.1 → nltk-3.3}/nltk/test/grammartestsuites.doctest +1 -1
  250. {nltk-3.2.1 → nltk-3.3}/nltk/test/index.doctest +1 -1
  251. {nltk-3.2.1 → nltk-3.3}/nltk/test/inference.doctest +1 -1
  252. {nltk-3.2.1 → nltk-3.3}/nltk/test/inference_fixt.py +1 -0
  253. {nltk-3.2.1 → nltk-3.3}/nltk/test/internals.doctest +1 -1
  254. {nltk-3.2.1 → nltk-3.3}/nltk/test/japanese.doctest +1 -1
  255. {nltk-3.2.1 → nltk-3.3}/nltk/test/logic.doctest +1 -1
  256. {nltk-3.2.1 → nltk-3.3}/nltk/test/metrics.doctest +3 -3
  257. {nltk-3.2.1 → nltk-3.3}/nltk/test/misc.doctest +1 -1
  258. {nltk-3.2.1 → nltk-3.3}/nltk/test/nonmonotonic.doctest +1 -1
  259. {nltk-3.2.1 → nltk-3.3}/nltk/test/nonmonotonic_fixt.py +1 -0
  260. {nltk-3.2.1 → nltk-3.3}/nltk/test/parse.doctest +1 -1
  261. {nltk-3.2.1 → nltk-3.3}/nltk/test/portuguese_en.doctest +1 -1
  262. {nltk-3.2.1 → nltk-3.3}/nltk/test/probability.doctest +1 -1
  263. {nltk-3.2.1 → nltk-3.3}/nltk/test/probability_fixt.py +2 -1
  264. {nltk-3.2.1 → nltk-3.3}/nltk/test/propbank.doctest +1 -1
  265. {nltk-3.2.1 → nltk-3.3}/nltk/test/relextract.doctest +1 -1
  266. {nltk-3.2.1 → nltk-3.3}/nltk/test/resolution.doctest +1 -1
  267. {nltk-3.2.1 → nltk-3.3}/nltk/test/runtests.py +35 -13
  268. {nltk-3.2.1 → nltk-3.3}/nltk/test/semantics.doctest +1 -1
  269. {nltk-3.2.1 → nltk-3.3}/nltk/test/sentiment.doctest +3 -3
  270. {nltk-3.2.1 → nltk-3.3}/nltk/test/sentiwordnet.doctest +1 -1
  271. {nltk-3.2.1 → nltk-3.3}/nltk/test/simple.doctest +1 -1
  272. {nltk-3.2.1 → nltk-3.3}/nltk/test/stem.doctest +2 -2
  273. {nltk-3.2.1 → nltk-3.3}/nltk/test/tag.doctest +1 -1
  274. {nltk-3.2.1 → nltk-3.3}/nltk/test/tokenize.doctest +109 -2
  275. {nltk-3.2.1 → nltk-3.3}/nltk/test/toolbox.doctest +1 -1
  276. {nltk-3.2.1 → nltk-3.3}/nltk/test/translate.doctest +1 -1
  277. {nltk-3.2.1 → nltk-3.3}/nltk/test/tree.doctest +1 -1
  278. {nltk-3.2.1 → nltk-3.3}/nltk/test/treeprettyprinter.doctest +1 -1
  279. {nltk-3.2.1 → nltk-3.3}/nltk/test/treetransforms.doctest +1 -1
  280. nltk-3.3/nltk/test/unit/test_aline.py +52 -0
  281. nltk-3.3/nltk/test/unit/test_chunk.py +56 -0
  282. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_classify.py +20 -18
  283. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_collocations.py +16 -7
  284. nltk-3.3/nltk/test/unit/test_concordance.py +98 -0
  285. nltk-3.3/nltk/test/unit/test_corenlp.py +1000 -0
  286. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_corpora.py +18 -1
  287. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_json2csv_corpus.py +4 -3
  288. nltk-3.3/nltk/test/unit/test_rte_classify.py +86 -0
  289. nltk-3.3/nltk/test/unit/test_senna.py +74 -0
  290. nltk-3.3/nltk/test/unit/test_stem.py +121 -0
  291. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_tgrep.py +4 -2
  292. nltk-3.3/nltk/test/unit/test_tokenize.py +168 -0
  293. nltk-3.3/nltk/test/unit/test_wordnet.py +134 -0
  294. nltk-3.3/nltk/test/unit/translate/test_bleu.py +244 -0
  295. nltk-3.3/nltk/test/unit/translate/test_gdfa.py +57 -0
  296. nltk-3.3/nltk/test/unit/translate/test_nist.py +37 -0
  297. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_stack_decoder.py +1 -1
  298. {nltk-3.2.1 → nltk-3.3}/nltk/test/util.doctest +1 -1
  299. {nltk-3.2.1 → nltk-3.3}/nltk/test/wordnet.doctest +17 -3
  300. {nltk-3.2.1 → nltk-3.3}/nltk/test/wordnet_fixt.py +1 -0
  301. {nltk-3.2.1 → nltk-3.3}/nltk/test/wordnet_lch.doctest +1 -1
  302. {nltk-3.2.1 → nltk-3.3}/nltk/test/wsd.doctest +1 -1
  303. {nltk-3.2.1 → nltk-3.3}/nltk/text.py +92 -39
  304. {nltk-3.2.1 → nltk-3.3}/nltk/tgrep.py +8 -7
  305. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/__init__.py +37 -15
  306. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/api.py +3 -4
  307. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/casual.py +10 -9
  308. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/mwe.py +2 -2
  309. nltk-3.3/nltk/tokenize/nist.py +173 -0
  310. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/punkt.py +186 -167
  311. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/regexp.py +13 -7
  312. nltk-3.3/nltk/tokenize/repp.py +151 -0
  313. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/sexpr.py +4 -6
  314. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/simple.py +7 -5
  315. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/stanford.py +18 -14
  316. nltk-3.3/nltk/tokenize/stanford_segmenter.py +246 -0
  317. {nltk-3.2.1 → nltk-3.3}/nltk/tokenize/texttiling.py +43 -45
  318. nltk-3.3/nltk/tokenize/toktok.py +155 -0
  319. nltk-3.3/nltk/tokenize/treebank.py +363 -0
  320. nltk-3.3/nltk/tokenize/util.py +261 -0
  321. {nltk-3.2.1 → nltk-3.3}/nltk/toolbox.py +4 -2
  322. {nltk-3.2.1 → nltk-3.3}/nltk/translate/__init__.py +1 -1
  323. {nltk-3.2.1 → nltk-3.3}/nltk/translate/api.py +50 -40
  324. {nltk-3.2.1 → nltk-3.3}/nltk/translate/bleu_score.py +221 -156
  325. nltk-3.3/nltk/translate/chrf_score.py +137 -0
  326. {nltk-3.2.1 → nltk-3.3}/nltk/translate/gale_church.py +32 -19
  327. {nltk-3.2.1 → nltk-3.3}/nltk/translate/gdfa.py +14 -10
  328. nltk-3.3/nltk/translate/gleu_score.py +193 -0
  329. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm1.py +17 -8
  330. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm2.py +4 -4
  331. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm4.py +1 -1
  332. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm5.py +1 -1
  333. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm_model.py +4 -1
  334. {nltk-3.2.1 → nltk-3.3}/nltk/translate/metrics.py +1 -1
  335. nltk-3.3/nltk/translate/nist_score.py +178 -0
  336. {nltk-3.2.1 → nltk-3.3}/nltk/translate/phrase_based.py +1 -1
  337. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ribes_score.py +1 -1
  338. {nltk-3.2.1 → nltk-3.3}/nltk/translate/stack_decoder.py +5 -1
  339. {nltk-3.2.1 → nltk-3.3}/nltk/tree.py +10 -8
  340. {nltk-3.2.1 → nltk-3.3}/nltk/treeprettyprinter.py +6 -6
  341. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/__init__.py +1 -1
  342. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/api.py +8 -3
  343. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/common.py +9 -3
  344. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/twitter_demo.py +1 -1
  345. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/twitterclient.py +1 -1
  346. {nltk-3.2.1 → nltk-3.3}/nltk/twitter/util.py +5 -4
  347. {nltk-3.2.1 → nltk-3.3}/nltk/util.py +41 -654
  348. {nltk-3.2.1 → nltk-3.3}/nltk/wsd.py +2 -2
  349. {nltk-3.2.1 → nltk-3.3/nltk.egg-info}/PKG-INFO +13 -5
  350. {nltk-3.2.1 → nltk-3.3}/nltk.egg-info/SOURCES.txt +23 -1
  351. nltk-3.3/nltk.egg-info/requires.txt +31 -0
  352. {nltk-3.2.1 → nltk-3.3}/setup.cfg +0 -1
  353. {nltk-3.2.1 → nltk-3.3}/setup.py +32 -5
  354. nltk-3.2.1/nltk/VERSION +0 -1
  355. nltk-3.2.1/nltk/compat.py +0 -706
  356. nltk-3.2.1/nltk/corpus/reader/pl196x.py +0 -278
  357. nltk-3.2.1/nltk/corpus/reader/verbnet.py +0 -408
  358. nltk-3.2.1/nltk/corpus/reader/wordlist.py +0 -37
  359. nltk-3.2.1/nltk/six.py +0 -868
  360. nltk-3.2.1/nltk/stem/porter.py +0 -692
  361. nltk-3.2.1/nltk/test/unit/test_stem.py +0 -36
  362. nltk-3.2.1/nltk/test/unit/test_tokenize.py +0 -23
  363. nltk-3.2.1/nltk/test/unit/translate/test_bleu.py +0 -138
  364. nltk-3.2.1/nltk/tokenize/stanford_segmenter.py +0 -157
  365. nltk-3.2.1/nltk/tokenize/treebank.py +0 -124
  366. nltk-3.2.1/nltk/tokenize/util.py +0 -90
  367. {nltk-3.2.1 → nltk-3.3}/INSTALL.txt +0 -0
  368. {nltk-3.2.1 → nltk-3.3}/MANIFEST.in +0 -0
  369. {nltk-3.2.1 → nltk-3.3}/nltk/classify/positivenaivebayes.py +0 -0
  370. {nltk-3.2.1 → nltk-3.3}/nltk/corpus/reader/udhr.py +0 -0
  371. {nltk-3.2.1 → nltk-3.3}/nltk/inference/mace.py +0 -0
  372. {nltk-3.2.1 → nltk-3.3}/nltk/lazyimport.py +0 -0
  373. {nltk-3.2.1 → nltk-3.3}/nltk/misc/babelfish.py +0 -0
  374. {nltk-3.2.1 → nltk-3.3}/nltk/tag/brill_trainer.py +0 -0
  375. {nltk-3.2.1 → nltk-3.3}/nltk/tbl/api.py +0 -0
  376. {nltk-3.2.1 → nltk-3.3}/nltk/test/bleu.doctest +0 -0
  377. {nltk-3.2.1 → nltk-3.3}/nltk/test/childes.doctest +0 -0
  378. {nltk-3.2.1 → nltk-3.3}/nltk/test/corpus_fixt.py +0 -0
  379. {nltk-3.2.1 → nltk-3.3}/nltk/test/paice.doctest +0 -0
  380. {nltk-3.2.1 → nltk-3.3}/nltk/test/portuguese_en_fixt.py +0 -0
  381. {nltk-3.2.1 → nltk-3.3}/nltk/test/segmentation_fixt.py +0 -0
  382. {nltk-3.2.1 → nltk-3.3}/nltk/test/semantics_fixt.py +0 -0
  383. {nltk-3.2.1 → nltk-3.3}/nltk/test/translate_fixt.py +0 -0
  384. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/__init__.py +0 -0
  385. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_2x_compat.py +0 -0
  386. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_corpus_views.py +0 -0
  387. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_hmm.py +0 -0
  388. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_naivebayes.py +0 -0
  389. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_seekable_unicode_stream_reader.py +0 -0
  390. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_tag.py +0 -0
  391. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/test_twitter_auth.py +0 -0
  392. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/__init__.py +0 -0
  393. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm1.py +0 -0
  394. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm2.py +0 -0
  395. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm3.py +0 -0
  396. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm4.py +0 -0
  397. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm5.py +0 -0
  398. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/translate/test_ibm_model.py +0 -0
  399. {nltk-3.2.1 → nltk-3.3}/nltk/test/unit/utils.py +0 -0
  400. {nltk-3.2.1 → nltk-3.3}/nltk/translate/ibm3.py +0 -0
  401. {nltk-3.2.1 → nltk-3.3}/nltk/treetransforms.py +0 -0
  402. {nltk-3.2.1 → nltk-3.3}/nltk.egg-info/dependency_links.txt +0 -0
  403. {nltk-3.2.1 → nltk-3.3}/nltk.egg-info/not-zip-safe +0 -0
  404. {nltk-3.2.1 → nltk-3.3}/nltk.egg-info/top_level.txt +0 -0
@@ -1,278 +0,0 @@
1
- # Natural Language Toolkit:
2
- #
3
- # Copyright (C) 2001-2016 NLTK Project
4
- # Author: Piotr Kasprzyk <p.j.kasprzyk@gmail.com>
5
- # URL: <http://nltk.org/>
6
- # For license information, see LICENSE.TXT
7
-
8
- import os
9
- import re
10
-
11
- from nltk import compat
12
- from nltk import tokenize, tree
13
-
14
- from nltk.corpus.reader.util import *
15
- from nltk.corpus.reader.api import *
16
- from nltk.corpus.reader.xmldocs import XMLCorpusReader
17
-
18
- # (?:something) -- non-capturing parentheses!
19
-
20
- PARA = re.compile(r'<p(?: [^>]*){0,1}>(.*?)</p>')
21
- SENT = re.compile(r'<s(?: [^>]*){0,1}>(.*?)</s>')
22
-
23
- TAGGEDWORD = re.compile(r'<([wc](?: [^>]*){0,1}>)(.*?)</[wc]>')
24
- WORD = re.compile(r'<[wc](?: [^>]*){0,1}>(.*?)</[wc]>')
25
-
26
- TYPE = re.compile(r'type="(.*?)"')
27
- ANA = re.compile(r'ana="(.*?)"')
28
-
29
- TEXTID = re.compile(r'text id="(.*?)"')
30
-
31
-
32
- class TEICorpusView(StreamBackedCorpusView):
33
- def __init__(self, corpus_file,
34
- tagged, group_by_sent, group_by_para,
35
- tagset=None, headLen=0, textids=None):
36
- self._tagged = tagged
37
- self._textids = textids
38
-
39
- self._group_by_sent = group_by_sent
40
- self._group_by_para = group_by_para
41
- # WARNING -- skip header
42
- StreamBackedCorpusView.__init__(self, corpus_file, startpos=headLen)
43
-
44
- _pagesize = 4096
45
-
46
- def read_block(self, stream):
47
- block = stream.readlines(self._pagesize)
48
- block = concat(block)
49
- while (block.count('<text id') > block.count('</text>')) \
50
- or block.count('<text id') == 0:
51
- tmp = stream.readline()
52
- if len(tmp) <= 0:
53
- break
54
- block += tmp
55
-
56
- block = block.replace('\n','')
57
-
58
- textids = TEXTID.findall(block)
59
- if self._textids:
60
- for tid in textids:
61
- if tid not in self._textids:
62
- beg = block.find(tid)-1
63
- end = block[beg: ].find('</text>')+len('</text>')
64
- block = block[ :beg]+block[beg+end: ]
65
-
66
- output = []
67
- for para_str in PARA.findall(block):
68
- para = []
69
- for sent_str in SENT.findall(para_str):
70
- if not self._tagged:
71
- sent = WORD.findall(sent_str)
72
- else:
73
- sent = list(map(self._parse_tag, TAGGEDWORD.findall(sent_str)))
74
- if self._group_by_sent:
75
- para.append(sent)
76
- else:
77
- para.extend(sent)
78
- if self._group_by_para:
79
- output.append(para)
80
- else:
81
- output.extend(para)
82
- return output
83
-
84
- def _parse_tag(self, tag_word_tuple):
85
- (tag, word) = tag_word_tuple
86
- if tag.startswith('w'):
87
- tag = ANA.search(tag).group(1)
88
- else: # tag.startswith('c')
89
- tag = TYPE.search(tag).group(1)
90
- return (word, tag)
91
-
92
-
93
- class Pl196xCorpusReader(CategorizedCorpusReader, XMLCorpusReader):
94
-
95
- headLen = 2770
96
-
97
- def __init__(self, *args, **kwargs):
98
- if 'textid_file' in kwargs: self._textids = kwargs['textid_file']
99
- else: self._textids = None
100
-
101
- XMLCorpusReader.__init__(self, *args)
102
- CategorizedCorpusReader.__init__(self, kwargs)
103
-
104
- self._init_textids()
105
-
106
- def _init_textids(self):
107
- self._f2t = defaultdict(list)
108
- self._t2f = defaultdict(list)
109
- if self._textids is not None:
110
- for line in self.open(self._textids).readlines():
111
- line = line.strip()
112
- file_id, text_ids = line.split(' ', 1)
113
- if file_id not in self.fileids():
114
- raise ValueError('In text_id mapping file %s: %s '
115
- 'not found' % (catfile, file_id))
116
- for text_id in text_ids.split(self._delimiter):
117
- self._add_textids(file_id, text_id)
118
-
119
- def _add_textids(self, file_id, text_id):
120
- self._f2t[file_id].append(text_id)
121
- self._t2f[text_id].append(file_id)
122
-
123
- def _resolve(self, fileids, categories, textids=None):
124
- tmp = None
125
- if fileids is not None:
126
- if not tmp:
127
- tmp = fileids, None
128
- else:
129
- raise ValueError('Specify only fileids, categories or textids')
130
- if categories is not None:
131
- if not tmp:
132
- tmp = self.fileids(categories), None
133
- else:
134
- raise ValueError('Specify only fileids, categories or textids')
135
- if textids is not None:
136
- if not tmp:
137
- if isinstance(textids, compat.string_types): textids = [textids]
138
- files = sum((self._t2f[t] for t in textids), [])
139
- tdict = dict()
140
- for f in files:
141
- tdict[f] = (set(self._f2t[f]) & set(textids))
142
- tmp = files, tdict
143
- else:
144
- raise ValueError('Specify only fileids, categories or textids')
145
- return None, None
146
-
147
- def decode_tag(self, tag):
148
- # to be implemented
149
- return tag
150
-
151
- def textids(self, fileids=None, categories=None):
152
- """
153
- In the pl196x corpus each category is stored in single
154
- file and thus both methods provide identical functionality. In order
155
- to accommodate finer granularity, a non-standard textids() method was
156
- implemented. All the main functions can be supplied with a list
157
- of required chunks---giving much more control to the user.
158
- """
159
- fileids, _ = self._resolve(fileids, categories)
160
- if fileids is None: return sorted(self._t2f)
161
-
162
- if isinstance(fileids, compat.string_types):
163
- fileids = [fileids]
164
- return sorted(sum((self._f2t[d] for d in fileids), []))
165
-
166
- def words(self, fileids=None, categories=None, textids=None):
167
- fileids, textids = self._resolve(fileids, categories, textids)
168
- if fileids is None: fileids = self._fileids
169
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
170
-
171
- if textids:
172
- return concat([TEICorpusView(self.abspath(fileid),
173
- False, False, False,
174
- headLen=self.headLen,
175
- textids=textids[fileid])
176
- for fileid in fileids])
177
- else:
178
- return concat([TEICorpusView(self.abspath(fileid),
179
- False, False, False,
180
- headLen=self.headLen)
181
- for fileid in fileids])
182
-
183
- def sents(self, fileids=None, categories=None, textids=None):
184
- fileids, textids = self._resolve(fileids, categories, textids)
185
- if fileids is None: fileids = self._fileids
186
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
187
-
188
- if textids:
189
- return concat([TEICorpusView(self.abspath(fileid),
190
- False, True, False,
191
- headLen=self.headLen,
192
- textids=textids[fileid])
193
- for fileid in fileids])
194
- else:
195
- return concat([TEICorpusView(self.abspath(fileid),
196
- False, True, False,
197
- headLen=self.headLen)
198
- for fileid in fileids])
199
-
200
- def paras(self, fileids=None, categories=None, textids=None):
201
- fileids, textids = self._resolve(fileids, categories, textids)
202
- if fileids is None: fileids = self._fileids
203
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
204
-
205
- if textids:
206
- return concat([TEICorpusView(self.abspath(fileid),
207
- False, True, True,
208
- headLen=self.headLen,
209
- textids=textids[fileid])
210
- for fileid in fileids])
211
- else:
212
- return concat([TEICorpusView(self.abspath(fileid),
213
- False, True, True,
214
- headLen=self.headLen)
215
- for fileid in fileids])
216
-
217
- def tagged_words(self, fileids=None, categories=None, textids=None):
218
- fileids, textids = self._resolve(fileids, categories, textids)
219
- if fileids is None: fileids = self._fileids
220
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
221
-
222
- if textids:
223
- return concat([TEICorpusView(self.abspath(fileid),
224
- True, False, False,
225
- headLen=self.headLen,
226
- textids=textids[fileid])
227
- for fileid in fileids])
228
- else:
229
- return concat([TEICorpusView(self.abspath(fileid),
230
- True, False, False,
231
- headLen=self.headLen)
232
- for fileid in fileids])
233
-
234
- def tagged_sents(self, fileids=None, categories=None, textids=None):
235
- fileids, textids = self._resolve(fileids, categories, textids)
236
- if fileids is None: fileids = self._fileids
237
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
238
-
239
- if textids:
240
- return concat([TEICorpusView(self.abspath(fileid),
241
- True, True, False,
242
- headLen=self.headLen,
243
- textids=textids[fileid])
244
- for fileid in fileids])
245
- else:
246
- return concat([TEICorpusView(self.abspath(fileid),
247
- True, True, False,
248
- headLen=self.headLen)
249
- for fileid in fileids])
250
-
251
- def tagged_paras(self, fileids=None, categories=None, textids=None):
252
- fileids, textids = self._resolve(fileids, categories, textids)
253
- if fileids is None: fileids = self._fileids
254
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
255
-
256
- if textids:
257
- return concat([TEICorpusView(self.abspath(fileid),
258
- True, True, True,
259
- headLen=self.headLen,
260
- textids=textids[fileid])
261
- for fileid in fileids])
262
- else:
263
- return concat([TEICorpusView(self.abspath(fileid),
264
- True, True, True,
265
- headLen=self.headLen)
266
- for fileid in fileids])
267
-
268
- def xml(self, fileids=None, categories=None):
269
- fileids, _ = self._resolve(fileids, categories)
270
- if len(fileids) == 1: return XMLCorpusReader.xml(self, fileids[0])
271
- else: raise TypeError('Expected a single file')
272
-
273
- def raw(self, fileids=None, categories=None):
274
- fileids, _ = self._resolve(fileids, categories)
275
- if fileids is None: fileids = self._fileids
276
- elif isinstance(fileids, compat.string_types): fileids = [fileids]
277
- return concat([self.open(f).read() for f in fileids])
278
-
@@ -1,408 +0,0 @@
1
- # Natural Language Toolkit: Verbnet Corpus Reader
2
- #
3
- # Copyright (C) 2001-2016 NLTK Project
4
- # Author: Edward Loper <edloper@gmail.com>
5
- # URL: <http://nltk.org/>
6
- # For license information, see LICENSE.TXT
7
-
8
- """
9
- An NLTK interface to the VerbNet verb lexicon
10
-
11
- For details about VerbNet see:
12
- http://verbs.colorado.edu/~mpalmer/projects/verbnet.html
13
- """
14
- from __future__ import unicode_literals
15
-
16
- import re
17
- import textwrap
18
- from collections import defaultdict
19
-
20
- from nltk import compat
21
- from nltk.corpus.reader.xmldocs import XMLCorpusReader
22
-
23
- class VerbnetCorpusReader(XMLCorpusReader):
24
- """
25
- An NLTK interface to the VerbNet verb lexicon.
26
-
27
- From the VerbNet site: "VerbNet (VN) (Kipper-Schuler 2006) is the largest
28
- on-line verb lexicon currently available for English. It is a hierarchical
29
- domain-independent, broad-coverage verb lexicon with mappings to other
30
- lexical resources such as WordNet (Miller, 1990; Fellbaum, 1998), Xtag
31
- (XTAG Research Group, 2001), and FrameNet (Baker et al., 1998)."
32
-
33
- For details about VerbNet see:
34
- http://verbs.colorado.edu/~mpalmer/projects/verbnet.html
35
- """
36
-
37
- # No unicode encoding param, since the data files are all XML.
38
- def __init__(self, root, fileids, wrap_etree=False):
39
- XMLCorpusReader.__init__(self, root, fileids, wrap_etree)
40
-
41
- self._lemma_to_class = defaultdict(list)
42
- """A dictionary mapping from verb lemma strings to lists of
43
- verbnet class identifiers."""
44
-
45
- self._wordnet_to_class = defaultdict(list)
46
- """A dictionary mapping from wordnet identifier strings to
47
- lists of verbnet class identifiers."""
48
-
49
- self._class_to_fileid = {}
50
- """A dictionary mapping from class identifiers to
51
- corresponding file identifiers. The keys of this dictionary
52
- provide a complete list of all classes and subclasses."""
53
-
54
- self._shortid_to_longid = {}
55
-
56
- # Initialize the dictionaries. Use the quick (regexp-based)
57
- # method instead of the slow (xml-based) method, because it
58
- # runs 2-30 times faster.
59
- self._quick_index()
60
-
61
- _LONGID_RE = re.compile(r'([^\-\.]*)-([\d+.\-]+)$')
62
- """Regular expression that matches (and decomposes) longids"""
63
-
64
- _SHORTID_RE = re.compile(r'[\d+.\-]+$')
65
- """Regular expression that matches shortids"""
66
-
67
- _INDEX_RE = re.compile(r'<MEMBER name="\??([^"]+)" wn="([^"]*)"[^>]+>|'
68
- r'<VNSUBCLASS ID="([^"]+)"/?>')
69
- """Regular expression used by ``_index()`` to quickly scan the corpus
70
- for basic information."""
71
-
72
- def lemmas(self, classid=None):
73
- """
74
- Return a list of all verb lemmas that appear in any class, or
75
- in the ``classid`` if specified.
76
- """
77
- if classid is None:
78
- return sorted(self._lemma_to_class.keys())
79
- else:
80
- # [xx] should this include subclass members?
81
- vnclass = self.vnclass(classid)
82
- return [member.get('name') for member in
83
- vnclass.findall('MEMBERS/MEMBER')]
84
-
85
- def wordnetids(self, classid=None):
86
- """
87
- Return a list of all wordnet identifiers that appear in any
88
- class, or in ``classid`` if specified.
89
- """
90
- if classid is None:
91
- return sorted(self._wordnet_to_class.keys())
92
- else:
93
- # [xx] should this include subclass members?
94
- vnclass = self.vnclass(classid)
95
- return sum([member.get('wn','').split() for member in
96
- vnclass.findall('MEMBERS/MEMBER')], [])
97
-
98
- def classids(self, lemma=None, wordnetid=None, fileid=None, classid=None):
99
- """
100
- Return a list of the verbnet class identifiers. If a file
101
- identifier is specified, then return only the verbnet class
102
- identifiers for classes (and subclasses) defined by that file.
103
- If a lemma is specified, then return only verbnet class
104
- identifiers for classes that contain that lemma as a member.
105
- If a wordnetid is specified, then return only identifiers for
106
- classes that contain that wordnetid as a member. If a classid
107
- is specified, then return only identifiers for subclasses of
108
- the specified verbnet class.
109
- """
110
- if len([x for x in [lemma, wordnetid, fileid, classid]
111
- if x is not None]) > 1:
112
- raise ValueError('Specify at most one of: fileid, wordnetid, '
113
- 'fileid, classid')
114
- if fileid is not None:
115
- return [c for (c,f) in self._class_to_fileid.items()
116
- if f == fileid]
117
- elif lemma is not None:
118
- return self._lemma_to_class[lemma]
119
- elif wordnetid is not None:
120
- return self._wordnet_to_class[wordnetid]
121
- elif classid is not None:
122
- xmltree = self.vnclass(classid)
123
- return [subclass.get('ID') for subclass in
124
- xmltree.findall('SUBCLASSES/VNSUBCLASS')]
125
- else:
126
- return sorted(self._class_to_fileid.keys())
127
-
128
- def vnclass(self, fileid_or_classid):
129
- """
130
- Return an ElementTree containing the xml for the specified
131
- verbnet class.
132
-
133
- :param fileid_or_classid: An identifier specifying which class
134
- should be returned. Can be a file identifier (such as
135
- ``'put-9.1.xml'``), or a verbnet class identifier (such as
136
- ``'put-9.1'``) or a short verbnet class identifier (such as
137
- ``'9.1'``).
138
- """
139
- # File identifier: just return the xml.
140
- if fileid_or_classid in self._fileids:
141
- return self.xml(fileid_or_classid)
142
-
143
- # Class identifier: get the xml, and find the right elt.
144
- classid = self.longid(fileid_or_classid)
145
- if classid in self._class_to_fileid:
146
- fileid = self._class_to_fileid[self.longid(classid)]
147
- tree = self.xml(fileid)
148
- if classid == tree.get('ID'):
149
- return tree
150
- else:
151
- for subclass in tree.findall('.//VNSUBCLASS'):
152
- if classid == subclass.get('ID'):
153
- return subclass
154
- else:
155
- assert False # we saw it during _index()!
156
-
157
- else:
158
- raise ValueError('Unknown identifier %s' % fileid_or_classid)
159
-
160
- def fileids(self, vnclass_ids=None):
161
- """
162
- Return a list of fileids that make up this corpus. If
163
- ``vnclass_ids`` is specified, then return the fileids that make
164
- up the specified verbnet class(es).
165
- """
166
- if vnclass_ids is None:
167
- return self._fileids
168
- elif isinstance(vnclass_ids, compat.string_types):
169
- return [self._class_to_fileid[self.longid(vnclass_ids)]]
170
- else:
171
- return [self._class_to_fileid[self.longid(vnclass_id)]
172
- for vnclass_id in vnclass_ids]
173
-
174
-
175
- ######################################################################
176
- #{ Index Initialization
177
- ######################################################################
178
-
179
- def _index(self):
180
- """
181
- Initialize the indexes ``_lemma_to_class``,
182
- ``_wordnet_to_class``, and ``_class_to_fileid`` by scanning
183
- through the corpus fileids. This is fast with cElementTree
184
- (<0.1 secs), but quite slow (>10 secs) with the python
185
- implementation of ElementTree.
186
- """
187
- for fileid in self._fileids:
188
- self._index_helper(self.xml(fileid), fileid)
189
-
190
- def _index_helper(self, xmltree, fileid):
191
- """Helper for ``_index()``"""
192
- vnclass = xmltree.get('ID')
193
- self._class_to_fileid[vnclass] = fileid
194
- self._shortid_to_longid[self.shortid(vnclass)] = vnclass
195
- for member in xmltree.findall('MEMBERS/MEMBER'):
196
- self._lemma_to_class[member.get('name')].append(vnclass)
197
- for wn in member.get('wn', '').split():
198
- self._wordnet_to_class[wn].append(vnclass)
199
- for subclass in xmltree.findall('SUBCLASSES/VNSUBCLASS'):
200
- self._index_helper(subclass, fileid)
201
-
202
- def _quick_index(self):
203
- """
204
- Initialize the indexes ``_lemma_to_class``,
205
- ``_wordnet_to_class``, and ``_class_to_fileid`` by scanning
206
- through the corpus fileids. This doesn't do proper xml parsing,
207
- but is good enough to find everything in the standard verbnet
208
- corpus -- and it runs about 30 times faster than xml parsing
209
- (with the python ElementTree; only 2-3 times faster with
210
- cElementTree).
211
- """
212
- # nb: if we got rid of wordnet_to_class, this would run 2-3
213
- # times faster.
214
- for fileid in self._fileids:
215
- vnclass = fileid[:-4] # strip the '.xml'
216
- self._class_to_fileid[vnclass] = fileid
217
- self._shortid_to_longid[self.shortid(vnclass)] = vnclass
218
- for m in self._INDEX_RE.finditer(self.open(fileid).read()):
219
- groups = m.groups()
220
- if groups[0] is not None:
221
- self._lemma_to_class[groups[0]].append(vnclass)
222
- for wn in groups[1].split():
223
- self._wordnet_to_class[wn].append(vnclass)
224
- elif groups[2] is not None:
225
- self._class_to_fileid[groups[2]] = fileid
226
- vnclass = groups[2] # for <MEMBER> elts.
227
- self._shortid_to_longid[self.shortid(vnclass)] = vnclass
228
- else:
229
- assert False, 'unexpected match condition'
230
-
231
- ######################################################################
232
- #{ Identifier conversion
233
- ######################################################################
234
-
235
- def longid(self, shortid):
236
- """Given a short verbnet class identifier (eg '37.10'), map it
237
- to a long id (eg 'confess-37.10'). If ``shortid`` is already a
238
- long id, then return it as-is"""
239
- if self._LONGID_RE.match(shortid):
240
- return shortid # it's already a longid.
241
- elif not self._SHORTID_RE.match(shortid):
242
- raise ValueError('vnclass identifier %r not found' % shortid)
243
- try:
244
- return self._shortid_to_longid[shortid]
245
- except KeyError:
246
- raise ValueError('vnclass identifier %r not found' % shortid)
247
-
248
- def shortid(self, longid):
249
- """Given a long verbnet class identifier (eg 'confess-37.10'),
250
- map it to a short id (eg '37.10'). If ``longid`` is already a
251
- short id, then return it as-is."""
252
- if self._SHORTID_RE.match(longid):
253
- return longid # it's already a shortid.
254
- m = self._LONGID_RE.match(longid)
255
- if m:
256
- return m.group(2)
257
- else:
258
- raise ValueError('vnclass identifier %r not found' % longid)
259
-
260
- ######################################################################
261
- #{ Pretty Printing
262
- ######################################################################
263
-
264
- def pprint(self, vnclass):
265
- """
266
- Return a string containing a pretty-printed representation of
267
- the given verbnet class.
268
-
269
- :param vnclass: A verbnet class identifier; or an ElementTree
270
- containing the xml contents of a verbnet class.
271
- """
272
- if isinstance(vnclass, compat.string_types):
273
- vnclass = self.vnclass(vnclass)
274
-
275
- s = vnclass.get('ID') + '\n'
276
- s += self.pprint_subclasses(vnclass, indent=' ') + '\n'
277
- s += self.pprint_members(vnclass, indent=' ') + '\n'
278
- s += ' Thematic roles:\n'
279
- s += self.pprint_themroles(vnclass, indent=' ') + '\n'
280
- s += ' Frames:\n'
281
- s += '\n'.join(self.pprint_frame(vnframe, indent=' ')
282
- for vnframe in vnclass.findall('FRAMES/FRAME'))
283
- return s
284
-
285
- def pprint_subclasses(self, vnclass, indent=''):
286
- """
287
- Return a string containing a pretty-printed representation of
288
- the given verbnet class's subclasses.
289
-
290
- :param vnclass: A verbnet class identifier; or an ElementTree
291
- containing the xml contents of a verbnet class.
292
- """
293
- if isinstance(vnclass, compat.string_types):
294
- vnclass = self.vnclass(vnclass)
295
-
296
- subclasses = [subclass.get('ID') for subclass in
297
- vnclass.findall('SUBCLASSES/VNSUBCLASS')]
298
- if not subclasses: subclasses = ['(none)']
299
- s = 'Subclasses: ' + ' '.join(subclasses)
300
- return textwrap.fill(s, 70, initial_indent=indent,
301
- subsequent_indent=indent+' ')
302
-
303
- def pprint_members(self, vnclass, indent=''):
304
- """
305
- Return a string containing a pretty-printed representation of
306
- the given verbnet class's member verbs.
307
-
308
- :param vnclass: A verbnet class identifier; or an ElementTree
309
- containing the xml contents of a verbnet class.
310
- """
311
- if isinstance(vnclass, compat.string_types):
312
- vnclass = self.vnclass(vnclass)
313
-
314
- members = [member.get('name') for member in
315
- vnclass.findall('MEMBERS/MEMBER')]
316
- if not members: members = ['(none)']
317
- s = 'Members: ' + ' '.join(members)
318
- return textwrap.fill(s, 70, initial_indent=indent,
319
- subsequent_indent=indent+' ')
320
-
321
- def pprint_themroles(self, vnclass, indent=''):
322
- """
323
- Return a string containing a pretty-printed representation of
324
- the given verbnet class's thematic roles.
325
-
326
- :param vnclass: A verbnet class identifier; or an ElementTree
327
- containing the xml contents of a verbnet class.
328
- """
329
- if isinstance(vnclass, compat.string_types):
330
- vnclass = self.vnclass(vnclass)
331
-
332
- pieces = []
333
- for themrole in vnclass.findall('THEMROLES/THEMROLE'):
334
- piece = indent + '* ' + themrole.get('type')
335
- modifiers = ['%(Value)s%(type)s' % restr.attrib
336
- for restr in themrole.findall('SELRESTRS/SELRESTR')]
337
- if modifiers:
338
- piece += '[%s]' % ' '.join(modifiers)
339
- pieces.append(piece)
340
-
341
- return '\n'.join(pieces)
342
-
343
- def pprint_frame(self, vnframe, indent=''):
344
- """
345
- Return a string containing a pretty-printed representation of
346
- the given verbnet frame.
347
-
348
- :param vnframe: An ElementTree containing the xml contents of
349
- a verbnet frame.
350
- """
351
- s = self.pprint_description(vnframe, indent) + '\n'
352
- s += self.pprint_syntax(vnframe, indent+' Syntax: ') + '\n'
353
- s += indent + ' Semantics:\n'
354
- s += self.pprint_semantics(vnframe, indent+' ')
355
- return s
356
-
357
- def pprint_description(self, vnframe, indent=''):
358
- """
359
- Return a string containing a pretty-printed representation of
360
- the given verbnet frame description.
361
-
362
- :param vnframe: An ElementTree containing the xml contents of
363
- a verbnet frame.
364
- """
365
- descr = vnframe.find('DESCRIPTION')
366
- s = indent + descr.attrib['primary']
367
- if descr.get('secondary', ''):
368
- s += ' (%s)' % descr.get('secondary')
369
- return s
370
-
371
- def pprint_syntax(self, vnframe, indent=''):
372
- """
373
- Return a string containing a pretty-printed representation of
374
- the given verbnet frame syntax.
375
-
376
- :param vnframe: An ElementTree containing the xml contents of
377
- a verbnet frame.
378
- """
379
- pieces = []
380
- for elt in vnframe.find('SYNTAX'):
381
- piece = elt.tag
382
- modifiers = []
383
- if 'value' in elt.attrib:
384
- modifiers.append(elt.get('value'))
385
- modifiers += ['%(Value)s%(type)s' % restr.attrib
386
- for restr in (elt.findall('SELRESTRS/SELRESTR') +
387
- elt.findall('SYNRESTRS/SYNRESTR'))]
388
- if modifiers:
389
- piece += '[%s]' % ' '.join(modifiers)
390
- pieces.append(piece)
391
-
392
- return indent + ' '.join(pieces)
393
-
394
- def pprint_semantics(self, vnframe, indent=''):
395
- """
396
- Return a string containing a pretty-printed representation of
397
- the given verbnet frame semantics.
398
-
399
- :param vnframe: An ElementTree containing the xml contents of
400
- a verbnet frame.
401
- """
402
- pieces = []
403
- for pred in vnframe.findall('SEMANTICS/PRED'):
404
- args = [arg.get('value') for arg in pred.findall('ARGS/ARG')]
405
- pieces.append('%s(%s)' % (pred.get('value'), ', '.join(args)))
406
- return '\n'.join('%s* %s' % (indent, piece) for piece in pieces)
407
-
408
-