natural 0.2.0 → 0.2.1

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 (382) hide show
  1. package/.editorconfig +7 -0
  2. package/README.md +3 -3
  3. package/lib/natural/inflectors/present_verb_inflector.js +1 -0
  4. package/lib/natural/phonetics/double_metaphone.js +9 -4
  5. package/lib/natural/phonetics/metaphone.js +4 -0
  6. package/lib/natural/phonetics/soundex.js +11 -7
  7. package/lib/natural/stemmers/porter_stemmer.js +90 -62
  8. package/lib/natural/stemmers/porter_stemmer_pt.js +302 -0
  9. package/lib/natural/stemmers/stemmer_pt.js +65 -0
  10. package/lib/natural/stemmers/token.js +163 -0
  11. package/lib/natural/util/stopwords_pt.js +136 -0
  12. package/package.json +2 -2
  13. package/spec/double_metaphone_spec.js +10 -1
  14. package/spec/metaphone_spec.js +5 -0
  15. package/spec/porter_stemmer_pt_spec.js +50 -0
  16. package/spec/porter_stemmer_spec.js +45 -0
  17. package/spec/present_verb_inflector_spec.js +3 -0
  18. package/spec/soundex_spec.js +13 -9
  19. package/spec/stemmer_token_spec.js +161 -0
  20. package/spec/test_data/snowball_pt.txt +32016 -0
  21. package/coverage/coverage.json +0 -1
  22. package/coverage/lcov-report/index.html +0 -584
  23. package/coverage/lcov-report/io_spec/index.html +0 -350
  24. package/coverage/lcov-report/io_spec/tfidf_spec.js.html +0 -469
  25. package/coverage/lcov-report/lib/natural/analyzers/index.html +0 -350
  26. package/coverage/lcov-report/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
  27. package/coverage/lcov-report/lib/natural/classifiers/bayes_classifier.js.html +0 -496
  28. package/coverage/lcov-report/lib/natural/classifiers/classifier.js.html +0 -823
  29. package/coverage/lcov-report/lib/natural/classifiers/index.html +0 -376
  30. package/coverage/lcov-report/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
  31. package/coverage/lcov-report/lib/natural/distance/dice_coefficient.js.html +0 -529
  32. package/coverage/lcov-report/lib/natural/distance/index.html +0 -376
  33. package/coverage/lcov-report/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
  34. package/coverage/lcov-report/lib/natural/distance/levenshtein_distance.js.html +0 -526
  35. package/coverage/lcov-report/lib/natural/index.html +0 -350
  36. package/coverage/lcov-report/lib/natural/index.js.html +0 -547
  37. package/coverage/lcov-report/lib/natural/inflectors/count_inflector.js.html +0 -484
  38. package/coverage/lcov-report/lib/natural/inflectors/form_set.js.html +0 -406
  39. package/coverage/lcov-report/lib/natural/inflectors/fr/count_inflector.js.html +0 -520
  40. package/coverage/lcov-report/lib/natural/inflectors/fr/index.html +0 -363
  41. package/coverage/lcov-report/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
  42. package/coverage/lcov-report/lib/natural/inflectors/index.html +0 -402
  43. package/coverage/lcov-report/lib/natural/inflectors/ja/index.html +0 -350
  44. package/coverage/lcov-report/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
  45. package/coverage/lcov-report/lib/natural/inflectors/noun_inflector.js.html +0 -637
  46. package/coverage/lcov-report/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
  47. package/coverage/lcov-report/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
  48. package/coverage/lcov-report/lib/natural/ngrams/index.html +0 -363
  49. package/coverage/lcov-report/lib/natural/ngrams/ngrams.js.html +0 -595
  50. package/coverage/lcov-report/lib/natural/ngrams/ngrams_zh.js.html +0 -571
  51. package/coverage/lcov-report/lib/natural/normalizers/index.html +0 -389
  52. package/coverage/lcov-report/lib/natural/normalizers/normalizer.js.html +0 -607
  53. package/coverage/lcov-report/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
  54. package/coverage/lcov-report/lib/natural/normalizers/normalizer_no.js.html +0 -496
  55. package/coverage/lcov-report/lib/natural/normalizers/remove_diacritics.js.html +0 -685
  56. package/coverage/lcov-report/lib/natural/phonetics/dm_soundex.js.html +0 -1063
  57. package/coverage/lcov-report/lib/natural/phonetics/double_metaphone.js.html +0 -1840
  58. package/coverage/lcov-report/lib/natural/phonetics/index.html +0 -402
  59. package/coverage/lcov-report/lib/natural/phonetics/metaphone.js.html +0 -886
  60. package/coverage/lcov-report/lib/natural/phonetics/phonetic.js.html +0 -484
  61. package/coverage/lcov-report/lib/natural/phonetics/soundex.js.html +0 -562
  62. package/coverage/lcov-report/lib/natural/spellcheck/index.html +0 -350
  63. package/coverage/lcov-report/lib/natural/spellcheck/spellcheck.js.html +0 -562
  64. package/coverage/lcov-report/lib/natural/stemmers/index.html +0 -571
  65. package/coverage/lcov-report/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
  66. package/coverage/lcov-report/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
  67. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer.js.html +0 -937
  68. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
  69. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
  70. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
  71. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
  72. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
  73. package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
  74. package/coverage/lcov-report/lib/natural/stemmers/stemmer.js.html +0 -505
  75. package/coverage/lcov-report/lib/natural/stemmers/stemmer_es.js.html +0 -496
  76. package/coverage/lcov-report/lib/natural/stemmers/stemmer_fa.js.html +0 -484
  77. package/coverage/lcov-report/lib/natural/stemmers/stemmer_fr.js.html +0 -496
  78. package/coverage/lcov-report/lib/natural/stemmers/stemmer_it.js.html +0 -427
  79. package/coverage/lcov-report/lib/natural/stemmers/stemmer_ja.js.html +0 -736
  80. package/coverage/lcov-report/lib/natural/stemmers/stemmer_no.js.html +0 -505
  81. package/coverage/lcov-report/lib/natural/stemmers/stemmer_pl.js.html +0 -496
  82. package/coverage/lcov-report/lib/natural/stemmers/stemmer_ru.js.html +0 -496
  83. package/coverage/lcov-report/lib/natural/tfidf/index.html +0 -350
  84. package/coverage/lcov-report/lib/natural/tfidf/tfidf.js.html +0 -904
  85. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
  86. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
  87. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
  88. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
  89. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
  90. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
  91. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
  92. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
  93. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
  94. package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
  95. package/coverage/lcov-report/lib/natural/tokenizers/index.html +0 -519
  96. package/coverage/lcov-report/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
  97. package/coverage/lcov-report/lib/natural/tokenizers/tokenizer.js.html +0 -475
  98. package/coverage/lcov-report/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
  99. package/coverage/lcov-report/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
  100. package/coverage/lcov-report/lib/natural/transliterators/ja/index.html +0 -350
  101. package/coverage/lcov-report/lib/natural/transliterators/ja/index.js.html +0 -2068
  102. package/coverage/lcov-report/lib/natural/trie/index.html +0 -350
  103. package/coverage/lcov-report/lib/natural/trie/trie.js.html +0 -1012
  104. package/coverage/lcov-report/lib/natural/util/bag.js.html +0 -472
  105. package/coverage/lcov-report/lib/natural/util/edge_weighted_digraph.js.html +0 -658
  106. package/coverage/lcov-report/lib/natural/util/index.html +0 -532
  107. package/coverage/lcov-report/lib/natural/util/longest_path_tree.js.html +0 -619
  108. package/coverage/lcov-report/lib/natural/util/shortest_path_tree.js.html +0 -622
  109. package/coverage/lcov-report/lib/natural/util/stopwords.js.html +0 -445
  110. package/coverage/lcov-report/lib/natural/util/stopwords_es.js.html +0 -430
  111. package/coverage/lcov-report/lib/natural/util/stopwords_fa.js.html +0 -436
  112. package/coverage/lcov-report/lib/natural/util/stopwords_fr.js.html +0 -904
  113. package/coverage/lcov-report/lib/natural/util/stopwords_it.js.html +0 -478
  114. package/coverage/lcov-report/lib/natural/util/stopwords_ja.js.html +0 -499
  115. package/coverage/lcov-report/lib/natural/util/stopwords_no.js.html +0 -445
  116. package/coverage/lcov-report/lib/natural/util/stopwords_pl.js.html +0 -508
  117. package/coverage/lcov-report/lib/natural/util/stopwords_ru.js.html +0 -445
  118. package/coverage/lcov-report/lib/natural/util/topological.js.html +0 -616
  119. package/coverage/lcov-report/lib/natural/util/utils.js.html +0 -679
  120. package/coverage/lcov-report/lib/natural/wordnet/data_file.js.html +0 -580
  121. package/coverage/lcov-report/lib/natural/wordnet/index.html +0 -389
  122. package/coverage/lcov-report/lib/natural/wordnet/index_file.js.html +0 -742
  123. package/coverage/lcov-report/lib/natural/wordnet/wordnet.js.html +0 -829
  124. package/coverage/lcov-report/lib/natural/wordnet/wordnet_file.js.html +0 -529
  125. package/coverage/lcov-report/natural/analyzers/index.html +0 -350
  126. package/coverage/lcov-report/natural/analyzers/sentence_analyzer.js.html +0 -850
  127. package/coverage/lcov-report/natural/classifiers/bayes_classifier.js.html +0 -496
  128. package/coverage/lcov-report/natural/classifiers/classifier.js.html +0 -823
  129. package/coverage/lcov-report/natural/classifiers/index.html +0 -376
  130. package/coverage/lcov-report/natural/classifiers/logistic_regression_classifier.js.html +0 -502
  131. package/coverage/lcov-report/natural/distance/dice_coefficient.js.html +0 -529
  132. package/coverage/lcov-report/natural/distance/index.html +0 -376
  133. package/coverage/lcov-report/natural/distance/jaro-winkler_distance.js.html +0 -670
  134. package/coverage/lcov-report/natural/distance/levenshtein_distance.js.html +0 -526
  135. package/coverage/lcov-report/natural/index.html +0 -350
  136. package/coverage/lcov-report/natural/index.js.html +0 -433
  137. package/coverage/lcov-report/natural/inflectors/count_inflector.js.html +0 -484
  138. package/coverage/lcov-report/natural/inflectors/form_set.js.html +0 -406
  139. package/coverage/lcov-report/natural/inflectors/fr/index.html +0 -350
  140. package/coverage/lcov-report/natural/inflectors/fr/noun_inflector.js.html +0 -1027
  141. package/coverage/lcov-report/natural/inflectors/index.html +0 -402
  142. package/coverage/lcov-report/natural/inflectors/ja/index.html +0 -350
  143. package/coverage/lcov-report/natural/inflectors/ja/noun_inflector.js.html +0 -727
  144. package/coverage/lcov-report/natural/inflectors/noun_inflector.js.html +0 -637
  145. package/coverage/lcov-report/natural/inflectors/present_verb_inflector.js.html +0 -544
  146. package/coverage/lcov-report/natural/inflectors/singular_plural_inflector.js.html +0 -649
  147. package/coverage/lcov-report/natural/lib/natural/analyzers/index.html +0 -350
  148. package/coverage/lcov-report/natural/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
  149. package/coverage/lcov-report/natural/lib/natural/classifiers/bayes_classifier.js.html +0 -496
  150. package/coverage/lcov-report/natural/lib/natural/classifiers/classifier.js.html +0 -823
  151. package/coverage/lcov-report/natural/lib/natural/classifiers/index.html +0 -376
  152. package/coverage/lcov-report/natural/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
  153. package/coverage/lcov-report/natural/lib/natural/distance/dice_coefficient.js.html +0 -529
  154. package/coverage/lcov-report/natural/lib/natural/distance/index.html +0 -376
  155. package/coverage/lcov-report/natural/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
  156. package/coverage/lcov-report/natural/lib/natural/distance/levenshtein_distance.js.html +0 -526
  157. package/coverage/lcov-report/natural/lib/natural/index.html +0 -350
  158. package/coverage/lcov-report/natural/lib/natural/index.js.html +0 -547
  159. package/coverage/lcov-report/natural/lib/natural/inflectors/count_inflector.js.html +0 -484
  160. package/coverage/lcov-report/natural/lib/natural/inflectors/form_set.js.html +0 -406
  161. package/coverage/lcov-report/natural/lib/natural/inflectors/fr/index.html +0 -350
  162. package/coverage/lcov-report/natural/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
  163. package/coverage/lcov-report/natural/lib/natural/inflectors/index.html +0 -402
  164. package/coverage/lcov-report/natural/lib/natural/inflectors/ja/index.html +0 -350
  165. package/coverage/lcov-report/natural/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
  166. package/coverage/lcov-report/natural/lib/natural/inflectors/noun_inflector.js.html +0 -637
  167. package/coverage/lcov-report/natural/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
  168. package/coverage/lcov-report/natural/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
  169. package/coverage/lcov-report/natural/lib/natural/ngrams/index.html +0 -363
  170. package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams.js.html +0 -595
  171. package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams_zh.js.html +0 -571
  172. package/coverage/lcov-report/natural/lib/natural/normalizers/index.html +0 -389
  173. package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer.js.html +0 -607
  174. package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
  175. package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_no.js.html +0 -496
  176. package/coverage/lcov-report/natural/lib/natural/normalizers/remove_diacritics.js.html +0 -685
  177. package/coverage/lcov-report/natural/lib/natural/phonetics/dm_soundex.js.html +0 -1063
  178. package/coverage/lcov-report/natural/lib/natural/phonetics/double_metaphone.js.html +0 -1840
  179. package/coverage/lcov-report/natural/lib/natural/phonetics/index.html +0 -402
  180. package/coverage/lcov-report/natural/lib/natural/phonetics/metaphone.js.html +0 -886
  181. package/coverage/lcov-report/natural/lib/natural/phonetics/phonetic.js.html +0 -484
  182. package/coverage/lcov-report/natural/lib/natural/phonetics/soundex.js.html +0 -562
  183. package/coverage/lcov-report/natural/lib/natural/stemmers/index.html +0 -571
  184. package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
  185. package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
  186. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer.js.html +0 -937
  187. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
  188. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
  189. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
  190. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
  191. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
  192. package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
  193. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer.js.html +0 -505
  194. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_es.js.html +0 -496
  195. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fa.js.html +0 -484
  196. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fr.js.html +0 -496
  197. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_it.js.html +0 -427
  198. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ja.js.html +0 -736
  199. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_no.js.html +0 -505
  200. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_pl.js.html +0 -496
  201. package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ru.js.html +0 -496
  202. package/coverage/lcov-report/natural/lib/natural/tfidf/index.html +0 -350
  203. package/coverage/lcov-report/natural/lib/natural/tfidf/tfidf.js.html +0 -904
  204. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
  205. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
  206. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
  207. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
  208. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
  209. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
  210. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
  211. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
  212. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
  213. package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
  214. package/coverage/lcov-report/natural/lib/natural/tokenizers/index.html +0 -519
  215. package/coverage/lcov-report/natural/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
  216. package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer.js.html +0 -475
  217. package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
  218. package/coverage/lcov-report/natural/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
  219. package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.html +0 -350
  220. package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.js.html +0 -2068
  221. package/coverage/lcov-report/natural/lib/natural/trie/index.html +0 -350
  222. package/coverage/lcov-report/natural/lib/natural/trie/trie.js.html +0 -1012
  223. package/coverage/lcov-report/natural/lib/natural/util/bag.js.html +0 -472
  224. package/coverage/lcov-report/natural/lib/natural/util/edge_weighted_digraph.js.html +0 -658
  225. package/coverage/lcov-report/natural/lib/natural/util/index.html +0 -532
  226. package/coverage/lcov-report/natural/lib/natural/util/longest_path_tree.js.html +0 -619
  227. package/coverage/lcov-report/natural/lib/natural/util/shortest_path_tree.js.html +0 -622
  228. package/coverage/lcov-report/natural/lib/natural/util/stopwords.js.html +0 -445
  229. package/coverage/lcov-report/natural/lib/natural/util/stopwords_es.js.html +0 -430
  230. package/coverage/lcov-report/natural/lib/natural/util/stopwords_fa.js.html +0 -436
  231. package/coverage/lcov-report/natural/lib/natural/util/stopwords_fr.js.html +0 -904
  232. package/coverage/lcov-report/natural/lib/natural/util/stopwords_it.js.html +0 -478
  233. package/coverage/lcov-report/natural/lib/natural/util/stopwords_ja.js.html +0 -499
  234. package/coverage/lcov-report/natural/lib/natural/util/stopwords_no.js.html +0 -445
  235. package/coverage/lcov-report/natural/lib/natural/util/stopwords_pl.js.html +0 -508
  236. package/coverage/lcov-report/natural/lib/natural/util/stopwords_ru.js.html +0 -445
  237. package/coverage/lcov-report/natural/lib/natural/util/topological.js.html +0 -616
  238. package/coverage/lcov-report/natural/lib/natural/util/utils.js.html +0 -679
  239. package/coverage/lcov-report/natural/lib/natural/wordnet/data_file.js.html +0 -580
  240. package/coverage/lcov-report/natural/lib/natural/wordnet/index.html +0 -389
  241. package/coverage/lcov-report/natural/lib/natural/wordnet/index_file.js.html +0 -742
  242. package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet.js.html +0 -829
  243. package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet_file.js.html +0 -529
  244. package/coverage/lcov-report/natural/ngrams/index.html +0 -363
  245. package/coverage/lcov-report/natural/ngrams/ngrams.js.html +0 -595
  246. package/coverage/lcov-report/natural/ngrams/ngrams_zh.js.html +0 -571
  247. package/coverage/lcov-report/natural/normalizers/index.html +0 -389
  248. package/coverage/lcov-report/natural/normalizers/normalizer.js.html +0 -607
  249. package/coverage/lcov-report/natural/normalizers/normalizer_ja.js.html +0 -2194
  250. package/coverage/lcov-report/natural/normalizers/normalizer_no.js.html +0 -496
  251. package/coverage/lcov-report/natural/normalizers/remove_diacritics.js.html +0 -685
  252. package/coverage/lcov-report/natural/phonetics/dm_soundex.js.html +0 -1063
  253. package/coverage/lcov-report/natural/phonetics/double_metaphone.js.html +0 -1840
  254. package/coverage/lcov-report/natural/phonetics/index.html +0 -402
  255. package/coverage/lcov-report/natural/phonetics/metaphone.js.html +0 -886
  256. package/coverage/lcov-report/natural/phonetics/phonetic.js.html +0 -484
  257. package/coverage/lcov-report/natural/phonetics/soundex.js.html +0 -562
  258. package/coverage/lcov-report/natural/stemmers/index.html +0 -571
  259. package/coverage/lcov-report/natural/stemmers/lancaster_rules.js.html +0 -2692
  260. package/coverage/lcov-report/natural/stemmers/lancaster_stemmer.js.html +0 -544
  261. package/coverage/lcov-report/natural/stemmers/porter_stemmer.js.html +0 -937
  262. package/coverage/lcov-report/natural/stemmers/porter_stemmer_es.js.html +0 -985
  263. package/coverage/lcov-report/natural/stemmers/porter_stemmer_fa.js.html +0 -418
  264. package/coverage/lcov-report/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
  265. package/coverage/lcov-report/natural/stemmers/porter_stemmer_it.js.html +0 -1018
  266. package/coverage/lcov-report/natural/stemmers/porter_stemmer_no.js.html +0 -784
  267. package/coverage/lcov-report/natural/stemmers/porter_stemmer_ru.js.html +0 -775
  268. package/coverage/lcov-report/natural/stemmers/stemmer.js.html +0 -505
  269. package/coverage/lcov-report/natural/stemmers/stemmer_es.js.html +0 -496
  270. package/coverage/lcov-report/natural/stemmers/stemmer_fa.js.html +0 -484
  271. package/coverage/lcov-report/natural/stemmers/stemmer_fr.js.html +0 -496
  272. package/coverage/lcov-report/natural/stemmers/stemmer_it.js.html +0 -427
  273. package/coverage/lcov-report/natural/stemmers/stemmer_ja.js.html +0 -736
  274. package/coverage/lcov-report/natural/stemmers/stemmer_no.js.html +0 -505
  275. package/coverage/lcov-report/natural/stemmers/stemmer_pl.js.html +0 -496
  276. package/coverage/lcov-report/natural/stemmers/stemmer_ru.js.html +0 -496
  277. package/coverage/lcov-report/natural/tfidf/index.html +0 -350
  278. package/coverage/lcov-report/natural/tfidf/tfidf.js.html +0 -904
  279. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
  280. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
  281. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
  282. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
  283. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
  284. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
  285. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
  286. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
  287. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
  288. package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
  289. package/coverage/lcov-report/natural/tokenizers/index.html +0 -519
  290. package/coverage/lcov-report/natural/tokenizers/regexp_tokenizer.js.html +0 -580
  291. package/coverage/lcov-report/natural/tokenizers/tokenizer.js.html +0 -475
  292. package/coverage/lcov-report/natural/tokenizers/tokenizer_ja.js.html +0 -1153
  293. package/coverage/lcov-report/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
  294. package/coverage/lcov-report/natural/transliterators/ja/index.html +0 -350
  295. package/coverage/lcov-report/natural/transliterators/ja/index.js.html +0 -2068
  296. package/coverage/lcov-report/natural/trie/index.html +0 -350
  297. package/coverage/lcov-report/natural/trie/trie.js.html +0 -1012
  298. package/coverage/lcov-report/natural/util/bag.js.html +0 -472
  299. package/coverage/lcov-report/natural/util/edge_weighted_digraph.js.html +0 -658
  300. package/coverage/lcov-report/natural/util/index.html +0 -532
  301. package/coverage/lcov-report/natural/util/longest_path_tree.js.html +0 -619
  302. package/coverage/lcov-report/natural/util/shortest_path_tree.js.html +0 -622
  303. package/coverage/lcov-report/natural/util/stopwords.js.html +0 -445
  304. package/coverage/lcov-report/natural/util/stopwords_es.js.html +0 -430
  305. package/coverage/lcov-report/natural/util/stopwords_fa.js.html +0 -436
  306. package/coverage/lcov-report/natural/util/stopwords_fr.js.html +0 -904
  307. package/coverage/lcov-report/natural/util/stopwords_it.js.html +0 -478
  308. package/coverage/lcov-report/natural/util/stopwords_ja.js.html +0 -499
  309. package/coverage/lcov-report/natural/util/stopwords_no.js.html +0 -445
  310. package/coverage/lcov-report/natural/util/stopwords_pl.js.html +0 -508
  311. package/coverage/lcov-report/natural/util/stopwords_ru.js.html +0 -445
  312. package/coverage/lcov-report/natural/util/topological.js.html +0 -616
  313. package/coverage/lcov-report/natural/util/utils.js.html +0 -679
  314. package/coverage/lcov-report/natural/wordnet/data_file.js.html +0 -580
  315. package/coverage/lcov-report/natural/wordnet/index.html +0 -389
  316. package/coverage/lcov-report/natural/wordnet/index_file.js.html +0 -742
  317. package/coverage/lcov-report/natural/wordnet/wordnet.js.html +0 -829
  318. package/coverage/lcov-report/natural/wordnet/wordnet_file.js.html +0 -529
  319. package/coverage/lcov-report/normalizers/index.html +0 -350
  320. package/coverage/lcov-report/normalizers/normalizer.js.html +0 -607
  321. package/coverage/lcov-report/prettify.css +0 -1
  322. package/coverage/lcov-report/prettify.js +0 -1
  323. package/coverage/lcov-report/spec/aggressive_tokenizer_es_spec.js.html +0 -430
  324. package/coverage/lcov-report/spec/aggressive_tokenizer_fr_spec.js.html +0 -625
  325. package/coverage/lcov-report/spec/aggressive_tokenizer_nl_spec.js.html +0 -430
  326. package/coverage/lcov-report/spec/aggressive_tokenizer_pt_spec.js.html +0 -490
  327. package/coverage/lcov-report/spec/aggressive_tokenizer_spec.js.html +0 -490
  328. package/coverage/lcov-report/spec/bayes_classifier_spec.js.html +0 -841
  329. package/coverage/lcov-report/spec/classifier_spec.js.html +0 -847
  330. package/coverage/lcov-report/spec/count_inflector_fr_spec.js.html +0 -454
  331. package/coverage/lcov-report/spec/count_inflector_spec.js.html +0 -538
  332. package/coverage/lcov-report/spec/dice_coefficient_spec.js.html +0 -400
  333. package/coverage/lcov-report/spec/double_metaphone_spec.js.html +0 -2548
  334. package/coverage/lcov-report/spec/index.html +0 -883
  335. package/coverage/lcov-report/spec/jaro-winkler_spec.js.html +0 -481
  336. package/coverage/lcov-report/spec/lancaster_stemmer_spec.js.html +0 -541
  337. package/coverage/lcov-report/spec/levenshtein_spec.js.html +0 -520
  338. package/coverage/lcov-report/spec/logistic_regression_classifier_spec.js.html +0 -688
  339. package/coverage/lcov-report/spec/longest_path_tree_spec.js.html +0 -628
  340. package/coverage/lcov-report/spec/metaphone_spec.js.html +0 -1063
  341. package/coverage/lcov-report/spec/ngram_spec.js.html +0 -712
  342. package/coverage/lcov-report/spec/ngram_zh_spec.js.html +0 -688
  343. package/coverage/lcov-report/spec/normalizer_ja_spec.js.html +0 -802
  344. package/coverage/lcov-report/spec/normalizer_no_spec.js.html +0 -655
  345. package/coverage/lcov-report/spec/normalizer_spec.js.html +0 -742
  346. package/coverage/lcov-report/spec/noun_inflector_fr_spec.js.html +0 -1750
  347. package/coverage/lcov-report/spec/noun_inflector_ja_spec.js.html +0 -658
  348. package/coverage/lcov-report/spec/noun_inflector_spec.js.html +0 -1132
  349. package/coverage/lcov-report/spec/porter_stemmer_fr_spec.js.html +0 -625
  350. package/coverage/lcov-report/spec/porter_stemmer_it_spec.js.html +0 -454
  351. package/coverage/lcov-report/spec/porter_stemmer_no_spec.js.html +0 -751
  352. package/coverage/lcov-report/spec/porter_stemmer_ru_spec.js.html +0 -511
  353. package/coverage/lcov-report/spec/porter_stemmer_spec.js.html +0 -733
  354. package/coverage/lcov-report/spec/present_verb_inflector_spec.js.html +0 -697
  355. package/coverage/lcov-report/spec/remove_diacritics_spec.js.html +0 -415
  356. package/coverage/lcov-report/spec/sentence_analyzer_spec.js.html +0 -1153
  357. package/coverage/lcov-report/spec/shortest_path_tree_spec.js.html +0 -718
  358. package/coverage/lcov-report/spec/soundex_spec.js.html +0 -784
  359. package/coverage/lcov-report/spec/spellcheck_spec.js.html +0 -598
  360. package/coverage/lcov-report/spec/stemmer_ja_spec.js.html +0 -520
  361. package/coverage/lcov-report/spec/tfidf_spec.js.html +0 -1288
  362. package/coverage/lcov-report/spec/tokenizer_ja_spec.js.html +0 -496
  363. package/coverage/lcov-report/spec/transliterator_ja_spec.js.html +0 -631
  364. package/coverage/lcov-report/spec/treebank_word_tokenizer_spec.js.html +0 -430
  365. package/coverage/lcov-report/spec/trie_spec.js.html +0 -940
  366. package/coverage/lcov-report/util/index.html +0 -350
  367. package/coverage/lcov-report/util/utils.js.html +0 -679
  368. package/coverage/lcov.info +0 -11094
  369. package/lib/natural/parsers/Agenda.js +0 -68
  370. package/lib/natural/parsers/CYKParser.js +0 -87
  371. package/lib/natural/parsers/CYK_Item.js +0 -159
  372. package/lib/natural/parsers/Chart.js +0 -211
  373. package/lib/natural/parsers/ChartParser.js +0 -60
  374. package/lib/natural/parsers/DoubleDottedItem.js +0 -226
  375. package/lib/natural/parsers/EarleyItem.js +0 -155
  376. package/lib/natural/parsers/EarleyParser.js +0 -58
  377. package/lib/natural/parsers/FunctionWordTagger.js +0 -84
  378. package/lib/natural/parsers/GoalItem.js +0 -206
  379. package/lib/natural/parsers/GrammarParser.js +0 -1289
  380. package/lib/natural/parsers/HeadCornerParser.js +0 -101
  381. package/lib/natural/parsers/LeftCornerParser.js +0 -58
  382. package/lib/natural/parsers/PEG-grammar-for-unification-grammar.txt +0 -345
package/coverage/lcov-report/spec/tfidf_spec.js.html DELETED
@@ -1,1288 +0,0 @@
1
- <!doctype html>
2
- <html lang="en">
3
- <head>
4
- <title>Code coverage report for spec/tfidf_spec.js</title>
5
- <meta charset="utf-8">
6
-
7
- <link rel="stylesheet" href="../prettify.css">
8
-
9
- <style>
10
- body, html {
11
- margin:0; padding: 0;
12
- }
13
- body {
14
- font-family: Helvetica Neue, Helvetica,Arial;
15
- font-size: 10pt;
16
- }
17
- div.header, div.footer {
18
- background: #eee;
19
- padding: 1em;
20
- }
21
- div.header {
22
- z-index: 100;
23
- position: fixed;
24
- top: 0;
25
- border-bottom: 1px solid #666;
26
- width: 100%;
27
- }
28
- div.footer {
29
- border-top: 1px solid #666;
30
- }
31
- div.body {
32
- margin-top: 10em;
33
- }
34
- div.meta {
35
- font-size: 90%;
36
- text-align: center;
37
- }
38
- h1, h2, h3 {
39
- font-weight: normal;
40
- }
41
- h1 {
42
- font-size: 12pt;
43
- }
44
- h2 {
45
- font-size: 10pt;
46
- }
47
- pre {
48
- font-family: Consolas, Menlo, Monaco, monospace;
49
- margin: 0;
50
- padding: 0;
51
- line-height: 14px;
52
- font-size: 14px;
53
- -moz-tab-size: 2;
54
- -o-tab-size: 2;
55
- tab-size: 2;
56
- }
57
-
58
- div.path { font-size: 110%; }
59
- div.path a:link, div.path a:visited { color: #000; }
60
- table.coverage { border-collapse: collapse; margin:0; padding: 0 }
61
-
62
- table.coverage td {
63
- margin: 0;
64
- padding: 0;
65
- color: #111;
66
- vertical-align: top;
67
- }
68
- table.coverage td.line-count {
69
- width: 50px;
70
- text-align: right;
71
- padding-right: 5px;
72
- }
73
- table.coverage td.line-coverage {
74
- color: #777 !important;
75
- text-align: right;
76
- border-left: 1px solid #666;
77
- border-right: 1px solid #666;
78
- }
79
-
80
- table.coverage td.text {
81
- }
82
-
83
- table.coverage td span.cline-any {
84
- display: inline-block;
85
- padding: 0 5px;
86
- width: 40px;
87
- }
88
- table.coverage td span.cline-neutral {
89
- background: #eee;
90
- }
91
- table.coverage td span.cline-yes {
92
- background: #b5d592;
93
- color: #999;
94
- }
95
- table.coverage td span.cline-no {
96
- background: #fc8c84;
97
- }
98
-
99
- .cstat-yes { color: #111; }
100
- .cstat-no { background: #fc8c84; color: #111; }
101
- .fstat-no { background: #ffc520; color: #111 !important; }
102
- .cbranch-no { background: yellow !important; color: #111; }
103
-
104
- .cstat-skip { background: #ddd; color: #111; }
105
- .fstat-skip { background: #ddd; color: #111 !important; }
106
- .cbranch-skip { background: #ddd !important; color: #111; }
107
-
108
- .missing-if-branch {
109
- display: inline-block;
110
- margin-right: 10px;
111
- position: relative;
112
- padding: 0 4px;
113
- background: black;
114
- color: yellow;
115
- }
116
-
117
- .skip-if-branch {
118
- display: none;
119
- margin-right: 10px;
120
- position: relative;
121
- padding: 0 4px;
122
- background: #ccc;
123
- color: white;
124
- }
125
-
126
- .missing-if-branch .typ, .skip-if-branch .typ {
127
- color: inherit !important;
128
- }
129
-
130
- .entity, .metric { font-weight: bold; }
131
- .metric { display: inline-block; border: 1px solid #333; padding: 0.3em; background: white; }
132
- .metric small { font-size: 80%; font-weight: normal; color: #666; }
133
-
134
- div.coverage-summary table { border-collapse: collapse; margin: 3em; font-size: 110%; }
135
- div.coverage-summary td, div.coverage-summary table th { margin: 0; padding: 0.25em 1em; border-top: 1px solid #666; border-bottom: 1px solid #666; }
136
- div.coverage-summary th { text-align: left; border: 1px solid #666; background: #eee; font-weight: normal; }
137
- div.coverage-summary th.file { border-right: none !important; }
138
- div.coverage-summary th.pic { border-left: none !important; text-align: right; }
139
- div.coverage-summary th.pct { border-right: none !important; }
140
- div.coverage-summary th.abs { border-left: none !important; text-align: right; }
141
- div.coverage-summary td.pct { text-align: right; border-left: 1px solid #666; }
142
- div.coverage-summary td.abs { text-align: right; font-size: 90%; color: #444; border-right: 1px solid #666; }
143
- div.coverage-summary td.file { text-align: right; border-left: 1px solid #666; white-space: nowrap; }
144
- div.coverage-summary td.pic { min-width: 120px !important; }
145
- div.coverage-summary a:link { text-decoration: none; color: #000; }
146
- div.coverage-summary a:visited { text-decoration: none; color: #333; }
147
- div.coverage-summary a:hover { text-decoration: underline; }
148
- div.coverage-summary tfoot td { border-top: 1px solid #666; }
149
-
150
- div.coverage-summary .yui3-datatable-sort-indicator, div.coverage-summary .dummy-sort-indicator {
151
- height: 10px;
152
- width: 7px;
153
- display: inline-block;
154
- margin-left: 0.5em;
155
- }
156
- div.coverage-summary .yui3-datatable-sort-indicator {
157
- background: url("https://yui-s.yahooapis.com/3.6.0/build/datatable-sort/assets/skins/sam/sort-arrow-sprite.png") no-repeat scroll 0 0 transparent;
158
- }
159
- div.coverage-summary .yui3-datatable-sorted .yui3-datatable-sort-indicator {
160
- background-position: 0 -20px;
161
- }
162
- div.coverage-summary .yui3-datatable-sorted-desc .yui3-datatable-sort-indicator {
163
- background-position: 0 -10px;
164
- }
165
-
166
- .high { background: #b5d592 !important; }
167
- .medium { background: #ffe87c !important; }
168
- .low { background: #fc8c84 !important; }
169
-
170
- span.cover-fill, span.cover-empty {
171
- display:inline-block;
172
- border:1px solid #444;
173
- background: white;
174
- height: 12px;
175
- }
176
- span.cover-fill {
177
- background: #ccc;
178
- border-right: 1px solid #444;
179
- }
180
- span.cover-empty {
181
- background: white;
182
- border-left: none;
183
- }
184
- span.cover-full {
185
- border-right: none !important;
186
- }
187
- pre.prettyprint {
188
- border: none !important;
189
- padding: 0 !important;
190
- margin: 0 !important;
191
- }
192
- .com { color: #999 !important; }
193
- .ignore-none { color: #999; font-weight: normal; }
194
-
195
- </style>
196
- </head>
197
- <body>
198
- <div class="header high">
199
- <h1>Code coverage report for <span class="entity">spec/tfidf_spec.js</span></h1>
200
- <h2>
201
-
202
- Statements: <span class="metric">100% <small>(155 / 155)</small></span> &nbsp;&nbsp;&nbsp;&nbsp;
203
-
204
-
205
- Branches: <span class="metric">100% <small>(6 / 6)</small></span> &nbsp;&nbsp;&nbsp;&nbsp;
206
-
207
-
208
- Functions: <span class="metric">100% <small>(37 / 37)</small></span> &nbsp;&nbsp;&nbsp;&nbsp;
209
-
210
-
211
- Lines: <span class="metric">100% <small>(152 / 152)</small></span> &nbsp;&nbsp;&nbsp;&nbsp;
212
-
213
- Ignored: <span class="metric"><span class="ignore-none">none</span></span> &nbsp;&nbsp;&nbsp;&nbsp;
214
- </h2>
215
- <div class="path"><a href="../index.html">All files</a> &#187; <a href="index.html">spec/</a> &#187; tfidf_spec.js</div>
216
- </div>
217
- <div class="body">
218
- <pre><table class="coverage">
219
- <tr><td class="line-count">1
220
- 2
221
- 3
222
- 4
223
- 5
224
- 6
225
- 7
226
- 8
227
- 9
228
- 10
229
- 11
230
- 12
231
- 13
232
- 14
233
- 15
234
- 16
235
- 17
236
- 18
237
- 19
238
- 20
239
- 21
240
- 22
241
- 23
242
- 24
243
- 25
244
- 26
245
- 27
246
- 28
247
- 29
248
- 30
249
- 31
250
- 32
251
- 33
252
- 34
253
- 35
254
- 36
255
- 37
256
- 38
257
- 39
258
- 40
259
- 41
260
- 42
261
- 43
262
- 44
263
- 45
264
- 46
265
- 47
266
- 48
267
- 49
268
- 50
269
- 51
270
- 52
271
- 53
272
- 54
273
- 55
274
- 56
275
- 57
276
- 58
277
- 59
278
- 60
279
- 61
280
- 62
281
- 63
282
- 64
283
- 65
284
- 66
285
- 67
286
- 68
287
- 69
288
- 70
289
- 71
290
- 72
291
- 73
292
- 74
293
- 75
294
- 76
295
- 77
296
- 78
297
- 79
298
- 80
299
- 81
300
- 82
301
- 83
302
- 84
303
- 85
304
- 86
305
- 87
306
- 88
307
- 89
308
- 90
309
- 91
310
- 92
311
- 93
312
- 94
313
- 95
314
- 96
315
- 97
316
- 98
317
- 99
318
- 100
319
- 101
320
- 102
321
- 103
322
- 104
323
- 105
324
- 106
325
- 107
326
- 108
327
- 109
328
- 110
329
- 111
330
- 112
331
- 113
332
- 114
333
- 115
334
- 116
335
- 117
336
- 118
337
- 119
338
- 120
339
- 121
340
- 122
341
- 123
342
- 124
343
- 125
344
- 126
345
- 127
346
- 128
347
- 129
348
- 130
349
- 131
350
- 132
351
- 133
352
- 134
353
- 135
354
- 136
355
- 137
356
- 138
357
- 139
358
- 140
359
- 141
360
- 142
361
- 143
362
- 144
363
- 145
364
- 146
365
- 147
366
- 148
367
- 149
368
- 150
369
- 151
370
- 152
371
- 153
372
- 154
373
- 155
374
- 156
375
- 157
376
- 158
377
- 159
378
- 160
379
- 161
380
- 162
381
- 163
382
- 164
383
- 165
384
- 166
385
- 167
386
- 168
387
- 169
388
- 170
389
- 171
390
- 172
391
- 173
392
- 174
393
- 175
394
- 176
395
- 177
396
- 178
397
- 179
398
- 180
399
- 181
400
- 182
401
- 183
402
- 184
403
- 185
404
- 186
405
- 187
406
- 188
407
- 189
408
- 190
409
- 191
410
- 192
411
- 193
412
- 194
413
- 195
414
- 196
415
- 197
416
- 198
417
- 199
418
- 200
419
- 201
420
- 202
421
- 203
422
- 204
423
- 205
424
- 206
425
- 207
426
- 208
427
- 209
428
- 210
429
- 211
430
- 212
431
- 213
432
- 214
433
- 215
434
- 216
435
- 217
436
- 218
437
- 219
438
- 220
439
- 221
440
- 222
441
- 223
442
- 224
443
- 225
444
- 226
445
- 227
446
- 228
447
- 229
448
- 230
449
- 231
450
- 232
451
- 233
452
- 234
453
- 235
454
- 236
455
- 237
456
- 238
457
- 239
458
- 240
459
- 241
460
- 242
461
- 243
462
- 244
463
- 245
464
- 246
465
- 247
466
- 248
467
- 249
468
- 250
469
- 251
470
- 252
471
- 253
472
- 254
473
- 255
474
- 256
475
- 257
476
- 258
477
- 259
478
- 260
479
- 261
480
- 262
481
- 263
482
- 264
483
- 265
484
- 266
485
- 267
486
- 268
487
- 269
488
- 270
489
- 271
490
- 272
491
- 273
492
- 274
493
- 275
494
- 276
495
- 277
496
- 278
497
- 279
498
- 280
499
- 281
500
- 282
501
- 283
502
- 284
503
- 285
504
- 286
505
- 287
506
- 288
507
- 289
508
- 290
509
- 291
510
- 292
511
- 293
512
- 294
513
- 295
514
- 296
515
- 297
516
- 298
517
- 299
518
- 300
519
- 301
520
- 302
521
- 303
522
- 304
523
- 305
524
- 306
525
- 307
526
- 308
527
- 309
528
- 310
529
- 311
530
- 312
531
- 313
532
- 314
533
- 315
534
- 316
535
- 317
536
- 318
537
- 319
538
- 320
539
- 321
540
- 322
541
- 323</td><td class="line-coverage"><span class="cline-any cline-neutral">&nbsp;</span>
542
- <span class="cline-any cline-neutral">&nbsp;</span>
543
- <span class="cline-any cline-neutral">&nbsp;</span>
544
- <span class="cline-any cline-neutral">&nbsp;</span>
545
- <span class="cline-any cline-neutral">&nbsp;</span>
546
- <span class="cline-any cline-neutral">&nbsp;</span>
547
- <span class="cline-any cline-neutral">&nbsp;</span>
548
- <span class="cline-any cline-neutral">&nbsp;</span>
549
- <span class="cline-any cline-neutral">&nbsp;</span>
550
- <span class="cline-any cline-neutral">&nbsp;</span>
551
- <span class="cline-any cline-neutral">&nbsp;</span>
552
- <span class="cline-any cline-neutral">&nbsp;</span>
553
- <span class="cline-any cline-neutral">&nbsp;</span>
554
- <span class="cline-any cline-neutral">&nbsp;</span>
555
- <span class="cline-any cline-neutral">&nbsp;</span>
556
- <span class="cline-any cline-neutral">&nbsp;</span>
557
- <span class="cline-any cline-neutral">&nbsp;</span>
558
- <span class="cline-any cline-neutral">&nbsp;</span>
559
- <span class="cline-any cline-neutral">&nbsp;</span>
560
- <span class="cline-any cline-neutral">&nbsp;</span>
561
- <span class="cline-any cline-neutral">&nbsp;</span>
562
- <span class="cline-any cline-neutral">&nbsp;</span>
563
- <span class="cline-any cline-yes">1</span>
564
- <span class="cline-any cline-yes">1</span>
565
- <span class="cline-any cline-neutral">&nbsp;</span>
566
- <span class="cline-any cline-yes">1</span>
567
- <span class="cline-any cline-yes">1</span>
568
- <span class="cline-any cline-yes">1</span>
569
- <span class="cline-any cline-yes">1</span>
570
- <span class="cline-any cline-yes">1</span>
571
- <span class="cline-any cline-yes">1</span>
572
- <span class="cline-any cline-neutral">&nbsp;</span>
573
- <span class="cline-any cline-neutral">&nbsp;</span>
574
- <span class="cline-any cline-neutral">&nbsp;</span>
575
- <span class="cline-any cline-yes">1</span>
576
- <span class="cline-any cline-yes">1</span>
577
- <span class="cline-any cline-yes">1</span>
578
- <span class="cline-any cline-yes">1</span>
579
- <span class="cline-any cline-yes">1</span>
580
- <span class="cline-any cline-neutral">&nbsp;</span>
581
- <span class="cline-any cline-yes">1</span>
582
- <span class="cline-any cline-yes">2</span>
583
- <span class="cline-any cline-yes">1</span>
584
- <span class="cline-any cline-neutral">&nbsp;</span>
585
- <span class="cline-any cline-yes">1</span>
586
- <span class="cline-any cline-neutral">&nbsp;</span>
587
- <span class="cline-any cline-neutral">&nbsp;</span>
588
- <span class="cline-any cline-neutral">&nbsp;</span>
589
- <span class="cline-any cline-yes">1</span>
590
- <span class="cline-any cline-yes">1</span>
591
- <span class="cline-any cline-neutral">&nbsp;</span>
592
- <span class="cline-any cline-neutral">&nbsp;</span>
593
- <span class="cline-any cline-neutral">&nbsp;</span>
594
- <span class="cline-any cline-neutral">&nbsp;</span>
595
- <span class="cline-any cline-neutral">&nbsp;</span>
596
- <span class="cline-any cline-yes">1</span>
597
- <span class="cline-any cline-yes">2</span>
598
- <span class="cline-any cline-yes">1</span>
599
- <span class="cline-any cline-neutral">&nbsp;</span>
600
- <span class="cline-any cline-yes">1</span>
601
- <span class="cline-any cline-neutral">&nbsp;</span>
602
- <span class="cline-any cline-neutral">&nbsp;</span>
603
- <span class="cline-any cline-neutral">&nbsp;</span>
604
- <span class="cline-any cline-yes">1</span>
605
- <span class="cline-any cline-yes">1</span>
606
- <span class="cline-any cline-neutral">&nbsp;</span>
607
- <span class="cline-any cline-neutral">&nbsp;</span>
608
- <span class="cline-any cline-neutral">&nbsp;</span>
609
- <span class="cline-any cline-neutral">&nbsp;</span>
610
- <span class="cline-any cline-neutral">&nbsp;</span>
611
- <span class="cline-any cline-yes">1</span>
612
- <span class="cline-any cline-yes">1</span>
613
- <span class="cline-any cline-neutral">&nbsp;</span>
614
- <span class="cline-any cline-neutral">&nbsp;</span>
615
- <span class="cline-any cline-yes">1</span>
616
- <span class="cline-any cline-yes">1</span>
617
- <span class="cline-any cline-yes">1</span>
618
- <span class="cline-any cline-yes">1</span>
619
- <span class="cline-any cline-yes">1</span>
620
- <span class="cline-any cline-neutral">&nbsp;</span>
621
- <span class="cline-any cline-yes">1</span>
622
- <span class="cline-any cline-yes">2</span>
623
- <span class="cline-any cline-yes">1</span>
624
- <span class="cline-any cline-neutral">&nbsp;</span>
625
- <span class="cline-any cline-yes">1</span>
626
- <span class="cline-any cline-neutral">&nbsp;</span>
627
- <span class="cline-any cline-neutral">&nbsp;</span>
628
- <span class="cline-any cline-neutral">&nbsp;</span>
629
- <span class="cline-any cline-neutral">&nbsp;</span>
630
- <span class="cline-any cline-yes">1</span>
631
- <span class="cline-any cline-yes">1</span>
632
- <span class="cline-any cline-yes">1</span>
633
- <span class="cline-any cline-yes">1</span>
634
- <span class="cline-any cline-yes">1</span>
635
- <span class="cline-any cline-neutral">&nbsp;</span>
636
- <span class="cline-any cline-neutral">&nbsp;</span>
637
- <span class="cline-any cline-yes">1</span>
638
- <span class="cline-any cline-yes">1</span>
639
- <span class="cline-any cline-yes">1</span>
640
- <span class="cline-any cline-neutral">&nbsp;</span>
641
- <span class="cline-any cline-neutral">&nbsp;</span>
642
- <span class="cline-any cline-neutral">&nbsp;</span>
643
- <span class="cline-any cline-yes">1</span>
644
- <span class="cline-any cline-neutral">&nbsp;</span>
645
- <span class="cline-any cline-neutral">&nbsp;</span>
646
- <span class="cline-any cline-yes">1</span>
647
- <span class="cline-any cline-yes">1</span>
648
- <span class="cline-any cline-neutral">&nbsp;</span>
649
- <span class="cline-any cline-neutral">&nbsp;</span>
650
- <span class="cline-any cline-neutral">&nbsp;</span>
651
- <span class="cline-any cline-neutral">&nbsp;</span>
652
- <span class="cline-any cline-neutral">&nbsp;</span>
653
- <span class="cline-any cline-neutral">&nbsp;</span>
654
- <span class="cline-any cline-neutral">&nbsp;</span>
655
- <span class="cline-any cline-neutral">&nbsp;</span>
656
- <span class="cline-any cline-yes">1</span>
657
- <span class="cline-any cline-yes">1</span>
658
- <span class="cline-any cline-neutral">&nbsp;</span>
659
- <span class="cline-any cline-yes">1</span>
660
- <span class="cline-any cline-yes">7</span>
661
- <span class="cline-any cline-neutral">&nbsp;</span>
662
- <span class="cline-any cline-neutral">&nbsp;</span>
663
- <span class="cline-any cline-neutral">&nbsp;</span>
664
- <span class="cline-any cline-yes">1</span>
665
- <span class="cline-any cline-yes">1</span>
666
- <span class="cline-any cline-yes">1</span>
667
- <span class="cline-any cline-yes">1</span>
668
- <span class="cline-any cline-yes">1</span>
669
- <span class="cline-any cline-neutral">&nbsp;</span>
670
- <span class="cline-any cline-neutral">&nbsp;</span>
671
- <span class="cline-any cline-neutral">&nbsp;</span>
672
- <span class="cline-any cline-yes">1</span>
673
- <span class="cline-any cline-neutral">&nbsp;</span>
674
- <span class="cline-any cline-yes">1</span>
675
- <span class="cline-any cline-neutral">&nbsp;</span>
676
- <span class="cline-any cline-yes">1</span>
677
- <span class="cline-any cline-yes">1</span>
678
- <span class="cline-any cline-yes">1</span>
679
- <span class="cline-any cline-yes">1</span>
680
- <span class="cline-any cline-yes">1</span>
681
- <span class="cline-any cline-neutral">&nbsp;</span>
682
- <span class="cline-any cline-yes">1</span>
683
- <span class="cline-any cline-neutral">&nbsp;</span>
684
- <span class="cline-any cline-neutral">&nbsp;</span>
685
- <span class="cline-any cline-yes">1</span>
686
- <span class="cline-any cline-neutral">&nbsp;</span>
687
- <span class="cline-any cline-yes">1</span>
688
- <span class="cline-any cline-yes">1</span>
689
- <span class="cline-any cline-yes">1</span>
690
- <span class="cline-any cline-yes">1</span>
691
- <span class="cline-any cline-yes">1</span>
692
- <span class="cline-any cline-yes">1</span>
693
- <span class="cline-any cline-yes">1</span>
694
- <span class="cline-any cline-neutral">&nbsp;</span>
695
- <span class="cline-any cline-yes">1</span>
696
- <span class="cline-any cline-neutral">&nbsp;</span>
697
- <span class="cline-any cline-neutral">&nbsp;</span>
698
- <span class="cline-any cline-yes">1</span>
699
- <span class="cline-any cline-yes">1</span>
700
- <span class="cline-any cline-yes">1</span>
701
- <span class="cline-any cline-yes">1</span>
702
- <span class="cline-any cline-yes">1</span>
703
- <span class="cline-any cline-neutral">&nbsp;</span>
704
- <span class="cline-any cline-neutral">&nbsp;</span>
705
- <span class="cline-any cline-neutral">&nbsp;</span>
706
- <span class="cline-any cline-yes">1</span>
707
- <span class="cline-any cline-neutral">&nbsp;</span>
708
- <span class="cline-any cline-yes">1</span>
709
- <span class="cline-any cline-neutral">&nbsp;</span>
710
- <span class="cline-any cline-neutral">&nbsp;</span>
711
- <span class="cline-any cline-neutral">&nbsp;</span>
712
- <span class="cline-any cline-neutral">&nbsp;</span>
713
- <span class="cline-any cline-neutral">&nbsp;</span>
714
- <span class="cline-any cline-neutral">&nbsp;</span>
715
- <span class="cline-any cline-yes">1</span>
716
- <span class="cline-any cline-yes">1</span>
717
- <span class="cline-any cline-yes">1</span>
718
- <span class="cline-any cline-yes">1</span>
719
- <span class="cline-any cline-yes">1</span>
720
- <span class="cline-any cline-neutral">&nbsp;</span>
721
- <span class="cline-any cline-yes">1</span>
722
- <span class="cline-any cline-yes">4</span>
723
- <span class="cline-any cline-neutral">&nbsp;</span>
724
- <span class="cline-any cline-neutral">&nbsp;</span>
725
- <span class="cline-any cline-yes">1</span>
726
- <span class="cline-any cline-yes">4</span>
727
- <span class="cline-any cline-neutral">&nbsp;</span>
728
- <span class="cline-any cline-neutral">&nbsp;</span>
729
- <span class="cline-any cline-neutral">&nbsp;</span>
730
- <span class="cline-any cline-neutral">&nbsp;</span>
731
- <span class="cline-any cline-yes">1</span>
732
- <span class="cline-any cline-yes">1</span>
733
- <span class="cline-any cline-yes">1</span>
734
- <span class="cline-any cline-neutral">&nbsp;</span>
735
- <span class="cline-any cline-yes">1</span>
736
- <span class="cline-any cline-neutral">&nbsp;</span>
737
- <span class="cline-any cline-neutral">&nbsp;</span>
738
- <span class="cline-any cline-neutral">&nbsp;</span>
739
- <span class="cline-any cline-neutral">&nbsp;</span>
740
- <span class="cline-any cline-neutral">&nbsp;</span>
741
- <span class="cline-any cline-yes">1</span>
742
- <span class="cline-any cline-neutral">&nbsp;</span>
743
- <span class="cline-any cline-yes">1</span>
744
- <span class="cline-any cline-neutral">&nbsp;</span>
745
- <span class="cline-any cline-neutral">&nbsp;</span>
746
- <span class="cline-any cline-yes">1</span>
747
- <span class="cline-any cline-yes">1</span>
748
- <span class="cline-any cline-neutral">&nbsp;</span>
749
- <span class="cline-any cline-neutral">&nbsp;</span>
750
- <span class="cline-any cline-yes">1</span>
751
- <span class="cline-any cline-neutral">&nbsp;</span>
752
- <span class="cline-any cline-neutral">&nbsp;</span>
753
- <span class="cline-any cline-yes">1</span>
754
- <span class="cline-any cline-yes">1</span>
755
- <span class="cline-any cline-neutral">&nbsp;</span>
756
- <span class="cline-any cline-neutral">&nbsp;</span>
757
- <span class="cline-any cline-yes">1</span>
758
- <span class="cline-any cline-neutral">&nbsp;</span>
759
- <span class="cline-any cline-neutral">&nbsp;</span>
760
- <span class="cline-any cline-neutral">&nbsp;</span>
761
- <span class="cline-any cline-yes">1</span>
762
- <span class="cline-any cline-neutral">&nbsp;</span>
763
- <span class="cline-any cline-yes">1</span>
764
- <span class="cline-any cline-neutral">&nbsp;</span>
765
- <span class="cline-any cline-yes">1</span>
766
- <span class="cline-any cline-yes">1</span>
767
- <span class="cline-any cline-yes">1</span>
768
- <span class="cline-any cline-yes">1</span>
769
- <span class="cline-any cline-neutral">&nbsp;</span>
770
- <span class="cline-any cline-yes">1</span>
771
- <span class="cline-any cline-neutral">&nbsp;</span>
772
- <span class="cline-any cline-neutral">&nbsp;</span>
773
- <span class="cline-any cline-neutral">&nbsp;</span>
774
- <span class="cline-any cline-neutral">&nbsp;</span>
775
- <span class="cline-any cline-neutral">&nbsp;</span>
776
- <span class="cline-any cline-neutral">&nbsp;</span>
777
- <span class="cline-any cline-yes">1</span>
778
- <span class="cline-any cline-yes">4</span>
779
- <span class="cline-any cline-neutral">&nbsp;</span>
780
- <span class="cline-any cline-neutral">&nbsp;</span>
781
- <span class="cline-any cline-yes">1</span>
782
- <span class="cline-any cline-yes">4</span>
783
- <span class="cline-any cline-neutral">&nbsp;</span>
784
- <span class="cline-any cline-neutral">&nbsp;</span>
785
- <span class="cline-any cline-neutral">&nbsp;</span>
786
- <span class="cline-any cline-yes">1</span>
787
- <span class="cline-any cline-neutral">&nbsp;</span>
788
- <span class="cline-any cline-neutral">&nbsp;</span>
789
- <span class="cline-any cline-neutral">&nbsp;</span>
790
- <span class="cline-any cline-yes">1</span>
791
- <span class="cline-any cline-yes">1</span>
792
- <span class="cline-any cline-neutral">&nbsp;</span>
793
- <span class="cline-any cline-neutral">&nbsp;</span>
794
- <span class="cline-any cline-yes">1</span>
795
- <span class="cline-any cline-yes">1</span>
796
- <span class="cline-any cline-neutral">&nbsp;</span>
797
- <span class="cline-any cline-neutral">&nbsp;</span>
798
- <span class="cline-any cline-yes">1</span>
799
- <span class="cline-any cline-neutral">&nbsp;</span>
800
- <span class="cline-any cline-neutral">&nbsp;</span>
801
- <span class="cline-any cline-yes">1</span>
802
- <span class="cline-any cline-yes">1</span>
803
- <span class="cline-any cline-neutral">&nbsp;</span>
804
- <span class="cline-any cline-neutral">&nbsp;</span>
805
- <span class="cline-any cline-yes">1</span>
806
- <span class="cline-any cline-neutral">&nbsp;</span>
807
- <span class="cline-any cline-neutral">&nbsp;</span>
808
- <span class="cline-any cline-neutral">&nbsp;</span>
809
- <span class="cline-any cline-yes">1</span>
810
- <span class="cline-any cline-yes">1</span>
811
- <span class="cline-any cline-neutral">&nbsp;</span>
812
- <span class="cline-any cline-yes">1</span>
813
- <span class="cline-any cline-yes">1</span>
814
- <span class="cline-any cline-yes">1</span>
815
- <span class="cline-any cline-yes">1</span>
816
- <span class="cline-any cline-neutral">&nbsp;</span>
817
- <span class="cline-any cline-yes">1</span>
818
- <span class="cline-any cline-neutral">&nbsp;</span>
819
- <span class="cline-any cline-yes">1</span>
820
- <span class="cline-any cline-yes">1</span>
821
- <span class="cline-any cline-yes">1</span>
822
- <span class="cline-any cline-neutral">&nbsp;</span>
823
- <span class="cline-any cline-yes">1</span>
824
- <span class="cline-any cline-yes">1</span>
825
- <span class="cline-any cline-yes">1</span>
826
- <span class="cline-any cline-neutral">&nbsp;</span>
827
- <span class="cline-any cline-yes">1</span>
828
- <span class="cline-any cline-yes">1</span>
829
- <span class="cline-any cline-yes">1</span>
830
- <span class="cline-any cline-neutral">&nbsp;</span>
831
- <span class="cline-any cline-yes">1</span>
832
- <span class="cline-any cline-yes">1</span>
833
- <span class="cline-any cline-yes">1</span>
834
- <span class="cline-any cline-neutral">&nbsp;</span>
835
- <span class="cline-any cline-neutral">&nbsp;</span>
836
- <span class="cline-any cline-yes">1</span>
837
- <span class="cline-any cline-yes">1</span>
838
- <span class="cline-any cline-neutral">&nbsp;</span>
839
- <span class="cline-any cline-yes">1</span>
840
- <span class="cline-any cline-yes">1</span>
841
- <span class="cline-any cline-neutral">&nbsp;</span>
842
- <span class="cline-any cline-neutral">&nbsp;</span>
843
- <span class="cline-any cline-neutral">&nbsp;</span>
844
- <span class="cline-any cline-yes">1</span>
845
- <span class="cline-any cline-neutral">&nbsp;</span>
846
- <span class="cline-any cline-yes">1</span>
847
- <span class="cline-any cline-neutral">&nbsp;</span>
848
- <span class="cline-any cline-yes">1</span>
849
- <span class="cline-any cline-yes">1</span>
850
- <span class="cline-any cline-neutral">&nbsp;</span>
851
- <span class="cline-any cline-yes">1</span>
852
- <span class="cline-any cline-neutral">&nbsp;</span>
853
- <span class="cline-any cline-neutral">&nbsp;</span>
854
- <span class="cline-any cline-neutral">&nbsp;</span>
855
- <span class="cline-any cline-yes">1</span>
856
- <span class="cline-any cline-neutral">&nbsp;</span>
857
- <span class="cline-any cline-yes">1</span>
858
- <span class="cline-any cline-neutral">&nbsp;</span>
859
- <span class="cline-any cline-yes">1</span>
860
- <span class="cline-any cline-neutral">&nbsp;</span>
861
- <span class="cline-any cline-neutral">&nbsp;</span>
862
- <span class="cline-any cline-neutral">&nbsp;</span>
863
- <span class="cline-any cline-neutral">&nbsp;</span></td><td class="text"><pre class="prettyprint lang-js">/*
864
- Copyright (c) 2011, Rob Ellis, Chris Umbel
865
- &nbsp;
866
- Permission is hereby granted, free of charge, to any person obtaining a copy
867
- of this software and associated documentation files (the "Software"), to deal
868
- in the Software without restriction, including without limitation the rights
869
- to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
870
- copies of the Software, and to permit persons to whom the Software is
871
- furnished to do so, subject to the following conditions:
872
- &nbsp;
873
- The above copyright notice and this permission notice shall be included in
874
- all copies or substantial portions of the Software.
875
- &nbsp;
876
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
877
- IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
878
- FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
879
- AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
880
- LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
881
- OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
882
- THE SOFTWARE.
883
- */
884
- &nbsp;
885
- var TfIdf = require('../lib/natural/tfidf/tfidf');
886
- var tfidf;
887
- &nbsp;
888
- describe('tfidf', function() {
889
- describe('stateless operations', function() {
890
- it('should tf', function() {
891
- expect(TfIdf.tf('document', { document : 2, one : 1 })).toBe(2);
892
- expect(TfIdf.tf('document', { greetings : 1, program : 1 })).toBe(0);
893
- expect(TfIdf.tf('program', { greetings : 1, program : 1 })).toBe(1);
894
- });
895
- });
896
- &nbsp;
897
- describe('keys', function() {
898
- it('should store and recall keys', function() {
899
- tfidf = new TfIdf();
900
- tfidf.addDocument('document one', 'un');
901
- tfidf.addDocument('document Two', 'deux');
902
- &nbsp;
903
- tfidf.tfidfs('two', function(i, tfidf, key) {
904
- if(i === 0)
905
- expect(key).toBe('un');
906
- else
907
- expect(key).toBe('deux');
908
- });
909
- });
910
- &nbsp;
911
- it('should handle a deserialized object passed to the constructor', function () {
912
- tfidf = new TfIdf({
913
- documents: [
914
- {__key: 'un', document: 1, one: 1},
915
- {__key: 'deux', document: 1, two: 1}
916
- ]
917
- });
918
- tfidf.tfidfs('two', function(i, tfidf, key) {
919
- if(i === 0)
920
- expect(key).toBe('un');
921
- else
922
- expect(key).toBe('deux');
923
- });
924
- });
925
- &nbsp;
926
- it('should work when called without a callback', function () {
927
- tfidf = new TfIdf({
928
- documents: [
929
- {__key: 'un', document: 1, one: 1},
930
- {__key: 'deux', document: 1, two: 1}
931
- ]
932
- });
933
- tfidfs = tfidf.tfidfs('two');
934
- expect(tfidfs[1]).toBe(Math.log( 2.0 / 1.0 ));
935
- });
936
- &nbsp;
937
- it('should work with the restoreCache flag set to true', function() {
938
- tfidf = new TfIdf();
939
- tfidf.addDocument('document one', 'un');
940
- expect(tfidf.idf("one")).toBe(Math.log( 1.0 / 1.0 ));
941
- tfidf.addDocument('document Two', 'deux', true);
942
- &nbsp;
943
- tfidf.tfidfs('two', function(i, tfidf, key) {
944
- if(i === 0)
945
- expect(key).toBe('un');
946
- else
947
- expect(key).toBe('deux');
948
- });
949
- });
950
- });
951
- &nbsp;
952
- describe('stateful operations', function() {
953
- beforeEach(function() {
954
- tfidf = new TfIdf();
955
- tfidf.addDocument('document one');
956
- tfidf.addDocument('document Two');
957
- });
958
- &nbsp;
959
- it('should list important terms', function() {
960
- var terms = tfidf.listTerms(0);
961
- expect(terms[0].tfidf).toBeGreaterThan(terms[1].tfidf);
962
- });
963
- });
964
- &nbsp;
965
- describe("special cases", function(){
966
- &nbsp;
967
- // In response to
968
- it("should handle reserved function names correctly in documents", function(){
969
- var reservedWords = [
970
- 'toString',
971
- 'toLocaleString',
972
- 'valueOf',
973
- 'hasOwnProperty',
974
- 'isPrototypeOf',
975
- 'propertyIsEnumerable',
976
- 'constructor'
977
- ];
978
- tfidf = new TfIdf();
979
- tfidf.addDocument(reservedWords.join(" "));
980
- &nbsp;
981
- for(var i in reservedWords) {
982
- expect(tfidf.tfidf(reservedWords[i], 0)).toBe(0);
983
- }
984
- });
985
- &nbsp;
986
- it('should handle an array passed to tfidf()', function() {
987
- tfidf = new TfIdf();
988
- var terms = ['this', 'document', 'is', 'about', 'poetry'];
989
- tfidf.addDocument(terms.join(" "));
990
- expect(tfidf.tfidf(terms, 0)).toBe(0);
991
- });
992
- });
993
- &nbsp;
994
- describe("correct calculations", function(){
995
- &nbsp;
996
- it("should compute idf correctly", function(){
997
- &nbsp;
998
- tfidf = new TfIdf();
999
- tfidf.addDocument('this document is about node.');
1000
- tfidf.addDocument('this document is about ruby.');
1001
- tfidf.addDocument('this document is about ruby and node.');
1002
- tfidf.addDocument('this document is about node. it has node examples');
1003
- &nbsp;
1004
- expect(tfidf.idf("node")).toBe(Math.log( 4.0 / 3.0 ));
1005
- });
1006
- &nbsp;
1007
- it("should compute idf correctly with non-string documents", function(){
1008
- &nbsp;
1009
- tfidf = new TfIdf();
1010
- tfidf.addDocument('this document is about node.');
1011
- tfidf.addDocument('this document is about ruby.');
1012
- tfidf.addDocument('this document is about ruby and node.');
1013
- tfidf.addDocument('this document is about node. it has node examples');
1014
- tfidf.addDocument({text: 'this document is about python'});
1015
- tfidf.addDocument(['this', 'document', 'is', 'about', 'node', 'and', 'JavaScript']);
1016
- &nbsp;
1017
- expect(tfidf.idf("node")).toBe(Math.log( 6.0 / 4.0 ));
1018
- });
1019
- &nbsp;
1020
- it("should compute tf correctly", function(){
1021
- expect(TfIdf.tf("node", {this:1, document:1, is:1, about:1, node:1})).toBe(1);
1022
- expect(TfIdf.tf("node", {this:1, document:1, is:1, about:1, ruby:1})).toBe(0);
1023
- expect(TfIdf.tf("node", {this:1, document:1, is:1, about:1, ruby:1, and:1, node:1})).toBe(1);
1024
- expect(TfIdf.tf("node", {this:1, document:1, is:1, about:1, node:2, it:1, has:1, examples:1})).toBe(2);
1025
- });
1026
- &nbsp;
1027
- // This is a test of the use case outlined in the readme.
1028
- it("should compute tf-idf correctly", function(){
1029
- &nbsp;
1030
- var correctCalculations = [
1031
- 1 * Math.log( 4.0 / 3.0 ),
1032
- 0,
1033
- 2 * Math.log( 4.0 / 3.0 ),
1034
- 1 * Math.log( 4.0 / 2.0 )
1035
- ];
1036
- &nbsp;
1037
- tfidf = new TfIdf();
1038
- tfidf.addDocument('this document is about node.', {node: 0, ruby:1});
1039
- tfidf.addDocument('this document is about ruby.', {node:1, ruby:3});
1040
- tfidf.addDocument('this document is about ruby and node.', {node:0, ruby:3});
1041
- tfidf.addDocument('this document is about node. it has node examples', {node:2, ruby:1});
1042
- &nbsp;
1043
- tfidf.tfidfs('node', function(i, measure, k) {
1044
- expect(measure).toBe(correctCalculations[k.node]);
1045
- });
1046
- &nbsp;
1047
- tfidf.tfidfs('ruby', function(i, measure, k) {
1048
- expect(measure).toBe(correctCalculations[k.ruby]);
1049
- });
1050
- &nbsp;
1051
- });
1052
- &nbsp;
1053
- it("should not return NaN if a term is not present in any documents", function() {
1054
- tfidf = new TfIdf();
1055
- tfidf.addDocument('this document is about node.');
1056
- &nbsp;
1057
- expect(tfidf.tfidf('ruby', 0)).toBe(0);
1058
- });
1059
- &nbsp;
1060
- // This test assures that tf-idf is computed correctly before and after a document is added
1061
- // Computes and tests a few tf-idfs, then adds a document and ensures that those terms tf-idf value
1062
- // is updated accordingly.
1063
- it("should update a terms tf-idf score after adding documents", function(){
1064
- &nbsp;
1065
- tfidf = new TfIdf();
1066
- &nbsp;
1067
- // Add 2 documents
1068
- tfidf.addDocument('this document is about node.', 0);
1069
- tfidf.addDocument('this document is about ruby.', 1);
1070
- &nbsp;
1071
- // check the tf-idf for 'node'
1072
- expect( tfidf.tfidf("node", 0) ).toBe( 1 * Math.log( 2.0 / 1.0 ) );
1073
- &nbsp;
1074
- // Add 2 more documents
1075
- tfidf.addDocument('this document is about ruby and node.');
1076
- tfidf.addDocument('this document is about node. it has node examples');
1077
- &nbsp;
1078
- // Ensure that the tf-idf in the same document has changed to reflect the new idf.
1079
- expect( tfidf.tfidf("node", 0) ).toBe( 1 * Math.log( 4.0 / 3.0 ) );
1080
- });
1081
- &nbsp;
1082
- // Test tf-idf computation on files loaded using readFileSync
1083
- it("should load documents from files", function(){
1084
- &nbsp;
1085
- tfidf = new TfIdf();
1086
- &nbsp;
1087
- tfidf.addFileSync("spec/test_data/tfidf_document1.txt", null, {node: 0, ruby:1});
1088
- tfidf.addFileSync("spec/test_data/tfidf_document2.txt", null, {node:1, ruby:3});
1089
- tfidf.addFileSync("spec/test_data/tfidf_document3.txt", null, {node:0, ruby:3});
1090
- tfidf.addFileSync("spec/test_data/tfidf_document4.txt", null, {node:2, ruby:1});
1091
- &nbsp;
1092
- var correctCalculations = [
1093
- 1 * Math.log( 4.0 / 3.0 ),
1094
- 0,
1095
- 2 * Math.log( 4.0 / 3.0 ),
1096
- 1 * Math.log( 4.0 / 2.0 )
1097
- ];
1098
- &nbsp;
1099
- tfidf.tfidfs('node', function(i, measure, k) {
1100
- expect(measure).toBe(correctCalculations[k.node]);
1101
- });
1102
- &nbsp;
1103
- tfidf.tfidfs('ruby', function(i, measure, k) {
1104
- expect(measure).toBe(correctCalculations[k.ruby]);
1105
- });
1106
- &nbsp;
1107
- // addFileSync with restoreCache flag set to true.
1108
- tfidf.addFileSync("spec/test_data/tfidf_document4.txt", null, {node:2, ruby:1}, true);
1109
- });
1110
- &nbsp;
1111
- // Test idf caching when adding documents from addFileSync
1112
- it("should update a terms tf-idf score after adding documents from addFileSync", function(){
1113
- tfidf = new TfIdf();
1114
- &nbsp;
1115
- // Add 2 documents
1116
- tfidf.addFileSync("spec/test_data/tfidf_document1.txt", null, 0);
1117
- tfidf.addFileSync("spec/test_data/tfidf_document2.txt", null, 1);
1118
- &nbsp;
1119
- // check the tf-idf for 'node'
1120
- expect( tfidf.tfidf("node", 0) ).toBe( 1 * Math.log( 2.0 / 1.0 ) );
1121
- &nbsp;
1122
- // Add 2 more documents
1123
- tfidf.addFileSync("spec/test_data/tfidf_document3.txt");
1124
- tfidf.addFileSync("spec/test_data/tfidf_document4.txt");
1125
- &nbsp;
1126
- // Ensure that the tf-idf in the same document has changed to reflect the new idf.
1127
- expect( tfidf.tfidf("node", 0) ).toBe( 1 * Math.log( 4.0 / 3.0 ) );
1128
- });
1129
- &nbsp;
1130
- // Test idf.setTokenizer
1131
- it('should allow for specific types of Tokenizers', function(){
1132
- tfidf = new TfIdf();
1133
- &nbsp;
1134
- tfidf.addDocument('this document isn\'t about node.', 0);
1135
- tfidf.addDocument('that doc is about node.', 1);
1136
- expect( tfidf.tfidf('n\'t', 0) ).toBe(0);
1137
- expect( tfidf.tfidf('isn', 0) ).toBe( 1 * Math.log( 2 / 1 ) );
1138
- &nbsp;
1139
- tfidf = new TfIdf();
1140
- &nbsp;
1141
- tfidf.addDocument('this document isn\'t about node.', 0);
1142
- tfidf.addDocument('this document isn\'t about node.', 1);
1143
- expect( tfidf.tfidf('isn', 0) ).toBe(0);
1144
- &nbsp;
1145
- tfidf = new TfIdf();
1146
- var TreebankWordTokenizer = require('../lib/natural/tokenizers/treebank_word_tokenizer');
1147
- var tokenizer = new TreebankWordTokenizer();
1148
- &nbsp;
1149
- tfidf.addDocument('this document isn\'t about node.', 0);
1150
- tfidf.setTokenizer(tokenizer);
1151
- tfidf.addDocument('this doc isn\'t about node.', 1);
1152
- &nbsp;
1153
- expect( tfidf.tfidf('isn', 0) ).toBe( 1 * Math.log( 2 / 1 ) );
1154
- expect( tfidf.tfidf('n\'t', 1) ).toBe( 1 * Math.log( 2 / 1 ) );
1155
- expect( tfidf.tfidf('isn', 1) ).toBe(0);
1156
- });
1157
- &nbsp;
1158
- it('should require a valid tokenizer when using setTokenizer', function(){
1159
- tfidf = new TfIdf();
1160
- &nbsp;
1161
- expect( function() { tfidf.setTokenizer(1); } ).toThrow(new Error('Expected a valid Tokenizer'));
1162
- expect( function() { tfidf.setTokenizer({}); } ).toThrow(new Error('Expected a valid Tokenizer'));
1163
- });
1164
- &nbsp;
1165
- // Test encoding for addFileSync
1166
- it('should use the specified encoding for addFileSync', function(){
1167
- &nbsp;
1168
- tfidf = new TfIdf();
1169
- &nbsp;
1170
- tfidf.addFileSync('spec/test_data/tfidf_document1.txt', 'base64');
1171
- tfidf.addFileSync('spec/test_data/tfidf_document1.txt', 'utf8');
1172
- &nbsp;
1173
- expect( tfidf.tfidf('dghpcybkb2n1bwvudcbpcybhym91dcbub2rllg', 0) ).toBe( 1 * Math.log( 2.0 / 1.0 ) );
1174
- });
1175
- &nbsp;
1176
- // Test encoding check for addFileSync
1177
- it('should require a valid encoding for addFileSync', function(){
1178
- &nbsp;
1179
- tfidf = new TfIdf();
1180
- &nbsp;
1181
- expect( function() { tfidf.addFileSync('spec/test_data/tfidf_document1.txt', 'foobar'); } ).toThrow(new Error('Invalid encoding: foobar'));
1182
- });
1183
- });
1184
- });
1185
- &nbsp;</pre></td></tr>
1186
- </table></pre>
1187
-
1188
- </div>
1189
- <div class="footer">
1190
- <div class="meta">Generated by <a href="http://istanbul-js.org/" target="_blank">istanbul</a> at Mon Dec 22 2014 18:17:21 GMT-0500 (EST)</div>
1191
- </div>
1192
-
1193
- <script src="../prettify.js"></script>
1194
-
1195
- <script src="https://yui-s.yahooapis.com/3.6.0/build/yui/yui-min.js"></script>
1196
- <script>
1197
-
1198
- YUI().use('datatable', function (Y) {
1199
-
1200
- var formatters = {
1201
- pct: function (o) {
1202
- o.className += o.record.get('classes')[o.column.key];
1203
- try {
1204
- return o.value.toFixed(2) + '%';
1205
- } catch (ex) { return o.value + '%'; }
1206
- },
1207
- html: function (o) {
1208
- o.className += o.record.get('classes')[o.column.key];
1209
- return o.record.get(o.column.key + '_html');
1210
- }
1211
- },
1212
- defaultFormatter = function (o) {
1213
- o.className += o.record.get('classes')[o.column.key];
1214
- return o.value;
1215
- };
1216
-
1217
- function getColumns(theadNode) {
1218
- var colNodes = theadNode.all('tr th'),
1219
- cols = [],
1220
- col;
1221
- colNodes.each(function (colNode) {
1222
- col = {
1223
- key: colNode.getAttribute('data-col'),
1224
- label: colNode.get('innerHTML') || ' ',
1225
- sortable: !colNode.getAttribute('data-nosort'),
1226
- className: colNode.getAttribute('class'),
1227
- type: colNode.getAttribute('data-type'),
1228
- allowHTML: colNode.getAttribute('data-html') === 'true' || colNode.getAttribute('data-fmt') === 'html'
1229
- };
1230
- col.formatter = formatters[colNode.getAttribute('data-fmt')] || defaultFormatter;
1231
- cols.push(col);
1232
- });
1233
- return cols;
1234
- }
1235
-
1236
- function getRowData(trNode, cols) {
1237
- var tdNodes = trNode.all('td'),
1238
- i,
1239
- row = { classes: {} },
1240
- node,
1241
- name;
1242
- for (i = 0; i < cols.length; i += 1) {
1243
- name = cols[i].key;
1244
- node = tdNodes.item(i);
1245
- row[name] = node.getAttribute('data-value') || node.get('innerHTML');
1246
- row[name + '_html'] = node.get('innerHTML');
1247
- row.classes[name] = node.getAttribute('class');
1248
- //Y.log('Name: ' + name + '; Value: ' + row[name]);
1249
- if (cols[i].type === 'number') { row[name] = row[name] * 1; }
1250
- }
1251
- //Y.log(row);
1252
- return row;
1253
- }
1254
-
1255
- function getData(tbodyNode, cols) {
1256
- var data = [];
1257
- tbodyNode.all('tr').each(function (trNode) {
1258
- data.push(getRowData(trNode, cols));
1259
- });
1260
- return data;
1261
- }
1262
-
1263
- function replaceTable(node) {
1264
- if (!node) { return; }
1265
- var cols = getColumns(node.one('thead')),
1266
- data = getData(node.one('tbody'), cols),
1267
- table,
1268
- parent = node.get('parentNode');
1269
-
1270
- table = new Y.DataTable({
1271
- columns: cols,
1272
- data: data,
1273
- sortBy: 'file'
1274
- });
1275
- parent.set('innerHTML', '');
1276
- table.render(parent);
1277
- }
1278
-
1279
- Y.on('domready', function () {
1280
- replaceTable(Y.one('div.coverage-summary table'));
1281
- if (typeof prettyPrint === 'function') {
1282
- prettyPrint();
1283
- }
1284
- });
1285
- });
1286
- </script>
1287
- </body>
1288
- </html>