semantic-compressor 2.4__tar.gz → 2.6__tar.gz

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 (203) hide show
  1. {semantic_compressor-2.4/semantic_compressor.egg-info → semantic_compressor-2.6}/PKG-INFO +5 -4
  2. semantic_compressor-2.6/compressor/semantic.py +696 -0
  3. {semantic_compressor-2.4 → semantic_compressor-2.6}/pyproject.toml +4 -4
  4. {semantic_compressor-2.4 → semantic_compressor-2.6/semantic_compressor.egg-info}/PKG-INFO +5 -4
  5. {semantic_compressor-2.4 → semantic_compressor-2.6}/semantic_compressor.egg-info/SOURCES.txt +14 -2
  6. {semantic_compressor-2.4 → semantic_compressor-2.6}/semantic_compressor.egg-info/requires.txt +2 -2
  7. {semantic_compressor-2.4 → semantic_compressor-2.6}/semantic_compressor.egg-info/top_level.txt +1 -0
  8. {semantic_compressor-2.4 → semantic_compressor-2.6}/setup.py +3 -3
  9. semantic_compressor-2.6/tests/__init__.py +0 -0
  10. semantic_compressor-2.6/tests/conftest.py +92 -0
  11. semantic_compressor-2.6/tests/test_abstractive_enhancements.py +302 -0
  12. semantic_compressor-2.6/tests/test_benchmark.py +187 -0
  13. semantic_compressor-2.6/tests/test_clean_text.py +90 -0
  14. semantic_compressor-2.6/tests/test_compression.py +123 -0
  15. semantic_compressor-2.6/tests/test_correct_spelling.py +30 -0
  16. semantic_compressor-2.6/tests/test_embeddings.py +74 -0
  17. semantic_compressor-2.6/tests/test_find_needle_in_haystack.py +147 -0
  18. semantic_compressor-2.6/tests/test_language_and_stemming.py +74 -0
  19. semantic_compressor-2.6/tests/test_ngrams.py +60 -0
  20. semantic_compressor-2.6/tests/test_semantic_embeddings.py +39 -0
  21. semantic_compressor-2.6/tests/test_tokenizer.py +70 -0
  22. semantic_compressor-2.4/compressor/resources/lid.176.ftz +0 -0
  23. semantic_compressor-2.4/compressor/semantic.py +0 -408
  24. {semantic_compressor-2.4 → semantic_compressor-2.6}/LICENSE +0 -0
  25. {semantic_compressor-2.4 → semantic_compressor-2.6}/MANIFEST.in +0 -0
  26. {semantic_compressor-2.4 → semantic_compressor-2.6}/README.md +0 -0
  27. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/__init__.py +0 -0
  28. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/minbpe/__init__.py +0 -0
  29. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/minbpe/base.py +0 -0
  30. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/minbpe/basic.py +0 -0
  31. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/minbpe/regex.py +0 -0
  32. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/__init__.py +0 -0
  33. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/en_stopwords.pkl +0 -0
  34. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/README +0 -0
  35. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/albanian +0 -0
  36. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/arabic +0 -0
  37. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/azerbaijani +0 -0
  38. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/basque +0 -0
  39. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/belarusian +0 -0
  40. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/bengali +0 -0
  41. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/catalan +0 -0
  42. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/chinese +0 -0
  43. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/danish +0 -0
  44. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/dutch +0 -0
  45. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/english +0 -0
  46. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/finnish +0 -0
  47. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/french +0 -0
  48. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/german +0 -0
  49. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/greek +0 -0
  50. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/hebrew +0 -0
  51. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/hinglish +0 -0
  52. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/hungarian +0 -0
  53. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/indonesian +0 -0
  54. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/italian +0 -0
  55. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/kazakh +0 -0
  56. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/nepali +0 -0
  57. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/norwegian +0 -0
  58. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/portuguese +0 -0
  59. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/romanian +0 -0
  60. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/russian +0 -0
  61. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/slovene +0 -0
  62. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/spanish +0 -0
  63. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/swedish +0 -0
  64. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/tajik +0 -0
  65. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/tamil +0 -0
  66. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords/turkish +0 -0
  67. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/stopwords.zip +0 -0
  68. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/wordnet.zip +0 -0
  69. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/words/README +0 -0
  70. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/words/en +0 -0
  71. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/corpora/words/en-basic +0 -0
  72. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step0.pt +0 -0
  73. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step1.pt +0 -0
  74. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step2.pt +0 -0
  75. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step3.pt +0 -0
  76. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step4.pt +0 -0
  77. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step5.pt +0 -0
  78. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp/step6.pt +0 -0
  79. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/stemmers/rslp.zip +0 -0
  80. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/taggers/averaged_perceptron_tagger/averaged_perceptron_tagger.pickle +0 -0
  81. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/.DS_Store +0 -0
  82. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/README +0 -0
  83. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/czech.pickle +0 -0
  84. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/danish.pickle +0 -0
  85. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/dutch.pickle +0 -0
  86. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/english.pickle +0 -0
  87. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/estonian.pickle +0 -0
  88. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/finnish.pickle +0 -0
  89. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/french.pickle +0 -0
  90. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/german.pickle +0 -0
  91. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/greek.pickle +0 -0
  92. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/italian.pickle +0 -0
  93. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/malayalam.pickle +0 -0
  94. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/norwegian.pickle +0 -0
  95. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/polish.pickle +0 -0
  96. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/portuguese.pickle +0 -0
  97. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/russian.pickle +0 -0
  98. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/slovene.pickle +0 -0
  99. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/spanish.pickle +0 -0
  100. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/swedish.pickle +0 -0
  101. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/PY3/turkish.pickle +0 -0
  102. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/README +0 -0
  103. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/czech.pickle +0 -0
  104. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/danish.pickle +0 -0
  105. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/dutch.pickle +0 -0
  106. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/english.pickle +0 -0
  107. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/estonian.pickle +0 -0
  108. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/finnish.pickle +0 -0
  109. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/french.pickle +0 -0
  110. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/german.pickle +0 -0
  111. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/greek.pickle +0 -0
  112. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/italian.pickle +0 -0
  113. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/malayalam.pickle +0 -0
  114. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/norwegian.pickle +0 -0
  115. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/polish.pickle +0 -0
  116. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/portuguese.pickle +0 -0
  117. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/russian.pickle +0 -0
  118. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/slovene.pickle +0 -0
  119. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/spanish.pickle +0 -0
  120. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/swedish.pickle +0 -0
  121. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt/turkish.pickle +0 -0
  122. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt.zip +0 -0
  123. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/README +0 -0
  124. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/czech/abbrev_types.txt +0 -0
  125. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/czech/collocations.tab +0 -0
  126. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/czech/ortho_context.tab +0 -0
  127. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/czech/sent_starters.txt +0 -0
  128. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/danish/abbrev_types.txt +0 -0
  129. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/danish/collocations.tab +0 -0
  130. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/danish/ortho_context.tab +0 -0
  131. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/danish/sent_starters.txt +0 -0
  132. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/dutch/abbrev_types.txt +0 -0
  133. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/dutch/collocations.tab +0 -0
  134. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/dutch/ortho_context.tab +0 -0
  135. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/dutch/sent_starters.txt +0 -0
  136. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/english/abbrev_types.txt +0 -0
  137. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/english/collocations.tab +0 -0
  138. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/english/ortho_context.tab +0 -0
  139. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/english/sent_starters.txt +0 -0
  140. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/estonian/abbrev_types.txt +0 -0
  141. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/estonian/collocations.tab +0 -0
  142. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/estonian/ortho_context.tab +0 -0
  143. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/estonian/sent_starters.txt +0 -0
  144. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/finnish/abbrev_types.txt +0 -0
  145. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/finnish/collocations.tab +0 -0
  146. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/finnish/ortho_context.tab +0 -0
  147. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/finnish/sent_starters.txt +0 -0
  148. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/french/abbrev_types.txt +0 -0
  149. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/french/collocations.tab +0 -0
  150. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/french/ortho_context.tab +0 -0
  151. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/french/sent_starters.txt +0 -0
  152. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/german/abbrev_types.txt +0 -0
  153. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/german/collocations.tab +0 -0
  154. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/german/ortho_context.tab +0 -0
  155. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/german/sent_starters.txt +0 -0
  156. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/greek/abbrev_types.txt +0 -0
  157. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/greek/collocations.tab +0 -0
  158. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/greek/ortho_context.tab +0 -0
  159. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/greek/sent_starters.txt +0 -0
  160. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/italian/abbrev_types.txt +0 -0
  161. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/italian/collocations.tab +0 -0
  162. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/italian/ortho_context.tab +0 -0
  163. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/italian/sent_starters.txt +0 -0
  164. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/malayalam/abbrev_types.txt +0 -0
  165. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/malayalam/collocations.tab +0 -0
  166. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/malayalam/ortho_context.tab +0 -0
  167. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/malayalam/sent_starters.txt +0 -0
  168. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/norwegian/abbrev_types.txt +0 -0
  169. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/norwegian/collocations.tab +0 -0
  170. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/norwegian/ortho_context.tab +0 -0
  171. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/norwegian/sent_starters.txt +0 -0
  172. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/polish/abbrev_types.txt +0 -0
  173. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/polish/collocations.tab +0 -0
  174. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/polish/ortho_context.tab +0 -0
  175. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/polish/sent_starters.txt +0 -0
  176. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/portuguese/abbrev_types.txt +0 -0
  177. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/portuguese/collocations.tab +0 -0
  178. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/portuguese/ortho_context.tab +0 -0
  179. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/portuguese/sent_starters.txt +0 -0
  180. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/russian/abbrev_types.txt +0 -0
  181. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/russian/collocations.tab +0 -0
  182. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/russian/ortho_context.tab +0 -0
  183. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/russian/sent_starters.txt +0 -0
  184. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/slovene/abbrev_types.txt +0 -0
  185. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/slovene/collocations.tab +0 -0
  186. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/slovene/ortho_context.tab +0 -0
  187. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/slovene/sent_starters.txt +0 -0
  188. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/spanish/abbrev_types.txt +0 -0
  189. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/spanish/collocations.tab +0 -0
  190. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/spanish/ortho_context.tab +0 -0
  191. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/spanish/sent_starters.txt +0 -0
  192. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/swedish/abbrev_types.txt +0 -0
  193. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/swedish/collocations.tab +0 -0
  194. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/swedish/ortho_context.tab +0 -0
  195. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/swedish/sent_starters.txt +0 -0
  196. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/turkish/abbrev_types.txt +0 -0
  197. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/turkish/collocations.tab +0 -0
  198. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/turkish/ortho_context.tab +0 -0
  199. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab/turkish/sent_starters.txt +0 -0
  200. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/nltk_data/tokenizers/punkt_tab.zip +0 -0
  201. {semantic_compressor-2.4 → semantic_compressor-2.6}/compressor/resources/pt_stopwords.pkl +0 -0
  202. {semantic_compressor-2.4 → semantic_compressor-2.6}/semantic_compressor.egg-info/dependency_links.txt +0 -0
  203. {semantic_compressor-2.4 → semantic_compressor-2.6}/setup.cfg +0 -0
{semantic_compressor-2.4/semantic_compressor.egg-info → semantic_compressor-2.6}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
- Metadata-Version: 2.2
1
+ Metadata-Version: 2.4
2
2
  Name: semantic_compressor
3
- Version: 2.4
3
+ Version: 2.6
4
4
  Author: Carlo Moro
5
5
  Author-email: Carlo Moro <cnmoro@gmail.com>
6
6
  Classifier: Programming Language :: Python :: 3
@@ -9,13 +9,14 @@ Classifier: Operating System :: OS Independent
9
9
  Requires-Python: >=3.7
10
10
  Description-Content-Type: text/markdown
11
11
  License-File: LICENSE
12
- Requires-Dist: numpy<2
12
+ Requires-Dist: numpy
13
13
  Requires-Dist: nltk
14
14
  Requires-Dist: scikit-learn
15
- Requires-Dist: fasttext
15
+ Requires-Dist: lingua-language-detector
16
16
  Requires-Dist: model2vec
17
17
  Requires-Dist: pyspellchecker
18
18
  Dynamic: author
19
+ Dynamic: license-file
19
20
  Dynamic: requires-python
20
21
 
21
22
  ```python