kreuzberg 4.0.0.pre.rc.6

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 (330) hide show
  1. checksums.yaml +7 -0
  2. data/.gitignore +14 -0
  3. data/.rspec +3 -0
  4. data/.rubocop.yaml +1 -0
  5. data/.rubocop.yml +538 -0
  6. data/Gemfile +8 -0
  7. data/Gemfile.lock +157 -0
  8. data/README.md +426 -0
  9. data/Rakefile +25 -0
  10. data/Steepfile +47 -0
  11. data/examples/async_patterns.rb +341 -0
  12. data/ext/kreuzberg_rb/extconf.rb +45 -0
  13. data/ext/kreuzberg_rb/native/Cargo.lock +6535 -0
  14. data/ext/kreuzberg_rb/native/Cargo.toml +44 -0
  15. data/ext/kreuzberg_rb/native/README.md +425 -0
  16. data/ext/kreuzberg_rb/native/build.rs +15 -0
  17. data/ext/kreuzberg_rb/native/include/ieeefp.h +11 -0
  18. data/ext/kreuzberg_rb/native/include/msvc_compat/strings.h +14 -0
  19. data/ext/kreuzberg_rb/native/include/strings.h +20 -0
  20. data/ext/kreuzberg_rb/native/include/unistd.h +47 -0
  21. data/ext/kreuzberg_rb/native/src/lib.rs +2998 -0
  22. data/extconf.rb +28 -0
  23. data/kreuzberg.gemspec +148 -0
  24. data/lib/kreuzberg/api_proxy.rb +142 -0
  25. data/lib/kreuzberg/cache_api.rb +46 -0
  26. data/lib/kreuzberg/cli.rb +55 -0
  27. data/lib/kreuzberg/cli_proxy.rb +127 -0
  28. data/lib/kreuzberg/config.rb +691 -0
  29. data/lib/kreuzberg/error_context.rb +32 -0
  30. data/lib/kreuzberg/errors.rb +118 -0
  31. data/lib/kreuzberg/extraction_api.rb +85 -0
  32. data/lib/kreuzberg/mcp_proxy.rb +186 -0
  33. data/lib/kreuzberg/ocr_backend_protocol.rb +113 -0
  34. data/lib/kreuzberg/post_processor_protocol.rb +86 -0
  35. data/lib/kreuzberg/result.rb +216 -0
  36. data/lib/kreuzberg/setup_lib_path.rb +80 -0
  37. data/lib/kreuzberg/validator_protocol.rb +89 -0
  38. data/lib/kreuzberg/version.rb +5 -0
  39. data/lib/kreuzberg.rb +103 -0
  40. data/sig/kreuzberg/internal.rbs +184 -0
  41. data/sig/kreuzberg.rbs +520 -0
  42. data/spec/binding/cache_spec.rb +227 -0
  43. data/spec/binding/cli_proxy_spec.rb +85 -0
  44. data/spec/binding/cli_spec.rb +55 -0
  45. data/spec/binding/config_spec.rb +345 -0
  46. data/spec/binding/config_validation_spec.rb +283 -0
  47. data/spec/binding/error_handling_spec.rb +213 -0
  48. data/spec/binding/errors_spec.rb +66 -0
  49. data/spec/binding/plugins/ocr_backend_spec.rb +307 -0
  50. data/spec/binding/plugins/postprocessor_spec.rb +269 -0
  51. data/spec/binding/plugins/validator_spec.rb +274 -0
  52. data/spec/fixtures/config.toml +39 -0
  53. data/spec/fixtures/config.yaml +41 -0
  54. data/spec/fixtures/invalid_config.toml +4 -0
  55. data/spec/smoke/package_spec.rb +178 -0
  56. data/spec/spec_helper.rb +42 -0
  57. data/vendor/kreuzberg/Cargo.toml +204 -0
  58. data/vendor/kreuzberg/README.md +175 -0
  59. data/vendor/kreuzberg/benches/otel_overhead.rs +48 -0
  60. data/vendor/kreuzberg/build.rs +474 -0
  61. data/vendor/kreuzberg/src/api/error.rs +81 -0
  62. data/vendor/kreuzberg/src/api/handlers.rs +199 -0
  63. data/vendor/kreuzberg/src/api/mod.rs +79 -0
  64. data/vendor/kreuzberg/src/api/server.rs +353 -0
  65. data/vendor/kreuzberg/src/api/types.rs +170 -0
  66. data/vendor/kreuzberg/src/cache/mod.rs +1167 -0
  67. data/vendor/kreuzberg/src/chunking/mod.rs +677 -0
  68. data/vendor/kreuzberg/src/core/batch_mode.rs +95 -0
  69. data/vendor/kreuzberg/src/core/config.rs +1032 -0
  70. data/vendor/kreuzberg/src/core/extractor.rs +1024 -0
  71. data/vendor/kreuzberg/src/core/io.rs +329 -0
  72. data/vendor/kreuzberg/src/core/mime.rs +605 -0
  73. data/vendor/kreuzberg/src/core/mod.rs +45 -0
  74. data/vendor/kreuzberg/src/core/pipeline.rs +984 -0
  75. data/vendor/kreuzberg/src/embeddings.rs +432 -0
  76. data/vendor/kreuzberg/src/error.rs +431 -0
  77. data/vendor/kreuzberg/src/extraction/archive.rs +954 -0
  78. data/vendor/kreuzberg/src/extraction/docx.rs +40 -0
  79. data/vendor/kreuzberg/src/extraction/email.rs +854 -0
  80. data/vendor/kreuzberg/src/extraction/excel.rs +688 -0
  81. data/vendor/kreuzberg/src/extraction/html.rs +553 -0
  82. data/vendor/kreuzberg/src/extraction/image.rs +368 -0
  83. data/vendor/kreuzberg/src/extraction/libreoffice.rs +563 -0
  84. data/vendor/kreuzberg/src/extraction/markdown.rs +213 -0
  85. data/vendor/kreuzberg/src/extraction/mod.rs +81 -0
  86. data/vendor/kreuzberg/src/extraction/office_metadata/app_properties.rs +398 -0
  87. data/vendor/kreuzberg/src/extraction/office_metadata/core_properties.rs +247 -0
  88. data/vendor/kreuzberg/src/extraction/office_metadata/custom_properties.rs +240 -0
  89. data/vendor/kreuzberg/src/extraction/office_metadata/mod.rs +130 -0
  90. data/vendor/kreuzberg/src/extraction/office_metadata/odt_properties.rs +287 -0
  91. data/vendor/kreuzberg/src/extraction/pptx.rs +3000 -0
  92. data/vendor/kreuzberg/src/extraction/structured.rs +490 -0
  93. data/vendor/kreuzberg/src/extraction/table.rs +328 -0
  94. data/vendor/kreuzberg/src/extraction/text.rs +269 -0
  95. data/vendor/kreuzberg/src/extraction/xml.rs +333 -0
  96. data/vendor/kreuzberg/src/extractors/archive.rs +446 -0
  97. data/vendor/kreuzberg/src/extractors/bibtex.rs +469 -0
  98. data/vendor/kreuzberg/src/extractors/docbook.rs +502 -0
  99. data/vendor/kreuzberg/src/extractors/docx.rs +367 -0
  100. data/vendor/kreuzberg/src/extractors/email.rs +143 -0
  101. data/vendor/kreuzberg/src/extractors/epub.rs +707 -0
  102. data/vendor/kreuzberg/src/extractors/excel.rs +343 -0
  103. data/vendor/kreuzberg/src/extractors/fictionbook.rs +491 -0
  104. data/vendor/kreuzberg/src/extractors/fictionbook.rs.backup2 +738 -0
  105. data/vendor/kreuzberg/src/extractors/html.rs +393 -0
  106. data/vendor/kreuzberg/src/extractors/image.rs +198 -0
  107. data/vendor/kreuzberg/src/extractors/jats.rs +1051 -0
  108. data/vendor/kreuzberg/src/extractors/jupyter.rs +367 -0
  109. data/vendor/kreuzberg/src/extractors/latex.rs +652 -0
  110. data/vendor/kreuzberg/src/extractors/markdown.rs +700 -0
  111. data/vendor/kreuzberg/src/extractors/mod.rs +365 -0
  112. data/vendor/kreuzberg/src/extractors/odt.rs +628 -0
  113. data/vendor/kreuzberg/src/extractors/opml.rs +634 -0
  114. data/vendor/kreuzberg/src/extractors/orgmode.rs +528 -0
  115. data/vendor/kreuzberg/src/extractors/pdf.rs +493 -0
  116. data/vendor/kreuzberg/src/extractors/pptx.rs +248 -0
  117. data/vendor/kreuzberg/src/extractors/rst.rs +576 -0
  118. data/vendor/kreuzberg/src/extractors/rtf.rs +810 -0
  119. data/vendor/kreuzberg/src/extractors/security.rs +484 -0
  120. data/vendor/kreuzberg/src/extractors/security_tests.rs +367 -0
  121. data/vendor/kreuzberg/src/extractors/structured.rs +140 -0
  122. data/vendor/kreuzberg/src/extractors/text.rs +260 -0
  123. data/vendor/kreuzberg/src/extractors/typst.rs +650 -0
  124. data/vendor/kreuzberg/src/extractors/xml.rs +135 -0
  125. data/vendor/kreuzberg/src/image/dpi.rs +164 -0
  126. data/vendor/kreuzberg/src/image/mod.rs +6 -0
  127. data/vendor/kreuzberg/src/image/preprocessing.rs +417 -0
  128. data/vendor/kreuzberg/src/image/resize.rs +89 -0
  129. data/vendor/kreuzberg/src/keywords/config.rs +154 -0
  130. data/vendor/kreuzberg/src/keywords/mod.rs +237 -0
  131. data/vendor/kreuzberg/src/keywords/processor.rs +267 -0
  132. data/vendor/kreuzberg/src/keywords/rake.rs +293 -0
  133. data/vendor/kreuzberg/src/keywords/types.rs +68 -0
  134. data/vendor/kreuzberg/src/keywords/yake.rs +163 -0
  135. data/vendor/kreuzberg/src/language_detection/mod.rs +942 -0
  136. data/vendor/kreuzberg/src/lib.rs +105 -0
  137. data/vendor/kreuzberg/src/mcp/mod.rs +32 -0
  138. data/vendor/kreuzberg/src/mcp/server.rs +1968 -0
  139. data/vendor/kreuzberg/src/ocr/cache.rs +469 -0
  140. data/vendor/kreuzberg/src/ocr/error.rs +37 -0
  141. data/vendor/kreuzberg/src/ocr/hocr.rs +216 -0
  142. data/vendor/kreuzberg/src/ocr/mod.rs +58 -0
  143. data/vendor/kreuzberg/src/ocr/processor.rs +863 -0
  144. data/vendor/kreuzberg/src/ocr/table/mod.rs +4 -0
  145. data/vendor/kreuzberg/src/ocr/table/tsv_parser.rs +144 -0
  146. data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +450 -0
  147. data/vendor/kreuzberg/src/ocr/types.rs +393 -0
  148. data/vendor/kreuzberg/src/ocr/utils.rs +47 -0
  149. data/vendor/kreuzberg/src/ocr/validation.rs +206 -0
  150. data/vendor/kreuzberg/src/panic_context.rs +154 -0
  151. data/vendor/kreuzberg/src/pdf/error.rs +122 -0
  152. data/vendor/kreuzberg/src/pdf/images.rs +139 -0
  153. data/vendor/kreuzberg/src/pdf/metadata.rs +346 -0
  154. data/vendor/kreuzberg/src/pdf/mod.rs +50 -0
  155. data/vendor/kreuzberg/src/pdf/rendering.rs +369 -0
  156. data/vendor/kreuzberg/src/pdf/table.rs +393 -0
  157. data/vendor/kreuzberg/src/pdf/text.rs +158 -0
  158. data/vendor/kreuzberg/src/plugins/extractor.rs +1013 -0
  159. data/vendor/kreuzberg/src/plugins/mod.rs +209 -0
  160. data/vendor/kreuzberg/src/plugins/ocr.rs +620 -0
  161. data/vendor/kreuzberg/src/plugins/processor.rs +642 -0
  162. data/vendor/kreuzberg/src/plugins/registry.rs +1337 -0
  163. data/vendor/kreuzberg/src/plugins/traits.rs +258 -0
  164. data/vendor/kreuzberg/src/plugins/validator.rs +956 -0
  165. data/vendor/kreuzberg/src/stopwords/mod.rs +1470 -0
  166. data/vendor/kreuzberg/src/text/mod.rs +19 -0
  167. data/vendor/kreuzberg/src/text/quality.rs +697 -0
  168. data/vendor/kreuzberg/src/text/string_utils.rs +217 -0
  169. data/vendor/kreuzberg/src/text/token_reduction/cjk_utils.rs +164 -0
  170. data/vendor/kreuzberg/src/text/token_reduction/config.rs +100 -0
  171. data/vendor/kreuzberg/src/text/token_reduction/core.rs +796 -0
  172. data/vendor/kreuzberg/src/text/token_reduction/filters.rs +902 -0
  173. data/vendor/kreuzberg/src/text/token_reduction/mod.rs +160 -0
  174. data/vendor/kreuzberg/src/text/token_reduction/semantic.rs +619 -0
  175. data/vendor/kreuzberg/src/text/token_reduction/simd_text.rs +147 -0
  176. data/vendor/kreuzberg/src/types.rs +903 -0
  177. data/vendor/kreuzberg/src/utils/mod.rs +17 -0
  178. data/vendor/kreuzberg/src/utils/quality.rs +959 -0
  179. data/vendor/kreuzberg/src/utils/string_utils.rs +381 -0
  180. data/vendor/kreuzberg/stopwords/af_stopwords.json +53 -0
  181. data/vendor/kreuzberg/stopwords/ar_stopwords.json +482 -0
  182. data/vendor/kreuzberg/stopwords/bg_stopwords.json +261 -0
  183. data/vendor/kreuzberg/stopwords/bn_stopwords.json +400 -0
  184. data/vendor/kreuzberg/stopwords/br_stopwords.json +1205 -0
  185. data/vendor/kreuzberg/stopwords/ca_stopwords.json +280 -0
  186. data/vendor/kreuzberg/stopwords/cs_stopwords.json +425 -0
  187. data/vendor/kreuzberg/stopwords/da_stopwords.json +172 -0
  188. data/vendor/kreuzberg/stopwords/de_stopwords.json +622 -0
  189. data/vendor/kreuzberg/stopwords/el_stopwords.json +849 -0
  190. data/vendor/kreuzberg/stopwords/en_stopwords.json +1300 -0
  191. data/vendor/kreuzberg/stopwords/eo_stopwords.json +175 -0
  192. data/vendor/kreuzberg/stopwords/es_stopwords.json +734 -0
  193. data/vendor/kreuzberg/stopwords/et_stopwords.json +37 -0
  194. data/vendor/kreuzberg/stopwords/eu_stopwords.json +100 -0
  195. data/vendor/kreuzberg/stopwords/fa_stopwords.json +801 -0
  196. data/vendor/kreuzberg/stopwords/fi_stopwords.json +849 -0
  197. data/vendor/kreuzberg/stopwords/fr_stopwords.json +693 -0
  198. data/vendor/kreuzberg/stopwords/ga_stopwords.json +111 -0
  199. data/vendor/kreuzberg/stopwords/gl_stopwords.json +162 -0
  200. data/vendor/kreuzberg/stopwords/gu_stopwords.json +226 -0
  201. data/vendor/kreuzberg/stopwords/ha_stopwords.json +41 -0
  202. data/vendor/kreuzberg/stopwords/he_stopwords.json +196 -0
  203. data/vendor/kreuzberg/stopwords/hi_stopwords.json +227 -0
  204. data/vendor/kreuzberg/stopwords/hr_stopwords.json +181 -0
  205. data/vendor/kreuzberg/stopwords/hu_stopwords.json +791 -0
  206. data/vendor/kreuzberg/stopwords/hy_stopwords.json +47 -0
  207. data/vendor/kreuzberg/stopwords/id_stopwords.json +760 -0
  208. data/vendor/kreuzberg/stopwords/it_stopwords.json +634 -0
  209. data/vendor/kreuzberg/stopwords/ja_stopwords.json +136 -0
  210. data/vendor/kreuzberg/stopwords/kn_stopwords.json +84 -0
  211. data/vendor/kreuzberg/stopwords/ko_stopwords.json +681 -0
  212. data/vendor/kreuzberg/stopwords/ku_stopwords.json +64 -0
  213. data/vendor/kreuzberg/stopwords/la_stopwords.json +51 -0
  214. data/vendor/kreuzberg/stopwords/lt_stopwords.json +476 -0
  215. data/vendor/kreuzberg/stopwords/lv_stopwords.json +163 -0
  216. data/vendor/kreuzberg/stopwords/ml_stopwords.json +1 -0
  217. data/vendor/kreuzberg/stopwords/mr_stopwords.json +101 -0
  218. data/vendor/kreuzberg/stopwords/ms_stopwords.json +477 -0
  219. data/vendor/kreuzberg/stopwords/ne_stopwords.json +490 -0
  220. data/vendor/kreuzberg/stopwords/nl_stopwords.json +415 -0
  221. data/vendor/kreuzberg/stopwords/no_stopwords.json +223 -0
  222. data/vendor/kreuzberg/stopwords/pl_stopwords.json +331 -0
  223. data/vendor/kreuzberg/stopwords/pt_stopwords.json +562 -0
  224. data/vendor/kreuzberg/stopwords/ro_stopwords.json +436 -0
  225. data/vendor/kreuzberg/stopwords/ru_stopwords.json +561 -0
  226. data/vendor/kreuzberg/stopwords/si_stopwords.json +193 -0
  227. data/vendor/kreuzberg/stopwords/sk_stopwords.json +420 -0
  228. data/vendor/kreuzberg/stopwords/sl_stopwords.json +448 -0
  229. data/vendor/kreuzberg/stopwords/so_stopwords.json +32 -0
  230. data/vendor/kreuzberg/stopwords/st_stopwords.json +33 -0
  231. data/vendor/kreuzberg/stopwords/sv_stopwords.json +420 -0
  232. data/vendor/kreuzberg/stopwords/sw_stopwords.json +76 -0
  233. data/vendor/kreuzberg/stopwords/ta_stopwords.json +129 -0
  234. data/vendor/kreuzberg/stopwords/te_stopwords.json +54 -0
  235. data/vendor/kreuzberg/stopwords/th_stopwords.json +118 -0
  236. data/vendor/kreuzberg/stopwords/tl_stopwords.json +149 -0
  237. data/vendor/kreuzberg/stopwords/tr_stopwords.json +506 -0
  238. data/vendor/kreuzberg/stopwords/uk_stopwords.json +75 -0
  239. data/vendor/kreuzberg/stopwords/ur_stopwords.json +519 -0
  240. data/vendor/kreuzberg/stopwords/vi_stopwords.json +647 -0
  241. data/vendor/kreuzberg/stopwords/yo_stopwords.json +62 -0
  242. data/vendor/kreuzberg/stopwords/zh_stopwords.json +796 -0
  243. data/vendor/kreuzberg/stopwords/zu_stopwords.json +31 -0
  244. data/vendor/kreuzberg/tests/api_extract_multipart.rs +52 -0
  245. data/vendor/kreuzberg/tests/api_tests.rs +966 -0
  246. data/vendor/kreuzberg/tests/archive_integration.rs +543 -0
  247. data/vendor/kreuzberg/tests/batch_orchestration.rs +556 -0
  248. data/vendor/kreuzberg/tests/batch_processing.rs +316 -0
  249. data/vendor/kreuzberg/tests/bibtex_parity_test.rs +421 -0
  250. data/vendor/kreuzberg/tests/concurrency_stress.rs +525 -0
  251. data/vendor/kreuzberg/tests/config_features.rs +598 -0
  252. data/vendor/kreuzberg/tests/config_loading_tests.rs +415 -0
  253. data/vendor/kreuzberg/tests/core_integration.rs +510 -0
  254. data/vendor/kreuzberg/tests/csv_integration.rs +414 -0
  255. data/vendor/kreuzberg/tests/docbook_extractor_tests.rs +498 -0
  256. data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +122 -0
  257. data/vendor/kreuzberg/tests/docx_vs_pandoc_comparison.rs +370 -0
  258. data/vendor/kreuzberg/tests/email_integration.rs +325 -0
  259. data/vendor/kreuzberg/tests/epub_native_extractor_tests.rs +275 -0
  260. data/vendor/kreuzberg/tests/error_handling.rs +393 -0
  261. data/vendor/kreuzberg/tests/fictionbook_extractor_tests.rs +228 -0
  262. data/vendor/kreuzberg/tests/format_integration.rs +159 -0
  263. data/vendor/kreuzberg/tests/helpers/mod.rs +142 -0
  264. data/vendor/kreuzberg/tests/html_table_test.rs +551 -0
  265. data/vendor/kreuzberg/tests/image_integration.rs +253 -0
  266. data/vendor/kreuzberg/tests/instrumentation_test.rs +139 -0
  267. data/vendor/kreuzberg/tests/jats_extractor_tests.rs +639 -0
  268. data/vendor/kreuzberg/tests/jupyter_extractor_tests.rs +704 -0
  269. data/vendor/kreuzberg/tests/keywords_integration.rs +479 -0
  270. data/vendor/kreuzberg/tests/keywords_quality.rs +509 -0
  271. data/vendor/kreuzberg/tests/latex_extractor_tests.rs +496 -0
  272. data/vendor/kreuzberg/tests/markdown_extractor_tests.rs +490 -0
  273. data/vendor/kreuzberg/tests/mime_detection.rs +428 -0
  274. data/vendor/kreuzberg/tests/ocr_configuration.rs +510 -0
  275. data/vendor/kreuzberg/tests/ocr_errors.rs +676 -0
  276. data/vendor/kreuzberg/tests/ocr_quality.rs +627 -0
  277. data/vendor/kreuzberg/tests/ocr_stress.rs +469 -0
  278. data/vendor/kreuzberg/tests/odt_extractor_tests.rs +695 -0
  279. data/vendor/kreuzberg/tests/opml_extractor_tests.rs +616 -0
  280. data/vendor/kreuzberg/tests/orgmode_extractor_tests.rs +822 -0
  281. data/vendor/kreuzberg/tests/pdf_integration.rs +43 -0
  282. data/vendor/kreuzberg/tests/pipeline_integration.rs +1411 -0
  283. data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +771 -0
  284. data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs +560 -0
  285. data/vendor/kreuzberg/tests/plugin_system.rs +921 -0
  286. data/vendor/kreuzberg/tests/plugin_validator_test.rs +783 -0
  287. data/vendor/kreuzberg/tests/registry_integration_tests.rs +586 -0
  288. data/vendor/kreuzberg/tests/rst_extractor_tests.rs +692 -0
  289. data/vendor/kreuzberg/tests/rtf_extractor_tests.rs +776 -0
  290. data/vendor/kreuzberg/tests/security_validation.rs +415 -0
  291. data/vendor/kreuzberg/tests/stopwords_integration_test.rs +888 -0
  292. data/vendor/kreuzberg/tests/test_fastembed.rs +609 -0
  293. data/vendor/kreuzberg/tests/typst_behavioral_tests.rs +1259 -0
  294. data/vendor/kreuzberg/tests/typst_extractor_tests.rs +647 -0
  295. data/vendor/kreuzberg/tests/xlsx_metadata_extraction_test.rs +87 -0
  296. data/vendor/rb-sys/.cargo-ok +1 -0
  297. data/vendor/rb-sys/.cargo_vcs_info.json +6 -0
  298. data/vendor/rb-sys/Cargo.lock +393 -0
  299. data/vendor/rb-sys/Cargo.toml +70 -0
  300. data/vendor/rb-sys/Cargo.toml.orig +57 -0
  301. data/vendor/rb-sys/LICENSE-APACHE +190 -0
  302. data/vendor/rb-sys/LICENSE-MIT +21 -0
  303. data/vendor/rb-sys/bin/release.sh +21 -0
  304. data/vendor/rb-sys/build/features.rs +108 -0
  305. data/vendor/rb-sys/build/main.rs +246 -0
  306. data/vendor/rb-sys/build/stable_api_config.rs +153 -0
  307. data/vendor/rb-sys/build/version.rs +48 -0
  308. data/vendor/rb-sys/readme.md +36 -0
  309. data/vendor/rb-sys/src/bindings.rs +21 -0
  310. data/vendor/rb-sys/src/hidden.rs +11 -0
  311. data/vendor/rb-sys/src/lib.rs +34 -0
  312. data/vendor/rb-sys/src/macros.rs +371 -0
  313. data/vendor/rb-sys/src/memory.rs +53 -0
  314. data/vendor/rb-sys/src/ruby_abi_version.rs +38 -0
  315. data/vendor/rb-sys/src/special_consts.rs +31 -0
  316. data/vendor/rb-sys/src/stable_api/compiled.c +179 -0
  317. data/vendor/rb-sys/src/stable_api/compiled.rs +257 -0
  318. data/vendor/rb-sys/src/stable_api/ruby_2_6.rs +316 -0
  319. data/vendor/rb-sys/src/stable_api/ruby_2_7.rs +316 -0
  320. data/vendor/rb-sys/src/stable_api/ruby_3_0.rs +324 -0
  321. data/vendor/rb-sys/src/stable_api/ruby_3_1.rs +317 -0
  322. data/vendor/rb-sys/src/stable_api/ruby_3_2.rs +315 -0
  323. data/vendor/rb-sys/src/stable_api/ruby_3_3.rs +326 -0
  324. data/vendor/rb-sys/src/stable_api/ruby_3_4.rs +327 -0
  325. data/vendor/rb-sys/src/stable_api.rs +261 -0
  326. data/vendor/rb-sys/src/symbol.rs +31 -0
  327. data/vendor/rb-sys/src/tracking_allocator.rs +332 -0
  328. data/vendor/rb-sys/src/utils.rs +89 -0
  329. data/vendor/rb-sys/src/value_type.rs +7 -0
  330. metadata +536 -0
data/vendor/kreuzberg/stopwords/br_stopwords.json ADDED
@@ -0,0 +1,1205 @@
1
+ [
2
+ "'blam",
3
+ "'d",
4
+ "'m",
5
+ "'r",
6
+ "'ta",
7
+ "'vat",
8
+ "'z",
9
+ "'zo",
10
+ "a",
11
+ "a:",
12
+ "aba",
13
+ "abalamour",
14
+ "abaoe",
15
+ "ac'hane",
16
+ "ac'hanoc'h",
17
+ "ac'hanomp",
18
+ "ac'hanon",
19
+ "ac'hanout",
20
+ "adal",
21
+ "adalek",
22
+ "adarre",
23
+ "ae",
24
+ "aec'h",
25
+ "aed",
26
+ "aemp",
27
+ "aen",
28
+ "aent",
29
+ "aes",
30
+ "afe",
31
+ "afec'h",
32
+ "afed",
33
+ "afemp",
34
+ "afen",
35
+ "afent",
36
+ "afes",
37
+ "ag",
38
+ "ah",
39
+ "aimp",
40
+ "aint",
41
+ "aio",
42
+ "aiou",
43
+ "aje",
44
+ "ajec'h",
45
+ "ajed",
46
+ "ajemp",
47
+ "ajen",
48
+ "ajent",
49
+ "ajes",
50
+ "al",
51
+ "alato",
52
+ "alies",
53
+ "aliesañ",
54
+ "alkent",
55
+ "all",
56
+ "allas",
57
+ "allo",
58
+ "allô",
59
+ "am",
60
+ "amañ",
61
+ "amzer",
62
+ "an",
63
+ "anezhañ",
64
+ "anezhe",
65
+ "anezhi",
66
+ "anezho",
67
+ "anvet",
68
+ "aon",
69
+ "aotren",
70
+ "ar",
71
+ "arall",
72
+ "araok",
73
+ "araoki",
74
+ "araozañ",
75
+ "araozo",
76
+ "araozoc'h",
77
+ "araozomp",
78
+ "araozon",
79
+ "araozor",
80
+ "araozout",
81
+ "arbenn",
82
+ "arre",
83
+ "atalek",
84
+ "atav",
85
+ "az",
86
+ "azalek",
87
+ "azirazañ",
88
+ "azirazi",
89
+ "azirazo",
90
+ "azirazoc'h",
91
+ "azirazomp",
92
+ "azirazon",
93
+ "azirazor",
94
+ "azirazout",
95
+ "b:",
96
+ "ba",
97
+ "ba'l",
98
+ "ba'n",
99
+ "ba'r",
100
+ "bad",
101
+ "bah",
102
+ "bal",
103
+ "ban",
104
+ "bar",
105
+ "bastañ",
106
+ "befe",
107
+ "bell",
108
+ "benaos",
109
+ "benn",
110
+ "bennag",
111
+ "bennak",
112
+ "bennozh",
113
+ "bep",
114
+ "bepred",
115
+ "berr",
116
+ "berzh",
117
+ "bet",
118
+ "betek",
119
+ "betra",
120
+ "bev",
121
+ "bevet",
122
+ "bez",
123
+ "bezañ",
124
+ "beze",
125
+ "bezent",
126
+ "bezet",
127
+ "bezh",
128
+ "bezit",
129
+ "bezomp",
130
+ "bihan",
131
+ "bije",
132
+ "biou",
133
+ "biskoazh",
134
+ "blam",
135
+ "bo",
136
+ "boa",
137
+ "bominapl",
138
+ "boudoudom",
139
+ "bouez",
140
+ "boull",
141
+ "boum",
142
+ "bout",
143
+ "bras",
144
+ "brasañ",
145
+ "brav",
146
+ "bravo",
147
+ "bremañ",
148
+ "bres",
149
+ "brokenn",
150
+ "bronn",
151
+ "brrr",
152
+ "brutal",
153
+ "buhezek",
154
+ "c'h:",
155
+ "c'haout",
156
+ "c'he",
157
+ "c'hem",
158
+ "c'herz",
159
+ "c'heñver",
160
+ "c'hichen",
161
+ "c'hiz",
162
+ "c'hoazh",
163
+ "c'horre",
164
+ "c'houde",
165
+ "c'houst",
166
+ "c'hreiz",
167
+ "c'hwec'h",
168
+ "c'hwec'hvet",
169
+ "c'hwezek",
170
+ "c'hwi",
171
+ "ch:",
172
+ "chaous",
173
+ "chik",
174
+ "chit",
175
+ "chom",
176
+ "chut",
177
+ "d'",
178
+ "d'al",
179
+ "d'an",
180
+ "d'ar",
181
+ "d'az",
182
+ "d'e",
183
+ "d'he",
184
+ "d'ho",
185
+ "d'hol",
186
+ "d'hon",
187
+ "d'hor",
188
+ "d'o",
189
+ "d'ober",
190
+ "d'ul",
191
+ "d'un",
192
+ "d'ur",
193
+ "d:",
194
+ "da",
195
+ "dak",
196
+ "daka",
197
+ "dal",
198
+ "dalbezh",
199
+ "dalc'hmat",
200
+ "dalit",
201
+ "damdost",
202
+ "damheñvel",
203
+ "damm",
204
+ "dan",
205
+ "danvez",
206
+ "dao",
207
+ "daol",
208
+ "daonet",
209
+ "daou",
210
+ "daoust",
211
+ "daouzek",
212
+ "daouzekvet",
213
+ "darn",
214
+ "dastrewiñ",
215
+ "dav",
216
+ "davedoc'h",
217
+ "davedomp",
218
+ "davedon",
219
+ "davedor",
220
+ "davedout",
221
+ "davet",
222
+ "davetañ",
223
+ "davete",
224
+ "daveti",
225
+ "daveto",
226
+ "defe",
227
+ "dehou",
228
+ "dek",
229
+ "dekvet",
230
+ "den",
231
+ "deoc'h",
232
+ "deomp",
233
+ "deor",
234
+ "derc'hel",
235
+ "deus",
236
+ "dez",
237
+ "deze",
238
+ "dezhañ",
239
+ "dezhe",
240
+ "dezhi",
241
+ "dezho",
242
+ "di",
243
+ "diabarzh",
244
+ "diagent",
245
+ "diar",
246
+ "diaraok",
247
+ "diavaez",
248
+ "dibaoe",
249
+ "dibaot",
250
+ "dibar",
251
+ "dic'halañ",
252
+ "didiac'h",
253
+ "dienn",
254
+ "difer",
255
+ "diganeoc'h",
256
+ "diganeomp",
257
+ "diganeor",
258
+ "diganimp",
259
+ "diganin",
260
+ "diganit",
261
+ "digant",
262
+ "digantañ",
263
+ "digante",
264
+ "diganti",
265
+ "diganto",
266
+ "digemmesk",
267
+ "diget",
268
+ "digor",
269
+ "digoret",
270
+ "dija",
271
+ "dije",
272
+ "dimp",
273
+ "din",
274
+ "dinaou",
275
+ "dindan",
276
+ "dindanañ",
277
+ "dindani",
278
+ "dindano",
279
+ "dindanoc'h",
280
+ "dindanomp",
281
+ "dindanon",
282
+ "dindanor",
283
+ "dindanout",
284
+ "dioutañ",
285
+ "dioute",
286
+ "diouti",
287
+ "diouto",
288
+ "diouzh",
289
+ "diouzhin",
290
+ "diouzhit",
291
+ "diouzhoc'h",
292
+ "diouzhomp",
293
+ "diouzhor",
294
+ "dirak",
295
+ "dirazañ",
296
+ "dirazi",
297
+ "dirazo",
298
+ "dirazoc'h",
299
+ "dirazomp",
300
+ "dirazon",
301
+ "dirazor",
302
+ "dirazout",
303
+ "disheñvel",
304
+ "dispar",
305
+ "distank",
306
+ "dister",
307
+ "disterañ",
308
+ "disterig",
309
+ "distro",
310
+ "dit",
311
+ "divaez",
312
+ "diwar",
313
+ "diwezhat",
314
+ "diwezhañ",
315
+ "do",
316
+ "doa",
317
+ "doare",
318
+ "dont",
319
+ "dost",
320
+ "doue",
321
+ "douetus",
322
+ "douez",
323
+ "doug",
324
+ "draou",
325
+ "draoñ",
326
+ "dre",
327
+ "drede",
328
+ "dreist",
329
+ "dreistañ",
330
+ "dreisti",
331
+ "dreisto",
332
+ "dreistoc'h",
333
+ "dreistomp",
334
+ "dreiston",
335
+ "dreistor",
336
+ "dreistout",
337
+ "drek",
338
+ "dreñv",
339
+ "dring",
340
+ "dro",
341
+ "du",
342
+ "e",
343
+ "e:",
344
+ "eas",
345
+ "ebet",
346
+ "ec'h",
347
+ "edo",
348
+ "edoc'h",
349
+ "edod",
350
+ "edomp",
351
+ "edon",
352
+ "edont",
353
+ "edos",
354
+ "eer",
355
+ "eeun",
356
+ "efed",
357
+ "egedoc'h",
358
+ "egedomp",
359
+ "egedon",
360
+ "egedor",
361
+ "egedout",
362
+ "eget",
363
+ "egetañ",
364
+ "egete",
365
+ "egeti",
366
+ "egeto",
367
+ "eh",
368
+ "eil",
369
+ "eilvet",
370
+ "eizh",
371
+ "eizhvet",
372
+ "ejoc'h",
373
+ "ejod",
374
+ "ejomp",
375
+ "ejont",
376
+ "ejout",
377
+ "el",
378
+ "em",
379
+ "emaint",
380
+ "emaoc'h",
381
+ "emaomp",
382
+ "emaon",
383
+ "emaout",
384
+ "emañ",
385
+ "eme",
386
+ "emeur",
387
+ "emezañ",
388
+ "emezi",
389
+ "emezo",
390
+ "emezoc'h",
391
+ "emezomp",
392
+ "emezon",
393
+ "emezout",
394
+ "emporzhiañ",
395
+ "en",
396
+ "end",
397
+ "endan",
398
+ "endra",
399
+ "enep",
400
+ "ennañ",
401
+ "enni",
402
+ "enno",
403
+ "ennoc'h",
404
+ "ennomp",
405
+ "ennon",
406
+ "ennor",
407
+ "ennout",
408
+ "enta",
409
+ "eo",
410
+ "eomp",
411
+ "eont",
412
+ "eor",
413
+ "eot",
414
+ "er",
415
+ "erbet",
416
+ "erfin",
417
+ "esa",
418
+ "esae",
419
+ "espar",
420
+ "estlamm",
421
+ "estrañj",
422
+ "eta",
423
+ "etre",
424
+ "etreoc'h",
425
+ "etrezo",
426
+ "etrezoc'h",
427
+ "etrezomp",
428
+ "etrezor",
429
+ "euh",
430
+ "eur",
431
+ "eus",
432
+ "evel",
433
+ "evelato",
434
+ "eveldoc'h",
435
+ "eveldomp",
436
+ "eveldon",
437
+ "eveldor",
438
+ "eveldout",
439
+ "evelkent",
440
+ "eveltañ",
441
+ "evelte",
442
+ "evelti",
443
+ "evelto",
444
+ "evidoc'h",
445
+ "evidomp",
446
+ "evidon",
447
+ "evidor",
448
+ "evidout",
449
+ "evit",
450
+ "evitañ",
451
+ "evite",
452
+ "eviti",
453
+ "evito",
454
+ "ez",
455
+ "eñ",
456
+ "f:",
457
+ "fac'h",
458
+ "fall",
459
+ "fed",
460
+ "feiz",
461
+ "fenn",
462
+ "fezh",
463
+ "fin",
464
+ "finsalvet",
465
+ "foei",
466
+ "fouilhezañ",
467
+ "g:",
468
+ "gallout",
469
+ "ganeoc'h",
470
+ "ganeomp",
471
+ "ganin",
472
+ "ganit",
473
+ "gant",
474
+ "gantañ",
475
+ "ganti",
476
+ "ganto",
477
+ "gaout",
478
+ "gast",
479
+ "gein",
480
+ "gellout",
481
+ "genndost",
482
+ "gentañ",
483
+ "ger",
484
+ "gerz",
485
+ "get",
486
+ "geñver",
487
+ "gichen",
488
+ "gin",
489
+ "giz",
490
+ "glan",
491
+ "gloev",
492
+ "goll",
493
+ "gorre",
494
+ "goude",
495
+ "gouez",
496
+ "gouezit",
497
+ "gouezomp",
498
+ "goulz",
499
+ "gounnar",
500
+ "gour",
501
+ "goust",
502
+ "gouze",
503
+ "gouzout",
504
+ "gra",
505
+ "grak",
506
+ "grec'h",
507
+ "greiz",
508
+ "grenn",
509
+ "greomp",
510
+ "grit",
511
+ "groñs",
512
+ "gutez",
513
+ "gwall",
514
+ "gwashoc'h",
515
+ "gwazh",
516
+ "gwech",
517
+ "gwechall",
518
+ "gwechoù",
519
+ "gwell",
520
+ "gwezh",
521
+ "gwezhall",
522
+ "gwezharall",
523
+ "gwezhoù",
524
+ "gwig",
525
+ "gwirionez",
526
+ "gwitibunan",
527
+ "gêr",
528
+ "h:",
529
+ "ha",
530
+ "hag",
531
+ "han",
532
+ "hanter",
533
+ "hanterc'hantad",
534
+ "hanterkantved",
535
+ "harz",
536
+ "hañ",
537
+ "hañval",
538
+ "he",
539
+ "hebioù",
540
+ "hec'h",
541
+ "hei",
542
+ "hein",
543
+ "hem",
544
+ "hemañ",
545
+ "hen",
546
+ "hend",
547
+ "henhont",
548
+ "henn",
549
+ "hennezh",
550
+ "hent",
551
+ "hep",
552
+ "hervez",
553
+ "hervezañ",
554
+ "hervezi",
555
+ "hervezo",
556
+ "hervezoc'h",
557
+ "hervezomp",
558
+ "hervezon",
559
+ "hervezor",
560
+ "hervezout",
561
+ "heul",
562
+ "heuliañ",
563
+ "hevelep",
564
+ "heverk",
565
+ "heñvel",
566
+ "heñvelat",
567
+ "heñvelañ",
568
+ "heñveliñ",
569
+ "heñveloc'h",
570
+ "heñvelout",
571
+ "hi",
572
+ "hilh",
573
+ "hini",
574
+ "hirie",
575
+ "hirio",
576
+ "hiziv",
577
+ "hiziviken",
578
+ "ho",
579
+ "hoaliñ",
580
+ "hoc'h",
581
+ "hogen",
582
+ "hogos",
583
+ "hogozik",
584
+ "hol",
585
+ "holl",
586
+ "holà",
587
+ "homañ",
588
+ "hon",
589
+ "honhont",
590
+ "honnezh",
591
+ "hont",
592
+ "hop",
593
+ "hopala",
594
+ "hor",
595
+ "hou",
596
+ "houp",
597
+ "hudu",
598
+ "hue",
599
+ "hui",
600
+ "hum",
601
+ "hurrah",
602
+ "i",
603
+ "i:",
604
+ "in",
605
+ "int",
606
+ "is",
607
+ "ispisial",
608
+ "isurzhiet",
609
+ "it",
610
+ "ivez",
611
+ "izelañ",
612
+ "j:",
613
+ "just",
614
+ "k:",
615
+ "kae",
616
+ "kaer",
617
+ "kalon",
618
+ "kalz",
619
+ "kant",
620
+ "kaout",
621
+ "kar",
622
+ "kazi",
623
+ "keid",
624
+ "kein",
625
+ "keit",
626
+ "kel",
627
+ "kellies",
628
+ "keloù",
629
+ "kement",
630
+ "ken",
631
+ "kenkent",
632
+ "kenkoulz",
633
+ "kenment",
634
+ "kent",
635
+ "kentañ",
636
+ "kentizh",
637
+ "kentoc'h",
638
+ "kentre",
639
+ "ker",
640
+ "kerkent",
641
+ "kerz",
642
+ "kerzh",
643
+ "ket",
644
+ "keta",
645
+ "keñver",
646
+ "keñverel",
647
+ "keñverius",
648
+ "kichen",
649
+ "kichenik",
650
+ "kit",
651
+ "kiz",
652
+ "klak",
653
+ "klek",
654
+ "klik",
655
+ "komprenet",
656
+ "komz",
657
+ "kont",
658
+ "korf",
659
+ "korre",
660
+ "koulskoude",
661
+ "koulz",
662
+ "koust",
663
+ "krak",
664
+ "krampouezh",
665
+ "krec'h",
666
+ "kreiz",
667
+ "kuit",
668
+ "kwir",
669
+ "l:",
670
+ "la",
671
+ "laez",
672
+ "laoskel",
673
+ "laouen",
674
+ "lavar",
675
+ "lavaret",
676
+ "lavarout",
677
+ "lec'h",
678
+ "lein",
679
+ "leizh",
680
+ "lerc'h",
681
+ "leun",
682
+ "leuskel",
683
+ "lew",
684
+ "lies",
685
+ "liesañ",
686
+ "lod",
687
+ "lusk",
688
+ "lâr",
689
+ "lârout",
690
+ "m:",
691
+ "ma",
692
+ "ma'z",
693
+ "mac'h",
694
+ "mac'hat",
695
+ "mac'hañ",
696
+ "mac'hoc'h",
697
+ "mad",
698
+ "maez",
699
+ "maksimal",
700
+ "mann",
701
+ "mar",
702
+ "mard",
703
+ "marg",
704
+ "marzh",
705
+ "mat",
706
+ "mañ",
707
+ "me",
708
+ "memes",
709
+ "memestra",
710
+ "merkapl",
711
+ "mersi",
712
+ "mes",
713
+ "mesk",
714
+ "met",
715
+ "meur",
716
+ "mil",
717
+ "minimal",
718
+ "moan",
719
+ "moaniaat",
720
+ "mod",
721
+ "mont",
722
+ "mout",
723
+ "mui",
724
+ "muiañ",
725
+ "muioc'h",
726
+ "n",
727
+ "n'",
728
+ "n:",
729
+ "na",
730
+ "nag",
731
+ "naontek",
732
+ "naturel",
733
+ "nav",
734
+ "navet",
735
+ "ne",
736
+ "nebeudig",
737
+ "nebeut",
738
+ "nebeutañ",
739
+ "nebeutoc'h",
740
+ "neketa",
741
+ "nemedoc'h",
742
+ "nemedomp",
743
+ "nemedon",
744
+ "nemedor",
745
+ "nemedout",
746
+ "nemet",
747
+ "nemetañ",
748
+ "nemete",
749
+ "nemeti",
750
+ "nemeto",
751
+ "nemeur",
752
+ "neoac'h",
753
+ "nepell",
754
+ "nerzh",
755
+ "nes",
756
+ "neseser",
757
+ "netra",
758
+ "neubeudoù",
759
+ "neuhe",
760
+ "neuze",
761
+ "nevez",
762
+ "newazh",
763
+ "nez",
764
+ "ni",
765
+ "nikun",
766
+ "niverus",
767
+ "nul",
768
+ "o",
769
+ "o:",
770
+ "oa",
771
+ "oac'h",
772
+ "oad",
773
+ "oamp",
774
+ "oan",
775
+ "oant",
776
+ "oar",
777
+ "oas",
778
+ "ober",
779
+ "oc'h",
780
+ "oc'ho",
781
+ "oc'hola",
782
+ "oc'hpenn",
783
+ "oh",
784
+ "ohe",
785
+ "ollé",
786
+ "olole",
787
+ "olé",
788
+ "omp",
789
+ "on",
790
+ "ordin",
791
+ "ordinal",
792
+ "ouejoc'h",
793
+ "ouejod",
794
+ "ouejomp",
795
+ "ouejont",
796
+ "ouejout",
797
+ "ouek",
798
+ "ouezas",
799
+ "ouezi",
800
+ "ouezimp",
801
+ "ouezin",
802
+ "ouezint",
803
+ "ouezis",
804
+ "ouezo",
805
+ "ouezoc'h",
806
+ "ouezor",
807
+ "ouf",
808
+ "oufe",
809
+ "oufec'h",
810
+ "oufed",
811
+ "oufemp",
812
+ "oufen",
813
+ "oufent",
814
+ "oufes",
815
+ "ouie",
816
+ "ouiec'h",
817
+ "ouied",
818
+ "ouiemp",
819
+ "ouien",
820
+ "ouient",
821
+ "ouies",
822
+ "ouije",
823
+ "ouijec'h",
824
+ "ouijed",
825
+ "ouijemp",
826
+ "ouijen",
827
+ "ouijent",
828
+ "ouijes",
829
+ "out",
830
+ "outañ",
831
+ "outi",
832
+ "outo",
833
+ "ouzer",
834
+ "ouzh",
835
+ "ouzhin",
836
+ "ouzhit",
837
+ "ouzhoc'h",
838
+ "ouzhomp",
839
+ "ouzhor",
840
+ "ouzhpenn",
841
+ "ouzhpennik",
842
+ "ouzoc'h",
843
+ "ouzomp",
844
+ "ouzon",
845
+ "ouzont",
846
+ "ouzout",
847
+ "p'",
848
+ "p:",
849
+ "pa",
850
+ "pad",
851
+ "padal",
852
+ "paf",
853
+ "pan",
854
+ "panevedeoc'h",
855
+ "panevedo",
856
+ "panevedomp",
857
+ "panevedon",
858
+ "panevedout",
859
+ "panevet",
860
+ "panevetañ",
861
+ "paneveti",
862
+ "pas",
863
+ "paseet",
864
+ "pe",
865
+ "peadra",
866
+ "peder",
867
+ "pedervet",
868
+ "pedervetvet",
869
+ "pefe",
870
+ "pegeit",
871
+ "pegement",
872
+ "pegen",
873
+ "pegiz",
874
+ "pegoulz",
875
+ "pehini",
876
+ "pelec'h",
877
+ "pell",
878
+ "pemod",
879
+ "pemp",
880
+ "pempved",
881
+ "pemzek",
882
+ "penaos",
883
+ "penn",
884
+ "peogwir",
885
+ "peotramant",
886
+ "pep",
887
+ "perak",
888
+ "perc'hennañ",
889
+ "pergen",
890
+ "permetiñ",
891
+ "peseurt",
892
+ "pet",
893
+ "petiaoul",
894
+ "petoare",
895
+ "petra",
896
+ "peur",
897
+ "peurgetket",
898
+ "peurheñvel",
899
+ "peurliesañ",
900
+ "peurvuiañ",
901
+ "peus",
902
+ "peustost",
903
+ "peuz",
904
+ "pevar",
905
+ "pevare",
906
+ "pevarevet",
907
+ "pevarzek",
908
+ "pez",
909
+ "peze",
910
+ "pezh",
911
+ "pff",
912
+ "pfft",
913
+ "pfut",
914
+ "picher",
915
+ "pif",
916
+ "pife",
917
+ "pign",
918
+ "pije",
919
+ "pikol",
920
+ "pitiaoul",
921
+ "piv",
922
+ "plaouf",
923
+ "plok",
924
+ "plouf",
925
+ "po",
926
+ "poa",
927
+ "poelladus",
928
+ "pof",
929
+ "pok",
930
+ "posupl",
931
+ "pouah",
932
+ "pourc'henn",
933
+ "prest",
934
+ "prestik",
935
+ "prim",
936
+ "prin",
937
+ "provostapl",
938
+ "pst",
939
+ "pu",
940
+ "pur",
941
+ "r:",
942
+ "ra",
943
+ "rae",
944
+ "raec'h",
945
+ "raed",
946
+ "raemp",
947
+ "raen",
948
+ "raent",
949
+ "raes",
950
+ "rafe",
951
+ "rafec'h",
952
+ "rafed",
953
+ "rafemp",
954
+ "rafen",
955
+ "rafent",
956
+ "rafes",
957
+ "rag",
958
+ "raimp",
959
+ "raint",
960
+ "raio",
961
+ "raje",
962
+ "rajec'h",
963
+ "rajed",
964
+ "rajemp",
965
+ "rajen",
966
+ "rajent",
967
+ "rajes",
968
+ "rak",
969
+ "ral",
970
+ "ran",
971
+ "rankout",
972
+ "raok",
973
+ "razh",
974
+ "re",
975
+ "reas",
976
+ "reer",
977
+ "regennoù",
978
+ "reiñ",
979
+ "rejoc'h",
980
+ "rejod",
981
+ "rejomp",
982
+ "rejont",
983
+ "rejout",
984
+ "rener",
985
+ "rentañ",
986
+ "reoc'h",
987
+ "reomp",
988
+ "reont",
989
+ "reor",
990
+ "reot",
991
+ "resis",
992
+ "ret",
993
+ "reve",
994
+ "rez",
995
+ "ri",
996
+ "rik",
997
+ "rin",
998
+ "ris",
999
+ "rit",
1000
+ "rouez",
1001
+ "s:",
1002
+ "sac'h",
1003
+ "sant",
1004
+ "sav",
1005
+ "sañset",
1006
+ "se",
1007
+ "sed",
1008
+ "seitek",
1009
+ "seizh",
1010
+ "seizhvet",
1011
+ "sell",
1012
+ "sellit",
1013
+ "ser",
1014
+ "setu",
1015
+ "seul",
1016
+ "seurt",
1017
+ "siwazh",
1018
+ "skignañ",
1019
+ "skoaz",
1020
+ "skouer",
1021
+ "sort",
1022
+ "souden",
1023
+ "souvitañ",
1024
+ "soñj",
1025
+ "speriañ",
1026
+ "spririñ",
1027
+ "stad",
1028
+ "stlabezañ",
1029
+ "stop",
1030
+ "stranañ",
1031
+ "strewiñ",
1032
+ "strishaat",
1033
+ "stumm",
1034
+ "sujed",
1035
+ "surtoud",
1036
+ "t:",
1037
+ "ta",
1038
+ "taer",
1039
+ "tailh",
1040
+ "tak",
1041
+ "tal",
1042
+ "talvoudegezh",
1043
+ "tamm",
1044
+ "tanav",
1045
+ "taol",
1046
+ "te",
1047
+ "techet",
1048
+ "teir",
1049
+ "teirvet",
1050
+ "telt",
1051
+ "teltenn",
1052
+ "teus",
1053
+ "teut",
1054
+ "teuteu",
1055
+ "ti",
1056
+ "tik",
1057
+ "toa",
1058
+ "tok",
1059
+ "tost",
1060
+ "tostig",
1061
+ "toud",
1062
+ "touesk",
1063
+ "touez",
1064
+ "toull",
1065
+ "tra",
1066
+ "trantenn",
1067
+ "traoñ",
1068
+ "trawalc'h",
1069
+ "tre",
1070
+ "trede",
1071
+ "tregont",
1072
+ "tremenet",
1073
+ "tri",
1074
+ "trivet",
1075
+ "triwec'h",
1076
+ "trizek",
1077
+ "tro",
1078
+ "trugarez",
1079
+ "trumm",
1080
+ "tsoin",
1081
+ "tsouin",
1082
+ "tu",
1083
+ "tud",
1084
+ "u:",
1085
+ "ugent",
1086
+ "uhel",
1087
+ "uhelañ",
1088
+ "ul",
1089
+ "un",
1090
+ "unan",
1091
+ "unanez",
1092
+ "unanig",
1093
+ "unnek",
1094
+ "unnekvet",
1095
+ "ur",
1096
+ "urzh",
1097
+ "us",
1098
+ "v:",
1099
+ "va",
1100
+ "vale",
1101
+ "van",
1102
+ "vare",
1103
+ "vat",
1104
+ "vefe",
1105
+ "vefec'h",
1106
+ "vefed",
1107
+ "vefemp",
1108
+ "vefen",
1109
+ "vefent",
1110
+ "vefes",
1111
+ "vesk",
1112
+ "vete",
1113
+ "vez",
1114
+ "vezan",
1115
+ "vezañ",
1116
+ "veze",
1117
+ "vezec'h",
1118
+ "vezed",
1119
+ "vezemp",
1120
+ "vezen",
1121
+ "vezent",
1122
+ "vezer",
1123
+ "vezes",
1124
+ "vezez",
1125
+ "vezit",
1126
+ "vezomp",
1127
+ "vezont",
1128
+ "vi",
1129
+ "vihan",
1130
+ "vihanañ",
1131
+ "vije",
1132
+ "vijec'h",
1133
+ "vijed",
1134
+ "vijemp",
1135
+ "vijen",
1136
+ "vijent",
1137
+ "vijes",
1138
+ "viken",
1139
+ "vimp",
1140
+ "vin",
1141
+ "vint",
1142
+ "vior",
1143
+ "viot",
1144
+ "virviken",
1145
+ "viskoazh",
1146
+ "vlan",
1147
+ "vlaou",
1148
+ "vo",
1149
+ "vod",
1150
+ "voe",
1151
+ "voec'h",
1152
+ "voed",
1153
+ "voemp",
1154
+ "voen",
1155
+ "voent",
1156
+ "voes",
1157
+ "vont",
1158
+ "vostapl",
1159
+ "vrac'h",
1160
+ "vrasañ",
1161
+ "vremañ",
1162
+ "w:",
1163
+ "walc'h",
1164
+ "war",
1165
+ "warnañ",
1166
+ "warni",
1167
+ "warno",
1168
+ "warnoc'h",
1169
+ "warnomp",
1170
+ "warnon",
1171
+ "warnor",
1172
+ "warnout",
1173
+ "wazh",
1174
+ "wech",
1175
+ "wechoù",
1176
+ "well",
1177
+ "y:",
1178
+ "you",
1179
+ "youadenn",
1180
+ "youc'hadenn",
1181
+ "youc'hou",
1182
+ "z:",
1183
+ "za",
1184
+ "zan",
1185
+ "zaw",
1186
+ "zeu",
1187
+ "zi",
1188
+ "ziar",
1189
+ "zigarez",
1190
+ "ziget",
1191
+ "zindan",
1192
+ "zioc'h",
1193
+ "ziouzh",
1194
+ "zirak",
1195
+ "zivout",
1196
+ "ziwar",
1197
+ "ziwezhañ",
1198
+ "zo",
1199
+ "zoken",
1200
+ "zokenoc'h",
1201
+ "zouesk",
1202
+ "zouez",
1203
+ "zro",
1204
+ "zu"
1205
+ ]