kreuzberg 4.0.0.rc2 → 4.0.1

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 (446) hide show
  1. checksums.yaml +4 -4
  2. data/.gitignore +14 -14
  3. data/.rspec +3 -3
  4. data/.rubocop.yaml +1 -1
  5. data/.rubocop.yml +543 -538
  6. data/Gemfile +8 -8
  7. data/Gemfile.lock +194 -6
  8. data/README.md +391 -426
  9. data/Rakefile +34 -25
  10. data/Steepfile +51 -47
  11. data/examples/async_patterns.rb +283 -341
  12. data/ext/kreuzberg_rb/extconf.rb +65 -45
  13. data/ext/kreuzberg_rb/native/.cargo/config.toml +23 -0
  14. data/ext/kreuzberg_rb/native/Cargo.lock +7619 -6535
  15. data/ext/kreuzberg_rb/native/Cargo.toml +75 -44
  16. data/ext/kreuzberg_rb/native/README.md +425 -425
  17. data/ext/kreuzberg_rb/native/build.rs +15 -15
  18. data/ext/kreuzberg_rb/native/include/ieeefp.h +11 -11
  19. data/ext/kreuzberg_rb/native/include/msvc_compat/strings.h +14 -14
  20. data/ext/kreuzberg_rb/native/include/strings.h +20 -20
  21. data/ext/kreuzberg_rb/native/include/unistd.h +47 -47
  22. data/ext/kreuzberg_rb/native/src/lib.rs +3802 -2998
  23. data/extconf.rb +60 -28
  24. data/kreuzberg.gemspec +199 -148
  25. data/lib/kreuzberg/api_proxy.rb +126 -142
  26. data/lib/kreuzberg/cache_api.rb +67 -46
  27. data/lib/kreuzberg/cli.rb +47 -55
  28. data/lib/kreuzberg/cli_proxy.rb +117 -127
  29. data/lib/kreuzberg/config.rb +936 -691
  30. data/lib/kreuzberg/error_context.rb +136 -32
  31. data/lib/kreuzberg/errors.rb +116 -118
  32. data/lib/kreuzberg/extraction_api.rb +313 -85
  33. data/lib/kreuzberg/mcp_proxy.rb +177 -186
  34. data/lib/kreuzberg/ocr_backend_protocol.rb +40 -113
  35. data/lib/kreuzberg/post_processor_protocol.rb +15 -86
  36. data/lib/kreuzberg/result.rb +334 -216
  37. data/lib/kreuzberg/setup_lib_path.rb +99 -80
  38. data/lib/kreuzberg/types.rb +170 -0
  39. data/lib/kreuzberg/validator_protocol.rb +16 -89
  40. data/lib/kreuzberg/version.rb +5 -5
  41. data/lib/kreuzberg.rb +96 -103
  42. data/lib/libpdfium.so +0 -0
  43. data/sig/kreuzberg/internal.rbs +184 -184
  44. data/sig/kreuzberg.rbs +561 -520
  45. data/spec/binding/async_operations_spec.rb +473 -0
  46. data/spec/binding/batch_operations_spec.rb +595 -0
  47. data/spec/binding/batch_spec.rb +359 -0
  48. data/spec/binding/cache_spec.rb +227 -227
  49. data/spec/binding/cli_proxy_spec.rb +85 -85
  50. data/spec/binding/cli_spec.rb +55 -55
  51. data/spec/binding/config_result_spec.rb +377 -0
  52. data/spec/binding/config_spec.rb +419 -345
  53. data/spec/binding/config_validation_spec.rb +377 -283
  54. data/spec/binding/embeddings_spec.rb +816 -0
  55. data/spec/binding/error_handling_spec.rb +399 -213
  56. data/spec/binding/error_recovery_spec.rb +488 -0
  57. data/spec/binding/errors_spec.rb +66 -66
  58. data/spec/binding/font_config_spec.rb +220 -0
  59. data/spec/binding/images_spec.rb +738 -0
  60. data/spec/binding/keywords_extraction_spec.rb +600 -0
  61. data/spec/binding/metadata_types_spec.rb +1228 -0
  62. data/spec/binding/pages_extraction_spec.rb +471 -0
  63. data/spec/binding/plugins/ocr_backend_spec.rb +307 -307
  64. data/spec/binding/plugins/postprocessor_spec.rb +269 -269
  65. data/spec/binding/plugins/validator_spec.rb +273 -274
  66. data/spec/binding/tables_spec.rb +641 -0
  67. data/spec/fixtures/config.toml +38 -39
  68. data/spec/fixtures/config.yaml +41 -41
  69. data/spec/fixtures/invalid_config.toml +3 -4
  70. data/spec/smoke/package_spec.rb +177 -178
  71. data/spec/spec_helper.rb +40 -42
  72. data/spec/unit/config/chunking_config_spec.rb +213 -0
  73. data/spec/unit/config/embedding_config_spec.rb +343 -0
  74. data/spec/unit/config/extraction_config_spec.rb +438 -0
  75. data/spec/unit/config/font_config_spec.rb +285 -0
  76. data/spec/unit/config/hierarchy_config_spec.rb +314 -0
  77. data/spec/unit/config/image_extraction_config_spec.rb +209 -0
  78. data/spec/unit/config/image_preprocessing_config_spec.rb +249 -0
  79. data/spec/unit/config/keyword_config_spec.rb +229 -0
  80. data/spec/unit/config/language_detection_config_spec.rb +258 -0
  81. data/spec/unit/config/ocr_config_spec.rb +171 -0
  82. data/spec/unit/config/page_config_spec.rb +221 -0
  83. data/spec/unit/config/pdf_config_spec.rb +267 -0
  84. data/spec/unit/config/postprocessor_config_spec.rb +290 -0
  85. data/spec/unit/config/tesseract_config_spec.rb +181 -0
  86. data/spec/unit/config/token_reduction_config_spec.rb +251 -0
  87. data/test/metadata_types_test.rb +959 -0
  88. data/vendor/Cargo.toml +61 -0
  89. data/vendor/kreuzberg/Cargo.toml +259 -204
  90. data/vendor/kreuzberg/README.md +263 -175
  91. data/vendor/kreuzberg/build.rs +782 -474
  92. data/vendor/kreuzberg/examples/bench_fixes.rs +71 -0
  93. data/vendor/kreuzberg/examples/test_pdfium_fork.rs +62 -0
  94. data/vendor/kreuzberg/src/api/error.rs +81 -81
  95. data/vendor/kreuzberg/src/api/handlers.rs +320 -199
  96. data/vendor/kreuzberg/src/api/mod.rs +94 -79
  97. data/vendor/kreuzberg/src/api/server.rs +518 -353
  98. data/vendor/kreuzberg/src/api/types.rs +206 -170
  99. data/vendor/kreuzberg/src/cache/mod.rs +1167 -1167
  100. data/vendor/kreuzberg/src/chunking/mod.rs +2303 -677
  101. data/vendor/kreuzberg/src/chunking/processor.rs +219 -0
  102. data/vendor/kreuzberg/src/core/batch_mode.rs +95 -95
  103. data/vendor/kreuzberg/src/core/batch_optimizations.rs +385 -0
  104. data/vendor/kreuzberg/src/core/config.rs +1914 -1032
  105. data/vendor/kreuzberg/src/core/config_validation.rs +949 -0
  106. data/vendor/kreuzberg/src/core/extractor.rs +1200 -1024
  107. data/vendor/kreuzberg/src/core/formats.rs +235 -0
  108. data/vendor/kreuzberg/src/core/io.rs +329 -329
  109. data/vendor/kreuzberg/src/core/mime.rs +605 -605
  110. data/vendor/kreuzberg/src/core/mod.rs +61 -45
  111. data/vendor/kreuzberg/src/core/pipeline.rs +1223 -984
  112. data/vendor/kreuzberg/src/core/server_config.rs +1220 -0
  113. data/vendor/kreuzberg/src/embeddings.rs +471 -432
  114. data/vendor/kreuzberg/src/error.rs +431 -431
  115. data/vendor/kreuzberg/src/extraction/archive.rs +959 -954
  116. data/vendor/kreuzberg/src/extraction/capacity.rs +263 -0
  117. data/vendor/kreuzberg/src/extraction/docx.rs +404 -40
  118. data/vendor/kreuzberg/src/extraction/email.rs +855 -854
  119. data/vendor/kreuzberg/src/extraction/excel.rs +697 -688
  120. data/vendor/kreuzberg/src/extraction/html.rs +1830 -553
  121. data/vendor/kreuzberg/src/extraction/image.rs +492 -368
  122. data/vendor/kreuzberg/src/extraction/libreoffice.rs +574 -563
  123. data/vendor/kreuzberg/src/extraction/markdown.rs +216 -213
  124. data/vendor/kreuzberg/src/extraction/mod.rs +93 -81
  125. data/vendor/kreuzberg/src/extraction/office_metadata/app_properties.rs +398 -398
  126. data/vendor/kreuzberg/src/extraction/office_metadata/core_properties.rs +247 -247
  127. data/vendor/kreuzberg/src/extraction/office_metadata/custom_properties.rs +240 -240
  128. data/vendor/kreuzberg/src/extraction/office_metadata/mod.rs +130 -130
  129. data/vendor/kreuzberg/src/extraction/office_metadata/odt_properties.rs +284 -287
  130. data/vendor/kreuzberg/src/extraction/pptx.rs +3102 -3000
  131. data/vendor/kreuzberg/src/extraction/structured.rs +491 -490
  132. data/vendor/kreuzberg/src/extraction/table.rs +329 -328
  133. data/vendor/kreuzberg/src/extraction/text.rs +277 -269
  134. data/vendor/kreuzberg/src/extraction/xml.rs +333 -333
  135. data/vendor/kreuzberg/src/extractors/archive.rs +447 -446
  136. data/vendor/kreuzberg/src/extractors/bibtex.rs +470 -469
  137. data/vendor/kreuzberg/src/extractors/docbook.rs +504 -502
  138. data/vendor/kreuzberg/src/extractors/docx.rs +400 -367
  139. data/vendor/kreuzberg/src/extractors/email.rs +157 -143
  140. data/vendor/kreuzberg/src/extractors/epub.rs +696 -707
  141. data/vendor/kreuzberg/src/extractors/excel.rs +385 -343
  142. data/vendor/kreuzberg/src/extractors/fictionbook.rs +492 -491
  143. data/vendor/kreuzberg/src/extractors/html.rs +419 -393
  144. data/vendor/kreuzberg/src/extractors/image.rs +219 -198
  145. data/vendor/kreuzberg/src/extractors/jats.rs +1054 -1051
  146. data/vendor/kreuzberg/src/extractors/jupyter.rs +368 -367
  147. data/vendor/kreuzberg/src/extractors/latex.rs +653 -652
  148. data/vendor/kreuzberg/src/extractors/markdown.rs +701 -700
  149. data/vendor/kreuzberg/src/extractors/mod.rs +429 -365
  150. data/vendor/kreuzberg/src/extractors/odt.rs +628 -628
  151. data/vendor/kreuzberg/src/extractors/opml.rs +635 -634
  152. data/vendor/kreuzberg/src/extractors/orgmode.rs +529 -528
  153. data/vendor/kreuzberg/src/extractors/pdf.rs +761 -493
  154. data/vendor/kreuzberg/src/extractors/pptx.rs +279 -248
  155. data/vendor/kreuzberg/src/extractors/rst.rs +577 -576
  156. data/vendor/kreuzberg/src/extractors/rtf.rs +809 -810
  157. data/vendor/kreuzberg/src/extractors/security.rs +484 -484
  158. data/vendor/kreuzberg/src/extractors/security_tests.rs +367 -367
  159. data/vendor/kreuzberg/src/extractors/structured.rs +142 -140
  160. data/vendor/kreuzberg/src/extractors/text.rs +265 -260
  161. data/vendor/kreuzberg/src/extractors/typst.rs +651 -650
  162. data/vendor/kreuzberg/src/extractors/xml.rs +147 -135
  163. data/vendor/kreuzberg/src/image/dpi.rs +164 -164
  164. data/vendor/kreuzberg/src/image/mod.rs +6 -6
  165. data/vendor/kreuzberg/src/image/preprocessing.rs +417 -417
  166. data/vendor/kreuzberg/src/image/resize.rs +89 -89
  167. data/vendor/kreuzberg/src/keywords/config.rs +154 -154
  168. data/vendor/kreuzberg/src/keywords/mod.rs +237 -237
  169. data/vendor/kreuzberg/src/keywords/processor.rs +275 -267
  170. data/vendor/kreuzberg/src/keywords/rake.rs +293 -293
  171. data/vendor/kreuzberg/src/keywords/types.rs +68 -68
  172. data/vendor/kreuzberg/src/keywords/yake.rs +163 -163
  173. data/vendor/kreuzberg/src/language_detection/mod.rs +985 -942
  174. data/vendor/kreuzberg/src/language_detection/processor.rs +218 -0
  175. data/vendor/kreuzberg/src/lib.rs +114 -105
  176. data/vendor/kreuzberg/src/mcp/mod.rs +35 -32
  177. data/vendor/kreuzberg/src/mcp/server.rs +2090 -1968
  178. data/vendor/kreuzberg/src/ocr/cache.rs +469 -469
  179. data/vendor/kreuzberg/src/ocr/error.rs +37 -37
  180. data/vendor/kreuzberg/src/ocr/hocr.rs +216 -216
  181. data/vendor/kreuzberg/src/ocr/language_registry.rs +520 -0
  182. data/vendor/kreuzberg/src/ocr/mod.rs +60 -58
  183. data/vendor/kreuzberg/src/ocr/processor.rs +858 -863
  184. data/vendor/kreuzberg/src/ocr/table/mod.rs +4 -4
  185. data/vendor/kreuzberg/src/ocr/table/tsv_parser.rs +144 -144
  186. data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +456 -450
  187. data/vendor/kreuzberg/src/ocr/types.rs +393 -393
  188. data/vendor/kreuzberg/src/ocr/utils.rs +47 -47
  189. data/vendor/kreuzberg/src/ocr/validation.rs +206 -206
  190. data/vendor/kreuzberg/src/panic_context.rs +154 -154
  191. data/vendor/kreuzberg/src/pdf/bindings.rs +306 -0
  192. data/vendor/kreuzberg/src/pdf/bundled.rs +408 -0
  193. data/vendor/kreuzberg/src/pdf/error.rs +214 -122
  194. data/vendor/kreuzberg/src/pdf/fonts.rs +358 -0
  195. data/vendor/kreuzberg/src/pdf/hierarchy.rs +903 -0
  196. data/vendor/kreuzberg/src/pdf/images.rs +139 -139
  197. data/vendor/kreuzberg/src/pdf/metadata.rs +509 -346
  198. data/vendor/kreuzberg/src/pdf/mod.rs +81 -50
  199. data/vendor/kreuzberg/src/pdf/rendering.rs +369 -369
  200. data/vendor/kreuzberg/src/pdf/table.rs +417 -393
  201. data/vendor/kreuzberg/src/pdf/text.rs +553 -158
  202. data/vendor/kreuzberg/src/plugins/extractor.rs +1042 -1013
  203. data/vendor/kreuzberg/src/plugins/mod.rs +212 -209
  204. data/vendor/kreuzberg/src/plugins/ocr.rs +637 -620
  205. data/vendor/kreuzberg/src/plugins/processor.rs +650 -642
  206. data/vendor/kreuzberg/src/plugins/registry.rs +1339 -1337
  207. data/vendor/kreuzberg/src/plugins/traits.rs +258 -258
  208. data/vendor/kreuzberg/src/plugins/validator.rs +967 -956
  209. data/vendor/kreuzberg/src/stopwords/mod.rs +1470 -1470
  210. data/vendor/kreuzberg/src/text/mod.rs +27 -19
  211. data/vendor/kreuzberg/src/text/quality.rs +710 -697
  212. data/vendor/kreuzberg/src/text/quality_processor.rs +231 -0
  213. data/vendor/kreuzberg/src/text/string_utils.rs +229 -217
  214. data/vendor/kreuzberg/src/text/token_reduction/cjk_utils.rs +164 -164
  215. data/vendor/kreuzberg/src/text/token_reduction/config.rs +100 -100
  216. data/vendor/kreuzberg/src/text/token_reduction/core.rs +832 -796
  217. data/vendor/kreuzberg/src/text/token_reduction/filters.rs +923 -902
  218. data/vendor/kreuzberg/src/text/token_reduction/mod.rs +160 -160
  219. data/vendor/kreuzberg/src/text/token_reduction/semantic.rs +619 -619
  220. data/vendor/kreuzberg/src/text/token_reduction/simd_text.rs +148 -147
  221. data/vendor/kreuzberg/src/text/utf8_validation.rs +193 -0
  222. data/vendor/kreuzberg/src/types.rs +1713 -903
  223. data/vendor/kreuzberg/src/utils/mod.rs +31 -17
  224. data/vendor/kreuzberg/src/utils/pool.rs +503 -0
  225. data/vendor/kreuzberg/src/utils/pool_sizing.rs +364 -0
  226. data/vendor/kreuzberg/src/utils/quality.rs +968 -959
  227. data/vendor/kreuzberg/src/utils/string_pool.rs +761 -0
  228. data/vendor/kreuzberg/src/utils/string_utils.rs +381 -381
  229. data/vendor/kreuzberg/stopwords/af_stopwords.json +53 -53
  230. data/vendor/kreuzberg/stopwords/ar_stopwords.json +482 -482
  231. data/vendor/kreuzberg/stopwords/bg_stopwords.json +261 -261
  232. data/vendor/kreuzberg/stopwords/bn_stopwords.json +400 -400
  233. data/vendor/kreuzberg/stopwords/br_stopwords.json +1205 -1205
  234. data/vendor/kreuzberg/stopwords/ca_stopwords.json +280 -280
  235. data/vendor/kreuzberg/stopwords/cs_stopwords.json +425 -425
  236. data/vendor/kreuzberg/stopwords/da_stopwords.json +172 -172
  237. data/vendor/kreuzberg/stopwords/de_stopwords.json +622 -622
  238. data/vendor/kreuzberg/stopwords/el_stopwords.json +849 -849
  239. data/vendor/kreuzberg/stopwords/en_stopwords.json +1300 -1300
  240. data/vendor/kreuzberg/stopwords/eo_stopwords.json +175 -175
  241. data/vendor/kreuzberg/stopwords/es_stopwords.json +734 -734
  242. data/vendor/kreuzberg/stopwords/et_stopwords.json +37 -37
  243. data/vendor/kreuzberg/stopwords/eu_stopwords.json +100 -100
  244. data/vendor/kreuzberg/stopwords/fa_stopwords.json +801 -801
  245. data/vendor/kreuzberg/stopwords/fi_stopwords.json +849 -849
  246. data/vendor/kreuzberg/stopwords/fr_stopwords.json +693 -693
  247. data/vendor/kreuzberg/stopwords/ga_stopwords.json +111 -111
  248. data/vendor/kreuzberg/stopwords/gl_stopwords.json +162 -162
  249. data/vendor/kreuzberg/stopwords/gu_stopwords.json +226 -226
  250. data/vendor/kreuzberg/stopwords/ha_stopwords.json +41 -41
  251. data/vendor/kreuzberg/stopwords/he_stopwords.json +196 -196
  252. data/vendor/kreuzberg/stopwords/hi_stopwords.json +227 -227
  253. data/vendor/kreuzberg/stopwords/hr_stopwords.json +181 -181
  254. data/vendor/kreuzberg/stopwords/hu_stopwords.json +791 -791
  255. data/vendor/kreuzberg/stopwords/hy_stopwords.json +47 -47
  256. data/vendor/kreuzberg/stopwords/id_stopwords.json +760 -760
  257. data/vendor/kreuzberg/stopwords/it_stopwords.json +634 -634
  258. data/vendor/kreuzberg/stopwords/ja_stopwords.json +136 -136
  259. data/vendor/kreuzberg/stopwords/kn_stopwords.json +84 -84
  260. data/vendor/kreuzberg/stopwords/ko_stopwords.json +681 -681
  261. data/vendor/kreuzberg/stopwords/ku_stopwords.json +64 -64
  262. data/vendor/kreuzberg/stopwords/la_stopwords.json +51 -51
  263. data/vendor/kreuzberg/stopwords/lt_stopwords.json +476 -476
  264. data/vendor/kreuzberg/stopwords/lv_stopwords.json +163 -163
  265. data/vendor/kreuzberg/stopwords/ml_stopwords.json +1 -1
  266. data/vendor/kreuzberg/stopwords/mr_stopwords.json +101 -101
  267. data/vendor/kreuzberg/stopwords/ms_stopwords.json +477 -477
  268. data/vendor/kreuzberg/stopwords/ne_stopwords.json +490 -490
  269. data/vendor/kreuzberg/stopwords/nl_stopwords.json +415 -415
  270. data/vendor/kreuzberg/stopwords/no_stopwords.json +223 -223
  271. data/vendor/kreuzberg/stopwords/pl_stopwords.json +331 -331
  272. data/vendor/kreuzberg/stopwords/pt_stopwords.json +562 -562
  273. data/vendor/kreuzberg/stopwords/ro_stopwords.json +436 -436
  274. data/vendor/kreuzberg/stopwords/ru_stopwords.json +561 -561
  275. data/vendor/kreuzberg/stopwords/si_stopwords.json +193 -193
  276. data/vendor/kreuzberg/stopwords/sk_stopwords.json +420 -420
  277. data/vendor/kreuzberg/stopwords/sl_stopwords.json +448 -448
  278. data/vendor/kreuzberg/stopwords/so_stopwords.json +32 -32
  279. data/vendor/kreuzberg/stopwords/st_stopwords.json +33 -33
  280. data/vendor/kreuzberg/stopwords/sv_stopwords.json +420 -420
  281. data/vendor/kreuzberg/stopwords/sw_stopwords.json +76 -76
  282. data/vendor/kreuzberg/stopwords/ta_stopwords.json +129 -129
  283. data/vendor/kreuzberg/stopwords/te_stopwords.json +54 -54
  284. data/vendor/kreuzberg/stopwords/th_stopwords.json +118 -118
  285. data/vendor/kreuzberg/stopwords/tl_stopwords.json +149 -149
  286. data/vendor/kreuzberg/stopwords/tr_stopwords.json +506 -506
  287. data/vendor/kreuzberg/stopwords/uk_stopwords.json +75 -75
  288. data/vendor/kreuzberg/stopwords/ur_stopwords.json +519 -519
  289. data/vendor/kreuzberg/stopwords/vi_stopwords.json +647 -647
  290. data/vendor/kreuzberg/stopwords/yo_stopwords.json +62 -62
  291. data/vendor/kreuzberg/stopwords/zh_stopwords.json +796 -796
  292. data/vendor/kreuzberg/stopwords/zu_stopwords.json +31 -31
  293. data/vendor/kreuzberg/tests/api_embed.rs +360 -0
  294. data/vendor/kreuzberg/tests/api_extract_multipart.rs +52 -52
  295. data/vendor/kreuzberg/tests/api_large_pdf_extraction.rs +471 -0
  296. data/vendor/kreuzberg/tests/api_large_pdf_extraction_diagnostics.rs +289 -0
  297. data/vendor/kreuzberg/tests/api_tests.rs +1472 -966
  298. data/vendor/kreuzberg/tests/archive_integration.rs +545 -543
  299. data/vendor/kreuzberg/tests/batch_orchestration.rs +587 -556
  300. data/vendor/kreuzberg/tests/batch_pooling_benchmark.rs +154 -0
  301. data/vendor/kreuzberg/tests/batch_processing.rs +328 -316
  302. data/vendor/kreuzberg/tests/bibtex_parity_test.rs +421 -421
  303. data/vendor/kreuzberg/tests/concurrency_stress.rs +541 -525
  304. data/vendor/kreuzberg/tests/config_features.rs +612 -598
  305. data/vendor/kreuzberg/tests/config_integration_test.rs +753 -0
  306. data/vendor/kreuzberg/tests/config_loading_tests.rs +416 -415
  307. data/vendor/kreuzberg/tests/core_integration.rs +519 -510
  308. data/vendor/kreuzberg/tests/csv_integration.rs +414 -414
  309. data/vendor/kreuzberg/tests/data/hierarchy_ground_truth.json +294 -0
  310. data/vendor/kreuzberg/tests/docbook_extractor_tests.rs +500 -498
  311. data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +122 -122
  312. data/vendor/kreuzberg/tests/docx_vs_pandoc_comparison.rs +370 -370
  313. data/vendor/kreuzberg/tests/email_integration.rs +327 -325
  314. data/vendor/kreuzberg/tests/epub_native_extractor_tests.rs +275 -275
  315. data/vendor/kreuzberg/tests/error_handling.rs +402 -393
  316. data/vendor/kreuzberg/tests/fictionbook_extractor_tests.rs +228 -228
  317. data/vendor/kreuzberg/tests/format_integration.rs +165 -159
  318. data/vendor/kreuzberg/tests/helpers/mod.rs +202 -142
  319. data/vendor/kreuzberg/tests/html_table_test.rs +551 -551
  320. data/vendor/kreuzberg/tests/image_integration.rs +255 -253
  321. data/vendor/kreuzberg/tests/instrumentation_test.rs +139 -139
  322. data/vendor/kreuzberg/tests/jats_extractor_tests.rs +639 -639
  323. data/vendor/kreuzberg/tests/jupyter_extractor_tests.rs +704 -704
  324. data/vendor/kreuzberg/tests/keywords_integration.rs +479 -479
  325. data/vendor/kreuzberg/tests/keywords_quality.rs +509 -509
  326. data/vendor/kreuzberg/tests/latex_extractor_tests.rs +496 -496
  327. data/vendor/kreuzberg/tests/markdown_extractor_tests.rs +490 -490
  328. data/vendor/kreuzberg/tests/mime_detection.rs +429 -428
  329. data/vendor/kreuzberg/tests/ocr_configuration.rs +514 -510
  330. data/vendor/kreuzberg/tests/ocr_errors.rs +698 -676
  331. data/vendor/kreuzberg/tests/ocr_language_registry.rs +191 -0
  332. data/vendor/kreuzberg/tests/ocr_quality.rs +629 -627
  333. data/vendor/kreuzberg/tests/ocr_stress.rs +469 -469
  334. data/vendor/kreuzberg/tests/odt_extractor_tests.rs +674 -695
  335. data/vendor/kreuzberg/tests/opml_extractor_tests.rs +616 -616
  336. data/vendor/kreuzberg/tests/orgmode_extractor_tests.rs +822 -822
  337. data/vendor/kreuzberg/tests/page_markers.rs +297 -0
  338. data/vendor/kreuzberg/tests/pdf_hierarchy_detection.rs +301 -0
  339. data/vendor/kreuzberg/tests/pdf_hierarchy_quality.rs +589 -0
  340. data/vendor/kreuzberg/tests/pdf_integration.rs +45 -43
  341. data/vendor/kreuzberg/tests/pdf_ocr_triggering.rs +301 -0
  342. data/vendor/kreuzberg/tests/pdf_text_merging.rs +475 -0
  343. data/vendor/kreuzberg/tests/pdfium_linking.rs +340 -0
  344. data/vendor/kreuzberg/tests/pipeline_integration.rs +1446 -1411
  345. data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +776 -771
  346. data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs +577 -560
  347. data/vendor/kreuzberg/tests/plugin_system.rs +927 -921
  348. data/vendor/kreuzberg/tests/plugin_validator_test.rs +783 -783
  349. data/vendor/kreuzberg/tests/registry_integration_tests.rs +587 -586
  350. data/vendor/kreuzberg/tests/rst_extractor_tests.rs +694 -692
  351. data/vendor/kreuzberg/tests/rtf_extractor_tests.rs +775 -776
  352. data/vendor/kreuzberg/tests/security_validation.rs +416 -415
  353. data/vendor/kreuzberg/tests/stopwords_integration_test.rs +888 -888
  354. data/vendor/kreuzberg/tests/test_fastembed.rs +631 -609
  355. data/vendor/kreuzberg/tests/typst_behavioral_tests.rs +1260 -1259
  356. data/vendor/kreuzberg/tests/typst_extractor_tests.rs +648 -647
  357. data/vendor/kreuzberg/tests/xlsx_metadata_extraction_test.rs +87 -87
  358. data/vendor/kreuzberg-ffi/Cargo.toml +67 -0
  359. data/vendor/kreuzberg-ffi/README.md +851 -0
  360. data/vendor/kreuzberg-ffi/benches/result_view_benchmark.rs +227 -0
  361. data/vendor/kreuzberg-ffi/build.rs +168 -0
  362. data/vendor/kreuzberg-ffi/cbindgen.toml +37 -0
  363. data/vendor/kreuzberg-ffi/kreuzberg-ffi.pc.in +12 -0
  364. data/vendor/kreuzberg-ffi/kreuzberg.h +3012 -0
  365. data/vendor/kreuzberg-ffi/src/batch_streaming.rs +588 -0
  366. data/vendor/kreuzberg-ffi/src/config.rs +1341 -0
  367. data/vendor/kreuzberg-ffi/src/error.rs +901 -0
  368. data/vendor/kreuzberg-ffi/src/extraction.rs +555 -0
  369. data/vendor/kreuzberg-ffi/src/helpers.rs +879 -0
  370. data/vendor/kreuzberg-ffi/src/lib.rs +977 -0
  371. data/vendor/kreuzberg-ffi/src/memory.rs +493 -0
  372. data/vendor/kreuzberg-ffi/src/mime.rs +329 -0
  373. data/vendor/kreuzberg-ffi/src/panic_shield.rs +265 -0
  374. data/vendor/kreuzberg-ffi/src/plugins/document_extractor.rs +442 -0
  375. data/vendor/kreuzberg-ffi/src/plugins/mod.rs +14 -0
  376. data/vendor/kreuzberg-ffi/src/plugins/ocr_backend.rs +628 -0
  377. data/vendor/kreuzberg-ffi/src/plugins/post_processor.rs +438 -0
  378. data/vendor/kreuzberg-ffi/src/plugins/validator.rs +329 -0
  379. data/vendor/kreuzberg-ffi/src/result.rs +510 -0
  380. data/vendor/kreuzberg-ffi/src/result_pool.rs +639 -0
  381. data/vendor/kreuzberg-ffi/src/result_view.rs +773 -0
  382. data/vendor/kreuzberg-ffi/src/string_intern.rs +568 -0
  383. data/vendor/kreuzberg-ffi/src/types.rs +363 -0
  384. data/vendor/kreuzberg-ffi/src/util.rs +210 -0
  385. data/vendor/kreuzberg-ffi/src/validation.rs +848 -0
  386. data/vendor/kreuzberg-ffi/tests.disabled/README.md +48 -0
  387. data/vendor/kreuzberg-ffi/tests.disabled/config_loading_tests.rs +299 -0
  388. data/vendor/kreuzberg-ffi/tests.disabled/config_tests.rs +346 -0
  389. data/vendor/kreuzberg-ffi/tests.disabled/extractor_tests.rs +232 -0
  390. data/vendor/kreuzberg-ffi/tests.disabled/plugin_registration_tests.rs +470 -0
  391. data/vendor/kreuzberg-tesseract/.commitlintrc.json +13 -0
  392. data/vendor/kreuzberg-tesseract/.crate-ignore +2 -0
  393. data/vendor/kreuzberg-tesseract/Cargo.lock +2933 -0
  394. data/vendor/kreuzberg-tesseract/Cargo.toml +57 -0
  395. data/vendor/{rb-sys/LICENSE-MIT → kreuzberg-tesseract/LICENSE} +22 -21
  396. data/vendor/kreuzberg-tesseract/README.md +399 -0
  397. data/vendor/kreuzberg-tesseract/build.rs +1127 -0
  398. data/vendor/kreuzberg-tesseract/patches/README.md +71 -0
  399. data/vendor/kreuzberg-tesseract/patches/tesseract.diff +199 -0
  400. data/vendor/kreuzberg-tesseract/src/api.rs +1371 -0
  401. data/vendor/kreuzberg-tesseract/src/choice_iterator.rs +77 -0
  402. data/vendor/kreuzberg-tesseract/src/enums.rs +297 -0
  403. data/vendor/kreuzberg-tesseract/src/error.rs +81 -0
  404. data/vendor/kreuzberg-tesseract/src/lib.rs +145 -0
  405. data/vendor/kreuzberg-tesseract/src/monitor.rs +57 -0
  406. data/vendor/kreuzberg-tesseract/src/mutable_iterator.rs +197 -0
  407. data/vendor/kreuzberg-tesseract/src/page_iterator.rs +253 -0
  408. data/vendor/kreuzberg-tesseract/src/result_iterator.rs +286 -0
  409. data/vendor/kreuzberg-tesseract/src/result_renderer.rs +183 -0
  410. data/vendor/kreuzberg-tesseract/tests/integration_test.rs +211 -0
  411. metadata +196 -45
  412. data/vendor/kreuzberg/benches/otel_overhead.rs +0 -48
  413. data/vendor/kreuzberg/src/extractors/fictionbook.rs.backup2 +0 -738
  414. data/vendor/rb-sys/.cargo-ok +0 -1
  415. data/vendor/rb-sys/.cargo_vcs_info.json +0 -6
  416. data/vendor/rb-sys/Cargo.lock +0 -393
  417. data/vendor/rb-sys/Cargo.toml +0 -70
  418. data/vendor/rb-sys/Cargo.toml.orig +0 -57
  419. data/vendor/rb-sys/LICENSE-APACHE +0 -190
  420. data/vendor/rb-sys/bin/release.sh +0 -21
  421. data/vendor/rb-sys/build/features.rs +0 -108
  422. data/vendor/rb-sys/build/main.rs +0 -246
  423. data/vendor/rb-sys/build/stable_api_config.rs +0 -153
  424. data/vendor/rb-sys/build/version.rs +0 -48
  425. data/vendor/rb-sys/readme.md +0 -36
  426. data/vendor/rb-sys/src/bindings.rs +0 -21
  427. data/vendor/rb-sys/src/hidden.rs +0 -11
  428. data/vendor/rb-sys/src/lib.rs +0 -34
  429. data/vendor/rb-sys/src/macros.rs +0 -371
  430. data/vendor/rb-sys/src/memory.rs +0 -53
  431. data/vendor/rb-sys/src/ruby_abi_version.rs +0 -38
  432. data/vendor/rb-sys/src/special_consts.rs +0 -31
  433. data/vendor/rb-sys/src/stable_api/compiled.c +0 -179
  434. data/vendor/rb-sys/src/stable_api/compiled.rs +0 -257
  435. data/vendor/rb-sys/src/stable_api/ruby_2_6.rs +0 -316
  436. data/vendor/rb-sys/src/stable_api/ruby_2_7.rs +0 -316
  437. data/vendor/rb-sys/src/stable_api/ruby_3_0.rs +0 -324
  438. data/vendor/rb-sys/src/stable_api/ruby_3_1.rs +0 -317
  439. data/vendor/rb-sys/src/stable_api/ruby_3_2.rs +0 -315
  440. data/vendor/rb-sys/src/stable_api/ruby_3_3.rs +0 -326
  441. data/vendor/rb-sys/src/stable_api/ruby_3_4.rs +0 -327
  442. data/vendor/rb-sys/src/stable_api.rs +0 -261
  443. data/vendor/rb-sys/src/symbol.rs +0 -31
  444. data/vendor/rb-sys/src/tracking_allocator.rs +0 -332
  445. data/vendor/rb-sys/src/utils.rs +0 -89
  446. data/vendor/rb-sys/src/value_type.rs +0 -7
@@ -1,153 +0,0 @@
1
- use rb_sys_build::{RbConfig, RubyEngine};
2
-
3
- use crate::{
4
- features::is_env_variable_defined,
5
- version::{MIN_SUPPORTED_STABLE_VERSION, Version},
6
- };
7
- use std::{convert::TryFrom, error::Error, path::Path};
8
-
9
- pub fn setup(rb_config: &RbConfig) -> Result<(), Box<dyn Error>> {
10
- let ruby_version = Version::current(rb_config);
11
- let ruby_engine = rb_config.ruby_engine();
12
- let strategy = Strategy::try_from((ruby_engine, ruby_version))?;
13
-
14
- strategy.apply()?;
15
-
16
- Ok(())
17
- }
18
-
19
- #[derive(Debug)]
20
- enum Strategy {
21
- RustOnly(Version),
22
- CompiledOnly,
23
- RustThenCompiled(Version),
24
- Testing(Version),
25
- }
26
-
27
- impl TryFrom<(RubyEngine, Version)> for Strategy {
28
- type Error = Box<dyn Error>;
29
-
30
- fn try_from((engine, current_ruby_version): (RubyEngine, Version)) -> Result<Self, Self::Error> {
31
- let mut strategy = None;
32
-
33
- match engine {
34
- RubyEngine::TruffleRuby => {
35
- return Ok(Strategy::CompiledOnly);
36
- }
37
- RubyEngine::JRuby => {
38
- return Err("JRuby is not supported".into());
39
- }
40
- RubyEngine::Mri => {}
41
- }
42
-
43
- if current_ruby_version.is_stable() {
44
- strategy = Some(Strategy::RustOnly(current_ruby_version));
45
- } else {
46
- maybe_warn_old_ruby_version(current_ruby_version);
47
- }
48
-
49
- if is_fallback_enabled() {
50
- strategy = Some(Strategy::RustThenCompiled(current_ruby_version));
51
- }
52
-
53
- if is_testing() {
54
- strategy = Some(Strategy::Testing(current_ruby_version));
55
- }
56
-
57
- if is_force_enabled() {
58
- strategy = Some(Strategy::CompiledOnly);
59
- }
60
-
61
- if let Some(strategy) = strategy {
62
- return Ok(strategy);
63
- }
64
-
65
- Err("Stable API is needed but could not find a candidate. Try enabling the `stable-api-compiled-fallback` feature in rb-sys.".into())
66
- }
67
- }
68
-
69
- impl Strategy {
70
- fn apply(self) -> Result<(), Box<dyn Error>> {
71
- println!("cargo:rustc-check-cfg=cfg(stable_api_include_rust_impl)");
72
- println!("cargo:rustc-check-cfg=cfg(stable_api_enable_compiled_mod)");
73
- println!("cargo:rustc-check-cfg=cfg(stable_api_export_compiled_as_api)");
74
- println!("cargo:rustc-check-cfg=cfg(stable_api_has_rust_impl)");
75
- match self {
76
- Strategy::RustOnly(current_ruby_version) => {
77
- if current_ruby_version.is_stable() {
78
- println!("cargo:rustc-cfg=stable_api_include_rust_impl");
79
- } else {
80
- return Err(format!("A stable Ruby API is needed but could not find a candidate. If you are using a stable version of Ruby, try upgrading rb-sys. Otherwise if you are testing against ruby-head or Ruby < {}, enable the `stable-api-compiled-fallback` feature in rb-sys.", MIN_SUPPORTED_STABLE_VERSION).into());
81
- }
82
- }
83
- Strategy::CompiledOnly => {
84
- compile()?;
85
- println!("cargo:rustc-cfg=stable_api_enable_compiled_mod");
86
- println!("cargo:rustc-cfg=stable_api_export_compiled_as_api");
87
- }
88
- Strategy::RustThenCompiled(current_ruby_version) => {
89
- if current_ruby_version.is_stable() {
90
- println!("cargo:rustc-cfg=stable_api_has_rust_impl");
91
- println!("cargo:rustc-cfg=stable_api_include_rust_impl");
92
- } else {
93
- compile()?;
94
- println!("cargo:rustc-cfg=stable_api_enable_compiled_mod");
95
- println!("cargo:rustc-cfg=stable_api_export_compiled_as_api");
96
- }
97
- }
98
- Strategy::Testing(current_ruby_version) => {
99
- compile()?;
100
-
101
- println!("cargo:rustc-cfg=stable_api_enable_compiled_mod");
102
-
103
- if current_ruby_version.is_stable() {
104
- println!("cargo:rustc-cfg=stable_api_include_rust_impl");
105
- } else {
106
- println!("cargo:rustc-cfg=stable_api_export_compiled_as_api");
107
- }
108
- }
109
- };
110
-
111
- Ok(())
112
- }
113
- }
114
-
115
- fn is_fallback_enabled() -> bool {
116
- println!("cargo:rerun-if-env-changed=RB_SYS_STABLE_API_COMPILED_FALLBACK");
117
-
118
- is_env_variable_defined("CARGO_FEATURE_STABLE_API_COMPILED_FALLBACK")
119
- || cfg!(rb_sys_use_stable_api_compiled_fallback)
120
- || is_env_variable_defined("RB_SYS_STABLE_API_COMPILED_FALLBACK")
121
- }
122
-
123
- fn is_force_enabled() -> bool {
124
- println!("cargo:rerun-if-env-changed=RB_SYS_STABLE_API_COMPILED_FORCE");
125
-
126
- is_env_variable_defined("CARGO_FEATURE_STABLE_API_COMPILED_FORCE")
127
- || cfg!(rb_sys_force_stable_api_compiled)
128
- || is_env_variable_defined("RB_SYS_STABLE_API_COMPILED_FORCE")
129
- }
130
-
131
- fn is_testing() -> bool {
132
- is_env_variable_defined("CARGO_FEATURE_STABLE_API_COMPILED_TESTING")
133
- }
134
-
135
- fn maybe_warn_old_ruby_version(current_ruby_version: Version) {
136
- if current_ruby_version < MIN_SUPPORTED_STABLE_VERSION {
137
- println!(
138
- "cargo:warning=Support for Ruby {} will be removed in a future release.",
139
- current_ruby_version
140
- );
141
- }
142
- }
143
-
144
- fn compile() -> Result<(), Box<dyn Error>> {
145
- eprintln!("INFO: Compiling the stable API compiled module");
146
- let mut build = rb_sys_build::cc::Build::new();
147
- let crate_dir = Path::new(env!("CARGO_MANIFEST_DIR"));
148
- let path = crate_dir.join("src").join("stable_api").join("compiled.c");
149
- eprintln!("cargo:rerun-if-changed={}", path.display());
150
-
151
- build.file(path);
152
- build.try_compile("compiled")
153
- }
@@ -1,48 +0,0 @@
1
- use crate::RbConfig;
2
-
3
- #[allow(dead_code)]
4
- pub const LATEST_STABLE_VERSION: Version = Version::new(3, 4);
5
- #[allow(dead_code)]
6
- pub const MIN_SUPPORTED_STABLE_VERSION: Version = Version::new(2, 6);
7
-
8
- #[derive(Debug, PartialEq, Eq, PartialOrd, Clone, Copy)]
9
- pub struct Version(u32, u32);
10
-
11
- impl Version {
12
- pub const fn new(major: u32, minor: u32) -> Self {
13
- Self(major, minor)
14
- }
15
-
16
- pub fn major(&self) -> u32 {
17
- self.0
18
- }
19
-
20
- pub fn minor(&self) -> u32 {
21
- self.1
22
- }
23
-
24
- pub fn current(rbconfig: &RbConfig) -> Version {
25
- match (rbconfig.get("MAJOR"), rbconfig.get("MINOR")) {
26
- (Some(major), Some(minor)) => Version::new(major.parse::<u32>().unwrap(), minor.parse::<u32>().unwrap()),
27
- _ => {
28
- // Try to parse out the first 3 components of the version string (for truffleruby)
29
- let version_string = rbconfig.get("ruby_version").expect("ruby_version");
30
- let mut parts = version_string.split('.').map(|s| s.parse::<u32>());
31
- let major = parts.next().expect("major").unwrap();
32
- let minor = parts.next().expect("minor").unwrap();
33
- Version::new(major, minor)
34
- }
35
- }
36
- }
37
-
38
- #[allow(dead_code)]
39
- pub fn is_stable(&self) -> bool {
40
- *self >= MIN_SUPPORTED_STABLE_VERSION && *self <= LATEST_STABLE_VERSION
41
- }
42
- }
43
-
44
- impl std::fmt::Display for Version {
45
- fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
46
- write!(f, "{}.{}", self.0, self.1)
47
- }
48
- }
@@ -1,36 +0,0 @@
1
- # rb-sys
2
-
3
- [![Join the discussion](https://img.shields.io/badge/slack-chat-blue.svg)](https://join.slack.com/t/oxidize-rb/shared_invite/zt-16zv5tqte-Vi7WfzxCesdo2TqF_RYBCw)
4
-
5
- Autogenerated Rust bindings for Ruby. Uses the [`rust-bindgen`](https://github.com/rust-lang/rust-bindgen) crate to
6
- generate bindings from the `ruby.h` header.
7
-
8
- ## Notice
9
-
10
- This is a very low-level library. If you are looking to write a gem in Rust, you should probably use
11
- https://github.com/matsadler/magnus crate, with the `rb-sys-interop` feature.
12
-
13
- If you actually _need_ raw/unsafe bindings to libruby, then this crate is for you!
14
-
15
- ## Documentation
16
-
17
- For comprehensive documentation, please refer to the [Ruby on Rust Book](https://oxidize-rb.github.io/rb-sys/), which
18
- includes:
19
-
20
- - [API Reference for rb-sys Features](https://oxidize-rb.github.io/rb-sys/api-reference/rb-sys-features.html)
21
- - [Usage examples and tutorials](https://oxidize-rb.github.io/rb-sys/introduction.html)
22
- - [Memory management guidance](https://oxidize-rb.github.io/rb-sys/memory-management.html)
23
-
24
- ## License
25
-
26
- Licensed under either of
27
-
28
- - Apache License, Version 2.0, ([LICENSE-APACHE](LICENSE-APACHE) or http://www.apache.org/licenses/LICENSE-2.0)
29
- - MIT license ([LICENSE-MIT](LICENSE-MIT) or http://opensource.org/licenses/MIT)
30
-
31
- at your option.
32
-
33
- ### Contribution
34
-
35
- Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as
36
- defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
@@ -1,21 +0,0 @@
1
- //! Raw bindings to libruby, generated by bindgen.
2
- //!
3
- //! This module contains the raw bindings to libruby, generated by bindgen.
4
-
5
- #![allow(non_upper_case_globals)]
6
- #![allow(non_camel_case_types)]
7
- #![allow(non_snake_case)]
8
- #![allow(unknown_lints)]
9
- #![allow(deref_nullptr)]
10
- #![warn(unknown_lints)]
11
- #![allow(clippy::all)]
12
- #![allow(rustdoc::broken_intra_doc_links)]
13
- #![allow(rustdoc::invalid_rust_codeblocks)]
14
- #![allow(rustdoc::invalid_html_tags)]
15
- #![allow(deprecated)]
16
- #![allow(dead_code)]
17
-
18
- include!(env!("RB_SYS_BINDINGS_PATH"));
19
-
20
- pub use uncategorized::*;
21
- pub use unstable::*;
@@ -1,11 +0,0 @@
1
- //! Hidden symbols from Ruby that we need to link to, not exposed to users.
2
- //!
3
- //! Note: Using these symbols is an absolute last resort. Try to use the
4
- //! official Ruby C API if at all possible.
5
-
6
- extern "C" {
7
- /// A pointer to the current Ruby VM.
8
- #[cfg(all(ruby_gt_2_4, ruby_lte_3_2))]
9
- #[cfg(ruby_engine = "mri")]
10
- pub(crate) static ruby_current_vm_ptr: *mut crate::ruby_vm_t;
11
- }
@@ -1,34 +0,0 @@
1
- #![allow(rustdoc::bare_urls)]
2
- #![doc = include_str!("../readme.md")]
3
-
4
- pub mod bindings;
5
- #[cfg(feature = "stable-api")]
6
- pub mod macros;
7
- pub mod memory;
8
- pub mod special_consts;
9
- #[cfg(feature = "stable-api")]
10
- pub mod stable_api;
11
- pub mod symbol;
12
- pub mod tracking_allocator;
13
- pub mod value_type;
14
-
15
- mod hidden;
16
- mod ruby_abi_version;
17
- mod utils;
18
-
19
- pub use bindings::*;
20
- #[cfg(feature = "stable-api")]
21
- pub use macros::*;
22
- pub use ruby_abi_version::*;
23
- pub use special_consts::*;
24
- #[cfg(feature = "stable-api")]
25
- pub use stable_api::StableApiDefinition;
26
- pub use value_type::*;
27
-
28
- #[deprecated(since = "0.9.79", note = "Use `VALUE` instead")]
29
- pub type Value = VALUE;
30
- #[deprecated(since = "0.9.79", note = "Use `VALUE` instead")]
31
- pub type RubyValue = VALUE;
32
-
33
- #[cfg(use_global_allocator)]
34
- set_global_tracking_allocator!();
@@ -1,371 +0,0 @@
1
- //! Implementation of Ruby macros.
2
- //!
3
- //! Since macros are rely on the C preprocessor, or defined as `inline` C
4
- //! functions, they are not available when linking libruby. In order to use the
5
- //! libruby macros from Rust, `rb-sys` implements them using the following
6
- //! strategies:
7
- //!
8
- //! 1. For stable versions of Ruby, the macros are implemented as Rust functions
9
- //! 2. For ruby-head, the macros are implemented as C functions that are linked
10
- //! into the crate.
11
-
12
- #![allow(rustdoc::broken_intra_doc_links)]
13
- #![allow(non_upper_case_globals)]
14
- #![allow(non_snake_case)]
15
-
16
- use crate::StableApiDefinition;
17
- use crate::VALUE;
18
- use crate::rb_data_type_t;
19
- use crate::ruby_value_type;
20
- use crate::stable_api::get_default as api;
21
- use std::ffi::c_void;
22
- use std::os::raw::{c_char, c_long};
23
-
24
- /// Emulates Ruby's "if" statement.
25
- ///
26
- /// - @param[in] obj An arbitrary ruby object.
27
- /// - @retval false `obj` is either ::RUBY_Qfalse or ::RUBY_Qnil.
28
- /// - @retval true Anything else.
29
- ///
30
- /// ```
31
- /// use rb_sys::special_consts::*;
32
- ///
33
- /// assert!(!TEST(Qfalse));
34
- /// assert!(!TEST(Qnil));
35
- /// assert!(TEST(Qtrue));
36
- /// ```
37
- #[inline]
38
- pub fn TEST<T: Into<VALUE>>(obj: T) -> bool {
39
- api().rb_test(obj.into())
40
- }
41
-
42
- /// Checks if the given object is nil.
43
- ///
44
- /// - @param[in] obj An arbitrary ruby object.
45
- /// - @retval true `obj` is ::RUBY_Qnil.
46
- /// - @retval false Anything else.
47
- ///
48
- /// ### Example
49
- ///
50
- /// ```
51
- /// use rb_sys::special_consts::*;
52
- ///
53
- /// assert!(NIL_P(Qnil));
54
- /// assert!(!NIL_P(Qtrue));
55
- /// ```
56
- #[inline]
57
- pub fn NIL_P<T: Into<VALUE>>(obj: T) -> bool {
58
- api().nil_p(obj.into())
59
- }
60
-
61
- /// Checks if the given object is a so-called Fixnum.
62
- ///
63
- /// - @param[in] obj An arbitrary ruby object.
64
- /// - @retval true `obj` is a Fixnum.
65
- /// - @retval false Anything else.
66
- /// - @note Fixnum was a thing in the 20th century, but it is rather an implementation detail today.
67
- #[inline]
68
- pub fn FIXNUM_P<T: Into<VALUE>>(obj: T) -> bool {
69
- api().fixnum_p(obj.into())
70
- }
71
-
72
- /// Checks if the given object is a static symbol.
73
- ///
74
- /// - @param[in] obj An arbitrary ruby object.
75
- /// - @retval true `obj` is a static symbol
76
- /// - @retval false Anything else.
77
- /// - @see RB_DYNAMIC_SYM_P()
78
- /// - @see RB_SYMBOL_P()
79
- /// - @note These days there are static and dynamic symbols, just like we once had Fixnum/Bignum back in the old days.
80
- #[inline]
81
- pub fn STATIC_SYM_P<T: Into<VALUE>>(obj: T) -> bool {
82
- api().static_sym_p(obj.into())
83
- }
84
-
85
- /// Get the backend storage of a Ruby array.
86
- ///
87
- /// ### Safety
88
- ///
89
- /// This function is unsafe because it dereferences a raw pointer and returns
90
- /// raw pointers to Ruby memory. The caller must ensure that the pointer stays live
91
- /// for the duration of usage the the underlying array (by either GC marking or
92
- /// keeping the RArray on the stack).
93
- ///
94
- /// - @param[in] a An object of ::RArray.
95
- /// - @return Its backend storage.
96
- #[inline]
97
- pub unsafe fn RARRAY_CONST_PTR<T: Into<VALUE>>(obj: T) -> *const VALUE {
98
- api().rarray_const_ptr(obj.into())
99
- }
100
-
101
- /// Get the length of a Ruby array.
102
- ///
103
- /// ### Safety
104
- ///
105
- /// This function is unsafe because it dereferences a raw pointer in order to
106
- /// access internal Ruby memory.
107
- ///
108
- /// - @param[in] a An object of ::RArray.
109
- /// - @return Its length.
110
- #[inline]
111
- pub unsafe fn RARRAY_LEN<T: Into<VALUE>>(obj: T) -> c_long {
112
- api().rarray_len(obj.into())
113
- }
114
-
115
- /// Get the length of a Ruby string.
116
- ///
117
- /// ### Safety
118
- ///
119
- /// This function is unsafe because it dereferences a raw pointer in order to
120
- /// access internal Ruby memory.
121
- ///
122
- /// - @param[in] a An object of ::RString.
123
- /// - @return Its length.
124
- #[inline]
125
- pub unsafe fn RSTRING_LEN<T: Into<VALUE>>(obj: T) -> c_long {
126
- api().rstring_len(obj.into())
127
- }
128
-
129
- /// Get the backend storage of a Ruby string.
130
- ///
131
- /// ### Safety
132
- ///
133
- /// This function is unsafe because it dereferences a raw pointer and returns
134
- /// raw pointers to Ruby memory.
135
- ///
136
- /// - @param[in] a An object of ::RString.
137
- /// - @return Its backend storage
138
- #[inline]
139
- pub unsafe fn RSTRING_PTR<T: Into<VALUE>>(obj: T) -> *const c_char {
140
- api().rstring_ptr(obj.into())
141
- }
142
-
143
- /// Checks if the given object is a so-called Flonum.
144
- ///
145
- /// @param[in] obj An arbitrary ruby object.
146
- /// @retval true `obj` is a Flonum.
147
- /// @retval false Anything else.
148
- /// @see RB_FLOAT_TYPE_P()
149
- /// @note These days there are Flonums and non-Flonum floats, just like we
150
- /// once had Fixnum/Bignum back in the old days.
151
- #[inline]
152
- pub fn FLONUM_P<T: Into<VALUE>>(#[allow(unused)] obj: T) -> bool {
153
- api().flonum_p(obj.into())
154
- }
155
-
156
- /// Checks if the given object is an immediate i.e. an object which has no
157
- /// corresponding storage inside of the object space.
158
- ///
159
- /// @param[in] obj An arbitrary ruby object.
160
- /// @retval true `obj` is a Flonum.
161
- /// @retval false Anything else.
162
- /// @see RB_FLOAT_TYPE_P()
163
- /// @note The concept of "immediate" is purely C specific.
164
- #[inline]
165
- pub fn IMMEDIATE_P<T: Into<VALUE>>(obj: T) -> bool {
166
- api().immediate_p(obj.into())
167
- }
168
-
169
- /// Checks if the given object is of enum ::ruby_special_consts.
170
- ///
171
- /// @param[in] obj An arbitrary ruby object.
172
- /// @retval true `obj` is a special constant.
173
- /// @retval false Anything else.
174
- ///
175
- /// ### Example
176
- ///
177
- /// ```
178
- /// use rb_sys::special_consts::*;
179
- ///
180
- /// assert!(SPECIAL_CONST_P(Qnil));
181
- /// assert!(SPECIAL_CONST_P(Qtrue));
182
- /// assert!(SPECIAL_CONST_P(Qfalse));
183
- /// ```
184
- #[inline]
185
- pub fn SPECIAL_CONST_P<T: Into<VALUE>>(obj: T) -> bool {
186
- api().special_const_p(obj.into())
187
- }
188
-
189
- /// Queries the type of the object.
190
- ///
191
- /// @param[in] obj Object in question.
192
- /// @pre `obj` must not be a special constant.
193
- /// @return The type of `obj`.
194
- ///
195
- /// # Safety
196
- /// This function is unsafe because it could dereference a raw pointer when
197
- /// attemping to access the underlying [`RBasic`] struct.
198
- #[inline]
199
- pub unsafe fn RB_BUILTIN_TYPE(obj: VALUE) -> ruby_value_type {
200
- api().builtin_type(obj)
201
- }
202
-
203
- /// Queries if the object is an instance of ::rb_cInteger.
204
- ///
205
- /// @param[in] obj Object in question.
206
- /// @retval true It is.
207
- /// @retval false It isn't.
208
- ///
209
- /// # Safety
210
- /// This function is unsafe because it could dereference a raw pointer when
211
- /// attemping to access the underlying [`RBasic`] struct.
212
- #[inline]
213
- pub unsafe fn RB_INTEGER_TYPE_P(obj: VALUE) -> bool {
214
- api().integer_type_p(obj)
215
- }
216
-
217
- /// Queries if the object is a dynamic symbol.
218
- ///
219
- /// @param[in] obj Object in question.
220
- /// @retval true It is.
221
- /// @retval false It isn't.
222
- ///
223
- /// # Safety
224
- /// This function is unsafe because it could dereference a raw pointer when
225
- /// attemping to access the underlying [`RBasic`] struct.
226
- #[inline]
227
- pub unsafe fn RB_DYNAMIC_SYM_P(obj: VALUE) -> bool {
228
- api().dynamic_sym_p(obj)
229
- }
230
-
231
- /// Queries if the object is an instance of ::rb_cSymbol.
232
- ///
233
- /// @param[in] obj Object in question.
234
- /// @retval true It is.
235
- /// @retval false It isn't.
236
- ///
237
- /// # Safety
238
- /// This function is unsafe because it could dereference a raw pointer when
239
- /// attemping to access the underlying [`RBasic`] struct.
240
- #[inline]
241
- pub unsafe fn RB_SYMBOL_P(obj: VALUE) -> bool {
242
- api().symbol_p(obj)
243
- }
244
-
245
- /// Identical to RB_BUILTIN_TYPE(), except it can also accept special constants.
246
- ///
247
- /// @param[in] obj Object in question.
248
- /// @return The type of `obj`.
249
- ///
250
- /// # Safety
251
- /// This function is unsafe because it could dereference a raw pointer when
252
- /// attemping to access the underlying [`RBasic`] struct.
253
- #[inline]
254
- pub unsafe fn RB_TYPE(value: VALUE) -> ruby_value_type {
255
- api().rb_type(value)
256
- }
257
-
258
- /// Queries if the given object is of given type.
259
- ///
260
- /// @param[in] obj An object.
261
- /// @param[in] t A type.
262
- /// @retval true `obj` is of type `t`.
263
- /// @retval false Otherwise.
264
- ///
265
- /// # Safety
266
- /// This function is unsafe because it could dereference a raw pointer when
267
- /// attemping to access the underlying [`RBasic`] struct.
268
- #[inline]
269
- #[cfg(ruby_engine = "mri")] // truffleruby provides its own implementation
270
- pub unsafe fn RB_TYPE_P(obj: VALUE, ty: ruby_value_type) -> bool {
271
- api().type_p(obj, ty)
272
- }
273
-
274
- /// Queries if the object is an instance of ::rb_cFloat.
275
- ///
276
- /// @param[in] obj Object in question.
277
- /// @retval true It is.
278
- /// @retval false It isn't.
279
- ///
280
- /// # Safety
281
- /// This function is unsafe because it could dereference a raw pointer when
282
- /// attemping to access the underlying [`RBasic`] struct.
283
- #[inline]
284
- pub unsafe fn RB_FLOAT_TYPE_P(obj: VALUE) -> bool {
285
- api().float_type_p(obj)
286
- }
287
-
288
- /// Checks if the given object is an RTypedData.
289
- ///
290
- /// @param[in] obj Object in question.
291
- /// @retval true It is an RTypedData.
292
- /// @retval false It isn't an RTypedData.
293
- ///
294
- /// # Safety
295
- /// This function is unsafe because it could dereference a raw pointer when
296
- /// accessing the underlying data structure.
297
- #[inline]
298
- pub unsafe fn RTYPEDDATA_P(obj: VALUE) -> bool {
299
- api().rtypeddata_p(obj)
300
- }
301
-
302
- /// Checks if the given RTypedData is embedded.
303
- ///
304
- /// @param[in] obj An RTypedData object.
305
- /// @retval true The data is embedded in the object itself.
306
- /// @retval false The data is stored separately.
307
- ///
308
- /// # Safety
309
- /// This function is unsafe because it could dereference a raw pointer when
310
- /// accessing the underlying data structure. The caller must ensure the object
311
- /// is a valid RTypedData.
312
- #[inline]
313
- pub unsafe fn RTYPEDDATA_EMBEDDED_P(obj: VALUE) -> bool {
314
- api().rtypeddata_embedded_p(obj)
315
- }
316
-
317
- /// Gets the data type information from an RTypedData object.
318
- ///
319
- /// @param[in] obj An RTypedData object.
320
- /// @return Pointer to the rb_data_type_t structure for this object.
321
- ///
322
- /// # Safety
323
- /// This function is unsafe because it dereferences a raw pointer to get
324
- /// access to the underlying data type. The caller must ensure the object
325
- /// is a valid RTypedData.
326
- #[inline]
327
- pub unsafe fn RTYPEDDATA_TYPE(obj: VALUE) -> *const rb_data_type_t {
328
- api().rtypeddata_type(obj)
329
- }
330
-
331
- /// Gets the data pointer from an RTypedData object.
332
- ///
333
- /// @param[in] obj An RTypedData object.
334
- /// @return Pointer to the wrapped C struct.
335
- ///
336
- /// # Safety
337
- /// This function is unsafe because it dereferences a raw pointer to get
338
- /// access to the underlying data. The caller must ensure the object
339
- /// is a valid RTypedData.
340
- #[inline]
341
- pub unsafe fn RTYPEDDATA_GET_DATA(obj: VALUE) -> *mut c_void {
342
- api().rtypeddata_get_data(obj)
343
- }
344
-
345
- /// Checks if the bignum is positive.
346
- ///
347
- /// @param[in] b An object of RBignum.
348
- /// @retval false `b` is less than zero.
349
- /// @retval true Otherwise.
350
- ///
351
- /// # Safety
352
- /// This function is unsafe because it could dereference a raw pointer when
353
- /// accessing the underlying bignum structure.
354
- #[inline]
355
- pub unsafe fn RBIGNUM_POSITIVE_P(b: VALUE) -> bool {
356
- api().bignum_positive_p(b)
357
- }
358
-
359
- /// Checks if the bignum is negative.
360
- ///
361
- /// @param[in] b An object of RBignum.
362
- /// @retval true `b` is less than zero.
363
- /// @retval false Otherwise.
364
- ///
365
- /// # Safety
366
- /// This function is unsafe because it could dereference a raw pointer when
367
- /// accessing the underlying bignum structure.
368
- #[inline]
369
- pub unsafe fn RBIGNUM_NEGATIVE_P(b: VALUE) -> bool {
370
- api().bignum_negative_p(b)
371
- }