kreuzberg 4.0.0.rc1 → 4.0.0.rc2

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 (342) hide show
  1. checksums.yaml +4 -4
  2. data/.gitignore +14 -8
  3. data/.rspec +3 -3
  4. data/.rubocop.yaml +1 -534
  5. data/.rubocop.yml +538 -0
  6. data/Gemfile +8 -9
  7. data/Gemfile.lock +9 -109
  8. data/README.md +426 -421
  9. data/Rakefile +25 -25
  10. data/Steepfile +47 -47
  11. data/examples/async_patterns.rb +341 -340
  12. data/ext/kreuzberg_rb/extconf.rb +45 -35
  13. data/ext/kreuzberg_rb/native/Cargo.lock +6535 -0
  14. data/ext/kreuzberg_rb/native/Cargo.toml +44 -36
  15. data/ext/kreuzberg_rb/native/README.md +425 -425
  16. data/ext/kreuzberg_rb/native/build.rs +15 -17
  17. data/ext/kreuzberg_rb/native/include/ieeefp.h +11 -11
  18. data/ext/kreuzberg_rb/native/include/msvc_compat/strings.h +14 -14
  19. data/ext/kreuzberg_rb/native/include/strings.h +20 -20
  20. data/ext/kreuzberg_rb/native/include/unistd.h +47 -47
  21. data/ext/kreuzberg_rb/native/src/lib.rs +2998 -2939
  22. data/extconf.rb +28 -28
  23. data/kreuzberg.gemspec +148 -105
  24. data/lib/kreuzberg/api_proxy.rb +142 -142
  25. data/lib/kreuzberg/cache_api.rb +46 -45
  26. data/lib/kreuzberg/cli.rb +55 -55
  27. data/lib/kreuzberg/cli_proxy.rb +127 -127
  28. data/lib/kreuzberg/config.rb +691 -684
  29. data/lib/kreuzberg/error_context.rb +32 -0
  30. data/lib/kreuzberg/errors.rb +118 -50
  31. data/lib/kreuzberg/extraction_api.rb +85 -84
  32. data/lib/kreuzberg/mcp_proxy.rb +186 -186
  33. data/lib/kreuzberg/ocr_backend_protocol.rb +113 -113
  34. data/lib/kreuzberg/post_processor_protocol.rb +86 -86
  35. data/lib/kreuzberg/result.rb +216 -216
  36. data/lib/kreuzberg/setup_lib_path.rb +80 -79
  37. data/lib/kreuzberg/validator_protocol.rb +89 -89
  38. data/lib/kreuzberg/version.rb +5 -5
  39. data/lib/kreuzberg.rb +103 -82
  40. data/sig/kreuzberg/internal.rbs +184 -184
  41. data/sig/kreuzberg.rbs +520 -468
  42. data/spec/binding/cache_spec.rb +227 -227
  43. data/spec/binding/cli_proxy_spec.rb +85 -87
  44. data/spec/binding/cli_spec.rb +55 -54
  45. data/spec/binding/config_spec.rb +345 -345
  46. data/spec/binding/config_validation_spec.rb +283 -283
  47. data/spec/binding/error_handling_spec.rb +213 -213
  48. data/spec/binding/errors_spec.rb +66 -66
  49. data/spec/binding/plugins/ocr_backend_spec.rb +307 -307
  50. data/spec/binding/plugins/postprocessor_spec.rb +269 -269
  51. data/spec/binding/plugins/validator_spec.rb +274 -274
  52. data/spec/fixtures/config.toml +39 -39
  53. data/spec/fixtures/config.yaml +41 -42
  54. data/spec/fixtures/invalid_config.toml +4 -4
  55. data/spec/smoke/package_spec.rb +178 -178
  56. data/spec/spec_helper.rb +42 -42
  57. data/vendor/kreuzberg/Cargo.toml +204 -134
  58. data/vendor/kreuzberg/README.md +175 -175
  59. data/vendor/kreuzberg/benches/otel_overhead.rs +48 -0
  60. data/vendor/kreuzberg/build.rs +474 -460
  61. data/vendor/kreuzberg/src/api/error.rs +81 -81
  62. data/vendor/kreuzberg/src/api/handlers.rs +199 -199
  63. data/vendor/kreuzberg/src/api/mod.rs +79 -79
  64. data/vendor/kreuzberg/src/api/server.rs +353 -353
  65. data/vendor/kreuzberg/src/api/types.rs +170 -170
  66. data/vendor/kreuzberg/src/cache/mod.rs +1167 -1143
  67. data/vendor/kreuzberg/src/chunking/mod.rs +677 -677
  68. data/vendor/kreuzberg/src/core/batch_mode.rs +95 -35
  69. data/vendor/kreuzberg/src/core/config.rs +1032 -1032
  70. data/vendor/kreuzberg/src/core/extractor.rs +1024 -903
  71. data/vendor/kreuzberg/src/core/io.rs +329 -327
  72. data/vendor/kreuzberg/src/core/mime.rs +605 -615
  73. data/vendor/kreuzberg/src/core/mod.rs +45 -42
  74. data/vendor/kreuzberg/src/core/pipeline.rs +984 -906
  75. data/vendor/kreuzberg/src/embeddings.rs +432 -323
  76. data/vendor/kreuzberg/src/error.rs +431 -431
  77. data/vendor/kreuzberg/src/extraction/archive.rs +954 -954
  78. data/vendor/kreuzberg/src/extraction/docx.rs +40 -40
  79. data/vendor/kreuzberg/src/extraction/email.rs +854 -854
  80. data/vendor/kreuzberg/src/extraction/excel.rs +688 -688
  81. data/vendor/kreuzberg/src/extraction/html.rs +553 -553
  82. data/vendor/kreuzberg/src/extraction/image.rs +368 -368
  83. data/vendor/kreuzberg/src/extraction/libreoffice.rs +563 -564
  84. data/vendor/kreuzberg/src/extraction/markdown.rs +213 -0
  85. data/vendor/kreuzberg/src/extraction/mod.rs +81 -77
  86. data/vendor/kreuzberg/src/extraction/office_metadata/app_properties.rs +398 -398
  87. data/vendor/kreuzberg/src/extraction/office_metadata/core_properties.rs +247 -247
  88. data/vendor/kreuzberg/src/extraction/office_metadata/custom_properties.rs +240 -240
  89. data/vendor/kreuzberg/src/extraction/office_metadata/mod.rs +130 -128
  90. data/vendor/kreuzberg/src/extraction/office_metadata/odt_properties.rs +287 -0
  91. data/vendor/kreuzberg/src/extraction/pptx.rs +3000 -3000
  92. data/vendor/kreuzberg/src/extraction/structured.rs +490 -490
  93. data/vendor/kreuzberg/src/extraction/table.rs +328 -328
  94. data/vendor/kreuzberg/src/extraction/text.rs +269 -269
  95. data/vendor/kreuzberg/src/extraction/xml.rs +333 -333
  96. data/vendor/kreuzberg/src/extractors/archive.rs +446 -425
  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 -479
  100. data/vendor/kreuzberg/src/extractors/email.rs +143 -129
  101. data/vendor/kreuzberg/src/extractors/epub.rs +707 -0
  102. data/vendor/kreuzberg/src/extractors/excel.rs +343 -344
  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 -410
  106. data/vendor/kreuzberg/src/extractors/image.rs +198 -195
  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 -268
  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 -496
  116. data/vendor/kreuzberg/src/extractors/pptx.rs +248 -234
  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 -126
  122. data/vendor/kreuzberg/src/extractors/text.rs +260 -242
  123. data/vendor/kreuzberg/src/extractors/typst.rs +650 -0
  124. data/vendor/kreuzberg/src/extractors/xml.rs +135 -128
  125. data/vendor/kreuzberg/src/image/dpi.rs +164 -164
  126. data/vendor/kreuzberg/src/image/mod.rs +6 -6
  127. data/vendor/kreuzberg/src/image/preprocessing.rs +417 -417
  128. data/vendor/kreuzberg/src/image/resize.rs +89 -89
  129. data/vendor/kreuzberg/src/keywords/config.rs +154 -154
  130. data/vendor/kreuzberg/src/keywords/mod.rs +237 -237
  131. data/vendor/kreuzberg/src/keywords/processor.rs +267 -267
  132. data/vendor/kreuzberg/src/keywords/rake.rs +293 -294
  133. data/vendor/kreuzberg/src/keywords/types.rs +68 -68
  134. data/vendor/kreuzberg/src/keywords/yake.rs +163 -163
  135. data/vendor/kreuzberg/src/language_detection/mod.rs +942 -942
  136. data/vendor/kreuzberg/src/lib.rs +105 -102
  137. data/vendor/kreuzberg/src/mcp/mod.rs +32 -32
  138. data/vendor/kreuzberg/src/mcp/server.rs +1968 -1966
  139. data/vendor/kreuzberg/src/ocr/cache.rs +469 -469
  140. data/vendor/kreuzberg/src/ocr/error.rs +37 -37
  141. data/vendor/kreuzberg/src/ocr/hocr.rs +216 -216
  142. data/vendor/kreuzberg/src/ocr/mod.rs +58 -58
  143. data/vendor/kreuzberg/src/ocr/processor.rs +863 -847
  144. data/vendor/kreuzberg/src/ocr/table/mod.rs +4 -4
  145. data/vendor/kreuzberg/src/ocr/table/tsv_parser.rs +144 -144
  146. data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +450 -450
  147. data/vendor/kreuzberg/src/ocr/types.rs +393 -393
  148. data/vendor/kreuzberg/src/ocr/utils.rs +47 -47
  149. data/vendor/kreuzberg/src/ocr/validation.rs +206 -206
  150. data/vendor/kreuzberg/src/panic_context.rs +154 -0
  151. data/vendor/kreuzberg/src/pdf/error.rs +122 -122
  152. data/vendor/kreuzberg/src/pdf/images.rs +139 -139
  153. data/vendor/kreuzberg/src/pdf/metadata.rs +346 -346
  154. data/vendor/kreuzberg/src/pdf/mod.rs +50 -50
  155. data/vendor/kreuzberg/src/pdf/rendering.rs +369 -369
  156. data/vendor/kreuzberg/src/pdf/table.rs +393 -420
  157. data/vendor/kreuzberg/src/pdf/text.rs +158 -161
  158. data/vendor/kreuzberg/src/plugins/extractor.rs +1013 -1010
  159. data/vendor/kreuzberg/src/plugins/mod.rs +209 -209
  160. data/vendor/kreuzberg/src/plugins/ocr.rs +620 -629
  161. data/vendor/kreuzberg/src/plugins/processor.rs +642 -641
  162. data/vendor/kreuzberg/src/plugins/registry.rs +1337 -1324
  163. data/vendor/kreuzberg/src/plugins/traits.rs +258 -258
  164. data/vendor/kreuzberg/src/plugins/validator.rs +956 -955
  165. data/vendor/kreuzberg/src/stopwords/mod.rs +1470 -1470
  166. data/vendor/kreuzberg/src/text/mod.rs +19 -19
  167. data/vendor/kreuzberg/src/text/quality.rs +697 -697
  168. data/vendor/kreuzberg/src/text/string_utils.rs +217 -217
  169. data/vendor/kreuzberg/src/text/token_reduction/cjk_utils.rs +164 -164
  170. data/vendor/kreuzberg/src/text/token_reduction/config.rs +100 -100
  171. data/vendor/kreuzberg/src/text/token_reduction/core.rs +796 -796
  172. data/vendor/kreuzberg/src/text/token_reduction/filters.rs +902 -902
  173. data/vendor/kreuzberg/src/text/token_reduction/mod.rs +160 -160
  174. data/vendor/kreuzberg/src/text/token_reduction/semantic.rs +619 -619
  175. data/vendor/kreuzberg/src/text/token_reduction/simd_text.rs +147 -147
  176. data/vendor/kreuzberg/src/types.rs +903 -873
  177. data/vendor/kreuzberg/src/utils/mod.rs +17 -17
  178. data/vendor/kreuzberg/src/utils/quality.rs +959 -959
  179. data/vendor/kreuzberg/src/utils/string_utils.rs +381 -381
  180. data/vendor/kreuzberg/stopwords/af_stopwords.json +53 -53
  181. data/vendor/kreuzberg/stopwords/ar_stopwords.json +482 -482
  182. data/vendor/kreuzberg/stopwords/bg_stopwords.json +261 -261
  183. data/vendor/kreuzberg/stopwords/bn_stopwords.json +400 -400
  184. data/vendor/kreuzberg/stopwords/br_stopwords.json +1205 -1205
  185. data/vendor/kreuzberg/stopwords/ca_stopwords.json +280 -280
  186. data/vendor/kreuzberg/stopwords/cs_stopwords.json +425 -425
  187. data/vendor/kreuzberg/stopwords/da_stopwords.json +172 -172
  188. data/vendor/kreuzberg/stopwords/de_stopwords.json +622 -622
  189. data/vendor/kreuzberg/stopwords/el_stopwords.json +849 -849
  190. data/vendor/kreuzberg/stopwords/en_stopwords.json +1300 -1300
  191. data/vendor/kreuzberg/stopwords/eo_stopwords.json +175 -175
  192. data/vendor/kreuzberg/stopwords/es_stopwords.json +734 -734
  193. data/vendor/kreuzberg/stopwords/et_stopwords.json +37 -37
  194. data/vendor/kreuzberg/stopwords/eu_stopwords.json +100 -100
  195. data/vendor/kreuzberg/stopwords/fa_stopwords.json +801 -801
  196. data/vendor/kreuzberg/stopwords/fi_stopwords.json +849 -849
  197. data/vendor/kreuzberg/stopwords/fr_stopwords.json +693 -693
  198. data/vendor/kreuzberg/stopwords/ga_stopwords.json +111 -111
  199. data/vendor/kreuzberg/stopwords/gl_stopwords.json +162 -162
  200. data/vendor/kreuzberg/stopwords/gu_stopwords.json +226 -226
  201. data/vendor/kreuzberg/stopwords/ha_stopwords.json +41 -41
  202. data/vendor/kreuzberg/stopwords/he_stopwords.json +196 -196
  203. data/vendor/kreuzberg/stopwords/hi_stopwords.json +227 -227
  204. data/vendor/kreuzberg/stopwords/hr_stopwords.json +181 -181
  205. data/vendor/kreuzberg/stopwords/hu_stopwords.json +791 -791
  206. data/vendor/kreuzberg/stopwords/hy_stopwords.json +47 -47
  207. data/vendor/kreuzberg/stopwords/id_stopwords.json +760 -760
  208. data/vendor/kreuzberg/stopwords/it_stopwords.json +634 -634
  209. data/vendor/kreuzberg/stopwords/ja_stopwords.json +136 -136
  210. data/vendor/kreuzberg/stopwords/kn_stopwords.json +84 -84
  211. data/vendor/kreuzberg/stopwords/ko_stopwords.json +681 -681
  212. data/vendor/kreuzberg/stopwords/ku_stopwords.json +64 -64
  213. data/vendor/kreuzberg/stopwords/la_stopwords.json +51 -51
  214. data/vendor/kreuzberg/stopwords/lt_stopwords.json +476 -476
  215. data/vendor/kreuzberg/stopwords/lv_stopwords.json +163 -163
  216. data/vendor/kreuzberg/stopwords/ml_stopwords.json +1 -1
  217. data/vendor/kreuzberg/stopwords/mr_stopwords.json +101 -101
  218. data/vendor/kreuzberg/stopwords/ms_stopwords.json +477 -477
  219. data/vendor/kreuzberg/stopwords/ne_stopwords.json +490 -490
  220. data/vendor/kreuzberg/stopwords/nl_stopwords.json +415 -415
  221. data/vendor/kreuzberg/stopwords/no_stopwords.json +223 -223
  222. data/vendor/kreuzberg/stopwords/pl_stopwords.json +331 -331
  223. data/vendor/kreuzberg/stopwords/pt_stopwords.json +562 -562
  224. data/vendor/kreuzberg/stopwords/ro_stopwords.json +436 -436
  225. data/vendor/kreuzberg/stopwords/ru_stopwords.json +561 -561
  226. data/vendor/kreuzberg/stopwords/si_stopwords.json +193 -193
  227. data/vendor/kreuzberg/stopwords/sk_stopwords.json +420 -420
  228. data/vendor/kreuzberg/stopwords/sl_stopwords.json +448 -448
  229. data/vendor/kreuzberg/stopwords/so_stopwords.json +32 -32
  230. data/vendor/kreuzberg/stopwords/st_stopwords.json +33 -33
  231. data/vendor/kreuzberg/stopwords/sv_stopwords.json +420 -420
  232. data/vendor/kreuzberg/stopwords/sw_stopwords.json +76 -76
  233. data/vendor/kreuzberg/stopwords/ta_stopwords.json +129 -129
  234. data/vendor/kreuzberg/stopwords/te_stopwords.json +54 -54
  235. data/vendor/kreuzberg/stopwords/th_stopwords.json +118 -118
  236. data/vendor/kreuzberg/stopwords/tl_stopwords.json +149 -149
  237. data/vendor/kreuzberg/stopwords/tr_stopwords.json +506 -506
  238. data/vendor/kreuzberg/stopwords/uk_stopwords.json +75 -75
  239. data/vendor/kreuzberg/stopwords/ur_stopwords.json +519 -519
  240. data/vendor/kreuzberg/stopwords/vi_stopwords.json +647 -647
  241. data/vendor/kreuzberg/stopwords/yo_stopwords.json +62 -62
  242. data/vendor/kreuzberg/stopwords/zh_stopwords.json +796 -796
  243. data/vendor/kreuzberg/stopwords/zu_stopwords.json +31 -31
  244. data/vendor/kreuzberg/tests/api_extract_multipart.rs +52 -0
  245. data/vendor/kreuzberg/tests/api_tests.rs +966 -966
  246. data/vendor/kreuzberg/tests/archive_integration.rs +543 -543
  247. data/vendor/kreuzberg/tests/batch_orchestration.rs +556 -542
  248. data/vendor/kreuzberg/tests/batch_processing.rs +316 -304
  249. data/vendor/kreuzberg/tests/bibtex_parity_test.rs +421 -0
  250. data/vendor/kreuzberg/tests/concurrency_stress.rs +525 -509
  251. data/vendor/kreuzberg/tests/config_features.rs +598 -580
  252. data/vendor/kreuzberg/tests/config_loading_tests.rs +415 -439
  253. data/vendor/kreuzberg/tests/core_integration.rs +510 -493
  254. data/vendor/kreuzberg/tests/csv_integration.rs +414 -424
  255. data/vendor/kreuzberg/tests/docbook_extractor_tests.rs +498 -0
  256. data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +122 -124
  257. data/vendor/kreuzberg/tests/docx_vs_pandoc_comparison.rs +370 -0
  258. data/vendor/kreuzberg/tests/email_integration.rs +325 -325
  259. data/vendor/kreuzberg/tests/epub_native_extractor_tests.rs +275 -0
  260. data/vendor/kreuzberg/tests/error_handling.rs +393 -393
  261. data/vendor/kreuzberg/tests/fictionbook_extractor_tests.rs +228 -0
  262. data/vendor/kreuzberg/tests/format_integration.rs +159 -159
  263. data/vendor/kreuzberg/tests/helpers/mod.rs +142 -142
  264. data/vendor/kreuzberg/tests/html_table_test.rs +551 -0
  265. data/vendor/kreuzberg/tests/image_integration.rs +253 -253
  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 -479
  270. data/vendor/kreuzberg/tests/keywords_quality.rs +509 -509
  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 -428
  274. data/vendor/kreuzberg/tests/ocr_configuration.rs +510 -510
  275. data/vendor/kreuzberg/tests/ocr_errors.rs +676 -676
  276. data/vendor/kreuzberg/tests/ocr_quality.rs +627 -627
  277. data/vendor/kreuzberg/tests/ocr_stress.rs +469 -469
  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 -43
  282. data/vendor/kreuzberg/tests/pipeline_integration.rs +1411 -1412
  283. data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +771 -771
  284. data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs +560 -561
  285. data/vendor/kreuzberg/tests/plugin_system.rs +921 -921
  286. data/vendor/kreuzberg/tests/plugin_validator_test.rs +783 -783
  287. data/vendor/kreuzberg/tests/registry_integration_tests.rs +586 -607
  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 -404
  291. data/vendor/kreuzberg/tests/stopwords_integration_test.rs +888 -888
  292. data/vendor/kreuzberg/tests/test_fastembed.rs +609 -609
  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 -87
  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 +90 -95
  331. data/pkg/kreuzberg-4.0.0.rc1.gem +0 -0
  332. data/spec/examples.txt +0 -104
  333. data/vendor/kreuzberg/src/bin/profile_extract.rs +0 -455
  334. data/vendor/kreuzberg/src/extraction/pandoc/batch.rs +0 -275
  335. data/vendor/kreuzberg/src/extraction/pandoc/mime_types.rs +0 -178
  336. data/vendor/kreuzberg/src/extraction/pandoc/mod.rs +0 -491
  337. data/vendor/kreuzberg/src/extraction/pandoc/server.rs +0 -496
  338. data/vendor/kreuzberg/src/extraction/pandoc/subprocess.rs +0 -1188
  339. data/vendor/kreuzberg/src/extraction/pandoc/version.rs +0 -162
  340. data/vendor/kreuzberg/src/extractors/pandoc.rs +0 -201
  341. data/vendor/kreuzberg/tests/chunking_offset_demo.rs +0 -92
  342. data/vendor/kreuzberg/tests/pandoc_integration.rs +0 -503
@@ -0,0 +1,327 @@
1
+ use super::StableApiDefinition;
2
+ use crate::{
3
+ VALUE, debug_ruby_assert_type,
4
+ internal::{RArray, RString, RTypedData},
5
+ ruby_value_type::RUBY_T_DATA,
6
+ value_type,
7
+ };
8
+ use std::{
9
+ ffi::c_void,
10
+ os::raw::{c_char, c_long},
11
+ ptr::NonNull,
12
+ time::Duration,
13
+ };
14
+
15
+ #[cfg(not(ruby_eq_3_4))]
16
+ compile_error!("This file should only be included in Ruby 3.4 builds");
17
+
18
+ pub struct Definition;
19
+
20
+ impl StableApiDefinition for Definition {
21
+ const VERSION_MAJOR: u32 = 3;
22
+ const VERSION_MINOR: u32 = 4;
23
+
24
+ #[inline]
25
+ unsafe fn rstring_len(&self, obj: VALUE) -> c_long {
26
+ assert!(self.type_p(obj, crate::ruby_value_type::RUBY_T_STRING));
27
+
28
+ let rstring: &RString = &*(obj as *const RString);
29
+ rstring.len
30
+ }
31
+
32
+ #[inline]
33
+ unsafe fn rstring_ptr(&self, obj: VALUE) -> *const c_char {
34
+ assert!(self.type_p(obj, crate::ruby_value_type::RUBY_T_STRING));
35
+
36
+ let rstring: &RString = &*(obj as *const RString);
37
+ let flags = rstring.basic.flags;
38
+ let is_heap = (flags & crate::ruby_rstring_flags::RSTRING_NOEMBED as VALUE) != 0;
39
+ let ptr = if !is_heap {
40
+ std::ptr::addr_of!(rstring.as_.embed.ary) as *const _
41
+ } else {
42
+ rstring.as_.heap.ptr
43
+ };
44
+
45
+ assert!(!ptr.is_null());
46
+
47
+ ptr
48
+ }
49
+
50
+ #[inline]
51
+ unsafe fn rarray_len(&self, obj: VALUE) -> c_long {
52
+ assert!(self.type_p(obj, value_type::RUBY_T_ARRAY));
53
+
54
+ let rarray: &RArray = &*(obj as *const RArray);
55
+ let flags = rarray.basic.flags;
56
+ let is_embedded = (flags & crate::ruby_rarray_flags::RARRAY_EMBED_FLAG as VALUE) != 0;
57
+
58
+ if is_embedded {
59
+ let mut f = rarray.basic.flags;
60
+ f &= crate::ruby_rarray_flags::RARRAY_EMBED_LEN_MASK as VALUE;
61
+ f >>= crate::ruby_rarray_consts::RARRAY_EMBED_LEN_SHIFT as VALUE;
62
+ f as c_long
63
+ } else {
64
+ rarray.as_.heap.len
65
+ }
66
+ }
67
+
68
+ #[inline]
69
+ unsafe fn rarray_const_ptr(&self, obj: VALUE) -> *const VALUE {
70
+ assert!(self.type_p(obj, value_type::RUBY_T_ARRAY));
71
+
72
+ let rarray: &RArray = &*(obj as *const RArray);
73
+ let flags = rarray.basic.flags;
74
+ let is_embedded = (flags & crate::ruby_rarray_flags::RARRAY_EMBED_FLAG as VALUE) != 0;
75
+ let ptr = if is_embedded {
76
+ std::ptr::addr_of!(rarray.as_.ary) as *const _
77
+ } else {
78
+ rarray.as_.heap.ptr
79
+ };
80
+
81
+ assert!(!ptr.is_null());
82
+
83
+ ptr
84
+ }
85
+
86
+ #[inline]
87
+ unsafe fn rbasic_class(&self, obj: VALUE) -> Option<NonNull<VALUE>> {
88
+ let rbasic = obj as *const crate::RBasic;
89
+
90
+ NonNull::<VALUE>::new((*rbasic).klass as _)
91
+ }
92
+
93
+ #[inline]
94
+ unsafe fn frozen_p(&self, obj: VALUE) -> bool {
95
+ if self.special_const_p(obj) {
96
+ true
97
+ } else {
98
+ let rbasic = obj as *const crate::RBasic;
99
+ ((*rbasic).flags & crate::ruby_fl_type::RUBY_FL_FREEZE as VALUE) != 0
100
+ }
101
+ }
102
+
103
+ #[inline]
104
+ fn special_const_p(&self, value: VALUE) -> bool {
105
+ let is_immediate = (value) & (crate::special_consts::IMMEDIATE_MASK as VALUE) != 0;
106
+ let test = (value & !(crate::Qnil as VALUE)) != 0;
107
+
108
+ is_immediate || !test
109
+ }
110
+
111
+ #[inline]
112
+ unsafe fn bignum_positive_p(&self, obj: VALUE) -> bool {
113
+ let rbasic = obj as *const crate::RBasic;
114
+
115
+ ((*rbasic).flags & crate::ruby_fl_type::RUBY_FL_USER1 as VALUE) != 0
116
+ }
117
+
118
+ #[inline]
119
+ unsafe fn builtin_type(&self, obj: VALUE) -> crate::ruby_value_type {
120
+ let rbasic = obj as *const crate::RBasic;
121
+ let ret: u32 = ((*rbasic).flags & crate::ruby_value_type::RUBY_T_MASK as VALUE) as _;
122
+
123
+ std::mem::transmute::<_, crate::ruby_value_type>(ret)
124
+ }
125
+
126
+ #[inline]
127
+ fn nil_p(&self, obj: VALUE) -> bool {
128
+ obj == (crate::Qnil as VALUE)
129
+ }
130
+
131
+ #[inline]
132
+ fn fixnum_p(&self, obj: VALUE) -> bool {
133
+ (obj & crate::FIXNUM_FLAG as VALUE) != 0
134
+ }
135
+
136
+ #[inline]
137
+ fn static_sym_p(&self, obj: VALUE) -> bool {
138
+ let mask = !(VALUE::MAX << crate::ruby_special_consts::RUBY_SPECIAL_SHIFT as VALUE);
139
+ (obj & mask) == crate::ruby_special_consts::RUBY_SYMBOL_FLAG as VALUE
140
+ }
141
+
142
+ #[inline]
143
+ fn flonum_p(&self, obj: VALUE) -> bool {
144
+ #[cfg(ruby_use_flonum = "true")]
145
+ let ret = (obj & crate::FLONUM_MASK as VALUE) == crate::FLONUM_FLAG as VALUE;
146
+
147
+ #[cfg(not(ruby_use_flonum = "true"))]
148
+ let ret = false;
149
+
150
+ ret
151
+ }
152
+
153
+ #[inline]
154
+ fn immediate_p(&self, obj: VALUE) -> bool {
155
+ (obj & crate::special_consts::IMMEDIATE_MASK as VALUE) != 0
156
+ }
157
+
158
+ #[inline]
159
+ fn rb_test(&self, obj: VALUE) -> bool {
160
+ (obj & !(crate::Qnil as VALUE)) != 0
161
+ }
162
+
163
+ #[inline]
164
+ unsafe fn type_p(&self, obj: VALUE, t: crate::ruby_value_type) -> bool {
165
+ use crate::ruby_special_consts::*;
166
+ use crate::ruby_value_type::*;
167
+
168
+ if t == RUBY_T_TRUE {
169
+ obj == RUBY_Qtrue as _
170
+ } else if t == RUBY_T_FALSE {
171
+ obj == RUBY_Qfalse as _
172
+ } else if t == RUBY_T_NIL {
173
+ obj == RUBY_Qnil as _
174
+ } else if t == RUBY_T_UNDEF {
175
+ obj == RUBY_Qundef as _
176
+ } else if t == RUBY_T_FIXNUM {
177
+ self.fixnum_p(obj)
178
+ } else if t == RUBY_T_SYMBOL {
179
+ self.symbol_p(obj)
180
+ } else if t == RUBY_T_FLOAT {
181
+ self.float_type_p(obj)
182
+ } else if self.special_const_p(obj) {
183
+ false
184
+ } else if t == self.builtin_type(obj) {
185
+ true
186
+ } else {
187
+ t == self.rb_type(obj)
188
+ }
189
+ }
190
+
191
+ #[inline]
192
+ unsafe fn symbol_p(&self, obj: VALUE) -> bool {
193
+ self.static_sym_p(obj) || self.dynamic_sym_p(obj)
194
+ }
195
+
196
+ #[inline]
197
+ unsafe fn float_type_p(&self, obj: VALUE) -> bool {
198
+ if self.flonum_p(obj) {
199
+ true
200
+ } else if self.special_const_p(obj) {
201
+ false
202
+ } else {
203
+ self.builtin_type(obj) == value_type::RUBY_T_FLOAT
204
+ }
205
+ }
206
+
207
+ #[inline]
208
+ unsafe fn rb_type(&self, obj: VALUE) -> crate::ruby_value_type {
209
+ use crate::ruby_special_consts::*;
210
+ use crate::ruby_value_type::*;
211
+
212
+ if !self.special_const_p(obj) {
213
+ self.builtin_type(obj)
214
+ } else if obj == RUBY_Qfalse as _ {
215
+ RUBY_T_FALSE
216
+ } else if obj == RUBY_Qnil as _ {
217
+ RUBY_T_NIL
218
+ } else if obj == RUBY_Qtrue as _ {
219
+ RUBY_T_TRUE
220
+ } else if obj == RUBY_Qundef as _ {
221
+ RUBY_T_UNDEF
222
+ } else if self.fixnum_p(obj) {
223
+ RUBY_T_FIXNUM
224
+ } else if self.static_sym_p(obj) {
225
+ RUBY_T_SYMBOL
226
+ } else {
227
+ debug_assert!(self.flonum_p(obj));
228
+ RUBY_T_FLOAT
229
+ }
230
+ }
231
+
232
+ #[inline]
233
+ unsafe fn dynamic_sym_p(&self, obj: VALUE) -> bool {
234
+ if self.special_const_p(obj) {
235
+ false
236
+ } else {
237
+ self.builtin_type(obj) == value_type::RUBY_T_SYMBOL
238
+ }
239
+ }
240
+
241
+ #[inline]
242
+ unsafe fn integer_type_p(&self, obj: VALUE) -> bool {
243
+ if self.fixnum_p(obj) {
244
+ true
245
+ } else if self.special_const_p(obj) {
246
+ false
247
+ } else {
248
+ self.builtin_type(obj) == value_type::RUBY_T_BIGNUM
249
+ }
250
+ }
251
+
252
+ #[inline]
253
+ unsafe fn rstring_interned_p(&self, obj: VALUE) -> bool {
254
+ assert!(self.type_p(obj, value_type::RUBY_T_STRING));
255
+
256
+ let rstring: &RString = &*(obj as *const RString);
257
+ let flags = rstring.basic.flags;
258
+
259
+ (flags & crate::ruby_rstring_flags::RSTRING_FSTR as VALUE) != 0
260
+ }
261
+
262
+ #[inline]
263
+ fn thread_sleep(&self, duration: Duration) {
264
+ let seconds = duration.as_secs() as _;
265
+ let microseconds = duration.subsec_micros() as _;
266
+
267
+ let time = crate::timeval {
268
+ tv_sec: seconds,
269
+ tv_usec: microseconds,
270
+ };
271
+
272
+ unsafe { crate::rb_thread_wait_for(time) }
273
+ }
274
+
275
+ #[inline]
276
+ unsafe fn rtypeddata_p(&self, obj: VALUE) -> bool {
277
+ debug_ruby_assert_type!(obj, RUBY_T_DATA, "rtypeddata_p called on non-T_DATA object");
278
+
279
+ // Access the RTypedData struct
280
+ let rdata = obj as *const RTypedData;
281
+ let typed_flag = (*rdata).typed_flag;
282
+ // Valid typed_flag values are 1, 2, or 3
283
+ typed_flag != 0 && typed_flag <= 3
284
+ }
285
+
286
+ #[inline]
287
+ unsafe fn rtypeddata_embedded_p(&self, obj: VALUE) -> bool {
288
+ debug_ruby_assert_type!(obj, RUBY_T_DATA, "rtypeddata_embedded_p called on non-T_DATA object");
289
+
290
+ let rdata = obj as *const RTypedData;
291
+ let typed_flag = (*rdata).typed_flag;
292
+ #[cfg(target_pointer_width = "64")]
293
+ const FLAG: u64 = crate::TYPED_DATA_EMBEDDED as u64;
294
+ #[cfg(target_pointer_width = "32")]
295
+ const FLAG: u32 = crate::TYPED_DATA_EMBEDDED as u32;
296
+
297
+ (typed_flag & FLAG) != 0
298
+ }
299
+
300
+ #[inline]
301
+ unsafe fn rtypeddata_type(&self, obj: VALUE) -> *const crate::rb_data_type_t {
302
+ debug_ruby_assert_type!(obj, RUBY_T_DATA, "rtypeddata_type called on non-T_DATA object");
303
+
304
+ let rdata = obj as *const RTypedData;
305
+ (*rdata).type_
306
+ }
307
+
308
+ #[inline]
309
+ unsafe fn rtypeddata_get_data(&self, obj: VALUE) -> *mut c_void {
310
+ debug_ruby_assert_type!(obj, RUBY_T_DATA, "rtypeddata_get_data called on non-T_DATA object");
311
+
312
+ if self.rtypeddata_embedded_p(obj) {
313
+ // For embedded data, calculate pointer based on struct layout
314
+ // The formula matches Ruby's implementation:
315
+ // embedded_typed_data_size = sizeof(RTypedData) - sizeof(void *)
316
+ const EMBEDDED_TYPED_DATA_SIZE: usize =
317
+ std::mem::size_of::<RTypedData>() - std::mem::size_of::<*mut c_void>();
318
+
319
+ // Return address after the header as the data pointer
320
+ (obj as *mut u8).add(EMBEDDED_TYPED_DATA_SIZE) as *mut c_void
321
+ } else {
322
+ // For non-embedded data, return the data field directly
323
+ let rdata = obj as *const RTypedData;
324
+ (*rdata).data
325
+ }
326
+ }
327
+ }
@@ -0,0 +1,261 @@
1
+ //! Stable ABI functions which provide access to Ruby internals that
2
+ //! is compatible across Ruby versions, and are guaranteed to be not break due
3
+ //! to Ruby binary changes.
4
+ //!
5
+ //! ### Goals
6
+ //!
7
+ //! 1. To provide access to Ruby internals that are not exposed by the libruby
8
+ //! (i.e. C macros and inline functions).
9
+ //! 2. Provide support for Ruby development versions, which can make breaking
10
+ //! changes without semantic versioning. We want to support these versions
11
+ //! to ensure Rust extensions don't prevent the Ruby core team from testing
12
+ //! changes in production.
13
+
14
+ use crate::VALUE;
15
+ use std::{
16
+ os::raw::{c_char, c_long},
17
+ ptr::NonNull,
18
+ time::Duration,
19
+ };
20
+
21
+ pub trait StableApiDefinition {
22
+ const VERSION_MAJOR: u32;
23
+ const VERSION_MINOR: u32;
24
+
25
+ fn version(&self) -> (u32, u32) {
26
+ (Self::VERSION_MAJOR, Self::VERSION_MINOR)
27
+ }
28
+
29
+ /// Get the length of a Ruby string (akin to `RSTRING_LEN`).
30
+ ///
31
+ /// # Safety
32
+ /// This function is unsafe because it dereferences a raw pointer to get
33
+ /// access to underlying Ruby data. The caller must ensure that the pointer
34
+ /// is valid.
35
+ unsafe fn rstring_len(&self, obj: VALUE) -> c_long;
36
+
37
+ /// Get a pointer to the bytes of a Ruby string (akin to `RSTRING_PTR`).
38
+ ///
39
+ /// # Safety
40
+ /// This function is unsafe because it dereferences a raw pointer to get
41
+ /// access to underlying Ruby data. The caller must ensure that the pointer
42
+ /// is valid.
43
+ unsafe fn rstring_ptr(&self, obj: VALUE) -> *const c_char;
44
+
45
+ /// Get the length of a Ruby array (akin to `RARRAY_LEN`).
46
+ ///
47
+ /// # Safety
48
+ /// This function is unsafe because it dereferences a raw pointer to get
49
+ /// access to underlying Ruby data. The caller must ensure that the pointer
50
+ /// is valid.
51
+ unsafe fn rarray_len(&self, obj: VALUE) -> c_long;
52
+
53
+ /// Get a pointer to the elements of a Ruby array (akin to `RARRAY_CONST_PTR`).
54
+ ///
55
+ /// # Safety
56
+ /// This function is unsafe because it dereferences a raw pointer to get
57
+ /// access to underlying Ruby data. The caller must ensure that the pointer
58
+ /// is valid.
59
+ unsafe fn rarray_const_ptr(&self, obj: VALUE) -> *const VALUE;
60
+
61
+ /// Get the class from a VALUE which contains an RBasic struct.
62
+ ///
63
+ /// `VALUE` is a valid pointer to a non-immediate object.
64
+ ///
65
+ /// # Safety
66
+ /// This function is unsafe because it dereferences a raw pointer to get
67
+ /// access to underlying RBasic struct. The caller must ensure that the
68
+ /// `VALUE` is a valid pointer to an RBasic struct.
69
+ unsafe fn rbasic_class(&self, obj: VALUE) -> Option<NonNull<VALUE>>;
70
+
71
+ /// Checks if the given object is frozen.
72
+ ///
73
+ /// `VALUE` is a valid pointer to a non-immediate object.
74
+ ///
75
+ /// # Safety
76
+ /// This function is unsafe because it may dereference a raw pointer to get
77
+ /// access to underlying RBasic struct. The caller must ensure that the
78
+ /// `VALUE` is a valid pointer to an RBasic struct.
79
+ unsafe fn frozen_p(&self, obj: VALUE) -> bool;
80
+
81
+ /// Tests if a bignum is positive.
82
+ ///
83
+ /// # Safety
84
+ /// This function is unsafe because it dereferences a raw pointer to get
85
+ /// access to underlying RBasic struct. The caller must ensure that the
86
+ /// `VALUE` is a valid pointer to a bignum.
87
+ unsafe fn bignum_positive_p(&self, obj: VALUE) -> bool;
88
+
89
+ /// Tests if a bignum is negative.
90
+ ///
91
+ /// # Safety
92
+ /// This function is unsafe because it dereferences a raw pointer to get
93
+ /// access to underlying RBasic struct. The caller must ensure that the
94
+ /// `VALUE` is a valid pointer to a bignum.
95
+ #[inline]
96
+ unsafe fn bignum_negative_p(&self, obj: VALUE) -> bool {
97
+ !self.bignum_positive_p(obj)
98
+ }
99
+
100
+ /// Tests if the given value is a special constant.
101
+ fn special_const_p(&self, value: VALUE) -> bool;
102
+
103
+ /// Queries the type of the object.
104
+ ///
105
+ /// # Note
106
+ /// The input `obj` must not be a special constant.
107
+ ///
108
+ /// # Safety
109
+ /// This function is unsafe because it could dereference a raw pointer when
110
+ /// attemping to access the underlying [`RBasic`] struct.
111
+ unsafe fn builtin_type(&self, obj: VALUE) -> crate::ruby_value_type;
112
+
113
+ /// Tests if the object's type is the given type.
114
+ ///
115
+ /// # Safety
116
+ /// This function is unsafe because it could dereference a raw pointer when
117
+ /// attemping to access the underlying [`RBasic`] struct.
118
+ unsafe fn type_p(&self, obj: VALUE, ty: crate::ruby_value_type) -> bool;
119
+
120
+ /// Checks if the given object is nil.
121
+ fn nil_p(&self, obj: VALUE) -> bool;
122
+
123
+ /// Checks if the given object is a so-called Fixnum.
124
+ fn fixnum_p(&self, obj: VALUE) -> bool;
125
+
126
+ /// Checks if the given object is a dynamic symbol.
127
+ ///
128
+ /// # Safety
129
+ /// This function is unsafe because it could dereference a raw pointer when
130
+ /// attemping to access the underlying [`RBasic`] struct.
131
+ unsafe fn dynamic_sym_p(&self, obj: VALUE) -> bool;
132
+
133
+ /// Checks if the given object is a static symbol.
134
+ fn static_sym_p(&self, obj: VALUE) -> bool;
135
+
136
+ /// Checks if the given object is a symbol.
137
+ ///
138
+ /// # Safety
139
+ /// This function is unsafe because it could dereference a raw pointer when
140
+ /// attemping to access the underlying [`RBasic`] struct.
141
+ unsafe fn symbol_p(&self, obj: VALUE) -> bool;
142
+
143
+ /// Checks if the given object is a so-called Flonum.
144
+ ///
145
+ /// # Safety
146
+ /// This function is unsafe because it could dereference a raw pointer when
147
+ /// attemping to access the underlying [`RBasic`] struct.
148
+ unsafe fn float_type_p(&self, obj: VALUE) -> bool;
149
+
150
+ /// Checks if the given object is an integer type
151
+ ///
152
+ /// # Safety
153
+ /// This function is unsafe because it could dereference a raw pointer when
154
+ /// attemping to access the underlying [`RBasic`] struct.
155
+ unsafe fn integer_type_p(&self, obj: VALUE) -> bool;
156
+
157
+ /// Checks if the given object is a so-called Flonum.
158
+ fn flonum_p(&self, obj: VALUE) -> bool;
159
+
160
+ /// Checks if the given object is an immediate i.e. an object which has
161
+ /// no corresponding storage inside of the object space.
162
+ fn immediate_p(&self, obj: VALUE) -> bool;
163
+
164
+ /// Emulates Ruby's "if" statement by testing if the given `obj` is neither `Qnil` or `Qfalse`.
165
+ ///
166
+ /// # Safety
167
+ /// This function is unsafe because it could dereference a raw pointer when
168
+ /// attemping to access the underlying [`RBasic`] struct.
169
+ fn rb_test(&self, ob: VALUE) -> bool;
170
+
171
+ /// Queries the type of the object. Identical to `StableApi.builtin_type`,
172
+ /// except it can also accept special constants.
173
+ ///
174
+ /// # Safety
175
+ /// This function is unsafe because it could dereference a raw pointer when
176
+ /// attemping to access the underlying [`RBasic`] struct.
177
+ unsafe fn rb_type(&self, obj: VALUE) -> crate::ruby_value_type;
178
+
179
+ /// Check if a Ruby string is interned (akin to `RSTRING_FSTR`).
180
+ ///
181
+ /// # Safety
182
+ /// This function is unsafe because it dereferences a raw pointer to get
183
+ /// access to underlying flags of the RString. The caller must ensure that
184
+ /// the `VALUE` is a valid pointer to an RString.
185
+ unsafe fn rstring_interned_p(&self, obj: VALUE) -> bool;
186
+
187
+ /// Blocks the current thread until the given duration has passed.
188
+ fn thread_sleep(&self, duration: Duration);
189
+
190
+ /// Checks if the given object is an RTypedData.
191
+ ///
192
+ /// # Safety
193
+ /// This function is unsafe because it dereferences a raw pointer to get
194
+ /// access to underlying Ruby data. The caller must ensure that the pointer
195
+ /// is valid and points to a T_DATA object.
196
+ unsafe fn rtypeddata_p(&self, obj: VALUE) -> bool;
197
+
198
+ /// Checks if the given RTypedData is embedded.
199
+ ///
200
+ /// # Safety
201
+ /// This function is unsafe because it dereferences a raw pointer to get
202
+ /// access to underlying Ruby data. The caller must ensure that the pointer
203
+ /// is valid and points to an RTypedData object.
204
+ unsafe fn rtypeddata_embedded_p(&self, obj: VALUE) -> bool;
205
+
206
+ /// Gets the data type from an RTypedData object.
207
+ ///
208
+ /// # Safety
209
+ /// This function is unsafe because it dereferences a raw pointer to get
210
+ /// access to underlying Ruby data. The caller must ensure that the pointer
211
+ /// is valid and points to an RTypedData object.
212
+ unsafe fn rtypeddata_type(&self, obj: VALUE) -> *const crate::rb_data_type_t;
213
+
214
+ /// Gets the data pointer from an RTypedData object.
215
+ ///
216
+ /// # Safety
217
+ /// This function is unsafe because it dereferences a raw pointer to get
218
+ /// access to underlying Ruby data. The caller must ensure that the pointer
219
+ /// is valid and points to an RTypedData object.
220
+ unsafe fn rtypeddata_get_data(&self, obj: VALUE) -> *mut std::ffi::c_void;
221
+ }
222
+
223
+ #[cfg(stable_api_enable_compiled_mod)]
224
+ mod compiled;
225
+ #[cfg(stable_api_export_compiled_as_api)]
226
+ use compiled as api;
227
+
228
+ #[cfg(stable_api_include_rust_impl)]
229
+ #[cfg_attr(ruby_eq_2_6, path = "stable_api/ruby_2_6.rs")]
230
+ #[cfg_attr(ruby_eq_2_7, path = "stable_api/ruby_2_7.rs")]
231
+ #[cfg_attr(ruby_eq_3_0, path = "stable_api/ruby_3_0.rs")]
232
+ #[cfg_attr(ruby_eq_3_1, path = "stable_api/ruby_3_1.rs")]
233
+ #[cfg_attr(ruby_eq_3_2, path = "stable_api/ruby_3_2.rs")]
234
+ #[cfg_attr(ruby_eq_3_3, path = "stable_api/ruby_3_3.rs")]
235
+ #[cfg_attr(ruby_eq_3_4, path = "stable_api/ruby_3_4.rs")]
236
+ mod rust;
237
+ #[cfg(not(stable_api_export_compiled_as_api))]
238
+ use rust as api;
239
+
240
+ impl std::fmt::Debug for api::Definition {
241
+ fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
242
+ f.debug_struct("StableApiDefinition")
243
+ .field("VERSION_MAJOR", &api::Definition::VERSION_MAJOR)
244
+ .field("VERSION_MINOR", &api::Definition::VERSION_MINOR)
245
+ .finish()
246
+ }
247
+ }
248
+
249
+ /// Get the default stable API definition for the current Ruby version.
250
+ pub const fn get_default() -> &'static api::Definition {
251
+ const API: api::Definition = api::Definition {};
252
+ &API
253
+ }
254
+
255
+ /// Get the fallback stable API definition for the current Ruby version, which
256
+ /// is compiled C code that is linked into to this crate.
257
+ #[cfg(stable_api_enable_compiled_mod)]
258
+ pub const fn get_compiled() -> &'static compiled::Definition {
259
+ const COMPILED_API: compiled::Definition = compiled::Definition {};
260
+ &COMPILED_API
261
+ }
@@ -0,0 +1,31 @@
1
+ /// Finds or creates a symbol for the given static string. This macro will
2
+ /// memoize the ID to avoid repeated calls to libruby. You should prefer this
3
+ /// macro over [`rb_intern3`] when the string is known at compile time.
4
+ ///
5
+ /// # Safety
6
+ ///
7
+ /// This macro is safe under two conditions:
8
+ /// - Ruby VM is initialized and that thus safe to call into libruby
9
+ /// - The first call to this macro will be done inside of a managed Ruby thread (i.e. not a native thread)
10
+ ///
11
+ /// # Example
12
+ ///
13
+ /// ```no_run
14
+ /// use rb_sys::{symbol::rb_intern, rb_funcall, rb_utf8_str_new};
15
+ ///
16
+ /// unsafe {
17
+ /// let reverse_id = rb_intern!("reverse");
18
+ /// let msg = rb_utf8_str_new("nice one".as_ptr() as *mut _, 4);
19
+ /// rb_funcall(msg, reverse_id, 0);
20
+ /// }
21
+ /// ```
22
+ #[macro_export]
23
+ macro_rules! rb_intern {
24
+ ($s:literal) => {{
25
+ static mut ID: $crate::ID = 0;
26
+ if ID == 0 {
27
+ ID = $crate::rb_intern3($s.as_ptr() as _, $s.len() as _, $crate::rb_utf8_encoding());
28
+ }
29
+ ID
30
+ }};
31
+ }