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
@@ -0,0 +1,628 @@
1
+ //! OCR Backend Plugin FFI
2
+ //!
3
+ //! Provides FFI bindings for registering and managing custom OCR backends.
4
+
5
+ use std::ffi::{CStr, CString};
6
+ use std::os::raw::c_char;
7
+ use std::ptr;
8
+ use std::sync::Arc;
9
+
10
+ use async_trait::async_trait;
11
+ use kreuzberg::core::config::OcrConfig;
12
+ use kreuzberg::plugins::registry::get_ocr_backend_registry;
13
+ use kreuzberg::plugins::{OcrBackend, Plugin};
14
+ use kreuzberg::types::ExtractionResult;
15
+ use kreuzberg::{KreuzbergError, Result};
16
+
17
+ use crate::helpers::{clear_last_error, set_last_error};
18
+ use crate::memory::kreuzberg_free_string;
19
+ // Macros are exported at the crate root due to #[macro_export]
20
+ use crate::{ffi_panic_guard, ffi_panic_guard_bool, ffi_panic_guard_i32};
21
+
22
+ /// Type alias for the OCR backend callback function.
23
+ ///
24
+ /// # Parameters
25
+ ///
26
+ /// - `image_bytes`: Raw image bytes
27
+ /// - `image_length`: Length of image data in bytes
28
+ /// - `config_json`: JSON-encoded OcrConfig (null-terminated string)
29
+ ///
30
+ /// # Returns
31
+ ///
32
+ /// Null-terminated string containing extracted text (must be freed by Rust via kreuzberg_free_string),
33
+ /// or NULL on error.
34
+ ///
35
+ /// # Safety
36
+ ///
37
+ /// The callback must:
38
+ /// - Not store the image_bytes pointer (it's only valid for the duration of the call)
39
+ /// - Return a valid null-terminated UTF-8 string allocated by the caller
40
+ /// - Return NULL on error (error message should be retrievable separately)
41
+ type OcrBackendCallback =
42
+ unsafe extern "C" fn(image_bytes: *const u8, image_length: usize, config_json: *const c_char) -> *mut c_char;
43
+
44
+ fn parse_languages_from_json(languages_json: *const c_char) -> std::result::Result<Option<Vec<String>>, String> {
45
+ if languages_json.is_null() {
46
+ return Ok(None);
47
+ }
48
+
49
+ let raw = unsafe { CStr::from_ptr(languages_json) }
50
+ .to_str()
51
+ .map_err(|e| format!("Invalid UTF-8 in languages JSON: {}", e))?;
52
+
53
+ if raw.trim().is_empty() {
54
+ return Ok(None);
55
+ }
56
+
57
+ let langs: Vec<String> = serde_json::from_str(raw).map_err(|e| format!("Failed to parse languages JSON: {}", e))?;
58
+
59
+ if langs.is_empty() {
60
+ return Ok(None);
61
+ }
62
+
63
+ let normalized = langs
64
+ .into_iter()
65
+ .map(|l| l.trim().to_string())
66
+ .filter(|l| !l.is_empty())
67
+ .collect::<Vec<_>>();
68
+
69
+ if normalized.is_empty() {
70
+ return Ok(None);
71
+ }
72
+
73
+ Ok(Some(normalized))
74
+ }
75
+
76
+ /// FFI wrapper for custom OCR backends registered from Java/C.
77
+ ///
78
+ /// This struct wraps a C function pointer and implements the OcrBackend trait,
79
+ /// allowing custom OCR implementations from FFI languages to be registered
80
+ /// and used within the Rust extraction pipeline.
81
+ struct FfiOcrBackend {
82
+ name: String,
83
+ callback: OcrBackendCallback,
84
+ supported_languages: Option<Vec<String>>,
85
+ }
86
+
87
+ impl FfiOcrBackend {
88
+ fn new(name: String, callback: OcrBackendCallback, supported_languages: Option<Vec<String>>) -> Self {
89
+ Self {
90
+ name,
91
+ callback,
92
+ supported_languages,
93
+ }
94
+ }
95
+ }
96
+
97
+ impl Plugin for FfiOcrBackend {
98
+ fn name(&self) -> &str {
99
+ &self.name
100
+ }
101
+
102
+ fn version(&self) -> String {
103
+ "ffi-1.0.0".to_string()
104
+ }
105
+
106
+ fn initialize(&self) -> Result<()> {
107
+ Ok(())
108
+ }
109
+
110
+ fn shutdown(&self) -> Result<()> {
111
+ Ok(())
112
+ }
113
+ }
114
+
115
+ #[async_trait]
116
+ impl OcrBackend for FfiOcrBackend {
117
+ async fn process_image(&self, image_bytes: &[u8], config: &OcrConfig) -> Result<ExtractionResult> {
118
+ let config_json = serde_json::to_string(config).map_err(|e| KreuzbergError::Validation {
119
+ message: format!("Failed to serialize OCR config: {}", e),
120
+ source: Some(Box::new(e)),
121
+ })?;
122
+
123
+ let callback = self.callback;
124
+ let image_data = image_bytes.to_vec();
125
+ let config_json_owned = config_json.clone();
126
+
127
+ let result_text = tokio::task::spawn_blocking(move || {
128
+ let config_cstring = CString::new(config_json_owned).map_err(|e| KreuzbergError::Validation {
129
+ message: format!("Failed to create C string from config JSON: {}", e),
130
+ source: Some(Box::new(e)),
131
+ })?;
132
+
133
+ let result_ptr = unsafe { callback(image_data.as_ptr(), image_data.len(), config_cstring.as_ptr()) };
134
+
135
+ if result_ptr.is_null() {
136
+ return Err(KreuzbergError::Ocr {
137
+ message: "OCR backend returned NULL (operation failed)".to_string(),
138
+ source: None,
139
+ });
140
+ }
141
+
142
+ let result_cstr = unsafe { CStr::from_ptr(result_ptr) };
143
+ let text = result_cstr
144
+ .to_str()
145
+ .map_err(|e| KreuzbergError::Ocr {
146
+ message: format!("OCR backend returned invalid UTF-8: {}", e),
147
+ source: Some(Box::new(e)),
148
+ })?
149
+ .to_string();
150
+
151
+ unsafe { kreuzberg_free_string(result_ptr) };
152
+
153
+ Ok(text)
154
+ })
155
+ .await
156
+ .map_err(|e| KreuzbergError::Ocr {
157
+ message: format!("OCR backend task panicked: {}", e),
158
+ source: Some(Box::new(e)),
159
+ })??;
160
+
161
+ Ok(ExtractionResult {
162
+ content: result_text,
163
+ mime_type: "text/plain".to_string(),
164
+ metadata: kreuzberg::types::Metadata::default(),
165
+ tables: vec![],
166
+ detected_languages: None,
167
+ chunks: None,
168
+ images: None,
169
+ pages: None,
170
+ })
171
+ }
172
+
173
+ fn supports_language(&self, _lang: &str) -> bool {
174
+ match &self.supported_languages {
175
+ Some(langs) => langs.iter().any(|candidate| candidate.eq_ignore_ascii_case(_lang)),
176
+ None => true,
177
+ }
178
+ }
179
+
180
+ fn backend_type(&self) -> kreuzberg::plugins::OcrBackendType {
181
+ kreuzberg::plugins::OcrBackendType::Custom
182
+ }
183
+ }
184
+
185
+ /// Register a custom OCR backend via FFI callback.
186
+ ///
187
+ /// # Safety
188
+ ///
189
+ /// - `name` must be a valid null-terminated C string
190
+ /// - `callback` must be a valid function pointer that:
191
+ /// - Does not store the image_bytes pointer
192
+ /// - Returns a null-terminated UTF-8 string or NULL on error
193
+ /// - The returned string must be freeable by kreuzberg_free_string
194
+ /// - Returns true on success, false on error (check kreuzberg_last_error)
195
+ ///
196
+ /// # Example (C)
197
+ ///
198
+ /// ```c
199
+ /// char* my_ocr_backend(const uint8_t* image_bytes, size_t image_length, const char* config_json) {
200
+ /// // Implement OCR logic here
201
+ /// // Return allocated string with result, or NULL on error
202
+ /// return strdup("Extracted text");
203
+ /// }
204
+ ///
205
+ /// bool success = kreuzberg_register_ocr_backend("my-ocr", my_ocr_backend);
206
+ /// if (!success) {
207
+ /// const char* error = kreuzberg_last_error();
208
+ /// printf("Failed to register: %s\n", error);
209
+ /// }
210
+ /// ```
211
+ #[unsafe(no_mangle)]
212
+ pub unsafe extern "C" fn kreuzberg_register_ocr_backend(name: *const c_char, callback: OcrBackendCallback) -> bool {
213
+ ffi_panic_guard_bool!("kreuzberg_register_ocr_backend", {
214
+ clear_last_error();
215
+
216
+ if name.is_null() {
217
+ set_last_error("Backend name cannot be NULL".to_string());
218
+ return false;
219
+ }
220
+
221
+ let name_str = match unsafe { CStr::from_ptr(name) }.to_str() {
222
+ Ok(s) => s,
223
+ Err(e) => {
224
+ set_last_error(format!("Invalid UTF-8 in backend name: {}", e));
225
+ return false;
226
+ }
227
+ };
228
+
229
+ if name_str.is_empty() {
230
+ set_last_error("Plugin name cannot be empty".to_string());
231
+ return false;
232
+ }
233
+
234
+ if name_str.chars().any(|c| c.is_whitespace()) {
235
+ set_last_error("Plugin name cannot contain whitespace".to_string());
236
+ return false;
237
+ }
238
+
239
+ let backend = Arc::new(FfiOcrBackend::new(name_str.to_string(), callback, None));
240
+
241
+ let registry = get_ocr_backend_registry();
242
+ let mut registry_guard = match registry.write() {
243
+ Ok(guard) => guard,
244
+ Err(e) => {
245
+ // ~keep: Lock poisoning indicates a panic in another thread holding the lock.
246
+ set_last_error(format!("Failed to acquire registry write lock: {}", e));
247
+ return false;
248
+ }
249
+ };
250
+
251
+ match registry_guard.register(backend) {
252
+ Ok(()) => true,
253
+ Err(e) => {
254
+ set_last_error(format!("Failed to register OCR backend: {}", e));
255
+ false
256
+ }
257
+ }
258
+ })
259
+ }
260
+
261
+ /// Register a custom OCR backend with explicit language support via FFI callback.
262
+ ///
263
+ /// # Safety
264
+ ///
265
+ /// - `languages_json` must be a null-terminated JSON array of language codes or NULL
266
+ /// - See `kreuzberg_register_ocr_backend` for additional safety notes.
267
+ #[unsafe(no_mangle)]
268
+ pub unsafe extern "C" fn kreuzberg_register_ocr_backend_with_languages(
269
+ name: *const c_char,
270
+ callback: OcrBackendCallback,
271
+ languages_json: *const c_char,
272
+ ) -> bool {
273
+ ffi_panic_guard_bool!("kreuzberg_register_ocr_backend_with_languages", {
274
+ clear_last_error();
275
+
276
+ if name.is_null() {
277
+ set_last_error("Backend name cannot be NULL".to_string());
278
+ return false;
279
+ }
280
+
281
+ let name_str = match unsafe { CStr::from_ptr(name) }.to_str() {
282
+ Ok(s) => s,
283
+ Err(e) => {
284
+ set_last_error(format!("Invalid UTF-8 in backend name: {}", e));
285
+ return false;
286
+ }
287
+ };
288
+
289
+ if name_str.is_empty() {
290
+ set_last_error("Plugin name cannot be empty".to_string());
291
+ return false;
292
+ }
293
+
294
+ if name_str.chars().any(|c| c.is_whitespace()) {
295
+ set_last_error("Plugin name cannot contain whitespace".to_string());
296
+ return false;
297
+ }
298
+
299
+ let supported_languages = match parse_languages_from_json(languages_json) {
300
+ Ok(langs) => langs,
301
+ Err(e) => {
302
+ set_last_error(e);
303
+ return false;
304
+ }
305
+ };
306
+
307
+ let backend = Arc::new(FfiOcrBackend::new(name_str.to_string(), callback, supported_languages));
308
+
309
+ let registry = get_ocr_backend_registry();
310
+ let mut registry_guard = match registry.write() {
311
+ Ok(guard) => guard,
312
+ Err(e) => {
313
+ // ~keep: Lock poisoning indicates a panic in another thread holding the lock.
314
+ set_last_error(format!("Failed to acquire registry write lock: {}", e));
315
+ return false;
316
+ }
317
+ };
318
+
319
+ match registry_guard.register(backend) {
320
+ Ok(()) => true,
321
+ Err(e) => {
322
+ set_last_error(format!("Failed to register OCR backend: {}", e));
323
+ false
324
+ }
325
+ }
326
+ })
327
+ }
328
+
329
+ /// Unregister an OCR backend by name.
330
+ ///
331
+ /// # Safety
332
+ ///
333
+ /// - `name` must be a valid null-terminated C string
334
+ /// - Returns true on success, false on error (check kreuzberg_last_error)
335
+ ///
336
+ /// # Example (C)
337
+ ///
338
+ /// ```c
339
+ /// bool success = kreuzberg_unregister_ocr_backend("custom-ocr");
340
+ /// if (!success) {
341
+ /// const char* error = kreuzberg_last_error();
342
+ /// printf("Failed to unregister: %s\n", error);
343
+ /// }
344
+ /// ```
345
+ #[unsafe(no_mangle)]
346
+ pub unsafe extern "C" fn kreuzberg_unregister_ocr_backend(name: *const c_char) -> bool {
347
+ ffi_panic_guard_bool!("kreuzberg_unregister_ocr_backend", {
348
+ clear_last_error();
349
+
350
+ if name.is_null() {
351
+ set_last_error("OCR backend name cannot be NULL".to_string());
352
+ return false;
353
+ }
354
+
355
+ let name_str = match unsafe { CStr::from_ptr(name) }.to_str() {
356
+ Ok(s) => s,
357
+ Err(e) => {
358
+ set_last_error(format!("Invalid UTF-8 in OCR backend name: {}", e));
359
+ return false;
360
+ }
361
+ };
362
+
363
+ if name_str.is_empty() {
364
+ set_last_error("OCR backend name cannot be empty".to_string());
365
+ return false;
366
+ }
367
+
368
+ if name_str.chars().any(|c| c.is_whitespace()) {
369
+ set_last_error("OCR backend name cannot contain whitespace".to_string());
370
+ return false;
371
+ }
372
+
373
+ match kreuzberg::plugins::unregister_ocr_backend(name_str) {
374
+ Ok(()) => true,
375
+ Err(e) => {
376
+ set_last_error(e.to_string());
377
+ false
378
+ }
379
+ }
380
+ })
381
+ }
382
+
383
+ /// List all registered OCR backends as a JSON array of names.
384
+ ///
385
+ /// # Safety
386
+ ///
387
+ /// - Returned string must be freed with `kreuzberg_free_string`.
388
+ /// - Returns NULL on error (check `kreuzberg_last_error`).
389
+ ///
390
+ /// # Example (C)
391
+ ///
392
+ /// ```c
393
+ /// char* backends = kreuzberg_list_ocr_backends();
394
+ /// if (backends == NULL) {
395
+ /// const char* error = kreuzberg_last_error();
396
+ /// printf("Failed to list backends: %s\n", error);
397
+ /// } else {
398
+ /// printf("OCR backends: %s\n", backends);
399
+ /// kreuzberg_free_string(backends);
400
+ /// }
401
+ /// ```
402
+ #[unsafe(no_mangle)]
403
+ pub unsafe extern "C" fn kreuzberg_list_ocr_backends() -> *mut c_char {
404
+ ffi_panic_guard!("kreuzberg_list_ocr_backends", {
405
+ clear_last_error();
406
+
407
+ match kreuzberg::plugins::list_ocr_backends() {
408
+ Ok(backends) => match serde_json::to_string(&backends) {
409
+ Ok(json) => match CString::new(json) {
410
+ Ok(cstr) => cstr.into_raw(),
411
+ Err(e) => {
412
+ set_last_error(format!("Failed to create C string: {}", e));
413
+ ptr::null_mut()
414
+ }
415
+ },
416
+ Err(e) => {
417
+ set_last_error(format!("Failed to serialize OCR backend list: {}", e));
418
+ ptr::null_mut()
419
+ }
420
+ },
421
+ Err(e) => {
422
+ set_last_error(e.to_string());
423
+ ptr::null_mut()
424
+ }
425
+ }
426
+ })
427
+ }
428
+
429
+ /// Clear all registered OCR backends.
430
+ ///
431
+ /// # Safety
432
+ ///
433
+ /// - Removes all registered OCR backends. Subsequent extractions will use only built-in backends.
434
+ /// - Returns true on success, false on error.
435
+ ///
436
+ /// # Example (C)
437
+ ///
438
+ /// ```c
439
+ /// bool success = kreuzberg_clear_ocr_backends();
440
+ /// if (!success) {
441
+ /// const char* error = kreuzberg_last_error();
442
+ /// printf("Failed to clear OCR backends: %s\n", error);
443
+ /// }
444
+ /// ```
445
+ #[unsafe(no_mangle)]
446
+ pub unsafe extern "C" fn kreuzberg_clear_ocr_backends() -> bool {
447
+ ffi_panic_guard_bool!("kreuzberg_clear_ocr_backends", {
448
+ clear_last_error();
449
+
450
+ match kreuzberg::plugins::clear_ocr_backends() {
451
+ Ok(()) => true,
452
+ Err(e) => {
453
+ set_last_error(e.to_string());
454
+ false
455
+ }
456
+ }
457
+ })
458
+ }
459
+
460
+ /// Get supported languages for an OCR backend.
461
+ ///
462
+ /// Returns a JSON array of supported language codes for the given backend.
463
+ /// Supported backends: "easyocr", "paddleocr", "tesseract"
464
+ ///
465
+ /// # Safety
466
+ ///
467
+ /// - The returned string must be freed with `kreuzberg_free_string`
468
+ /// - Returns NULL if backend not found or on error (check `kreuzberg_last_error`)
469
+ ///
470
+ /// # Example (C)
471
+ ///
472
+ /// ```c
473
+ /// char* languages = kreuzberg_get_ocr_languages("easyocr");
474
+ /// if (languages != NULL) {
475
+ /// printf("EasyOCR languages: %s\n", languages);
476
+ /// kreuzberg_free_string(languages);
477
+ /// }
478
+ /// ```
479
+ #[unsafe(no_mangle)]
480
+ pub unsafe extern "C" fn kreuzberg_get_ocr_languages(backend: *const c_char) -> *mut c_char {
481
+ ffi_panic_guard!("kreuzberg_get_ocr_languages", {
482
+ clear_last_error();
483
+
484
+ if backend.is_null() {
485
+ set_last_error("Backend name cannot be NULL".to_string());
486
+ return ptr::null_mut();
487
+ }
488
+
489
+ let backend_str = match unsafe { CStr::from_ptr(backend) }.to_str() {
490
+ Ok(s) => s,
491
+ Err(e) => {
492
+ set_last_error(format!("Invalid UTF-8 in backend name: {}", e));
493
+ return ptr::null_mut();
494
+ }
495
+ };
496
+
497
+ use kreuzberg::ocr::LanguageRegistry;
498
+ let registry = LanguageRegistry::global();
499
+
500
+ match registry.get_supported_languages(backend_str) {
501
+ Some(languages) => match serde_json::to_string(languages) {
502
+ Ok(json) => match CString::new(json) {
503
+ Ok(cstr) => cstr.into_raw(),
504
+ Err(e) => {
505
+ set_last_error(format!("Failed to serialize language list: {}", e));
506
+ ptr::null_mut()
507
+ }
508
+ },
509
+ Err(e) => {
510
+ set_last_error(format!("Failed to serialize language list: {}", e));
511
+ ptr::null_mut()
512
+ }
513
+ },
514
+ None => {
515
+ set_last_error(format!("Unknown OCR backend: '{}'", backend_str));
516
+ ptr::null_mut()
517
+ }
518
+ }
519
+ })
520
+ }
521
+
522
+ /// Check if a language is supported by an OCR backend.
523
+ ///
524
+ /// Returns 1 (true) if the language is supported, 0 (false) otherwise.
525
+ ///
526
+ /// # Arguments
527
+ ///
528
+ /// * `backend` - Backend name (e.g., "easyocr", "paddleocr", "tesseract")
529
+ /// * `language` - Language code to check
530
+ ///
531
+ /// # Returns
532
+ ///
533
+ /// 1 if supported, 0 if not supported or backend not found.
534
+ ///
535
+ /// # Example (C)
536
+ ///
537
+ /// ```c
538
+ /// int is_supported = kreuzberg_is_language_supported("easyocr", "en");
539
+ /// if (is_supported) {
540
+ /// printf("English is supported by EasyOCR\n");
541
+ /// }
542
+ /// ```
543
+ ///
544
+ /// # Safety
545
+ ///
546
+ /// - `backend` and `language` must be valid pointers to valid UTF-8 C strings.
547
+ /// - Both pointers can be checked for NULL; returns 0 if either is NULL.
548
+ /// - The C strings must remain valid for the duration of the function call.
549
+ #[unsafe(no_mangle)]
550
+ pub unsafe extern "C" fn kreuzberg_is_language_supported(backend: *const c_char, language: *const c_char) -> i32 {
551
+ ffi_panic_guard_i32!("kreuzberg_is_language_supported", {
552
+ clear_last_error();
553
+
554
+ if backend.is_null() || language.is_null() {
555
+ set_last_error("Backend and language parameters cannot be NULL".to_string());
556
+ return 0;
557
+ }
558
+
559
+ let backend_str = match unsafe { CStr::from_ptr(backend) }.to_str() {
560
+ Ok(s) => s,
561
+ Err(e) => {
562
+ set_last_error(format!("Invalid UTF-8 in backend name: {}", e));
563
+ return 0;
564
+ }
565
+ };
566
+
567
+ let language_str = match unsafe { CStr::from_ptr(language) }.to_str() {
568
+ Ok(s) => s,
569
+ Err(e) => {
570
+ set_last_error(format!("Invalid UTF-8 in language code: {}", e));
571
+ return 0;
572
+ }
573
+ };
574
+
575
+ use kreuzberg::ocr::LanguageRegistry;
576
+ let registry = LanguageRegistry::global();
577
+
578
+ if registry.is_language_supported(backend_str, language_str) {
579
+ 1
580
+ } else {
581
+ 0
582
+ }
583
+ })
584
+ }
585
+
586
+ /// Get list of all registered OCR backends with language support.
587
+ ///
588
+ /// Returns a JSON object mapping backend names to language counts.
589
+ /// Example: `{"easyocr": 80, "paddleocr": 14, "tesseract": 100}`
590
+ ///
591
+ /// # Safety
592
+ ///
593
+ /// - The returned string must be freed with `kreuzberg_free_string`
594
+ /// - Returns NULL on error (check `kreuzberg_last_error`)
595
+ ///
596
+ /// # Example (C)
597
+ ///
598
+ /// ```c
599
+ /// char* backends = kreuzberg_list_ocr_backends_with_languages();
600
+ /// if (backends != NULL) {
601
+ /// printf("Available backends: %s\n", backends);
602
+ /// kreuzberg_free_string(backends);
603
+ /// }
604
+ /// ```
605
+ #[unsafe(no_mangle)]
606
+ pub unsafe extern "C" fn kreuzberg_list_ocr_backends_with_languages() -> *mut c_char {
607
+ ffi_panic_guard!("kreuzberg_list_ocr_backends_with_languages", {
608
+ clear_last_error();
609
+
610
+ use kreuzberg::ocr::LanguageRegistry;
611
+ let registry = LanguageRegistry::global();
612
+ let backends = registry.get_backends();
613
+
614
+ let mut backend_map = serde_json::json!({});
615
+ for backend in backends {
616
+ let count = registry.get_language_count(&backend);
617
+ backend_map[&backend] = serde_json::json!(count);
618
+ }
619
+
620
+ match CString::new(backend_map.to_string()) {
621
+ Ok(cstr) => cstr.into_raw(),
622
+ Err(e) => {
623
+ set_last_error(format!("Failed to serialize backend list: {}", e));
624
+ ptr::null_mut()
625
+ }
626
+ }
627
+ })
628
+ }