kreuzberg 4.0.0.pre.rc.29 → 4.0.0.rc1

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 (321) hide show
  1. checksums.yaml +4 -4
  2. data/.gitignore +0 -6
  3. data/.rubocop.yaml +534 -1
  4. data/Gemfile +2 -1
  5. data/Gemfile.lock +28 -116
  6. data/README.md +269 -629
  7. data/Rakefile +0 -9
  8. data/Steepfile +4 -8
  9. data/examples/async_patterns.rb +58 -1
  10. data/ext/kreuzberg_rb/extconf.rb +5 -35
  11. data/ext/kreuzberg_rb/native/Cargo.toml +16 -55
  12. data/ext/kreuzberg_rb/native/build.rs +14 -12
  13. data/ext/kreuzberg_rb/native/include/ieeefp.h +1 -1
  14. data/ext/kreuzberg_rb/native/include/msvc_compat/strings.h +1 -1
  15. data/ext/kreuzberg_rb/native/include/strings.h +2 -2
  16. data/ext/kreuzberg_rb/native/include/unistd.h +1 -1
  17. data/ext/kreuzberg_rb/native/src/lib.rs +34 -897
  18. data/extconf.rb +6 -38
  19. data/kreuzberg.gemspec +20 -114
  20. data/lib/kreuzberg/api_proxy.rb +18 -2
  21. data/lib/kreuzberg/cache_api.rb +0 -22
  22. data/lib/kreuzberg/cli.rb +10 -2
  23. data/lib/kreuzberg/cli_proxy.rb +10 -0
  24. data/lib/kreuzberg/config.rb +22 -274
  25. data/lib/kreuzberg/errors.rb +7 -73
  26. data/lib/kreuzberg/extraction_api.rb +8 -237
  27. data/lib/kreuzberg/mcp_proxy.rb +11 -2
  28. data/lib/kreuzberg/ocr_backend_protocol.rb +73 -0
  29. data/lib/kreuzberg/post_processor_protocol.rb +71 -0
  30. data/lib/kreuzberg/result.rb +33 -151
  31. data/lib/kreuzberg/setup_lib_path.rb +2 -22
  32. data/lib/kreuzberg/validator_protocol.rb +73 -0
  33. data/lib/kreuzberg/version.rb +1 -1
  34. data/lib/kreuzberg.rb +13 -27
  35. data/pkg/kreuzberg-4.0.0.rc1.gem +0 -0
  36. data/sig/kreuzberg.rbs +12 -105
  37. data/spec/binding/cache_spec.rb +22 -22
  38. data/spec/binding/cli_proxy_spec.rb +4 -2
  39. data/spec/binding/cli_spec.rb +11 -12
  40. data/spec/binding/config_spec.rb +0 -74
  41. data/spec/binding/config_validation_spec.rb +6 -100
  42. data/spec/binding/error_handling_spec.rb +97 -283
  43. data/spec/binding/plugins/ocr_backend_spec.rb +8 -8
  44. data/spec/binding/plugins/postprocessor_spec.rb +11 -11
  45. data/spec/binding/plugins/validator_spec.rb +13 -12
  46. data/spec/examples.txt +104 -0
  47. data/spec/fixtures/config.toml +1 -0
  48. data/spec/fixtures/config.yaml +1 -0
  49. data/spec/fixtures/invalid_config.toml +1 -0
  50. data/spec/smoke/package_spec.rb +3 -2
  51. data/spec/spec_helper.rb +3 -1
  52. data/vendor/kreuzberg/Cargo.toml +67 -192
  53. data/vendor/kreuzberg/README.md +9 -97
  54. data/vendor/kreuzberg/build.rs +194 -516
  55. data/vendor/kreuzberg/src/api/handlers.rs +9 -130
  56. data/vendor/kreuzberg/src/api/mod.rs +3 -18
  57. data/vendor/kreuzberg/src/api/server.rs +71 -236
  58. data/vendor/kreuzberg/src/api/types.rs +7 -43
  59. data/vendor/kreuzberg/src/bin/profile_extract.rs +455 -0
  60. data/vendor/kreuzberg/src/cache/mod.rs +3 -27
  61. data/vendor/kreuzberg/src/chunking/mod.rs +79 -1705
  62. data/vendor/kreuzberg/src/core/batch_mode.rs +0 -60
  63. data/vendor/kreuzberg/src/core/config.rs +23 -905
  64. data/vendor/kreuzberg/src/core/extractor.rs +106 -403
  65. data/vendor/kreuzberg/src/core/io.rs +2 -4
  66. data/vendor/kreuzberg/src/core/mime.rs +12 -2
  67. data/vendor/kreuzberg/src/core/mod.rs +3 -22
  68. data/vendor/kreuzberg/src/core/pipeline.rs +78 -395
  69. data/vendor/kreuzberg/src/embeddings.rs +21 -169
  70. data/vendor/kreuzberg/src/error.rs +2 -2
  71. data/vendor/kreuzberg/src/extraction/archive.rs +31 -36
  72. data/vendor/kreuzberg/src/extraction/docx.rs +1 -365
  73. data/vendor/kreuzberg/src/extraction/email.rs +11 -12
  74. data/vendor/kreuzberg/src/extraction/excel.rs +129 -138
  75. data/vendor/kreuzberg/src/extraction/html.rs +170 -1447
  76. data/vendor/kreuzberg/src/extraction/image.rs +14 -138
  77. data/vendor/kreuzberg/src/extraction/libreoffice.rs +3 -13
  78. data/vendor/kreuzberg/src/extraction/mod.rs +5 -21
  79. data/vendor/kreuzberg/src/extraction/office_metadata/mod.rs +0 -2
  80. data/vendor/kreuzberg/src/extraction/pandoc/batch.rs +275 -0
  81. data/vendor/kreuzberg/src/extraction/pandoc/mime_types.rs +178 -0
  82. data/vendor/kreuzberg/src/extraction/pandoc/mod.rs +491 -0
  83. data/vendor/kreuzberg/src/extraction/pandoc/server.rs +496 -0
  84. data/vendor/kreuzberg/src/extraction/pandoc/subprocess.rs +1188 -0
  85. data/vendor/kreuzberg/src/extraction/pandoc/version.rs +162 -0
  86. data/vendor/kreuzberg/src/extraction/pptx.rs +94 -196
  87. data/vendor/kreuzberg/src/extraction/structured.rs +4 -5
  88. data/vendor/kreuzberg/src/extraction/table.rs +1 -2
  89. data/vendor/kreuzberg/src/extraction/text.rs +10 -18
  90. data/vendor/kreuzberg/src/extractors/archive.rs +0 -22
  91. data/vendor/kreuzberg/src/extractors/docx.rs +148 -69
  92. data/vendor/kreuzberg/src/extractors/email.rs +9 -37
  93. data/vendor/kreuzberg/src/extractors/excel.rs +40 -81
  94. data/vendor/kreuzberg/src/extractors/html.rs +173 -182
  95. data/vendor/kreuzberg/src/extractors/image.rs +8 -32
  96. data/vendor/kreuzberg/src/extractors/mod.rs +10 -171
  97. data/vendor/kreuzberg/src/extractors/pandoc.rs +201 -0
  98. data/vendor/kreuzberg/src/extractors/pdf.rs +64 -329
  99. data/vendor/kreuzberg/src/extractors/pptx.rs +34 -79
  100. data/vendor/kreuzberg/src/extractors/structured.rs +0 -16
  101. data/vendor/kreuzberg/src/extractors/text.rs +7 -30
  102. data/vendor/kreuzberg/src/extractors/xml.rs +8 -27
  103. data/vendor/kreuzberg/src/keywords/processor.rs +1 -9
  104. data/vendor/kreuzberg/src/keywords/rake.rs +1 -0
  105. data/vendor/kreuzberg/src/language_detection/mod.rs +51 -94
  106. data/vendor/kreuzberg/src/lib.rs +5 -17
  107. data/vendor/kreuzberg/src/mcp/mod.rs +1 -4
  108. data/vendor/kreuzberg/src/mcp/server.rs +21 -145
  109. data/vendor/kreuzberg/src/ocr/mod.rs +0 -2
  110. data/vendor/kreuzberg/src/ocr/processor.rs +8 -19
  111. data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +0 -2
  112. data/vendor/kreuzberg/src/pdf/error.rs +1 -93
  113. data/vendor/kreuzberg/src/pdf/metadata.rs +100 -263
  114. data/vendor/kreuzberg/src/pdf/mod.rs +2 -33
  115. data/vendor/kreuzberg/src/pdf/rendering.rs +12 -12
  116. data/vendor/kreuzberg/src/pdf/table.rs +64 -61
  117. data/vendor/kreuzberg/src/pdf/text.rs +24 -416
  118. data/vendor/kreuzberg/src/plugins/extractor.rs +8 -40
  119. data/vendor/kreuzberg/src/plugins/mod.rs +0 -3
  120. data/vendor/kreuzberg/src/plugins/ocr.rs +14 -22
  121. data/vendor/kreuzberg/src/plugins/processor.rs +1 -10
  122. data/vendor/kreuzberg/src/plugins/registry.rs +0 -15
  123. data/vendor/kreuzberg/src/plugins/validator.rs +8 -20
  124. data/vendor/kreuzberg/src/stopwords/mod.rs +2 -2
  125. data/vendor/kreuzberg/src/text/mod.rs +0 -8
  126. data/vendor/kreuzberg/src/text/quality.rs +15 -28
  127. data/vendor/kreuzberg/src/text/string_utils.rs +10 -22
  128. data/vendor/kreuzberg/src/text/token_reduction/core.rs +50 -86
  129. data/vendor/kreuzberg/src/text/token_reduction/filters.rs +16 -37
  130. data/vendor/kreuzberg/src/text/token_reduction/simd_text.rs +1 -2
  131. data/vendor/kreuzberg/src/types.rs +67 -907
  132. data/vendor/kreuzberg/src/utils/mod.rs +0 -14
  133. data/vendor/kreuzberg/src/utils/quality.rs +3 -12
  134. data/vendor/kreuzberg/tests/api_tests.rs +0 -506
  135. data/vendor/kreuzberg/tests/archive_integration.rs +0 -2
  136. data/vendor/kreuzberg/tests/batch_orchestration.rs +12 -57
  137. data/vendor/kreuzberg/tests/batch_processing.rs +8 -32
  138. data/vendor/kreuzberg/tests/chunking_offset_demo.rs +92 -0
  139. data/vendor/kreuzberg/tests/concurrency_stress.rs +8 -40
  140. data/vendor/kreuzberg/tests/config_features.rs +1 -33
  141. data/vendor/kreuzberg/tests/config_loading_tests.rs +39 -16
  142. data/vendor/kreuzberg/tests/core_integration.rs +9 -35
  143. data/vendor/kreuzberg/tests/csv_integration.rs +81 -71
  144. data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +25 -23
  145. data/vendor/kreuzberg/tests/email_integration.rs +1 -3
  146. data/vendor/kreuzberg/tests/error_handling.rs +34 -43
  147. data/vendor/kreuzberg/tests/format_integration.rs +1 -7
  148. data/vendor/kreuzberg/tests/helpers/mod.rs +0 -60
  149. data/vendor/kreuzberg/tests/image_integration.rs +0 -2
  150. data/vendor/kreuzberg/tests/mime_detection.rs +16 -17
  151. data/vendor/kreuzberg/tests/ocr_configuration.rs +0 -4
  152. data/vendor/kreuzberg/tests/ocr_errors.rs +0 -22
  153. data/vendor/kreuzberg/tests/ocr_quality.rs +0 -2
  154. data/vendor/kreuzberg/tests/pandoc_integration.rs +503 -0
  155. data/vendor/kreuzberg/tests/pdf_integration.rs +0 -2
  156. data/vendor/kreuzberg/tests/pipeline_integration.rs +2 -36
  157. data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +0 -5
  158. data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs +1 -17
  159. data/vendor/kreuzberg/tests/plugin_system.rs +0 -6
  160. data/vendor/kreuzberg/tests/registry_integration_tests.rs +22 -2
  161. data/vendor/kreuzberg/tests/security_validation.rs +1 -13
  162. data/vendor/kreuzberg/tests/test_fastembed.rs +23 -45
  163. metadata +25 -171
  164. data/.rubocop.yml +0 -543
  165. data/ext/kreuzberg_rb/native/.cargo/config.toml +0 -23
  166. data/ext/kreuzberg_rb/native/Cargo.lock +0 -7619
  167. data/lib/kreuzberg/error_context.rb +0 -136
  168. data/lib/kreuzberg/types.rb +0 -170
  169. data/lib/libpdfium.so +0 -0
  170. data/spec/binding/async_operations_spec.rb +0 -473
  171. data/spec/binding/batch_operations_spec.rb +0 -595
  172. data/spec/binding/batch_spec.rb +0 -359
  173. data/spec/binding/config_result_spec.rb +0 -377
  174. data/spec/binding/embeddings_spec.rb +0 -816
  175. data/spec/binding/error_recovery_spec.rb +0 -488
  176. data/spec/binding/font_config_spec.rb +0 -220
  177. data/spec/binding/images_spec.rb +0 -738
  178. data/spec/binding/keywords_extraction_spec.rb +0 -600
  179. data/spec/binding/metadata_types_spec.rb +0 -1228
  180. data/spec/binding/pages_extraction_spec.rb +0 -471
  181. data/spec/binding/tables_spec.rb +0 -641
  182. data/spec/unit/config/chunking_config_spec.rb +0 -213
  183. data/spec/unit/config/embedding_config_spec.rb +0 -343
  184. data/spec/unit/config/extraction_config_spec.rb +0 -438
  185. data/spec/unit/config/font_config_spec.rb +0 -285
  186. data/spec/unit/config/hierarchy_config_spec.rb +0 -314
  187. data/spec/unit/config/image_extraction_config_spec.rb +0 -209
  188. data/spec/unit/config/image_preprocessing_config_spec.rb +0 -249
  189. data/spec/unit/config/keyword_config_spec.rb +0 -229
  190. data/spec/unit/config/language_detection_config_spec.rb +0 -258
  191. data/spec/unit/config/ocr_config_spec.rb +0 -171
  192. data/spec/unit/config/page_config_spec.rb +0 -221
  193. data/spec/unit/config/pdf_config_spec.rb +0 -267
  194. data/spec/unit/config/postprocessor_config_spec.rb +0 -290
  195. data/spec/unit/config/tesseract_config_spec.rb +0 -181
  196. data/spec/unit/config/token_reduction_config_spec.rb +0 -251
  197. data/test/metadata_types_test.rb +0 -959
  198. data/vendor/Cargo.toml +0 -61
  199. data/vendor/kreuzberg/examples/bench_fixes.rs +0 -71
  200. data/vendor/kreuzberg/examples/test_pdfium_fork.rs +0 -62
  201. data/vendor/kreuzberg/src/chunking/processor.rs +0 -219
  202. data/vendor/kreuzberg/src/core/batch_optimizations.rs +0 -385
  203. data/vendor/kreuzberg/src/core/config_validation.rs +0 -949
  204. data/vendor/kreuzberg/src/core/formats.rs +0 -235
  205. data/vendor/kreuzberg/src/core/server_config.rs +0 -1220
  206. data/vendor/kreuzberg/src/extraction/capacity.rs +0 -263
  207. data/vendor/kreuzberg/src/extraction/markdown.rs +0 -216
  208. data/vendor/kreuzberg/src/extraction/office_metadata/odt_properties.rs +0 -284
  209. data/vendor/kreuzberg/src/extractors/bibtex.rs +0 -470
  210. data/vendor/kreuzberg/src/extractors/docbook.rs +0 -504
  211. data/vendor/kreuzberg/src/extractors/epub.rs +0 -696
  212. data/vendor/kreuzberg/src/extractors/fictionbook.rs +0 -492
  213. data/vendor/kreuzberg/src/extractors/jats.rs +0 -1054
  214. data/vendor/kreuzberg/src/extractors/jupyter.rs +0 -368
  215. data/vendor/kreuzberg/src/extractors/latex.rs +0 -653
  216. data/vendor/kreuzberg/src/extractors/markdown.rs +0 -701
  217. data/vendor/kreuzberg/src/extractors/odt.rs +0 -628
  218. data/vendor/kreuzberg/src/extractors/opml.rs +0 -635
  219. data/vendor/kreuzberg/src/extractors/orgmode.rs +0 -529
  220. data/vendor/kreuzberg/src/extractors/rst.rs +0 -577
  221. data/vendor/kreuzberg/src/extractors/rtf.rs +0 -809
  222. data/vendor/kreuzberg/src/extractors/security.rs +0 -484
  223. data/vendor/kreuzberg/src/extractors/security_tests.rs +0 -367
  224. data/vendor/kreuzberg/src/extractors/typst.rs +0 -651
  225. data/vendor/kreuzberg/src/language_detection/processor.rs +0 -218
  226. data/vendor/kreuzberg/src/ocr/language_registry.rs +0 -520
  227. data/vendor/kreuzberg/src/panic_context.rs +0 -154
  228. data/vendor/kreuzberg/src/pdf/bindings.rs +0 -306
  229. data/vendor/kreuzberg/src/pdf/bundled.rs +0 -408
  230. data/vendor/kreuzberg/src/pdf/fonts.rs +0 -358
  231. data/vendor/kreuzberg/src/pdf/hierarchy.rs +0 -903
  232. data/vendor/kreuzberg/src/text/quality_processor.rs +0 -231
  233. data/vendor/kreuzberg/src/text/utf8_validation.rs +0 -193
  234. data/vendor/kreuzberg/src/utils/pool.rs +0 -503
  235. data/vendor/kreuzberg/src/utils/pool_sizing.rs +0 -364
  236. data/vendor/kreuzberg/src/utils/string_pool.rs +0 -761
  237. data/vendor/kreuzberg/tests/api_embed.rs +0 -360
  238. data/vendor/kreuzberg/tests/api_extract_multipart.rs +0 -52
  239. data/vendor/kreuzberg/tests/api_large_pdf_extraction.rs +0 -471
  240. data/vendor/kreuzberg/tests/api_large_pdf_extraction_diagnostics.rs +0 -289
  241. data/vendor/kreuzberg/tests/batch_pooling_benchmark.rs +0 -154
  242. data/vendor/kreuzberg/tests/bibtex_parity_test.rs +0 -421
  243. data/vendor/kreuzberg/tests/config_integration_test.rs +0 -753
  244. data/vendor/kreuzberg/tests/data/hierarchy_ground_truth.json +0 -294
  245. data/vendor/kreuzberg/tests/docbook_extractor_tests.rs +0 -500
  246. data/vendor/kreuzberg/tests/docx_vs_pandoc_comparison.rs +0 -370
  247. data/vendor/kreuzberg/tests/epub_native_extractor_tests.rs +0 -275
  248. data/vendor/kreuzberg/tests/fictionbook_extractor_tests.rs +0 -228
  249. data/vendor/kreuzberg/tests/html_table_test.rs +0 -551
  250. data/vendor/kreuzberg/tests/instrumentation_test.rs +0 -139
  251. data/vendor/kreuzberg/tests/jats_extractor_tests.rs +0 -639
  252. data/vendor/kreuzberg/tests/jupyter_extractor_tests.rs +0 -704
  253. data/vendor/kreuzberg/tests/latex_extractor_tests.rs +0 -496
  254. data/vendor/kreuzberg/tests/markdown_extractor_tests.rs +0 -490
  255. data/vendor/kreuzberg/tests/ocr_language_registry.rs +0 -191
  256. data/vendor/kreuzberg/tests/odt_extractor_tests.rs +0 -674
  257. data/vendor/kreuzberg/tests/opml_extractor_tests.rs +0 -616
  258. data/vendor/kreuzberg/tests/orgmode_extractor_tests.rs +0 -822
  259. data/vendor/kreuzberg/tests/page_markers.rs +0 -297
  260. data/vendor/kreuzberg/tests/pdf_hierarchy_detection.rs +0 -301
  261. data/vendor/kreuzberg/tests/pdf_hierarchy_quality.rs +0 -589
  262. data/vendor/kreuzberg/tests/pdf_ocr_triggering.rs +0 -301
  263. data/vendor/kreuzberg/tests/pdf_text_merging.rs +0 -475
  264. data/vendor/kreuzberg/tests/pdfium_linking.rs +0 -340
  265. data/vendor/kreuzberg/tests/rst_extractor_tests.rs +0 -694
  266. data/vendor/kreuzberg/tests/rtf_extractor_tests.rs +0 -775
  267. data/vendor/kreuzberg/tests/typst_behavioral_tests.rs +0 -1260
  268. data/vendor/kreuzberg/tests/typst_extractor_tests.rs +0 -648
  269. data/vendor/kreuzberg-ffi/Cargo.toml +0 -67
  270. data/vendor/kreuzberg-ffi/README.md +0 -851
  271. data/vendor/kreuzberg-ffi/benches/result_view_benchmark.rs +0 -227
  272. data/vendor/kreuzberg-ffi/build.rs +0 -168
  273. data/vendor/kreuzberg-ffi/cbindgen.toml +0 -37
  274. data/vendor/kreuzberg-ffi/kreuzberg-ffi.pc.in +0 -12
  275. data/vendor/kreuzberg-ffi/kreuzberg.h +0 -3012
  276. data/vendor/kreuzberg-ffi/src/batch_streaming.rs +0 -588
  277. data/vendor/kreuzberg-ffi/src/config.rs +0 -1341
  278. data/vendor/kreuzberg-ffi/src/error.rs +0 -901
  279. data/vendor/kreuzberg-ffi/src/extraction.rs +0 -555
  280. data/vendor/kreuzberg-ffi/src/helpers.rs +0 -879
  281. data/vendor/kreuzberg-ffi/src/lib.rs +0 -977
  282. data/vendor/kreuzberg-ffi/src/memory.rs +0 -493
  283. data/vendor/kreuzberg-ffi/src/mime.rs +0 -329
  284. data/vendor/kreuzberg-ffi/src/panic_shield.rs +0 -265
  285. data/vendor/kreuzberg-ffi/src/plugins/document_extractor.rs +0 -442
  286. data/vendor/kreuzberg-ffi/src/plugins/mod.rs +0 -14
  287. data/vendor/kreuzberg-ffi/src/plugins/ocr_backend.rs +0 -628
  288. data/vendor/kreuzberg-ffi/src/plugins/post_processor.rs +0 -438
  289. data/vendor/kreuzberg-ffi/src/plugins/validator.rs +0 -329
  290. data/vendor/kreuzberg-ffi/src/result.rs +0 -510
  291. data/vendor/kreuzberg-ffi/src/result_pool.rs +0 -639
  292. data/vendor/kreuzberg-ffi/src/result_view.rs +0 -773
  293. data/vendor/kreuzberg-ffi/src/string_intern.rs +0 -568
  294. data/vendor/kreuzberg-ffi/src/types.rs +0 -363
  295. data/vendor/kreuzberg-ffi/src/util.rs +0 -210
  296. data/vendor/kreuzberg-ffi/src/validation.rs +0 -848
  297. data/vendor/kreuzberg-ffi/tests.disabled/README.md +0 -48
  298. data/vendor/kreuzberg-ffi/tests.disabled/config_loading_tests.rs +0 -299
  299. data/vendor/kreuzberg-ffi/tests.disabled/config_tests.rs +0 -346
  300. data/vendor/kreuzberg-ffi/tests.disabled/extractor_tests.rs +0 -232
  301. data/vendor/kreuzberg-ffi/tests.disabled/plugin_registration_tests.rs +0 -470
  302. data/vendor/kreuzberg-tesseract/.commitlintrc.json +0 -13
  303. data/vendor/kreuzberg-tesseract/.crate-ignore +0 -2
  304. data/vendor/kreuzberg-tesseract/Cargo.lock +0 -2933
  305. data/vendor/kreuzberg-tesseract/Cargo.toml +0 -57
  306. data/vendor/kreuzberg-tesseract/LICENSE +0 -22
  307. data/vendor/kreuzberg-tesseract/README.md +0 -399
  308. data/vendor/kreuzberg-tesseract/build.rs +0 -1127
  309. data/vendor/kreuzberg-tesseract/patches/README.md +0 -71
  310. data/vendor/kreuzberg-tesseract/patches/tesseract.diff +0 -199
  311. data/vendor/kreuzberg-tesseract/src/api.rs +0 -1371
  312. data/vendor/kreuzberg-tesseract/src/choice_iterator.rs +0 -77
  313. data/vendor/kreuzberg-tesseract/src/enums.rs +0 -297
  314. data/vendor/kreuzberg-tesseract/src/error.rs +0 -81
  315. data/vendor/kreuzberg-tesseract/src/lib.rs +0 -145
  316. data/vendor/kreuzberg-tesseract/src/monitor.rs +0 -57
  317. data/vendor/kreuzberg-tesseract/src/mutable_iterator.rs +0 -197
  318. data/vendor/kreuzberg-tesseract/src/page_iterator.rs +0 -253
  319. data/vendor/kreuzberg-tesseract/src/result_iterator.rs +0 -286
  320. data/vendor/kreuzberg-tesseract/src/result_renderer.rs +0 -183
  321. data/vendor/kreuzberg-tesseract/tests/integration_test.rs +0 -211
@@ -1,77 +0,0 @@
1
- use crate::api::TessDeleteText;
2
- use crate::error::{Result, TesseractError};
3
- use std::ffi::CStr;
4
- use std::os::raw::{c_char, c_float, c_int, c_void};
5
- use std::sync::{Arc, Mutex};
6
-
7
- pub struct ChoiceIterator {
8
- handle: Arc<Mutex<*mut c_void>>,
9
- }
10
-
11
- unsafe impl Send for ChoiceIterator {}
12
- unsafe impl Sync for ChoiceIterator {}
13
-
14
- impl ChoiceIterator {
15
- /// Creates a new instance of the ChoiceIterator.
16
- ///
17
- /// # Arguments
18
- ///
19
- /// * `handle` - Pointer to the ChoiceIterator.
20
- pub fn new(handle: *mut c_void) -> Self {
21
- ChoiceIterator {
22
- handle: Arc::new(Mutex::new(handle)),
23
- }
24
- }
25
-
26
- /// Gets the next choice.
27
- ///
28
- /// # Returns
29
- ///
30
- /// Returns `true` if the next choice is successful, otherwise returns `false`.
31
- pub fn next(&self) -> Result<bool> {
32
- let handle = self.handle.lock().map_err(|_| TesseractError::MutexLockError)?;
33
- Ok(unsafe { TessChoiceIteratorNext(*handle) != 0 })
34
- }
35
-
36
- /// Gets the UTF-8 text for the current choice.
37
- ///
38
- /// # Returns
39
- ///
40
- /// Returns the UTF-8 text as a `String` if successful, otherwise returns an error.
41
- pub fn get_utf8_text(&self) -> Result<String> {
42
- let handle = self.handle.lock().map_err(|_| TesseractError::MutexLockError)?;
43
- let text_ptr = unsafe { TessChoiceIteratorGetUTF8Text(*handle) };
44
- if text_ptr.is_null() {
45
- return Err(TesseractError::NullPointerError);
46
- }
47
- let c_str = unsafe { CStr::from_ptr(text_ptr) };
48
- let result = c_str.to_str()?.to_owned();
49
- unsafe { TessDeleteText(text_ptr) };
50
- Ok(result)
51
- }
52
-
53
- /// Gets the confidence of the current choice.
54
- ///
55
- /// # Returns
56
- ///
57
- /// Returns the confidence as a `f32`.
58
- pub fn confidence(&self) -> Result<f32> {
59
- let handle = self.handle.lock().map_err(|_| TesseractError::MutexLockError)?;
60
- Ok(unsafe { TessChoiceIteratorConfidence(*handle) })
61
- }
62
- }
63
-
64
- impl Drop for ChoiceIterator {
65
- fn drop(&mut self) {
66
- if let Ok(handle) = self.handle.lock() {
67
- unsafe { TessChoiceIteratorDelete(*handle) };
68
- }
69
- }
70
- }
71
-
72
- unsafe extern "C" {
73
- fn TessChoiceIteratorDelete(handle: *mut c_void);
74
- fn TessChoiceIteratorNext(handle: *mut c_void) -> c_int;
75
- fn TessChoiceIteratorGetUTF8Text(handle: *mut c_void) -> *mut c_char;
76
- fn TessChoiceIteratorConfidence(handle: *mut c_void) -> c_float;
77
- }
@@ -1,297 +0,0 @@
1
- #[repr(C)]
2
- #[derive(Debug, Clone, Copy, PartialEq)]
3
- #[allow(non_camel_case_types)]
4
- pub enum TessPageSegMode {
5
- PSM_OSD_ONLY = 0,
6
- PSM_AUTO_OSD = 1,
7
- PSM_AUTO_ONLY = 2,
8
- PSM_AUTO = 3,
9
- PSM_SINGLE_COLUMN = 4,
10
- PSM_SINGLE_BLOCK_VERT_TEXT = 5,
11
- PSM_SINGLE_BLOCK = 6,
12
- PSM_SINGLE_LINE = 7,
13
- PSM_SINGLE_WORD = 8,
14
- PSM_CIRCLE_WORD = 9,
15
- PSM_SINGLE_CHAR = 10,
16
- PSM_SPARSE_TEXT = 11,
17
- PSM_SPARSE_TEXT_OSD = 12,
18
- PSM_RAW_LINE = 13,
19
- PSM_COUNT = 14,
20
- }
21
-
22
- impl TessPageSegMode {
23
- pub fn from_int(value: i32) -> Self {
24
- match value {
25
- 0 => TessPageSegMode::PSM_OSD_ONLY,
26
- 1 => TessPageSegMode::PSM_AUTO_OSD,
27
- 2 => TessPageSegMode::PSM_AUTO_ONLY,
28
- 3 => TessPageSegMode::PSM_AUTO,
29
- 4 => TessPageSegMode::PSM_SINGLE_COLUMN,
30
- 5 => TessPageSegMode::PSM_SINGLE_BLOCK_VERT_TEXT,
31
- 6 => TessPageSegMode::PSM_SINGLE_BLOCK,
32
- 7 => TessPageSegMode::PSM_SINGLE_LINE,
33
- 8 => TessPageSegMode::PSM_SINGLE_WORD,
34
- 9 => TessPageSegMode::PSM_CIRCLE_WORD,
35
- 10 => TessPageSegMode::PSM_SINGLE_CHAR,
36
- 11 => TessPageSegMode::PSM_SPARSE_TEXT,
37
- 12 => TessPageSegMode::PSM_SPARSE_TEXT_OSD,
38
- 13 => TessPageSegMode::PSM_RAW_LINE,
39
- 14 => TessPageSegMode::PSM_COUNT,
40
- _ => TessPageSegMode::PSM_AUTO,
41
- }
42
- }
43
- }
44
-
45
- #[repr(C)]
46
- #[derive(Debug, Clone, Copy, PartialEq)]
47
- #[allow(non_camel_case_types)]
48
- pub enum TessPageIteratorLevel {
49
- RIL_BLOCK = 0,
50
- RIL_PARA = 1,
51
- RIL_TEXTLINE = 2,
52
- RIL_WORD = 3,
53
- RIL_SYMBOL = 4,
54
- }
55
-
56
- impl TessPageIteratorLevel {
57
- pub fn from_int(value: i32) -> Self {
58
- match value {
59
- 0 => TessPageIteratorLevel::RIL_BLOCK,
60
- 1 => TessPageIteratorLevel::RIL_PARA,
61
- 2 => TessPageIteratorLevel::RIL_TEXTLINE,
62
- 3 => TessPageIteratorLevel::RIL_WORD,
63
- 4 => TessPageIteratorLevel::RIL_SYMBOL,
64
- _ => TessPageIteratorLevel::RIL_BLOCK,
65
- }
66
- }
67
- }
68
-
69
- #[repr(C)]
70
- #[derive(Debug, Clone, Copy, PartialEq)]
71
- #[allow(non_camel_case_types)]
72
- pub enum TessPolyBlockType {
73
- PT_UNKNOWN = 0,
74
- PT_FLOWING_TEXT = 1,
75
- PT_HEADING_TEXT = 2,
76
- PT_PULLOUT_TEXT = 3,
77
- PT_EQUATION = 4,
78
- PT_INLINE_EQUATION = 5,
79
- PT_TABLE = 6,
80
- PT_VERTICAL_TEXT = 7,
81
- PT_CAPTION_TEXT = 8,
82
- PT_FLOWING_IMAGE = 9,
83
- PT_HEADING_IMAGE = 10,
84
- PT_PULLOUT_IMAGE = 11,
85
- PT_HORZ_LINE = 12,
86
- PT_VERT_LINE = 13,
87
- PT_NOISE = 14,
88
- PT_COUNT = 15,
89
- }
90
-
91
- impl TessPolyBlockType {
92
- pub fn from_int(value: i32) -> Self {
93
- match value {
94
- 0 => TessPolyBlockType::PT_UNKNOWN,
95
- 1 => TessPolyBlockType::PT_FLOWING_TEXT,
96
- 2 => TessPolyBlockType::PT_HEADING_TEXT,
97
- 3 => TessPolyBlockType::PT_PULLOUT_TEXT,
98
- 4 => TessPolyBlockType::PT_EQUATION,
99
- 5 => TessPolyBlockType::PT_INLINE_EQUATION,
100
- 6 => TessPolyBlockType::PT_TABLE,
101
- 7 => TessPolyBlockType::PT_VERTICAL_TEXT,
102
- 8 => TessPolyBlockType::PT_CAPTION_TEXT,
103
- 9 => TessPolyBlockType::PT_FLOWING_IMAGE,
104
- 10 => TessPolyBlockType::PT_HEADING_IMAGE,
105
- 11 => TessPolyBlockType::PT_PULLOUT_IMAGE,
106
- 12 => TessPolyBlockType::PT_HORZ_LINE,
107
- 13 => TessPolyBlockType::PT_VERT_LINE,
108
- 14 => TessPolyBlockType::PT_NOISE,
109
- 15 => TessPolyBlockType::PT_COUNT,
110
- _ => TessPolyBlockType::PT_UNKNOWN,
111
- }
112
- }
113
- }
114
-
115
- #[repr(C)]
116
- #[derive(Debug, Clone, Copy, PartialEq)]
117
- #[allow(non_camel_case_types)]
118
- pub enum TessOrientation {
119
- ORIENTATION_PAGE_UP = 0,
120
- ORIENTATION_PAGE_RIGHT = 1,
121
- ORIENTATION_PAGE_DOWN = 2,
122
- ORIENTATION_PAGE_LEFT = 3,
123
- }
124
-
125
- impl TessOrientation {
126
- pub fn from_int(value: i32) -> Self {
127
- match value {
128
- 0 => TessOrientation::ORIENTATION_PAGE_UP,
129
- 1 => TessOrientation::ORIENTATION_PAGE_RIGHT,
130
- 2 => TessOrientation::ORIENTATION_PAGE_DOWN,
131
- 3 => TessOrientation::ORIENTATION_PAGE_LEFT,
132
- _ => TessOrientation::ORIENTATION_PAGE_UP,
133
- }
134
- }
135
- }
136
-
137
- #[repr(C)]
138
- #[derive(Debug, Clone, Copy, PartialEq)]
139
- #[allow(non_camel_case_types)]
140
- pub enum TessParagraphJustification {
141
- JUSTIFICATION_UNKNOWN = 0,
142
- JUSTIFICATION_LEFT = 1,
143
- JUSTIFICATION_CENTER = 2,
144
- JUSTIFICATION_RIGHT = 3,
145
- }
146
-
147
- impl TessParagraphJustification {
148
- pub fn from_int(value: i32) -> Self {
149
- match value {
150
- 0 => TessParagraphJustification::JUSTIFICATION_UNKNOWN,
151
- 1 => TessParagraphJustification::JUSTIFICATION_LEFT,
152
- 2 => TessParagraphJustification::JUSTIFICATION_CENTER,
153
- 3 => TessParagraphJustification::JUSTIFICATION_RIGHT,
154
- _ => TessParagraphJustification::JUSTIFICATION_UNKNOWN,
155
- }
156
- }
157
- }
158
-
159
- #[repr(C)]
160
- #[derive(Debug, Clone, Copy, PartialEq)]
161
- #[allow(non_camel_case_types)]
162
- pub enum TessWritingDirection {
163
- WRITING_DIRECTION_LEFT_TO_RIGHT = 0,
164
- WRITING_DIRECTION_RIGHT_TO_LEFT = 1,
165
- WRITING_DIRECTION_TOP_TO_BOTTOM = 2,
166
- }
167
-
168
- impl TessWritingDirection {
169
- pub fn from_int(value: i32) -> Self {
170
- match value {
171
- 0 => TessWritingDirection::WRITING_DIRECTION_LEFT_TO_RIGHT,
172
- 1 => TessWritingDirection::WRITING_DIRECTION_RIGHT_TO_LEFT,
173
- 2 => TessWritingDirection::WRITING_DIRECTION_TOP_TO_BOTTOM,
174
- _ => TessWritingDirection::WRITING_DIRECTION_LEFT_TO_RIGHT,
175
- }
176
- }
177
- }
178
-
179
- #[repr(C)]
180
- #[derive(Debug, Clone, Copy, PartialEq)]
181
- #[allow(non_camel_case_types)]
182
- pub enum TessTextlineOrder {
183
- TEXTLINE_ORDER_LEFT_TO_RIGHT = 0,
184
- TEXTLINE_ORDER_RIGHT_TO_LEFT = 1,
185
- TEXTLINE_ORDER_TOP_TO_BOTTOM = 2,
186
- }
187
-
188
- impl TessTextlineOrder {
189
- pub fn from_int(value: i32) -> Self {
190
- match value {
191
- 0 => TessTextlineOrder::TEXTLINE_ORDER_LEFT_TO_RIGHT,
192
- 1 => TessTextlineOrder::TEXTLINE_ORDER_RIGHT_TO_LEFT,
193
- 2 => TessTextlineOrder::TEXTLINE_ORDER_TOP_TO_BOTTOM,
194
- _ => TessTextlineOrder::TEXTLINE_ORDER_LEFT_TO_RIGHT,
195
- }
196
- }
197
- }
198
-
199
- #[cfg(test)]
200
- mod tests {
201
- use super::*;
202
-
203
- #[test]
204
- fn test_page_seg_mode_from_int() {
205
- assert_eq!(TessPageSegMode::from_int(0), TessPageSegMode::PSM_OSD_ONLY);
206
- assert_eq!(TessPageSegMode::from_int(3), TessPageSegMode::PSM_AUTO);
207
- assert_eq!(TessPageSegMode::from_int(10), TessPageSegMode::PSM_SINGLE_CHAR);
208
- assert_eq!(TessPageSegMode::from_int(999), TessPageSegMode::PSM_AUTO);
209
- }
210
-
211
- #[test]
212
- fn test_page_seg_mode_conversion() {
213
- let mode = TessPageSegMode::PSM_SINGLE_LINE;
214
- assert_eq!(mode as i32, 7);
215
- }
216
-
217
- #[test]
218
- fn test_page_iterator_level_from_int() {
219
- assert_eq!(TessPageIteratorLevel::from_int(0), TessPageIteratorLevel::RIL_BLOCK);
220
- assert_eq!(TessPageIteratorLevel::from_int(3), TessPageIteratorLevel::RIL_WORD);
221
- assert_eq!(TessPageIteratorLevel::from_int(-1), TessPageIteratorLevel::RIL_BLOCK);
222
- }
223
-
224
- #[test]
225
- fn test_poly_block_type_from_int() {
226
- assert_eq!(TessPolyBlockType::from_int(1), TessPolyBlockType::PT_FLOWING_TEXT);
227
- assert_eq!(TessPolyBlockType::from_int(6), TessPolyBlockType::PT_TABLE);
228
- assert_eq!(TessPolyBlockType::from_int(100), TessPolyBlockType::PT_UNKNOWN);
229
- }
230
-
231
- #[test]
232
- fn test_orientation_from_int() {
233
- assert_eq!(TessOrientation::from_int(0), TessOrientation::ORIENTATION_PAGE_UP);
234
- assert_eq!(TessOrientation::from_int(2), TessOrientation::ORIENTATION_PAGE_DOWN);
235
- assert_eq!(TessOrientation::from_int(5), TessOrientation::ORIENTATION_PAGE_UP);
236
- }
237
-
238
- #[test]
239
- fn test_paragraph_justification_from_int() {
240
- assert_eq!(
241
- TessParagraphJustification::from_int(1),
242
- TessParagraphJustification::JUSTIFICATION_LEFT
243
- );
244
- assert_eq!(
245
- TessParagraphJustification::from_int(3),
246
- TessParagraphJustification::JUSTIFICATION_RIGHT
247
- );
248
- assert_eq!(
249
- TessParagraphJustification::from_int(-1),
250
- TessParagraphJustification::JUSTIFICATION_UNKNOWN
251
- );
252
- }
253
-
254
- #[test]
255
- fn test_writing_direction_from_int() {
256
- assert_eq!(
257
- TessWritingDirection::from_int(0),
258
- TessWritingDirection::WRITING_DIRECTION_LEFT_TO_RIGHT
259
- );
260
- assert_eq!(
261
- TessWritingDirection::from_int(1),
262
- TessWritingDirection::WRITING_DIRECTION_RIGHT_TO_LEFT
263
- );
264
- assert_eq!(
265
- TessWritingDirection::from_int(10),
266
- TessWritingDirection::WRITING_DIRECTION_LEFT_TO_RIGHT
267
- );
268
- }
269
-
270
- #[test]
271
- fn test_textline_order_from_int() {
272
- assert_eq!(
273
- TessTextlineOrder::from_int(0),
274
- TessTextlineOrder::TEXTLINE_ORDER_LEFT_TO_RIGHT
275
- );
276
- assert_eq!(
277
- TessTextlineOrder::from_int(2),
278
- TessTextlineOrder::TEXTLINE_ORDER_TOP_TO_BOTTOM
279
- );
280
- assert_eq!(
281
- TessTextlineOrder::from_int(99),
282
- TessTextlineOrder::TEXTLINE_ORDER_LEFT_TO_RIGHT
283
- );
284
- }
285
-
286
- #[test]
287
- fn test_enums_are_copy() {
288
- fn assert_copy<T: Copy>() {}
289
- assert_copy::<TessPageSegMode>();
290
- assert_copy::<TessPageIteratorLevel>();
291
- assert_copy::<TessPolyBlockType>();
292
- assert_copy::<TessOrientation>();
293
- assert_copy::<TessParagraphJustification>();
294
- assert_copy::<TessWritingDirection>();
295
- assert_copy::<TessTextlineOrder>();
296
- }
297
- }
@@ -1,81 +0,0 @@
1
- use std::str::Utf8Error;
2
- use thiserror::Error;
3
-
4
- /// Errors that can occur when using the Tesseract API.
5
- #[derive(Error, Debug)]
6
- pub enum TesseractError {
7
- #[error("Failed to initialize Tesseract")]
8
- InitError,
9
- #[error("Failed to set image")]
10
- SetImageError,
11
- #[error("OCR operation failed")]
12
- OcrError,
13
- #[error("Invalid UTF-8 in Tesseract output")]
14
- Utf8Error(#[from] Utf8Error),
15
- #[error("Failed to lock mutex")]
16
- MutexLockError,
17
- #[error("Failed to set variable")]
18
- SetVariableError,
19
- #[error("Failed to get variable")]
20
- GetVariableError,
21
- #[error("Null pointer error")]
22
- NullPointerError,
23
- #[error("Invalid parameter")]
24
- InvalidParameterError,
25
- #[error("Layout analysis failed")]
26
- AnalyseLayoutError,
27
- #[error("Page processing failed")]
28
- ProcessPagesError,
29
- #[error("I/O error")]
30
- IoError,
31
- #[error("Mutex error")]
32
- MutexError,
33
- #[error("Invalid dimensions")]
34
- InvalidDimensions,
35
- #[error("Invalid bytes per pixel")]
36
- InvalidBytesPerPixel,
37
- #[error("Invalid bytes per line")]
38
- InvalidBytesPerLine,
39
- #[error("Invalid image data")]
40
- InvalidImageData,
41
- #[error("Uninitialized error")]
42
- UninitializedError,
43
- }
44
-
45
- /// Result type for Tesseract operations.
46
- pub type Result<T> = std::result::Result<T, TesseractError>;
47
-
48
- #[cfg(test)]
49
- mod tests {
50
- use super::*;
51
-
52
- #[test]
53
- fn test_error_display() {
54
- let error = TesseractError::InitError;
55
- assert_eq!(error.to_string(), "Failed to initialize Tesseract");
56
-
57
- let error = TesseractError::SetImageError;
58
- assert_eq!(error.to_string(), "Failed to set image");
59
-
60
- let error = TesseractError::OcrError;
61
- assert_eq!(error.to_string(), "OCR operation failed");
62
- }
63
-
64
- #[test]
65
- fn test_utf8_error_conversion() {
66
- let invalid_utf8 = vec![0xFF, 0xFE];
67
- let utf8_error = std::str::from_utf8(&invalid_utf8).unwrap_err();
68
- let tess_error: TesseractError = utf8_error.into();
69
-
70
- match tess_error {
71
- TesseractError::Utf8Error(_) => {}
72
- _ => panic!("Expected Utf8Error variant"),
73
- }
74
- }
75
-
76
- #[test]
77
- fn test_error_is_send_sync() {
78
- fn assert_send_sync<T: Send + Sync>() {}
79
- assert_send_sync::<TesseractError>();
80
- }
81
- }
@@ -1,145 +0,0 @@
1
- #![cfg_attr(not(feature = "build-tesseract"), allow(unused_variables, dead_code))]
2
- #![allow(clippy::arc_with_non_send_sync)]
3
- #![allow(clippy::missing_transmute_annotations)]
4
- #![allow(clippy::type_complexity)]
5
- #![allow(clippy::new_without_default)]
6
- #![allow(clippy::not_unsafe_ptr_arg_deref)]
7
- #![allow(clippy::cmp_null)]
8
-
9
- //! # tesseract-rs
10
- //!
11
- //! `tesseract-rs` provides safe Rust bindings for Tesseract OCR with built-in compilation
12
- //! of Tesseract and Leptonica libraries. This crate aims to make OCR functionality
13
- //! easily accessible in Rust projects while handling the complexity of interfacing
14
- //! with the underlying C++ libraries.
15
- //!
16
- //! ## Usage
17
- //!
18
- //! Here's a basic example of how to use `tesseract-rs`:
19
- //!
20
- //! ```rust
21
- //! use std::path::PathBuf;
22
- //! use std::error::Error;
23
- //! use kreuzberg_tesseract::TesseractAPI;
24
- //!
25
- //! fn get_default_tessdata_dir() -> PathBuf {
26
- //! if cfg!(target_os = "macos") {
27
- //! let home_dir = std::env::var("HOME").expect("HOME environment variable not set");
28
- //! PathBuf::from(home_dir)
29
- //! .join("Library")
30
- //! .join("Application Support")
31
- //! .join("tesseract-rs")
32
- //! .join("tessdata")
33
- //! } else if cfg!(target_os = "linux") {
34
- //! let home_dir = std::env::var("HOME").expect("HOME environment variable not set");
35
- //! PathBuf::from(home_dir)
36
- //! .join(".tesseract-rs")
37
- //! .join("tessdata")
38
- //! } else if cfg!(target_os = "windows") {
39
- //! PathBuf::from(std::env::var("APPDATA").expect("APPDATA environment variable not set"))
40
- //! .join("tesseract-rs")
41
- //! .join("tessdata")
42
- //! } else {
43
- //! panic!("Unsupported operating system");
44
- //! }
45
- //! }
46
- //!
47
- //! fn get_tessdata_dir() -> PathBuf {
48
- //! match std::env::var("TESSDATA_PREFIX") {
49
- //! Ok(dir) => {
50
- //! let path = PathBuf::from(dir);
51
- //! println!("Using TESSDATA_PREFIX directory: {:?}", path);
52
- //! path
53
- //! }
54
- //! Err(_) => {
55
- //! let default_dir = get_default_tessdata_dir();
56
- //! println!(
57
- //! "TESSDATA_PREFIX not set, using default directory: {:?}",
58
- //! default_dir
59
- //! );
60
- //! default_dir
61
- //! }
62
- //! }
63
- //! }
64
- //!
65
- //! fn main() -> Result<(), Box<dyn Error>> {
66
- //! let api = TesseractAPI::new();
67
- //!
68
- //! // Get tessdata directory (uses default location or TESSDATA_PREFIX if set)
69
- //! let tessdata_dir = get_tessdata_dir();
70
- //! api.init(tessdata_dir.to_str().unwrap(), "eng")?;
71
- //!
72
- //! let width = 24;
73
- //! let height = 24;
74
- //! let bytes_per_pixel = 1;
75
- //! let bytes_per_line = width * bytes_per_pixel;
76
- //!
77
- //! // Initialize image data with all white pixels
78
- //! let mut image_data = vec![255u8; width * height];
79
- //!
80
- //! // Draw number 9 with clearer distinction
81
- //! for y in 4..19 {
82
- //! for x in 7..17 {
83
- //! // Top bar
84
- //! if y == 4 && x >= 8 && x <= 15 {
85
- //! image_data[y * width + x] = 0;
86
- //! }
87
- //! // Top curve left side
88
- //! if y >= 4 && y <= 10 && x == 7 {
89
- //! image_data[y * width + x] = 0;
90
- //! }
91
- //! // Top curve right side
92
- //! if y >= 4 && y <= 11 && x == 16 {
93
- //! image_data[y * width + x] = 0;
94
- //! }
95
- //! // Middle bar
96
- //! if y == 11 && x >= 8 && x <= 15 {
97
- //! image_data[y * width + x] = 0;
98
- //! }
99
- //! // Bottom right vertical line
100
- //! if y >= 11 && y <= 18 && x == 16 {
101
- //! image_data[y * width + x] = 0;
102
- //! }
103
- //! // Bottom bar
104
- //! if y == 18 && x >= 8 && x <= 15 {
105
- //! image_data[y * width + x] = 0;
106
- //! }
107
- //! }
108
- //! }
109
- //!
110
- //! // Set the image data
111
- //! api.set_image(&image_data, width.try_into().unwrap(), height.try_into().unwrap(), bytes_per_pixel.try_into().unwrap(), bytes_per_line.try_into().unwrap())?;
112
- //!
113
- //! // Set whitelist for digits only
114
- //! api.set_variable("tessedit_char_whitelist", "0123456789")?;
115
- //!
116
- //! // Set PSM mode to single character
117
- //! api.set_variable("tessedit_pageseg_mode", "10")?;
118
- //!
119
- //! // Get the recognized text
120
- //! let text = api.get_utf8_text()?;
121
- //! println!("Recognized text: {}", text.trim());
122
- //!
123
- //! Ok(())
124
- //! }
125
- //! ```
126
- pub use error::{Result, TesseractError};
127
- mod error;
128
- mod page_iterator;
129
- pub use page_iterator::PageIterator;
130
- mod result_iterator;
131
- pub use result_iterator::ResultIterator;
132
- mod choice_iterator;
133
- pub use choice_iterator::ChoiceIterator;
134
- mod monitor;
135
- pub use monitor::TessMonitor;
136
- mod result_renderer;
137
- pub use result_renderer::TessResultRenderer;
138
- mod mutable_iterator;
139
- pub use mutable_iterator::MutableIterator;
140
- mod enums;
141
- pub use enums::{
142
- TessOrientation, TessPageIteratorLevel, TessPageSegMode, TessPolyBlockType, TessTextlineOrder, TessWritingDirection,
143
- };
144
- mod api;
145
- pub use api::TesseractAPI;
@@ -1,57 +0,0 @@
1
- use std::os::raw::{c_int, c_void};
2
- use std::sync::{Arc, Mutex};
3
-
4
- pub struct TessMonitor {
5
- handle: Arc<Mutex<*mut c_void>>,
6
- }
7
-
8
- unsafe impl Send for TessMonitor {}
9
- unsafe impl Sync for TessMonitor {}
10
-
11
- impl TessMonitor {
12
- /// Creates a new instance of the TessMonitor.
13
- ///
14
- /// # Returns
15
- ///
16
- /// Returns the new instance of the TessMonitor.
17
- pub fn new() -> Self {
18
- let handle = unsafe { TessMonitorCreate() };
19
- TessMonitor {
20
- handle: Arc::new(Mutex::new(handle)),
21
- }
22
- }
23
-
24
- /// Sets the deadline for the monitor.
25
- ///
26
- /// # Arguments
27
- ///
28
- /// * `deadline` - Deadline in milliseconds.
29
- pub fn set_deadline(&self, deadline: i32) {
30
- let handle = self.handle.lock().unwrap();
31
- unsafe { TessMonitorSetDeadlineMSecs(*handle, deadline) };
32
- }
33
-
34
- /// Gets the progress of the monitor.
35
- ///
36
- /// # Returns
37
- ///
38
- /// Returns the progress as an `i32`.
39
- pub fn get_progress(&self) -> i32 {
40
- let handle = self.handle.lock().unwrap();
41
- unsafe { TessMonitorGetProgress(*handle) }
42
- }
43
- }
44
-
45
- impl Drop for TessMonitor {
46
- fn drop(&mut self) {
47
- let handle = self.handle.lock().unwrap();
48
- unsafe { TessMonitorDelete(*handle) };
49
- }
50
- }
51
-
52
- unsafe extern "C" {
53
- pub fn TessMonitorCreate() -> *mut c_void;
54
- pub fn TessMonitorDelete(monitor: *mut c_void);
55
- pub fn TessMonitorSetDeadlineMSecs(monitor: *mut c_void, deadline: c_int);
56
- pub fn TessMonitorGetProgress(monitor: *mut c_void) -> c_int;
57
- }