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
@@ -1,761 +0,0 @@
1
- //! String interning/pooling for frequently used strings.
2
- //!
3
- //! This module provides thread-safe string interning to reduce memory allocations
4
- //! for strings that appear repeatedly across documents (MIME types, language codes, format field names).
5
- //!
6
- //! # Performance
7
- //!
8
- //! String interning provides 0.1-0.3% improvement by:
9
- //! - Deduplicating repeated strings (e.g., "application/pdf" appears 1000s of times)
10
- //! - Reducing allocation overhead for commonly used strings
11
- //! - Enabling pointer comparisons instead of string comparisons
12
- //!
13
- //! # Thread Safety
14
- //!
15
- //! The intern pool uses a `DashMap` for lock-free concurrent access. Multiple threads
16
- //! can insert and lookup strings simultaneously without contention.
17
- //!
18
- //! # Example
19
- //!
20
- //! ```rust,ignore
21
- //! use kreuzberg::utils::string_pool::intern_mime_type;
22
- //!
23
- //! let mime1 = intern_mime_type("application/pdf");
24
- //! let mime2 = intern_mime_type("application/pdf");
25
- //! // Both mime1 and mime2 point to the same interned string
26
- //! assert_eq!(mime1, mime2);
27
- //! ```
28
-
29
- use once_cell::sync::Lazy;
30
- use std::collections::VecDeque;
31
- use std::sync::Arc;
32
- use std::sync::atomic::{AtomicBool, Ordering};
33
-
34
- #[cfg(feature = "pool-metrics")]
35
- use std::sync::atomic::AtomicUsize;
36
-
37
- /// A reference to an interned string stored in an Arc.
38
- ///
39
- /// This wraps an Arc<String> and provides convenient access to the string content.
40
- /// Multiple calls with the same string content will share the same Arc, reducing memory usage.
41
- #[derive(Clone)]
42
- pub struct InternedString(Arc<String>);
43
-
44
- impl InternedString {
45
- /// Get the string content.
46
- pub fn as_str(&self) -> &str {
47
- self.0.as_str()
48
- }
49
- }
50
-
51
- impl AsRef<str> for InternedString {
52
- fn as_ref(&self) -> &str {
53
- self.as_str()
54
- }
55
- }
56
-
57
- impl std::fmt::Display for InternedString {
58
- fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
59
- write!(f, "{}", self.as_str())
60
- }
61
- }
62
-
63
- impl std::fmt::Debug for InternedString {
64
- fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
65
- f.debug_tuple("InternedString").field(&self.as_str()).finish()
66
- }
67
- }
68
-
69
- impl PartialEq for InternedString {
70
- fn eq(&self, other: &Self) -> bool {
71
- Arc::ptr_eq(&self.0, &other.0) || self.as_str() == other.as_str()
72
- }
73
- }
74
-
75
- impl Eq for InternedString {}
76
-
77
- impl std::hash::Hash for InternedString {
78
- fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
79
- self.as_str().hash(state);
80
- }
81
- }
82
-
83
- impl std::ops::Deref for InternedString {
84
- type Target = str;
85
-
86
- fn deref(&self) -> &Self::Target {
87
- self.as_str()
88
- }
89
- }
90
-
91
- /// String pool for MIME types.
92
- ///
93
- /// Lazily initializes with all known MIME types from `kreuzberg::core::mime`.
94
- /// Pre-interning is deferred until first access to reduce startup memory usage.
95
- struct MimeStringPool {
96
- pool: dashmap::DashMap<String, Arc<String>>,
97
- initialized: AtomicBool,
98
- }
99
-
100
- impl MimeStringPool {
101
- /// Create a new MIME string pool.
102
- /// Pre-interning is deferred until first `get_or_intern()` call.
103
- fn new() -> Self {
104
- MimeStringPool {
105
- pool: dashmap::DashMap::new(),
106
- initialized: AtomicBool::new(false),
107
- }
108
- }
109
-
110
- /// Ensure all known MIME types are pre-interned (one-time initialization).
111
- #[inline]
112
- fn ensure_initialized(&self) {
113
- if self.initialized.load(Ordering::Acquire) {
114
- return;
115
- }
116
-
117
- let mime_types = vec![
118
- "text/html",
119
- "text/markdown",
120
- "text/x-markdown",
121
- "text/plain",
122
- "application/pdf",
123
- "application/vnd.openxmlformats-officedocument.presentationml.presentation",
124
- "application/msword",
125
- "application/vnd.ms-powerpoint",
126
- "message/rfc822",
127
- "application/vnd.ms-outlook",
128
- "application/json",
129
- "text/json",
130
- "application/x-yaml",
131
- "text/yaml",
132
- "text/x-yaml",
133
- "application/yaml",
134
- "application/toml",
135
- "text/toml",
136
- "application/xml",
137
- "text/xml",
138
- "image/svg+xml",
139
- "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
140
- "application/vnd.ms-excel",
141
- "application/vnd.ms-excel.sheet.macroEnabled.12",
142
- "application/vnd.ms-excel.sheet.binary.macroEnabled.12",
143
- "application/vnd.ms-excel.addin.macroEnabled.12",
144
- "application/vnd.ms-excel.template.macroEnabled.12",
145
- "application/vnd.oasis.opendocument.spreadsheet",
146
- "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
147
- "application/vnd.oasis.opendocument.text",
148
- "image/bmp",
149
- "image/gif",
150
- "image/jp2",
151
- "image/jpeg",
152
- "image/jpm",
153
- "image/jpx",
154
- "image/mj2",
155
- "image/pjpeg",
156
- "image/png",
157
- "image/tiff",
158
- "image/webp",
159
- "image/x-bmp",
160
- "image/x-ms-bmp",
161
- "image/x-portable-anymap",
162
- "image/x-portable-bitmap",
163
- "image/x-portable-graymap",
164
- "image/x-portable-pixmap",
165
- "image/x-tiff",
166
- "application/csl+json",
167
- "application/docbook+xml",
168
- "application/epub+zip",
169
- "application/rtf",
170
- "application/x-biblatex",
171
- "application/x-bibtex",
172
- "application/x-endnote+xml",
173
- "application/x-fictionbook+xml",
174
- "application/x-ipynb+json",
175
- "application/x-jats+xml",
176
- "application/x-latex",
177
- "application/xml+opml",
178
- "application/x-opml+xml",
179
- "application/x-research-info-systems",
180
- "application/x-typst",
181
- "text/csv",
182
- "text/tab-separated-values",
183
- "text/troff",
184
- "text/x-commonmark",
185
- "text/x-dokuwiki",
186
- "text/x-gfm",
187
- "text/x-markdown-extra",
188
- "text/x-mdoc",
189
- "text/x-multimarkdown",
190
- "text/x-opml",
191
- "text/x-org",
192
- "text/x-pod",
193
- "text/x-rst",
194
- "application/zip",
195
- "application/x-zip-compressed",
196
- "application/x-tar",
197
- "application/tar",
198
- "application/x-gtar",
199
- "application/x-ustar",
200
- "application/gzip",
201
- "application/x-7z-compressed",
202
- ];
203
-
204
- for mime_type in mime_types {
205
- self.pool.insert(mime_type.to_string(), Arc::new(mime_type.to_string()));
206
- }
207
-
208
- let _ = self
209
- .initialized
210
- .compare_exchange(false, true, Ordering::Release, Ordering::Relaxed);
211
- }
212
-
213
- /// Get or intern a MIME type string.
214
- /// Ensures pre-interned MIME types are initialized on first call.
215
- fn get_or_intern(&self, mime_type: &str) -> Arc<String> {
216
- self.ensure_initialized();
217
-
218
- if let Some(entry) = self.pool.get(mime_type) {
219
- Arc::clone(&*entry)
220
- } else {
221
- let arc_string = Arc::new(mime_type.to_string());
222
- self.pool.insert(mime_type.to_string(), Arc::clone(&arc_string));
223
- arc_string
224
- }
225
- }
226
- }
227
-
228
- /// String pool for language codes.
229
- ///
230
- /// Lazily initializes with common ISO 639 language codes.
231
- /// Pre-interning is deferred until first access to reduce startup memory usage.
232
- struct LanguageStringPool {
233
- pool: dashmap::DashMap<String, Arc<String>>,
234
- initialized: AtomicBool,
235
- }
236
-
237
- impl LanguageStringPool {
238
- /// Create a new language string pool.
239
- /// Pre-interning is deferred until first `get_or_intern()` call.
240
- fn new() -> Self {
241
- LanguageStringPool {
242
- pool: dashmap::DashMap::new(),
243
- initialized: AtomicBool::new(false),
244
- }
245
- }
246
-
247
- /// Ensure all known language codes are pre-interned (one-time initialization).
248
- #[inline]
249
- fn ensure_initialized(&self) {
250
- if self.initialized.load(Ordering::Acquire) {
251
- return;
252
- }
253
-
254
- let lang_codes = vec![
255
- "en", "es", "fr", "de", "it", "pt", "ru", "ja", "ko", "zh", "ar", "hi", "th", "tr", "pl", "nl", "sv", "no",
256
- "da", "fi", "cs", "hu", "ro", "el", "he", "fa", "ur", "vi", "id", "ms", "bn", "pa", "te", "mr", "ta", "gu",
257
- "kn", "ml", "or", "uk", "bg", "sr", "hr", "sl", "sk", "et", "lv", "lt", "sq", "mk", "ka", "hy", "eo",
258
- "ast", "ca", "eu", "gl", "cy", "gd", "ga",
259
- ];
260
-
261
- for code in lang_codes {
262
- self.pool.insert(code.to_string(), Arc::new(code.to_string()));
263
- }
264
-
265
- let _ = self
266
- .initialized
267
- .compare_exchange(false, true, Ordering::Release, Ordering::Relaxed);
268
- }
269
-
270
- /// Get or intern a language code string.
271
- /// Ensures pre-interned language codes are initialized on first call.
272
- fn get_or_intern(&self, lang_code: &str) -> Arc<String> {
273
- self.ensure_initialized();
274
-
275
- if let Some(entry) = self.pool.get(lang_code) {
276
- Arc::clone(&*entry)
277
- } else {
278
- let arc_string = Arc::new(lang_code.to_string());
279
- self.pool.insert(lang_code.to_string(), Arc::clone(&arc_string));
280
- arc_string
281
- }
282
- }
283
- }
284
-
285
- /// Configuration for the string buffer pool.
286
- pub struct PoolConfig {
287
- /// Maximum buffers per size bucket
288
- pub max_buffers_per_size: usize,
289
- /// Initial capacity for new buffers
290
- pub initial_capacity: usize,
291
- /// Maximum capacity before discarding
292
- pub max_capacity_before_discard: usize,
293
- }
294
-
295
- impl Default for PoolConfig {
296
- fn default() -> Self {
297
- Self {
298
- max_buffers_per_size: 4,
299
- initial_capacity: 4096,
300
- max_capacity_before_discard: 65536,
301
- }
302
- }
303
- }
304
-
305
- /// Thread-safe reusable string buffer pool.
306
- ///
307
- /// This pool allows allocation and reuse of String buffers to reduce memory allocations
308
- /// during document extraction. Buffers are returned to the pool with cleared contents
309
- /// but preserved capacity, ready for reuse.
310
- ///
311
- /// # Thread Safety
312
- ///
313
- /// The pool uses DashMap for lock-free concurrent access. Multiple threads can
314
- /// acquire and release buffers simultaneously.
315
- ///
316
- /// # Usage
317
- ///
318
- /// ```rust,ignore
319
- /// use kreuzberg::utils::string_pool::STRING_BUFFER_POOL;
320
- ///
321
- /// // Acquire a buffer from the pool
322
- /// let mut buffer = STRING_BUFFER_POOL.acquire();
323
- /// buffer.push_str("some content");
324
- /// // Automatically returned to pool when dropped
325
- /// drop(buffer);
326
- /// ```
327
- pub struct StringBufferPool {
328
- pool: dashmap::DashMap<usize, VecDeque<String>>,
329
- config: PoolConfig,
330
- #[cfg(feature = "pool-metrics")]
331
- acquire_count: AtomicUsize,
332
- #[cfg(feature = "pool-metrics")]
333
- reuse_count: AtomicUsize,
334
- }
335
-
336
- impl StringBufferPool {
337
- /// Create a new string buffer pool with given configuration.
338
- pub fn new(config: PoolConfig) -> Self {
339
- StringBufferPool {
340
- pool: dashmap::DashMap::new(),
341
- config,
342
- #[cfg(feature = "pool-metrics")]
343
- acquire_count: AtomicUsize::new(0),
344
- #[cfg(feature = "pool-metrics")]
345
- reuse_count: AtomicUsize::new(0),
346
- }
347
- }
348
-
349
- /// Find the appropriate bucket size for a given capacity.
350
- fn find_bucket(&self, capacity: usize) -> usize {
351
- if capacity <= 1024 {
352
- 1024
353
- } else if capacity <= 4096 {
354
- 4096
355
- } else if capacity <= 16384 {
356
- 16384
357
- } else if capacity <= 65536 {
358
- 65536
359
- } else {
360
- 262144
361
- }
362
- }
363
-
364
- /// Try to acquire a buffer from a specific bucket, returning it if found.
365
- fn try_acquire_from_bucket(&self, bucket: usize) -> Option<String> {
366
- if let Some(mut entry) = self.pool.get_mut(&bucket) {
367
- entry.pop_front()
368
- } else {
369
- None
370
- }
371
- }
372
-
373
- /// Acquire a string buffer from the pool, or allocate a new one if pool is exhausted.
374
- ///
375
- /// The returned buffer is automatically returned to the pool when dropped.
376
- /// Must be called with the pool wrapped in Arc.
377
- pub fn acquire(self: Arc<Self>) -> PooledString {
378
- #[cfg(feature = "pool-metrics")]
379
- self.acquire_count.fetch_add(1, Ordering::Relaxed);
380
-
381
- let default_bucket = self.config.initial_capacity;
382
- if let Some(buffer) = self.try_acquire_from_bucket(default_bucket) {
383
- #[cfg(feature = "pool-metrics")]
384
- self.reuse_count.fetch_add(1, Ordering::Relaxed);
385
- return PooledString { buffer, pool: self };
386
- }
387
-
388
- for &bucket in &[1024, 16384, 65536] {
389
- if let Some(buffer) = self.try_acquire_from_bucket(bucket) {
390
- #[cfg(feature = "pool-metrics")]
391
- self.reuse_count.fetch_add(1, Ordering::Relaxed);
392
- return PooledString { buffer, pool: self };
393
- }
394
- }
395
-
396
- PooledString {
397
- buffer: String::with_capacity(self.config.initial_capacity),
398
- pool: self,
399
- }
400
- }
401
-
402
- /// Return a buffer to the pool for reuse.
403
- pub fn release(&self, mut buffer: String) {
404
- if buffer.capacity() > self.config.max_capacity_before_discard {
405
- return;
406
- }
407
-
408
- let bucket = self.find_bucket(buffer.capacity());
409
- buffer.clear();
410
-
411
- if let Some(mut queue) = self.pool.get_mut(&bucket) {
412
- if queue.len() < self.config.max_buffers_per_size {
413
- queue.push_back(buffer);
414
- }
415
- } else {
416
- let mut queue = VecDeque::with_capacity(self.config.max_buffers_per_size);
417
- queue.push_back(buffer);
418
- self.pool.insert(bucket, queue);
419
- }
420
- }
421
-
422
- /// Get the current pool size across all buckets.
423
- #[allow(dead_code)]
424
- pub fn size(&self) -> usize {
425
- self.pool.iter().map(|entry| entry.value().len()).sum()
426
- }
427
-
428
- /// Get buffer reuse metrics (only available with `pool-metrics` feature).
429
- #[cfg(feature = "pool-metrics")]
430
- pub fn metrics(&self) -> StringBufferPoolMetrics {
431
- let acquire = self.acquire_count.load(Ordering::Relaxed);
432
- let reuse = self.reuse_count.load(Ordering::Relaxed);
433
- let hit_rate = if acquire == 0 {
434
- 0.0
435
- } else {
436
- (reuse as f64 / acquire as f64) * 100.0
437
- };
438
-
439
- StringBufferPoolMetrics {
440
- total_acquires: acquire,
441
- total_reuses: reuse,
442
- hit_rate,
443
- }
444
- }
445
- }
446
-
447
- /// Metrics for StringBufferPool (only available with `pool-metrics` feature).
448
- #[cfg(feature = "pool-metrics")]
449
- #[derive(Debug, Clone, Copy)]
450
- pub struct StringBufferPoolMetrics {
451
- /// Total number of acquire calls
452
- pub total_acquires: usize,
453
- /// Total number of buffer reuses from pool
454
- pub total_reuses: usize,
455
- /// Hit rate as percentage (0.0-100.0)
456
- pub hit_rate: f64,
457
- }
458
-
459
- /// RAII wrapper for a pooled string buffer.
460
- ///
461
- /// Automatically returns the buffer to the pool when dropped.
462
- pub struct PooledString {
463
- buffer: String,
464
- pool: Arc<StringBufferPool>,
465
- }
466
-
467
- impl PooledString {
468
- /// Get mutable access to the underlying string buffer.
469
- pub fn buffer_mut(&mut self) -> &mut String {
470
- &mut self.buffer
471
- }
472
-
473
- /// Get immutable access to the underlying string buffer.
474
- pub fn as_str(&self) -> &str {
475
- self.buffer.as_str()
476
- }
477
- }
478
-
479
- impl std::ops::Deref for PooledString {
480
- type Target = String;
481
-
482
- fn deref(&self) -> &Self::Target {
483
- &self.buffer
484
- }
485
- }
486
-
487
- impl std::ops::DerefMut for PooledString {
488
- fn deref_mut(&mut self) -> &mut Self::Target {
489
- &mut self.buffer
490
- }
491
- }
492
-
493
- impl Drop for PooledString {
494
- fn drop(&mut self) {
495
- let buffer = std::mem::take(&mut self.buffer);
496
- self.pool.release(buffer);
497
- }
498
- }
499
-
500
- impl std::fmt::Display for PooledString {
501
- fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
502
- write!(f, "{}", self.buffer)
503
- }
504
- }
505
-
506
- impl std::fmt::Debug for PooledString {
507
- fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
508
- f.debug_tuple("PooledString").field(&self.buffer).finish()
509
- }
510
- }
511
-
512
- /// Global MIME type string pool.
513
- static MIME_POOL: Lazy<MimeStringPool> = Lazy::new(MimeStringPool::new);
514
-
515
- /// Global language code string pool.
516
- static LANGUAGE_POOL: Lazy<LanguageStringPool> = Lazy::new(LanguageStringPool::new);
517
-
518
- /// Global string buffer pool for temporary allocations during extraction.
519
- pub static STRING_BUFFER_POOL: Lazy<Arc<StringBufferPool>> =
520
- Lazy::new(|| Arc::new(StringBufferPool::new(PoolConfig::default())));
521
-
522
- /// Get or intern a MIME type string.
523
- ///
524
- /// Returns an `InternedString` that is guaranteed to be deduplicated with any other
525
- /// intern call for the same MIME type. This reduces memory usage and allows
526
- /// fast pointer-based comparisons.
527
- ///
528
- /// # Arguments
529
- ///
530
- /// * `mime_type` - The MIME type string to intern
531
- ///
532
- /// # Returns
533
- ///
534
- /// An `InternedString` pointing to the deduplicated string
535
- ///
536
- /// # Example
537
- ///
538
- /// ```rust,ignore
539
- /// let pdf1 = intern_mime_type("application/pdf");
540
- /// let pdf2 = intern_mime_type("application/pdf");
541
- /// assert_eq!(pdf1, pdf2); // Same pointer
542
- /// ```
543
- pub fn intern_mime_type(mime_type: &str) -> InternedString {
544
- InternedString(MIME_POOL.get_or_intern(mime_type))
545
- }
546
-
547
- /// Get or intern a language code string.
548
- ///
549
- /// Returns an `InternedString` that is guaranteed to be deduplicated with any other
550
- /// intern call for the same language code.
551
- ///
552
- /// # Arguments
553
- ///
554
- /// * `lang_code` - The language code to intern (e.g., "en", "es", "fr")
555
- ///
556
- /// # Returns
557
- ///
558
- /// An `InternedString` pointing to the deduplicated string
559
- ///
560
- /// # Example
561
- ///
562
- /// ```rust,ignore
563
- /// let en1 = intern_language_code("en");
564
- /// let en2 = intern_language_code("en");
565
- /// assert_eq!(en1, en2); // Same pointer
566
- /// ```
567
- pub fn intern_language_code(lang_code: &str) -> InternedString {
568
- InternedString(LANGUAGE_POOL.get_or_intern(lang_code))
569
- }
570
-
571
- /// Acquire a string buffer from the global pool.
572
- ///
573
- /// The returned buffer is automatically returned to the pool when dropped.
574
- ///
575
- /// # Example
576
- ///
577
- /// ```rust,ignore
578
- /// let mut buffer = acquire_string_buffer();
579
- /// buffer.push_str("content");
580
- /// // Automatically returned to pool when buffer goes out of scope
581
- /// ```
582
- pub fn acquire_string_buffer() -> PooledString {
583
- Arc::clone(&*STRING_BUFFER_POOL).acquire()
584
- }
585
-
586
- #[cfg(test)]
587
- mod tests {
588
- use super::*;
589
-
590
- #[test]
591
- fn test_mime_type_deduplication() {
592
- let mime1 = intern_mime_type("application/pdf");
593
- let mime2 = intern_mime_type("application/pdf");
594
-
595
- assert_eq!(mime1, mime2);
596
- assert!(Arc::ptr_eq(&mime1.0, &mime2.0));
597
- }
598
-
599
- #[test]
600
- fn test_language_code_deduplication() {
601
- let en1 = intern_language_code("en");
602
- let en2 = intern_language_code("en");
603
-
604
- assert_eq!(en1, en2);
605
- assert!(Arc::ptr_eq(&en1.0, &en2.0));
606
- }
607
-
608
- #[test]
609
- fn test_interned_string_display() {
610
- let mime = intern_mime_type("text/html");
611
- assert_eq!(format!("{}", mime), "text/html");
612
- }
613
-
614
- #[test]
615
- fn test_interned_string_deref() {
616
- let mime = intern_mime_type("application/json");
617
- assert_eq!(&*mime, "application/json");
618
- assert_eq!(mime.as_ref(), "application/json");
619
- assert_eq!(mime.as_str(), "application/json");
620
- }
621
-
622
- #[test]
623
- fn test_preinterned_mime_types() {
624
- let pdf = intern_mime_type("application/pdf");
625
- assert_eq!(pdf.as_str(), "application/pdf");
626
-
627
- let html = intern_mime_type("text/html");
628
- assert_eq!(html.as_str(), "text/html");
629
-
630
- let json = intern_mime_type("application/json");
631
- assert_eq!(json.as_str(), "application/json");
632
- }
633
-
634
- #[test]
635
- fn test_preinterned_language_codes() {
636
- let en = intern_language_code("en");
637
- assert_eq!(en.as_str(), "en");
638
-
639
- let es = intern_language_code("es");
640
- assert_eq!(es.as_str(), "es");
641
-
642
- let fr = intern_language_code("fr");
643
- assert_eq!(fr.as_str(), "fr");
644
- }
645
-
646
- #[test]
647
- fn test_concurrent_interning() {
648
- use std::sync::Arc;
649
- use std::thread;
650
-
651
- let mime = "application/pdf";
652
- let results = Arc::new(std::sync::Mutex::new(Vec::new()));
653
-
654
- let handles: Vec<_> = (0..10)
655
- .map(|_| {
656
- let results = Arc::clone(&results);
657
- thread::spawn(move || {
658
- let interned = intern_mime_type(mime);
659
- results.lock().unwrap().push(interned);
660
- })
661
- })
662
- .collect();
663
-
664
- for handle in handles {
665
- handle.join().unwrap();
666
- }
667
-
668
- let interned_strings = results.lock().unwrap();
669
- assert_eq!(interned_strings.len(), 10);
670
-
671
- let first_arc = &interned_strings[0].0;
672
- for interned in &*interned_strings {
673
- assert!(
674
- Arc::ptr_eq(&interned.0, first_arc),
675
- "All interned strings should share the same Arc"
676
- );
677
- }
678
- }
679
-
680
- #[test]
681
- fn test_interned_string_hash() {
682
- let mime1 = intern_mime_type("application/pdf");
683
- let mime2 = intern_mime_type("application/pdf");
684
-
685
- use std::collections::HashSet;
686
- let mut set = HashSet::new();
687
- set.insert(mime1);
688
- set.insert(mime2);
689
-
690
- assert_eq!(set.len(), 1);
691
- }
692
-
693
- #[test]
694
- fn test_interned_string_clone() {
695
- let mime1 = intern_mime_type("text/html");
696
- let mime2 = mime1.clone();
697
-
698
- assert_eq!(mime1, mime2);
699
- assert!(Arc::ptr_eq(&mime1.0, &mime2.0));
700
- }
701
-
702
- #[test]
703
- fn test_buffer_pool_acquire_and_release() {
704
- let config = PoolConfig::default();
705
- let pool = Arc::new(StringBufferPool::new(config));
706
-
707
- let mut buffer = pool.clone().acquire();
708
- buffer.push_str("test content");
709
- let capacity = buffer.capacity();
710
-
711
- drop(buffer);
712
-
713
- let buffer2 = pool.clone().acquire();
714
- assert_eq!(buffer2.capacity(), capacity);
715
- assert!(buffer2.is_empty());
716
- }
717
-
718
- #[test]
719
- fn test_buffer_pool_size() {
720
- let config = PoolConfig::default();
721
- let pool = Arc::new(StringBufferPool::new(config));
722
-
723
- assert_eq!(pool.size(), 0);
724
-
725
- let buffer1 = pool.clone().acquire();
726
- drop(buffer1);
727
- assert_eq!(pool.size(), 1);
728
-
729
- let buffer2 = pool.clone().acquire();
730
- drop(buffer2);
731
- assert_eq!(pool.size(), 1);
732
- }
733
-
734
- #[test]
735
- fn test_buffer_pool_global() {
736
- let buffer1 = acquire_string_buffer();
737
- drop(buffer1);
738
-
739
- let buffer2 = acquire_string_buffer();
740
- assert!(buffer2.capacity() >= 4096);
741
- }
742
-
743
- #[test]
744
- fn test_pooled_string_deref() {
745
- let mut buffer = acquire_string_buffer();
746
- buffer.push_str("hello");
747
-
748
- assert_eq!(&*buffer, "hello");
749
- assert_eq!(buffer.as_str(), "hello");
750
- assert!(!buffer.is_empty());
751
- }
752
-
753
- #[test]
754
- fn test_pooled_string_deref_mut() {
755
- let mut buffer = acquire_string_buffer();
756
- buffer.push_str("test");
757
-
758
- buffer.buffer_mut().push_str(" more");
759
- assert_eq!(buffer.as_str(), "test more");
760
- }
761
- }