@llamaindex/liteparse 1.0.0
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.
- package/LICENSE +201 -0
- package/README.md +339 -0
- package/dist/cli/parse.d.ts +4 -0
- package/dist/cli/parse.d.ts.map +1 -0
- package/dist/cli/parse.js +401 -0
- package/dist/cli/parse.js.map +1 -0
- package/dist/src/conversion/convertToPdf.d.ts +47 -0
- package/dist/src/conversion/convertToPdf.d.ts.map +1 -0
- package/dist/src/conversion/convertToPdf.js +337 -0
- package/dist/src/conversion/convertToPdf.js.map +1 -0
- package/dist/src/conversion/convertToPdf.test.d.ts +2 -0
- package/dist/src/conversion/convertToPdf.test.d.ts.map +1 -0
- package/dist/src/conversion/convertToPdf.test.js +208 -0
- package/dist/src/conversion/convertToPdf.test.js.map +1 -0
- package/dist/src/core/config.d.ts +4 -0
- package/dist/src/core/config.d.ts.map +1 -0
- package/dist/src/core/config.js +25 -0
- package/dist/src/core/config.js.map +1 -0
- package/dist/src/core/config.test.d.ts +2 -0
- package/dist/src/core/config.test.d.ts.map +1 -0
- package/dist/src/core/config.test.js +21 -0
- package/dist/src/core/config.test.js.map +1 -0
- package/dist/src/core/parser.d.ts +83 -0
- package/dist/src/core/parser.d.ts.map +1 -0
- package/dist/src/core/parser.js +333 -0
- package/dist/src/core/parser.js.map +1 -0
- package/dist/src/core/parser.test.d.ts +2 -0
- package/dist/src/core/parser.test.d.ts.map +1 -0
- package/dist/src/core/parser.test.js +537 -0
- package/dist/src/core/parser.test.js.map +1 -0
- package/dist/src/core/types.d.ts +287 -0
- package/dist/src/core/types.d.ts.map +1 -0
- package/dist/src/core/types.js +2 -0
- package/dist/src/core/types.js.map +1 -0
- package/dist/src/engines/ocr/http-simple.d.ts +19 -0
- package/dist/src/engines/ocr/http-simple.d.ts.map +1 -0
- package/dist/src/engines/ocr/http-simple.js +63 -0
- package/dist/src/engines/ocr/http-simple.js.map +1 -0
- package/dist/src/engines/ocr/http-simple.test.d.ts +2 -0
- package/dist/src/engines/ocr/http-simple.test.d.ts.map +1 -0
- package/dist/src/engines/ocr/http-simple.test.js +108 -0
- package/dist/src/engines/ocr/http-simple.test.js.map +1 -0
- package/dist/src/engines/ocr/interface.d.ts +15 -0
- package/dist/src/engines/ocr/interface.d.ts.map +1 -0
- package/dist/src/engines/ocr/interface.js +2 -0
- package/dist/src/engines/ocr/interface.js.map +1 -0
- package/dist/src/engines/ocr/tesseract.d.ts +19 -0
- package/dist/src/engines/ocr/tesseract.d.ts.map +1 -0
- package/dist/src/engines/ocr/tesseract.js +112 -0
- package/dist/src/engines/ocr/tesseract.js.map +1 -0
- package/dist/src/engines/ocr/tesseract.test.d.ts +2 -0
- package/dist/src/engines/ocr/tesseract.test.d.ts.map +1 -0
- package/dist/src/engines/ocr/tesseract.test.js +84 -0
- package/dist/src/engines/ocr/tesseract.test.js.map +1 -0
- package/dist/src/engines/pdf/interface.d.ts +79 -0
- package/dist/src/engines/pdf/interface.d.ts.map +1 -0
- package/dist/src/engines/pdf/interface.js +2 -0
- package/dist/src/engines/pdf/interface.js.map +1 -0
- package/dist/src/engines/pdf/pdfium-renderer.d.ts +11 -0
- package/dist/src/engines/pdf/pdfium-renderer.d.ts.map +1 -0
- package/dist/src/engines/pdf/pdfium-renderer.js +64 -0
- package/dist/src/engines/pdf/pdfium-renderer.js.map +1 -0
- package/dist/src/engines/pdf/pdfium-renderer.test.d.ts +2 -0
- package/dist/src/engines/pdf/pdfium-renderer.test.d.ts.map +1 -0
- package/dist/src/engines/pdf/pdfium-renderer.test.js +76 -0
- package/dist/src/engines/pdf/pdfium-renderer.test.js.map +1 -0
- package/dist/src/engines/pdf/pdfjs.d.ts +13 -0
- package/dist/src/engines/pdf/pdfjs.d.ts.map +1 -0
- package/dist/src/engines/pdf/pdfjs.js +538 -0
- package/dist/src/engines/pdf/pdfjs.js.map +1 -0
- package/dist/src/engines/pdf/pdfjs.test.d.ts +2 -0
- package/dist/src/engines/pdf/pdfjs.test.d.ts.map +1 -0
- package/dist/src/engines/pdf/pdfjs.test.js +220 -0
- package/dist/src/engines/pdf/pdfjs.test.js.map +1 -0
- package/dist/src/engines/pdf/pdfjsImporter.d.ts +5 -0
- package/dist/src/engines/pdf/pdfjsImporter.d.ts.map +1 -0
- package/dist/src/engines/pdf/pdfjsImporter.js +9 -0
- package/dist/src/engines/pdf/pdfjsImporter.js.map +1 -0
- package/dist/src/index.d.ts +3 -0
- package/dist/src/index.d.ts.map +1 -0
- package/dist/src/index.js +5 -0
- package/dist/src/index.js.map +1 -0
- package/dist/src/lib.d.ts +17 -0
- package/dist/src/lib.d.ts.map +1 -0
- package/dist/src/lib.js +16 -0
- package/dist/src/lib.js.map +1 -0
- package/dist/src/output/json.d.ts +10 -0
- package/dist/src/output/json.d.ts.map +1 -0
- package/dist/src/output/json.js +31 -0
- package/dist/src/output/json.js.map +1 -0
- package/dist/src/output/json.test.d.ts +2 -0
- package/dist/src/output/json.test.d.ts.map +1 -0
- package/dist/src/output/json.test.js +136 -0
- package/dist/src/output/json.test.js.map +1 -0
- package/dist/src/output/text.d.ts +10 -0
- package/dist/src/output/text.d.ts.map +1 -0
- package/dist/src/output/text.js +17 -0
- package/dist/src/output/text.js.map +1 -0
- package/dist/src/output/text.test.d.ts +2 -0
- package/dist/src/output/text.test.d.ts.map +1 -0
- package/dist/src/output/text.test.js +65 -0
- package/dist/src/output/text.test.js.map +1 -0
- package/dist/src/processing/bbox.d.ts +20 -0
- package/dist/src/processing/bbox.d.ts.map +1 -0
- package/dist/src/processing/bbox.js +258 -0
- package/dist/src/processing/bbox.js.map +1 -0
- package/dist/src/processing/bbox.test.d.ts +2 -0
- package/dist/src/processing/bbox.test.d.ts.map +1 -0
- package/dist/src/processing/bbox.test.js +334 -0
- package/dist/src/processing/bbox.test.js.map +1 -0
- package/dist/src/processing/cleanText.d.ts +6 -0
- package/dist/src/processing/cleanText.d.ts.map +1 -0
- package/dist/src/processing/cleanText.js +73 -0
- package/dist/src/processing/cleanText.js.map +1 -0
- package/dist/src/processing/cleanText.test.d.ts +2 -0
- package/dist/src/processing/cleanText.test.d.ts.map +1 -0
- package/dist/src/processing/cleanText.test.js +46 -0
- package/dist/src/processing/cleanText.test.js.map +1 -0
- package/dist/src/processing/grid.d.ts +7 -0
- package/dist/src/processing/grid.d.ts.map +1 -0
- package/dist/src/processing/grid.js +13 -0
- package/dist/src/processing/grid.js.map +1 -0
- package/dist/src/processing/gridProjection.d.ts +18 -0
- package/dist/src/processing/gridProjection.d.ts.map +1 -0
- package/dist/src/processing/gridProjection.js +1392 -0
- package/dist/src/processing/gridProjection.js.map +1 -0
- package/dist/src/processing/gridProjection.test.d.ts +2 -0
- package/dist/src/processing/gridProjection.test.d.ts.map +1 -0
- package/dist/src/processing/gridProjection.test.js +464 -0
- package/dist/src/processing/gridProjection.test.js.map +1 -0
- package/dist/src/processing/markupUtils.d.ts +7 -0
- package/dist/src/processing/markupUtils.d.ts.map +1 -0
- package/dist/src/processing/markupUtils.js +25 -0
- package/dist/src/processing/markupUtils.js.map +1 -0
- package/dist/src/processing/markupUtils.test.d.ts +2 -0
- package/dist/src/processing/markupUtils.test.d.ts.map +1 -0
- package/dist/src/processing/markupUtils.test.js +26 -0
- package/dist/src/processing/markupUtils.test.js.map +1 -0
- package/dist/src/processing/ocrUtils.d.ts +24 -0
- package/dist/src/processing/ocrUtils.d.ts.map +1 -0
- package/dist/src/processing/ocrUtils.js +79 -0
- package/dist/src/processing/ocrUtils.js.map +1 -0
- package/dist/src/processing/octUtils.test.d.ts +2 -0
- package/dist/src/processing/octUtils.test.d.ts.map +1 -0
- package/dist/src/processing/octUtils.test.js +72 -0
- package/dist/src/processing/octUtils.test.js.map +1 -0
- package/dist/src/processing/textUtils.d.ts +20 -0
- package/dist/src/processing/textUtils.d.ts.map +1 -0
- package/dist/src/processing/textUtils.js +142 -0
- package/dist/src/processing/textUtils.js.map +1 -0
- package/dist/src/processing/textUtils.test.d.ts +2 -0
- package/dist/src/processing/textUtils.test.d.ts.map +1 -0
- package/dist/src/processing/textUtils.test.js +45 -0
- package/dist/src/processing/textUtils.test.js.map +1 -0
- package/dist/src/vendor/pdfjs/LICENSE +177 -0
- package/dist/src/vendor/pdfjs/README.md +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78ms-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/78ms-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/83pv-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90ms-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90ms-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90msp-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90msp-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90pv-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/90pv-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Add-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Add-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Add-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Add-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-0.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-1.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-3.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-4.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-5.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-6.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-CNS1-UCS2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-0.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-1.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-3.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-4.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-5.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-GB1-UCS2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-0.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-1.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-3.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-4.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-5.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-6.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Japan1-UCS2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Korea1-0.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Korea1-1.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Korea1-2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Adobe-Korea1-UCS2.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/B5pc-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/B5pc-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS1-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS1-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS2-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/CNS2-V.bcmap +3 -0
- package/dist/src/vendor/pdfjs/cmaps/ETHK-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/ETHK-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/ETen-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/ETen-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/ETenms-B5-H.bcmap +3 -0
- package/dist/src/vendor/pdfjs/cmaps/ETenms-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Ext-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Ext-RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Ext-RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Ext-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GB-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GB-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GB-H.bcmap +4 -0
- package/dist/src/vendor/pdfjs/cmaps/GB-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBK-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBK-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBK2K-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBK2K-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBKp-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBKp-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBT-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBT-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBT-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBT-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBTpc-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBTpc-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBpc-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/GBpc-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKdla-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKdla-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKdlb-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKdlb-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKgccs-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKgccs-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKm314-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKm314-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKm471-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKm471-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKscs-B5-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/HKscs-B5-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Hankaku.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Hiragana.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-Johab-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-Johab-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCms-UHC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCms-UHC-HW-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCms-UHC-HW-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCms-UHC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCpc-EUC-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/KSCpc-EUC-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Katakana.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/LICENSE +36 -0
- package/dist/src/vendor/pdfjs/cmaps/NWP-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/NWP-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/RKSJ-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/RKSJ-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/Roman.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UCS2-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UCS2-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF16-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF16-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF8-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniCNS-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UCS2-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UCS2-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF16-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF16-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF8-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniGB-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UCS2-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UCS2-HW-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UCS2-HW-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UCS2-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF16-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF16-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF8-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF16-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF16-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF8-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJIS2004-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISPro-UCS2-HW-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISPro-UCS2-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISPro-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISX0213-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISX0213-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISX02132004-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniJISX02132004-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UCS2-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UCS2-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF16-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF16-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF32-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF32-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF8-H.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/UniKS-UTF8-V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/V.bcmap +0 -0
- package/dist/src/vendor/pdfjs/cmaps/WP-Symbol.bcmap +0 -0
- package/dist/src/vendor/pdfjs/pdf.mjs +19481 -0
- package/dist/src/vendor/pdfjs/pdf.mjs.map +1 -0
- package/dist/src/vendor/pdfjs/pdf.sandbox.mjs +210 -0
- package/dist/src/vendor/pdfjs/pdf.sandbox.mjs.map +1 -0
- package/dist/src/vendor/pdfjs/pdf.worker.mjs +56001 -0
- package/dist/src/vendor/pdfjs/pdf.worker.mjs.map +1 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitDingbats.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitFixed.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitFixedBold.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitFixedBoldItalic.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitFixedItalic.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitSerif.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitSerifBold.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitSerifBoldItalic.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitSerifItalic.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/FoxitSymbol.pfb +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LICENSE_FOXIT +27 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LICENSE_LIBERATION +102 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LiberationSans-Bold.ttf +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LiberationSans-BoldItalic.ttf +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LiberationSans-Italic.ttf +0 -0
- package/dist/src/vendor/pdfjs/standard_fonts/LiberationSans-Regular.ttf +0 -0
- package/package.json +89 -0
- package/src/vendor/pdfjs/LICENSE +177 -0
- package/src/vendor/pdfjs/README.md +0 -0
- package/src/vendor/pdfjs/cmaps/78-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78ms-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/78ms-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/83pv-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90ms-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90ms-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90msp-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90msp-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90pv-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/90pv-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Add-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Add-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Add-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Add-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-0.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-1.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-3.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-4.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-5.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-6.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-CNS1-UCS2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-0.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-1.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-3.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-4.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-5.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-GB1-UCS2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-0.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-1.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-3.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-4.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-5.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-6.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Japan1-UCS2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Korea1-0.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Korea1-1.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Korea1-2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Adobe-Korea1-UCS2.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/B5pc-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/B5pc-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS1-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS1-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS2-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/CNS2-V.bcmap +3 -0
- package/src/vendor/pdfjs/cmaps/ETHK-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/ETHK-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/ETen-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/ETen-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/ETenms-B5-H.bcmap +3 -0
- package/src/vendor/pdfjs/cmaps/ETenms-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Ext-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Ext-RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Ext-RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Ext-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GB-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GB-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GB-H.bcmap +4 -0
- package/src/vendor/pdfjs/cmaps/GB-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBK-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBK-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBK2K-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBK2K-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBKp-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBKp-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBT-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBT-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBT-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBT-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBTpc-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBTpc-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBpc-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/GBpc-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKdla-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKdla-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKdlb-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKdlb-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKgccs-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKgccs-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKm314-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKm314-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKm471-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKm471-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKscs-B5-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/HKscs-B5-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Hankaku.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Hiragana.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-Johab-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-Johab-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCms-UHC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCms-UHC-HW-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCms-UHC-HW-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCms-UHC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCpc-EUC-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/KSCpc-EUC-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Katakana.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/LICENSE +36 -0
- package/src/vendor/pdfjs/cmaps/NWP-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/NWP-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/RKSJ-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/RKSJ-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/Roman.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UCS2-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UCS2-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF16-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF16-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF8-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniCNS-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UCS2-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UCS2-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF16-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF16-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF8-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniGB-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UCS2-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UCS2-HW-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UCS2-HW-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UCS2-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF16-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF16-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF8-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF16-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF16-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF8-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJIS2004-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISPro-UCS2-HW-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISPro-UCS2-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISPro-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISX0213-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISX0213-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISX02132004-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniJISX02132004-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UCS2-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UCS2-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF16-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF16-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF32-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF32-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF8-H.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/UniKS-UTF8-V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/V.bcmap +0 -0
- package/src/vendor/pdfjs/cmaps/WP-Symbol.bcmap +0 -0
- package/src/vendor/pdfjs/pdf.mjs +19481 -0
- package/src/vendor/pdfjs/pdf.mjs.map +1 -0
- package/src/vendor/pdfjs/pdf.sandbox.mjs +210 -0
- package/src/vendor/pdfjs/pdf.sandbox.mjs.map +1 -0
- package/src/vendor/pdfjs/pdf.worker.mjs +56001 -0
- package/src/vendor/pdfjs/pdf.worker.mjs.map +1 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitDingbats.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitFixed.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitFixedBold.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitFixedBoldItalic.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitFixedItalic.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitSerif.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitSerifBold.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitSerifBoldItalic.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitSerifItalic.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/FoxitSymbol.pfb +0 -0
- package/src/vendor/pdfjs/standard_fonts/LICENSE_FOXIT +27 -0
- package/src/vendor/pdfjs/standard_fonts/LICENSE_LIBERATION +102 -0
- package/src/vendor/pdfjs/standard_fonts/LiberationSans-Bold.ttf +0 -0
- package/src/vendor/pdfjs/standard_fonts/LiberationSans-BoldItalic.ttf +0 -0
- package/src/vendor/pdfjs/standard_fonts/LiberationSans-Italic.ttf +0 -0
- package/src/vendor/pdfjs/standard_fonts/LiberationSans-Regular.ttf +0 -0
|
@@ -0,0 +1,20 @@
|
|
|
1
|
+
import { TextItem, BoundingBox, ProjectionTextBox, LiteParseConfig } from "../core/types.js";
|
|
2
|
+
import { PageData, Image } from "../engines/pdf/interface.js";
|
|
3
|
+
/**
|
|
4
|
+
* Filters images that should not be OCR'd based on various criteria.
|
|
5
|
+
* Returns the filtered array of images that should be processed.
|
|
6
|
+
*/
|
|
7
|
+
export declare function filterImagesForOCR(images: Image[], page: {
|
|
8
|
+
width: number;
|
|
9
|
+
height: number;
|
|
10
|
+
}): Image[];
|
|
11
|
+
/**
|
|
12
|
+
* Build projection text boxes from page data, including OCR results
|
|
13
|
+
* This is the complete implementation from buildBbox.ts
|
|
14
|
+
*/
|
|
15
|
+
export declare function buildBbox(pageData: PageData, config: LiteParseConfig): ProjectionTextBox[];
|
|
16
|
+
/**
|
|
17
|
+
* Build bounding boxes from text items
|
|
18
|
+
*/
|
|
19
|
+
export declare function buildBoundingBoxes(textItems: TextItem[]): BoundingBox[];
|
|
20
|
+
//# sourceMappingURL=bbox.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bbox.d.ts","sourceRoot":"","sources":["../../../src/processing/bbox.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,QAAQ,EACR,WAAW,EACX,iBAAiB,EAEjB,eAAe,EAChB,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,6BAA6B,CAAC;AAgC9D;;;GAGG;AACH,wBAAgB,kBAAkB,CAChC,MAAM,EAAE,KAAK,EAAE,EACf,IAAI,EAAE;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAA;CAAE,GACtC,KAAK,EAAE,CA2DT;AAsFD;;;GAGG;AACH,wBAAgB,SAAS,CAAC,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,eAAe,GAAG,iBAAiB,EAAE,CAmH1F;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CAAC,SAAS,EAAE,QAAQ,EAAE,GAAG,WAAW,EAAE,CAiBvE"}
|
|
@@ -0,0 +1,258 @@
|
|
|
1
|
+
import { parseImageOcrBlocks } from "./ocrUtils.js";
|
|
2
|
+
import { cleanOcrTableArtifacts } from "./textUtils.js";
|
|
3
|
+
const OCR_CONFIDENCE_THRESHOLD = 0.1;
|
|
4
|
+
/**
|
|
5
|
+
* Minimum overlap ratio (0-1) required to consider an OCR block as duplicate of existing text.
|
|
6
|
+
* An OCR block is filtered out if:
|
|
7
|
+
* - Total overlap with all text items covers more than this ratio of the OCR block area
|
|
8
|
+
* - OR the OCR block covers more than this ratio of any single text item
|
|
9
|
+
*/
|
|
10
|
+
const OCR_OVERLAP_THRESHOLD = 0.5;
|
|
11
|
+
/**
|
|
12
|
+
* Maximum number of embedded images to process for OCR per page.
|
|
13
|
+
* Keeps the largest images when limit is exceeded.
|
|
14
|
+
*/
|
|
15
|
+
const MAX_IMAGES_PER_PAGE = 10;
|
|
16
|
+
/**
|
|
17
|
+
* Minimum image dimensions for OCR processing
|
|
18
|
+
*/
|
|
19
|
+
const MIN_IMAGE_DIMENSION = 12;
|
|
20
|
+
const MIN_IMAGE_AREA = 200;
|
|
21
|
+
/**
|
|
22
|
+
* Minimum rendered image dimensions for OCR processing
|
|
23
|
+
*/
|
|
24
|
+
const MIN_RENDERED_DIMENSION = 6;
|
|
25
|
+
const MIN_RENDERED_AREA = 200;
|
|
26
|
+
/**
|
|
27
|
+
* Filters images that should not be OCR'd based on various criteria.
|
|
28
|
+
* Returns the filtered array of images that should be processed.
|
|
29
|
+
*/
|
|
30
|
+
export function filterImagesForOCR(images, page) {
|
|
31
|
+
// Filter images that start with g_ or pattern_ (generated/pattern images)
|
|
32
|
+
let filtered = images.filter((image) => !image.type?.startsWith("g_") && !image.type?.startsWith("pattern_"));
|
|
33
|
+
// Limit to max images per page, keeping the largest ones
|
|
34
|
+
if (filtered.length > MAX_IMAGES_PER_PAGE) {
|
|
35
|
+
filtered.sort((a, b) => b.width * b.height - a.width * a.height);
|
|
36
|
+
filtered = filtered.slice(0, MAX_IMAGES_PER_PAGE);
|
|
37
|
+
}
|
|
38
|
+
// Apply additional filtering criteria
|
|
39
|
+
filtered = filtered.filter((image) => {
|
|
40
|
+
// Ignore layout extracted images
|
|
41
|
+
if (image.type?.includes("layout_")) {
|
|
42
|
+
return false;
|
|
43
|
+
}
|
|
44
|
+
// Get image coords (use image dimensions if coords not set)
|
|
45
|
+
const coords = image.coords || {
|
|
46
|
+
x: image.x,
|
|
47
|
+
y: image.y,
|
|
48
|
+
w: image.width,
|
|
49
|
+
h: image.height,
|
|
50
|
+
};
|
|
51
|
+
// Skip images that are out of viewport
|
|
52
|
+
if (coords.x + coords.w < 0 || // left of page
|
|
53
|
+
coords.y + coords.h < 0 || // above page
|
|
54
|
+
coords.x > page.width || // right of page
|
|
55
|
+
coords.y > page.height // below page
|
|
56
|
+
) {
|
|
57
|
+
return false;
|
|
58
|
+
}
|
|
59
|
+
// Skip small images (raw dimensions)
|
|
60
|
+
if (image.width < MIN_IMAGE_DIMENSION ||
|
|
61
|
+
image.height < MIN_IMAGE_DIMENSION ||
|
|
62
|
+
image.width * image.height < MIN_IMAGE_AREA) {
|
|
63
|
+
return false;
|
|
64
|
+
}
|
|
65
|
+
// Skip images that render too small in the viewport
|
|
66
|
+
if (coords.w < MIN_RENDERED_DIMENSION ||
|
|
67
|
+
coords.h < MIN_RENDERED_DIMENSION ||
|
|
68
|
+
coords.w * coords.h < MIN_RENDERED_AREA) {
|
|
69
|
+
return false;
|
|
70
|
+
}
|
|
71
|
+
return true;
|
|
72
|
+
});
|
|
73
|
+
return filtered;
|
|
74
|
+
}
|
|
75
|
+
/**
|
|
76
|
+
* Checks if two bounding boxes overlap and returns the overlap area.
|
|
77
|
+
*/
|
|
78
|
+
function getOverlapArea(box1, box2) {
|
|
79
|
+
const left = Math.max(box1.x, box2.x);
|
|
80
|
+
const right = Math.min(box1.x + box1.w, box2.x + box2.w);
|
|
81
|
+
const top = Math.max(box1.y, box2.y);
|
|
82
|
+
const bottom = Math.min(box1.y + box1.h, box2.y + box2.h);
|
|
83
|
+
if (left >= right || top >= bottom) {
|
|
84
|
+
return 0;
|
|
85
|
+
}
|
|
86
|
+
return (right - left) * (bottom - top);
|
|
87
|
+
}
|
|
88
|
+
/**
|
|
89
|
+
* Filters out OCR blocks that significantly overlap with already-extracted text items.
|
|
90
|
+
* This prevents duplicate text when both document text extraction and OCR detect the same content.
|
|
91
|
+
* Prefers document-extracted text over OCR text.
|
|
92
|
+
*
|
|
93
|
+
* An OCR block is rejected if:
|
|
94
|
+
* 1. The total overlap with all text items covers more than 50% of the OCR block area
|
|
95
|
+
* 2. OR the OCR block covers more than 50% of any single text item's area
|
|
96
|
+
*/
|
|
97
|
+
function filterOcrBlocksOverlappingWithText(ocrBlocks, textItems) {
|
|
98
|
+
if (!textItems.length || !ocrBlocks.length) {
|
|
99
|
+
return ocrBlocks;
|
|
100
|
+
}
|
|
101
|
+
return ocrBlocks.filter((ocrBlock) => {
|
|
102
|
+
const ocrBox = {
|
|
103
|
+
x: ocrBlock.x,
|
|
104
|
+
y: ocrBlock.y,
|
|
105
|
+
w: ocrBlock.w,
|
|
106
|
+
h: ocrBlock.h,
|
|
107
|
+
};
|
|
108
|
+
const ocrArea = ocrBlock.w * ocrBlock.h;
|
|
109
|
+
if (ocrArea <= 0) {
|
|
110
|
+
return false;
|
|
111
|
+
}
|
|
112
|
+
let totalOverlapArea = 0;
|
|
113
|
+
// Check overlap with each text item
|
|
114
|
+
for (const textItem of textItems) {
|
|
115
|
+
const textBox = {
|
|
116
|
+
x: textItem.x,
|
|
117
|
+
y: textItem.y,
|
|
118
|
+
w: textItem.w,
|
|
119
|
+
h: textItem.h,
|
|
120
|
+
};
|
|
121
|
+
const textItemArea = textItem.w * textItem.h;
|
|
122
|
+
const overlapArea = getOverlapArea(ocrBox, textBox);
|
|
123
|
+
if (overlapArea > 0) {
|
|
124
|
+
// Accumulate total overlap for condition 1
|
|
125
|
+
totalOverlapArea += overlapArea;
|
|
126
|
+
// Condition 2: Reject if OCR block covers more than 50% of any single text item
|
|
127
|
+
if (textItemArea > 0 && overlapArea / textItemArea >= OCR_OVERLAP_THRESHOLD) {
|
|
128
|
+
return false;
|
|
129
|
+
}
|
|
130
|
+
}
|
|
131
|
+
}
|
|
132
|
+
// Condition 1: Reject if total overlap covers more than 50% of the OCR block
|
|
133
|
+
const totalOverlapRatio = totalOverlapArea / ocrArea;
|
|
134
|
+
if (totalOverlapRatio >= OCR_OVERLAP_THRESHOLD) {
|
|
135
|
+
return false;
|
|
136
|
+
}
|
|
137
|
+
return true;
|
|
138
|
+
});
|
|
139
|
+
}
|
|
140
|
+
/**
|
|
141
|
+
* Build projection text boxes from page data, including OCR results
|
|
142
|
+
* This is the complete implementation from buildBbox.ts
|
|
143
|
+
*/
|
|
144
|
+
export function buildBbox(pageData, config) {
|
|
145
|
+
const lines = [];
|
|
146
|
+
// Process all extracted text items
|
|
147
|
+
for (const item of pageData.textItems) {
|
|
148
|
+
const line = {
|
|
149
|
+
x: item.x,
|
|
150
|
+
y: item.y,
|
|
151
|
+
rx: item.rx || 0,
|
|
152
|
+
ry: item.ry || 0,
|
|
153
|
+
w: Math.round(item.w || item.width),
|
|
154
|
+
h: Math.round(item.h || item.height),
|
|
155
|
+
r: item.r || 0,
|
|
156
|
+
str: item.str,
|
|
157
|
+
strLength: [...item.str].length, // Handle multi-byte characters correctly
|
|
158
|
+
pageBbox: {
|
|
159
|
+
x: item.x,
|
|
160
|
+
y: item.y,
|
|
161
|
+
w: item.w || item.width,
|
|
162
|
+
h: item.h || item.height,
|
|
163
|
+
},
|
|
164
|
+
vgap: item.vgap,
|
|
165
|
+
isPlaceholder: item.isPlaceholder,
|
|
166
|
+
};
|
|
167
|
+
lines.push(line);
|
|
168
|
+
}
|
|
169
|
+
// Process OCR data if images are present
|
|
170
|
+
if (pageData.images.length && config.ocrEnabled) {
|
|
171
|
+
// Filter images that should be processed for OCR
|
|
172
|
+
const imagesToProcess = filterImagesForOCR(pageData.images, {
|
|
173
|
+
width: pageData.width,
|
|
174
|
+
height: pageData.height,
|
|
175
|
+
});
|
|
176
|
+
// Collect text item bounding boxes for overlap checking
|
|
177
|
+
const textItemBoxes = pageData.textItems.map((item) => ({
|
|
178
|
+
x: item.x,
|
|
179
|
+
y: item.y,
|
|
180
|
+
w: item.w || item.width,
|
|
181
|
+
h: item.h || item.height,
|
|
182
|
+
}));
|
|
183
|
+
// Collect text content for content-based deduplication (normalized lowercase)
|
|
184
|
+
const existingTextContent = new Set(pageData.textItems.map((item) => item.str.trim().toLowerCase()).filter((s) => s.length > 0));
|
|
185
|
+
for (const image of imagesToProcess) {
|
|
186
|
+
// Parse OCR blocks from image
|
|
187
|
+
let ocrData = parseImageOcrBlocks(image);
|
|
188
|
+
// Filter by confidence threshold
|
|
189
|
+
ocrData = ocrData.filter((block) => parseFloat(block.confidence.toString()) >= OCR_CONFIDENCE_THRESHOLD);
|
|
190
|
+
// Filter out OCR blocks that overlap with already-extracted text
|
|
191
|
+
ocrData = filterOcrBlocksOverlappingWithText(ocrData, textItemBoxes);
|
|
192
|
+
// Filter out OCR blocks whose text content already exists in native PDF text
|
|
193
|
+
// This catches duplicates that are at different positions (e.g., watermarks, repeated headers)
|
|
194
|
+
ocrData = ocrData.filter((block) => {
|
|
195
|
+
const ocrText = block.c.trim().toLowerCase();
|
|
196
|
+
return ocrText.length > 0 && !existingTextContent.has(ocrText);
|
|
197
|
+
});
|
|
198
|
+
const ocrParsed = [];
|
|
199
|
+
for (const block of ocrData) {
|
|
200
|
+
const confidenceRounded = Math.round(parseFloat(block.confidence.toString()) * 1000) / 1000;
|
|
201
|
+
// Clean OCR artifacts from table border misreads
|
|
202
|
+
const cleanedText = cleanOcrTableArtifacts(block.c);
|
|
203
|
+
// Skip if cleaning removed all content
|
|
204
|
+
if (cleanedText.length === 0) {
|
|
205
|
+
continue;
|
|
206
|
+
}
|
|
207
|
+
const line = {
|
|
208
|
+
fromOCR: true,
|
|
209
|
+
x: block.x,
|
|
210
|
+
y: block.y,
|
|
211
|
+
w: block.w,
|
|
212
|
+
h: block.h,
|
|
213
|
+
r: image.originalOrientationAngle || 0,
|
|
214
|
+
str: cleanedText,
|
|
215
|
+
strLength: [...cleanedText].length,
|
|
216
|
+
pageBbox: {
|
|
217
|
+
x: block.x,
|
|
218
|
+
y: block.y,
|
|
219
|
+
w: block.w,
|
|
220
|
+
h: block.h,
|
|
221
|
+
},
|
|
222
|
+
};
|
|
223
|
+
lines.push(line);
|
|
224
|
+
ocrParsed.push({
|
|
225
|
+
x: block.rx,
|
|
226
|
+
y: block.ry,
|
|
227
|
+
w: block.rw,
|
|
228
|
+
h: block.rh,
|
|
229
|
+
confidence: confidenceRounded,
|
|
230
|
+
text: cleanedText,
|
|
231
|
+
});
|
|
232
|
+
}
|
|
233
|
+
if (ocrParsed.length) {
|
|
234
|
+
image.ocrParsed = ocrParsed;
|
|
235
|
+
}
|
|
236
|
+
}
|
|
237
|
+
}
|
|
238
|
+
return lines;
|
|
239
|
+
}
|
|
240
|
+
/**
|
|
241
|
+
* Build bounding boxes from text items
|
|
242
|
+
*/
|
|
243
|
+
export function buildBoundingBoxes(textItems) {
|
|
244
|
+
const bboxes = [];
|
|
245
|
+
for (const item of textItems) {
|
|
246
|
+
if (item.str.trim() === "") {
|
|
247
|
+
continue;
|
|
248
|
+
}
|
|
249
|
+
bboxes.push({
|
|
250
|
+
x1: item.x,
|
|
251
|
+
y1: item.y,
|
|
252
|
+
x2: item.x + (item.w || item.width),
|
|
253
|
+
y2: item.y + (item.h || item.height),
|
|
254
|
+
});
|
|
255
|
+
}
|
|
256
|
+
return bboxes;
|
|
257
|
+
}
|
|
258
|
+
//# sourceMappingURL=bbox.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bbox.js","sourceRoot":"","sources":["../../../src/processing/bbox.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,mBAAmB,EAAY,MAAM,eAAe,CAAC;AAC9D,OAAO,EAAE,sBAAsB,EAAE,MAAM,gBAAgB,CAAC;AAExD,MAAM,wBAAwB,GAAG,GAAG,CAAC;AAErC;;;;;GAKG;AACH,MAAM,qBAAqB,GAAG,GAAG,CAAC;AAElC;;;GAGG;AACH,MAAM,mBAAmB,GAAG,EAAE,CAAC;AAE/B;;GAEG;AACH,MAAM,mBAAmB,GAAG,EAAE,CAAC;AAC/B,MAAM,cAAc,GAAG,GAAG,CAAC;AAE3B;;GAEG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAC;AACjC,MAAM,iBAAiB,GAAG,GAAG,CAAC;AAE9B;;;GAGG;AACH,MAAM,UAAU,kBAAkB,CAChC,MAAe,EACf,IAAuC;IAEvC,0EAA0E;IAC1E,IAAI,QAAQ,GAAG,MAAM,CAAC,MAAM,CAC1B,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,IAAI,EAAE,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,UAAU,CAAC,UAAU,CAAC,CAChF,CAAC;IAEF,yDAAyD;IACzD,IAAI,QAAQ,CAAC,MAAM,GAAG,mBAAmB,EAAE,CAAC;QAC1C,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,KAAK,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC;QACjE,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,mBAAmB,CAAC,CAAC;IACpD,CAAC;IAED,sCAAsC;IACtC,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE;QACnC,iCAAiC;QACjC,IAAI,KAAK,CAAC,IAAI,EAAE,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;YACpC,OAAO,KAAK,CAAC;QACf,CAAC;QAED,4DAA4D;QAC5D,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,IAAI;YAC7B,CAAC,EAAE,KAAK,CAAC,CAAC;YACV,CAAC,EAAE,KAAK,CAAC,CAAC;YACV,CAAC,EAAE,KAAK,CAAC,KAAK;YACd,CAAC,EAAE,KAAK,CAAC,MAAM;SAChB,CAAC;QAEF,uCAAuC;QACvC,IACE,MAAM,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,IAAI,eAAe;YAC1C,MAAM,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,IAAI,aAAa;YACxC,MAAM,CAAC,CAAC,GAAG,IAAI,CAAC,KAAK,IAAI,gBAAgB;YACzC,MAAM,CAAC,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,aAAa;UACpC,CAAC;YACD,OAAO,KAAK,CAAC;QACf,CAAC;QAED,qCAAqC;QACrC,IACE,KAAK,CAAC,KAAK,GAAG,mBAAmB;YACjC,KAAK,CAAC,MAAM,GAAG,mBAAmB;YAClC,KAAK,CAAC,KAAK,GAAG,KAAK,CAAC,MAAM,GAAG,cAAc,EAC3C,CAAC;YACD,OAAO,KAAK,CAAC;QACf,CAAC;QAED,oDAAoD;QACpD,IACE,MAAM,CAAC,CAAC,GAAG,sBAAsB;YACjC,MAAM,CAAC,CAAC,GAAG,sBAAsB;YACjC,MAAM,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,iBAAiB,EACvC,CAAC;YACD,OAAO,KAAK,CAAC;QACf,CAAC;QAED,OAAO,IAAI,CAAC;IACd,CAAC,CAAC,CAAC;IAEH,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;GAEG;AACH,SAAS,cAAc,CACrB,IAAoD,EACpD,IAAoD;IAEpD,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC;IACtC,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;IACzD,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC;IACrC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;IAE1D,IAAI,IAAI,IAAI,KAAK,IAAI,GAAG,IAAI,MAAM,EAAE,CAAC;QACnC,OAAO,CAAC,CAAC;IACX,CAAC;IAED,OAAO,CAAC,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,GAAG,GAAG,CAAC,CAAC;AACzC,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,kCAAkC,CACzC,SAAqB,EACrB,SAAgE;IAEhE,IAAI,CAAC,SAAS,CAAC,MAAM,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,CAAC;QAC3C,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,OAAO,SAAS,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE;QACnC,MAAM,MAAM,GAAG;YACb,CAAC,EAAE,QAAQ,CAAC,CAAC;YACb,CAAC,EAAE,QAAQ,CAAC,CAAC;YACb,CAAC,EAAE,QAAQ,CAAC,CAAC;YACb,CAAC,EAAE,QAAQ,CAAC,CAAC;SACd,CAAC;QACF,MAAM,OAAO,GAAG,QAAQ,CAAC,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC;QAExC,IAAI,OAAO,IAAI,CAAC,EAAE,CAAC;YACjB,OAAO,KAAK,CAAC;QACf,CAAC;QAED,IAAI,gBAAgB,GAAG,CAAC,CAAC;QAEzB,oCAAoC;QACpC,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YACjC,MAAM,OAAO,GAAG;gBACd,CAAC,EAAE,QAAQ,CAAC,CAAC;gBACb,CAAC,EAAE,QAAQ,CAAC,CAAC;gBACb,CAAC,EAAE,QAAQ,CAAC,CAAC;gBACb,CAAC,EAAE,QAAQ,CAAC,CAAC;aACd,CAAC;YACF,MAAM,YAAY,GAAG,QAAQ,CAAC,CAAC,GAAG,QAAQ,CAAC,CAAC,CAAC;YAE7C,MAAM,WAAW,GAAG,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;YAEpD,IAAI,WAAW,GAAG,CAAC,EAAE,CAAC;gBACpB,2CAA2C;gBAC3C,gBAAgB,IAAI,WAAW,CAAC;gBAEhC,gFAAgF;gBAChF,IAAI,YAAY,GAAG,CAAC,IAAI,WAAW,GAAG,YAAY,IAAI,qBAAqB,EAAE,CAAC;oBAC5E,OAAO,KAAK,CAAC;gBACf,CAAC;YACH,CAAC;QACH,CAAC;QAED,6EAA6E;QAC7E,MAAM,iBAAiB,GAAG,gBAAgB,GAAG,OAAO,CAAC;QACrD,IAAI,iBAAiB,IAAI,qBAAqB,EAAE,CAAC;YAC/C,OAAO,KAAK,CAAC;QACf,CAAC;QAED,OAAO,IAAI,CAAC;IACd,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,SAAS,CAAC,QAAkB,EAAE,MAAuB;IACnE,MAAM,KAAK,GAAwB,EAAE,CAAC;IAEtC,mCAAmC;IACnC,KAAK,MAAM,IAAI,IAAI,QAAQ,CAAC,SAAS,EAAE,CAAC;QACtC,MAAM,IAAI,GAAsB;YAC9B,CAAC,EAAE,IAAI,CAAC,CAAC;YACT,CAAC,EAAE,IAAI,CAAC,CAAC;YACT,EAAE,EAAE,IAAI,CAAC,EAAE,IAAI,CAAC;YAChB,EAAE,EAAE,IAAI,CAAC,EAAE,IAAI,CAAC;YAChB,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC;YACnC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC;YACpC,CAAC,EAAE,IAAI,CAAC,CAAC,IAAI,CAAC;YACd,GAAG,EAAE,IAAI,CAAC,GAAG;YACb,SAAS,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,yCAAyC;YAC1E,QAAQ,EAAE;gBACR,CAAC,EAAE,IAAI,CAAC,CAAC;gBACT,CAAC,EAAE,IAAI,CAAC,CAAC;gBACT,CAAC,EAAE,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK;gBACvB,CAAC,EAAE,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,MAAM;aACzB;YACD,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,aAAa,EAAE,IAAI,CAAC,aAAa;SAClC,CAAC;QAEF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACnB,CAAC;IAED,yCAAyC;IACzC,IAAI,QAAQ,CAAC,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC,UAAU,EAAE,CAAC;QAChD,iDAAiD;QACjD,MAAM,eAAe,GAAG,kBAAkB,CAAC,QAAQ,CAAC,MAAM,EAAE;YAC1D,KAAK,EAAE,QAAQ,CAAC,KAAK;YACrB,MAAM,EAAE,QAAQ,CAAC,MAAM;SACxB,CAAC,CAAC;QAEH,wDAAwD;QACxD,MAAM,aAAa,GAAG,QAAQ,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;YACtD,CAAC,EAAE,IAAI,CAAC,CAAC;YACT,CAAC,EAAE,IAAI,CAAC,CAAC;YACT,CAAC,EAAE,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK;YACvB,CAAC,EAAE,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,MAAM;SACzB,CAAC,CAAC,CAAC;QAEJ,8EAA8E;QAC9E,MAAM,mBAAmB,GAAG,IAAI,GAAG,CACjC,QAAQ,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAC5F,CAAC;QAEF,KAAK,MAAM,KAAK,IAAI,eAAe,EAAE,CAAC;YACpC,8BAA8B;YAC9B,IAAI,OAAO,GAAG,mBAAmB,CAAC,KAAK,CAAC,CAAC;YAEzC,iCAAiC;YACjC,OAAO,GAAG,OAAO,CAAC,MAAM,CACtB,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,IAAI,wBAAwB,CAC/E,CAAC;YAEF,iEAAiE;YACjE,OAAO,GAAG,kCAAkC,CAAC,OAAO,EAAE,aAAa,CAAC,CAAC;YAErE,6EAA6E;YAC7E,+FAA+F;YAC/F,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE;gBACjC,MAAM,OAAO,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;gBAC7C,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;YACjE,CAAC,CAAC,CAAC;YAEH,MAAM,SAAS,GAAc,EAAE,CAAC;YAChC,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;gBAC5B,MAAM,iBAAiB,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC;gBAE5F,iDAAiD;gBACjD,MAAM,WAAW,GAAG,sBAAsB,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;gBAEpD,uCAAuC;gBACvC,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;oBAC7B,SAAS;gBACX,CAAC;gBAED,MAAM,IAAI,GAAsB;oBAC9B,OAAO,EAAE,IAAI;oBACb,CAAC,EAAE,KAAK,CAAC,CAAC;oBACV,CAAC,EAAE,KAAK,CAAC,CAAC;oBACV,CAAC,EAAE,KAAK,CAAC,CAAC;oBACV,CAAC,EAAE,KAAK,CAAC,CAAC;oBACV,CAAC,EAAE,KAAK,CAAC,wBAAwB,IAAI,CAAC;oBACtC,GAAG,EAAE,WAAW;oBAChB,SAAS,EAAE,CAAC,GAAG,WAAW,CAAC,CAAC,MAAM;oBAClC,QAAQ,EAAE;wBACR,CAAC,EAAE,KAAK,CAAC,CAAC;wBACV,CAAC,EAAE,KAAK,CAAC,CAAC;wBACV,CAAC,EAAE,KAAK,CAAC,CAAC;wBACV,CAAC,EAAE,KAAK,CAAC,CAAC;qBACX;iBACF,CAAC;gBACF,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;gBAEjB,SAAS,CAAC,IAAI,CAAC;oBACb,CAAC,EAAE,KAAK,CAAC,EAAE;oBACX,CAAC,EAAE,KAAK,CAAC,EAAE;oBACX,CAAC,EAAE,KAAK,CAAC,EAAE;oBACX,CAAC,EAAE,KAAK,CAAC,EAAE;oBACX,UAAU,EAAE,iBAAiB;oBAC7B,IAAI,EAAE,WAAW;iBAClB,CAAC,CAAC;YACL,CAAC;YAED,IAAI,SAAS,CAAC,MAAM,EAAE,CAAC;gBACrB,KAAK,CAAC,SAAS,GAAG,SAAS,CAAC;YAC9B,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,SAAqB;IACtD,MAAM,MAAM,GAAkB,EAAE,CAAC;IAEjC,KAAK,MAAM,IAAI,IAAI,SAAS,EAAE,CAAC;QAC7B,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;YAC3B,SAAS;QACX,CAAC;QAED,MAAM,CAAC,IAAI,CAAC;YACV,EAAE,EAAE,IAAI,CAAC,CAAC;YACV,EAAE,EAAE,IAAI,CAAC,CAAC;YACV,EAAE,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,KAAK,CAAC;YACnC,EAAE,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,CAAC,MAAM,CAAC;SACrC,CAAC,CAAC;IACL,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bbox.test.d.ts","sourceRoot":"","sources":["../../../src/processing/bbox.test.ts"],"names":[],"mappings":""}
|
|
@@ -0,0 +1,334 @@
|
|
|
1
|
+
import { expect, describe, it } from "vitest";
|
|
2
|
+
import { buildBbox, buildBoundingBoxes, filterImagesForOCR } from "./bbox";
|
|
3
|
+
import { DEFAULT_CONFIG } from "../core/config";
|
|
4
|
+
describe("test filterImagesForOCR", () => {
|
|
5
|
+
it("test valid image", () => {
|
|
6
|
+
const images = [
|
|
7
|
+
{
|
|
8
|
+
type: "photo",
|
|
9
|
+
width: 200,
|
|
10
|
+
height: 200,
|
|
11
|
+
x: 10,
|
|
12
|
+
y: 10,
|
|
13
|
+
coords: { x: 10, y: 10, w: 200, h: 200 },
|
|
14
|
+
},
|
|
15
|
+
];
|
|
16
|
+
const page = { width: 1000, height: 1000 };
|
|
17
|
+
const result = filterImagesForOCR(images, page);
|
|
18
|
+
expect(result).toStrictEqual(images);
|
|
19
|
+
});
|
|
20
|
+
it("test filter on patterns", () => {
|
|
21
|
+
const images = [
|
|
22
|
+
{ type: "g_background", width: 200, height: 200, x: 0, y: 0 },
|
|
23
|
+
{ type: "pattern_stripe", width: 200, height: 200, x: 0, y: 0 },
|
|
24
|
+
{
|
|
25
|
+
type: "photo",
|
|
26
|
+
width: 200,
|
|
27
|
+
height: 200,
|
|
28
|
+
x: 0,
|
|
29
|
+
y: 0,
|
|
30
|
+
coords: { x: 0, y: 0, w: 200, h: 200 },
|
|
31
|
+
},
|
|
32
|
+
];
|
|
33
|
+
const page = { width: 1000, height: 1000 };
|
|
34
|
+
const result = filterImagesForOCR(images, page);
|
|
35
|
+
expect(result).toStrictEqual([images[2]]);
|
|
36
|
+
});
|
|
37
|
+
it("test filter layout", () => {
|
|
38
|
+
const images = [
|
|
39
|
+
{
|
|
40
|
+
type: "layout_header",
|
|
41
|
+
width: 300,
|
|
42
|
+
height: 300,
|
|
43
|
+
x: 0,
|
|
44
|
+
y: 0,
|
|
45
|
+
coords: { x: 0, y: 0, w: 300, h: 300 },
|
|
46
|
+
},
|
|
47
|
+
];
|
|
48
|
+
const page = { width: 1000, height: 1000 };
|
|
49
|
+
const result = filterImagesForOCR(images, page);
|
|
50
|
+
expect(result.length).toBe(0);
|
|
51
|
+
});
|
|
52
|
+
it("test out of viewport", () => {
|
|
53
|
+
const images = [
|
|
54
|
+
{
|
|
55
|
+
type: "photo",
|
|
56
|
+
width: 200,
|
|
57
|
+
height: 200,
|
|
58
|
+
coords: { x: 1100, y: 0, w: 200, h: 200 },
|
|
59
|
+
x: 0,
|
|
60
|
+
y: 0,
|
|
61
|
+
},
|
|
62
|
+
];
|
|
63
|
+
const page = { width: 1000, height: 1000 };
|
|
64
|
+
const result = filterImagesForOCR(images, page);
|
|
65
|
+
expect(result.length).toBe(0);
|
|
66
|
+
});
|
|
67
|
+
it("test min dimensions", () => {
|
|
68
|
+
const images = [
|
|
69
|
+
{ type: "photo", width: 5, height: 5, x: 0, y: 0, coords: { x: 0, y: 0, w: 5, h: 5 } },
|
|
70
|
+
];
|
|
71
|
+
const page = { width: 1000, height: 1000 };
|
|
72
|
+
const result = filterImagesForOCR(images, page);
|
|
73
|
+
expect(result.length).toBe(0);
|
|
74
|
+
});
|
|
75
|
+
it("test max images per page", () => {
|
|
76
|
+
const images = [
|
|
77
|
+
{
|
|
78
|
+
type: "photo",
|
|
79
|
+
width: 100,
|
|
80
|
+
height: 100,
|
|
81
|
+
x: 0,
|
|
82
|
+
y: 0,
|
|
83
|
+
coords: { x: 0, y: 0, w: 100, h: 100 },
|
|
84
|
+
},
|
|
85
|
+
{
|
|
86
|
+
type: "photo",
|
|
87
|
+
width: 100,
|
|
88
|
+
height: 100,
|
|
89
|
+
x: 0,
|
|
90
|
+
y: 0,
|
|
91
|
+
coords: { x: 0, y: 0, w: 500, h: 500 },
|
|
92
|
+
},
|
|
93
|
+
{
|
|
94
|
+
type: "photo",
|
|
95
|
+
width: 100,
|
|
96
|
+
height: 100,
|
|
97
|
+
x: 0,
|
|
98
|
+
y: 0,
|
|
99
|
+
coords: { x: 0, y: 0, w: 300, h: 300 },
|
|
100
|
+
},
|
|
101
|
+
{
|
|
102
|
+
type: "photo",
|
|
103
|
+
width: 100,
|
|
104
|
+
height: 100,
|
|
105
|
+
x: 0,
|
|
106
|
+
y: 0,
|
|
107
|
+
coords: { x: 0, y: 0, w: 100, h: 100 },
|
|
108
|
+
},
|
|
109
|
+
{
|
|
110
|
+
type: "photo",
|
|
111
|
+
width: 100,
|
|
112
|
+
height: 100,
|
|
113
|
+
x: 0,
|
|
114
|
+
y: 0,
|
|
115
|
+
coords: { x: 0, y: 0, w: 500, h: 500 },
|
|
116
|
+
},
|
|
117
|
+
{
|
|
118
|
+
type: "photo",
|
|
119
|
+
width: 100,
|
|
120
|
+
height: 100,
|
|
121
|
+
x: 0,
|
|
122
|
+
y: 0,
|
|
123
|
+
coords: { x: 0, y: 0, w: 300, h: 300 },
|
|
124
|
+
},
|
|
125
|
+
{
|
|
126
|
+
type: "photo",
|
|
127
|
+
width: 100,
|
|
128
|
+
height: 100,
|
|
129
|
+
x: 0,
|
|
130
|
+
y: 0,
|
|
131
|
+
coords: { x: 0, y: 0, w: 100, h: 100 },
|
|
132
|
+
},
|
|
133
|
+
{
|
|
134
|
+
type: "photo",
|
|
135
|
+
width: 100,
|
|
136
|
+
height: 100,
|
|
137
|
+
x: 0,
|
|
138
|
+
y: 0,
|
|
139
|
+
coords: { x: 0, y: 0, w: 500, h: 500 },
|
|
140
|
+
},
|
|
141
|
+
{
|
|
142
|
+
type: "photo",
|
|
143
|
+
width: 100,
|
|
144
|
+
height: 100,
|
|
145
|
+
x: 0,
|
|
146
|
+
y: 0,
|
|
147
|
+
coords: { x: 0, y: 0, w: 300, h: 300 },
|
|
148
|
+
},
|
|
149
|
+
{
|
|
150
|
+
type: "photo",
|
|
151
|
+
width: 100,
|
|
152
|
+
height: 100,
|
|
153
|
+
x: 0,
|
|
154
|
+
y: 0,
|
|
155
|
+
coords: { x: 0, y: 0, w: 100, h: 100 },
|
|
156
|
+
},
|
|
157
|
+
{
|
|
158
|
+
type: "photo",
|
|
159
|
+
width: 100,
|
|
160
|
+
height: 100,
|
|
161
|
+
x: 0,
|
|
162
|
+
y: 0,
|
|
163
|
+
coords: { x: 0, y: 0, w: 500, h: 500 },
|
|
164
|
+
},
|
|
165
|
+
{
|
|
166
|
+
type: "photo",
|
|
167
|
+
width: 100,
|
|
168
|
+
height: 100,
|
|
169
|
+
x: 0,
|
|
170
|
+
y: 0,
|
|
171
|
+
coords: { x: 0, y: 0, w: 300, h: 300 },
|
|
172
|
+
},
|
|
173
|
+
];
|
|
174
|
+
const page = { width: 1000, height: 1000 };
|
|
175
|
+
const result = filterImagesForOCR(images, page);
|
|
176
|
+
expect(result).toStrictEqual(images.slice(0, 10));
|
|
177
|
+
});
|
|
178
|
+
});
|
|
179
|
+
describe("test buildBox", () => {
|
|
180
|
+
it("test with OCR disabled", () => {
|
|
181
|
+
const pageData = {
|
|
182
|
+
pageNum: 1,
|
|
183
|
+
width: 612,
|
|
184
|
+
height: 792,
|
|
185
|
+
textItems: [
|
|
186
|
+
{ str: "Hello World", x: 50, y: 100, width: 120, height: 14, w: 120, h: 14 },
|
|
187
|
+
{ str: "Some body text", x: 50, y: 130, width: 200, height: 14, w: 200, h: 14 },
|
|
188
|
+
],
|
|
189
|
+
images: [],
|
|
190
|
+
};
|
|
191
|
+
const config = { ...DEFAULT_CONFIG, ocrEnabled: false };
|
|
192
|
+
const expectedOutput = [
|
|
193
|
+
{
|
|
194
|
+
x: 50,
|
|
195
|
+
y: 100,
|
|
196
|
+
rx: 0,
|
|
197
|
+
ry: 0,
|
|
198
|
+
w: 120,
|
|
199
|
+
h: 14,
|
|
200
|
+
r: 0,
|
|
201
|
+
str: "Hello World",
|
|
202
|
+
strLength: 11,
|
|
203
|
+
pageBbox: { x: 50, y: 100, w: 120, h: 14 },
|
|
204
|
+
vgap: undefined,
|
|
205
|
+
isPlaceholder: undefined,
|
|
206
|
+
},
|
|
207
|
+
{
|
|
208
|
+
x: 50,
|
|
209
|
+
y: 130,
|
|
210
|
+
rx: 0,
|
|
211
|
+
ry: 0,
|
|
212
|
+
w: 200,
|
|
213
|
+
h: 14,
|
|
214
|
+
r: 0,
|
|
215
|
+
str: "Some body text",
|
|
216
|
+
strLength: 14,
|
|
217
|
+
pageBbox: { x: 50, y: 130, w: 200, h: 14 },
|
|
218
|
+
vgap: undefined,
|
|
219
|
+
isPlaceholder: undefined,
|
|
220
|
+
},
|
|
221
|
+
];
|
|
222
|
+
const result = buildBbox(pageData, config);
|
|
223
|
+
expect(result).toStrictEqual(expectedOutput);
|
|
224
|
+
});
|
|
225
|
+
it("test with OCR enabled", () => {
|
|
226
|
+
const pageData = {
|
|
227
|
+
pageNum: 1,
|
|
228
|
+
width: 612,
|
|
229
|
+
height: 792,
|
|
230
|
+
textItems: [
|
|
231
|
+
// Native PDF text (top-left)
|
|
232
|
+
{ str: "Hello World", x: 50, y: 100, width: 120, height: 14, w: 120, h: 14 },
|
|
233
|
+
],
|
|
234
|
+
images: [
|
|
235
|
+
{
|
|
236
|
+
x: 0,
|
|
237
|
+
y: 200,
|
|
238
|
+
width: 612,
|
|
239
|
+
height: 400,
|
|
240
|
+
originalOrientationAngle: 0,
|
|
241
|
+
// parseImageOcrBlocks() reads this internally:
|
|
242
|
+
ocrRaw: [
|
|
243
|
+
// Block A: no spatial overlap with native text, unique content → KEPT
|
|
244
|
+
[
|
|
245
|
+
[
|
|
246
|
+
[50, 50],
|
|
247
|
+
[250, 50],
|
|
248
|
+
[250, 70],
|
|
249
|
+
[50, 70],
|
|
250
|
+
],
|
|
251
|
+
"Scanned paragraph text",
|
|
252
|
+
0.95,
|
|
253
|
+
// resolved by parseImageOcrBlocks to absolute page coords:
|
|
254
|
+
// x:50, y:50, w:200, h:20, rx/ry/rw/rh for rotated coords
|
|
255
|
+
],
|
|
256
|
+
// Block B: text already exists in native items ("hello world") → FILTERED (content dedup)
|
|
257
|
+
[
|
|
258
|
+
[
|
|
259
|
+
[50, 0],
|
|
260
|
+
[170, 0],
|
|
261
|
+
[170, 14],
|
|
262
|
+
[50, 14],
|
|
263
|
+
],
|
|
264
|
+
"Hello World",
|
|
265
|
+
0.97,
|
|
266
|
+
// x:50, y:200, w:120, h:14 — also overlaps native text box
|
|
267
|
+
],
|
|
268
|
+
// Block C: low confidence → FILTERED (below threshold)
|
|
269
|
+
[
|
|
270
|
+
[
|
|
271
|
+
[300, 100],
|
|
272
|
+
[500, 100],
|
|
273
|
+
[500, 120],
|
|
274
|
+
[300, 120],
|
|
275
|
+
],
|
|
276
|
+
"Low confidence text",
|
|
277
|
+
0.05,
|
|
278
|
+
],
|
|
279
|
+
],
|
|
280
|
+
},
|
|
281
|
+
],
|
|
282
|
+
};
|
|
283
|
+
const config = { ...DEFAULT_CONFIG, ocrEnabled: true };
|
|
284
|
+
const expectedOutput = [
|
|
285
|
+
// ── Native text item ──────────────────────────────────────────────
|
|
286
|
+
{
|
|
287
|
+
x: 50,
|
|
288
|
+
y: 100,
|
|
289
|
+
rx: 0,
|
|
290
|
+
ry: 0,
|
|
291
|
+
w: 120,
|
|
292
|
+
h: 14,
|
|
293
|
+
r: 0,
|
|
294
|
+
str: "Hello World",
|
|
295
|
+
strLength: 11,
|
|
296
|
+
pageBbox: { x: 50, y: 100, w: 120, h: 14 },
|
|
297
|
+
vgap: undefined,
|
|
298
|
+
isPlaceholder: undefined,
|
|
299
|
+
},
|
|
300
|
+
// ── OCR block A (passed all filters) ─────────────────────────────
|
|
301
|
+
{
|
|
302
|
+
fromOCR: true,
|
|
303
|
+
x: 50,
|
|
304
|
+
y: 50,
|
|
305
|
+
w: 200,
|
|
306
|
+
h: 20,
|
|
307
|
+
r: 0,
|
|
308
|
+
str: "Scanned paragraph text",
|
|
309
|
+
strLength: 22,
|
|
310
|
+
pageBbox: { x: 50, y: 50, w: 200, h: 20 },
|
|
311
|
+
},
|
|
312
|
+
// Block B removed: spatial overlap >50% of native text item AND content dedup match
|
|
313
|
+
// Block C removed: confidence 0.40 < OCR_CONFIDENCE_THRESHOLD (0.5)
|
|
314
|
+
];
|
|
315
|
+
const result = buildBbox(pageData, config);
|
|
316
|
+
expect(result).toStrictEqual(expectedOutput);
|
|
317
|
+
});
|
|
318
|
+
});
|
|
319
|
+
describe("test buildBoundingBoxes", () => {
|
|
320
|
+
it("test buildBoundingBoxes success", () => {
|
|
321
|
+
const textItems = [
|
|
322
|
+
{ str: "Hello", x: 50, y: 100, width: 60, height: 14, w: 60, h: 14 },
|
|
323
|
+
{ str: " ", x: 50, y: 120, width: 30, height: 14, w: 30, h: 14 },
|
|
324
|
+
{ str: "World", x: 50, y: 140, width: 80, height: 14, w: 80, h: 14 },
|
|
325
|
+
];
|
|
326
|
+
const expectedOutput = [
|
|
327
|
+
{ x1: 50, y1: 100, x2: 110, y2: 114 },
|
|
328
|
+
{ x1: 50, y1: 140, x2: 130, y2: 154 },
|
|
329
|
+
];
|
|
330
|
+
const result = buildBoundingBoxes(textItems);
|
|
331
|
+
expect(result).toStrictEqual(expectedOutput);
|
|
332
|
+
});
|
|
333
|
+
});
|
|
334
|
+
//# sourceMappingURL=bbox.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"bbox.test.js","sourceRoot":"","sources":["../../../src/processing/bbox.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,EAAE,MAAM,QAAQ,CAAC;AAC9C,OAAO,EAAE,SAAS,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,MAAM,QAAQ,CAAC;AAE3E,OAAO,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAGhD,QAAQ,CAAC,yBAAyB,EAAE,GAAG,EAAE;IACvC,EAAE,CAAC,kBAAkB,EAAE,GAAG,EAAE;QAC1B,MAAM,MAAM,GAAG;YACb;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,EAAE;gBACL,MAAM,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACzC;SACF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC;IACvC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,yBAAyB,EAAE,GAAG,EAAE;QACjC,MAAM,MAAM,GAAG;YACb,EAAE,IAAI,EAAE,cAAc,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE;YAC7D,EAAE,IAAI,EAAE,gBAAgB,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,GAAG,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE;YAC/D;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;SACF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC5C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,oBAAoB,EAAE,GAAG,EAAE;QAC5B,MAAM,MAAM,GAAG;YACb;gBACE,IAAI,EAAE,eAAe;gBACrB,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;SACF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAChC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sBAAsB,EAAE,GAAG,EAAE;QAC9B,MAAM,MAAM,GAAG;YACb;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,MAAM,EAAE,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;gBACzC,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;aACL;SACF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAChC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,qBAAqB,EAAE,GAAG,EAAE;QAC7B,MAAM,MAAM,GAAG;YACb,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,EAAE;SACvF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAChC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,0BAA0B,EAAE,GAAG,EAAE;QAClC,MAAM,MAAM,GAAG;YACb;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;YACD;gBACE,IAAI,EAAE,OAAO;gBACb,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,CAAC,EAAE,CAAC;gBACJ,CAAC,EAAE,CAAC;gBACJ,MAAM,EAAE,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE;aACvC;SACF,CAAC;QACF,MAAM,IAAI,GAAG,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QAC3C,MAAM,MAAM,GAAG,kBAAkB,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;QAChD,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC;AAEH,QAAQ,CAAC,eAAe,EAAE,GAAG,EAAE;IAC7B,EAAE,CAAC,wBAAwB,EAAE,GAAG,EAAE;QAChC,MAAM,QAAQ,GAAG;YACf,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE;gBACT,EAAE,GAAG,EAAE,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;gBAC5E,EAAE,GAAG,EAAE,gBAAgB,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;aAChF;YACD,MAAM,EAAE,EAAE;SACX,CAAC;QACF,MAAM,MAAM,GAAoB,EAAE,GAAG,cAAc,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC;QAEzE,MAAM,cAAc,GAAG;YACrB;gBACE,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,GAAG;gBACN,EAAE,EAAE,CAAC;gBACL,EAAE,EAAE,CAAC;gBACL,CAAC,EAAE,GAAG;gBACN,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,CAAC;gBACJ,GAAG,EAAE,aAAa;gBAClB,SAAS,EAAE,EAAE;gBACb,QAAQ,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;gBAC1C,IAAI,EAAE,SAAS;gBACf,aAAa,EAAE,SAAS;aACzB;YACD;gBACE,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,GAAG;gBACN,EAAE,EAAE,CAAC;gBACL,EAAE,EAAE,CAAC;gBACL,CAAC,EAAE,GAAG;gBACN,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,CAAC;gBACJ,GAAG,EAAE,gBAAgB;gBACrB,SAAS,EAAE,EAAE;gBACb,QAAQ,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;gBAC1C,IAAI,EAAE,SAAS;gBACf,aAAa,EAAE,SAAS;aACzB;SACF,CAAC;QAEF,MAAM,MAAM,GAAG,SAAS,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;QAC3C,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,cAAc,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,uBAAuB,EAAE,GAAG,EAAE;QAC/B,MAAM,QAAQ,GAAG;YACf,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE;gBACT,6BAA6B;gBAC7B,EAAE,GAAG,EAAE,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,GAAG,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;aAC7E;YACD,MAAM,EAAE;gBACN;oBACE,CAAC,EAAE,CAAC;oBACJ,CAAC,EAAE,GAAG;oBACN,KAAK,EAAE,GAAG;oBACV,MAAM,EAAE,GAAG;oBACX,wBAAwB,EAAE,CAAC;oBAC3B,+CAA+C;oBAC/C,MAAM,EAAE;wBACN,sEAAsE;wBACtE;4BACE;gCACE,CAAC,EAAE,EAAE,EAAE,CAAC;gCACR,CAAC,GAAG,EAAE,EAAE,CAAC;gCACT,CAAC,GAAG,EAAE,EAAE,CAAC;gCACT,CAAC,EAAE,EAAE,EAAE,CAAC;6BACT;4BACD,wBAAwB;4BACxB,IAAI;4BACJ,2DAA2D;4BAC3D,0DAA0D;yBACtC;wBACtB,0FAA0F;wBAC1F;4BACE;gCACE,CAAC,EAAE,EAAE,CAAC,CAAC;gCACP,CAAC,GAAG,EAAE,CAAC,CAAC;gCACR,CAAC,GAAG,EAAE,EAAE,CAAC;gCACT,CAAC,EAAE,EAAE,EAAE,CAAC;6BACT;4BACD,aAAa;4BACb,IAAI;4BACJ,2DAA2D;yBACvC;wBACtB,uDAAuD;wBACvD;4BACE;gCACE,CAAC,GAAG,EAAE,GAAG,CAAC;gCACV,CAAC,GAAG,EAAE,GAAG,CAAC;gCACV,CAAC,GAAG,EAAE,GAAG,CAAC;gCACV,CAAC,GAAG,EAAE,GAAG,CAAC;6BACX;4BACD,qBAAqB;4BACrB,IAAI;yBACgB;qBACvB;iBACF;aACF;SACF,CAAC;QACF,MAAM,MAAM,GAAoB,EAAE,GAAG,cAAc,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC;QAExE,MAAM,cAAc,GAAG;YACrB,qEAAqE;YACrE;gBACE,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,GAAG;gBACN,EAAE,EAAE,CAAC;gBACL,EAAE,EAAE,CAAC;gBACL,CAAC,EAAE,GAAG;gBACN,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,CAAC;gBACJ,GAAG,EAAE,aAAa;gBAClB,SAAS,EAAE,EAAE;gBACb,QAAQ,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;gBAC1C,IAAI,EAAE,SAAS;gBACf,aAAa,EAAE,SAAS;aACzB;YAED,oEAAoE;YACpE;gBACE,OAAO,EAAE,IAAI;gBACb,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,GAAG;gBACN,CAAC,EAAE,EAAE;gBACL,CAAC,EAAE,CAAC;gBACJ,GAAG,EAAE,wBAAwB;gBAC7B,SAAS,EAAE,EAAE;gBACb,QAAQ,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE;aAC1C;YAED,oFAAoF;YACpF,oEAAoE;SACrE,CAAC;QAEF,MAAM,MAAM,GAAG,SAAS,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;QAC3C,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,cAAc,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC;AAEH,QAAQ,CAAC,yBAAyB,EAAE,GAAG,EAAE;IACvC,EAAE,CAAC,iCAAiC,EAAE,GAAG,EAAE;QACzC,MAAM,SAAS,GAAG;YAChB,EAAE,GAAG,EAAE,OAAO,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE;YACpE,EAAE,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE;YAClE,EAAE,GAAG,EAAE,OAAO,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE;SACrE,CAAC;QAEF,MAAM,cAAc,GAAG;YACrB,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE;YACrC,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE;SACtC,CAAC;QAEF,MAAM,MAAM,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAC;QAC7C,MAAM,CAAC,MAAM,CAAC,CAAC,aAAa,CAAC,cAAc,CAAC,CAAC;IAC/C,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -0,0 +1,6 @@
|
|
|
1
|
+
import { ParsedPage, LiteParseConfig } from "../core/types.js";
|
|
2
|
+
/**
|
|
3
|
+
* Clean raw text output - removes margins, null characters
|
|
4
|
+
*/
|
|
5
|
+
export declare function cleanRawText(pages: ParsedPage[], _config: LiteParseConfig): void;
|
|
6
|
+
//# sourceMappingURL=cleanText.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"cleanText.d.ts","sourceRoot":"","sources":["../../../src/processing/cleanText.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,MAAM,kBAAkB,CAAC;AA2E/D;;GAEG;AACH,wBAAgB,YAAY,CAAC,KAAK,EAAE,UAAU,EAAE,EAAE,OAAO,EAAE,eAAe,GAAG,IAAI,CAShF"}
|