@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 +1,537 @@
|
|
|
1
|
+
import { vi, describe, it, expect } from "vitest";
|
|
2
|
+
const { mockPages, mockPdfDocument, mockParsedPages, mockBoundingBoxes, mockPdfConvertedResult, mockOcrResults, mockParseResultJson, mockParsedPagesOcr, mockParseResultJsonOcr, mockTextItems, mockParsedPagesWithBoundingBoxes, mockParsedPagesWithBoundingBoxesOcr, mockScreenshotResults, } = vi.hoisted(() => {
|
|
3
|
+
const mockPdfConvertedResult = {
|
|
4
|
+
pdfPath: "/tmp/converted.pdf",
|
|
5
|
+
originalExtension: ".docx",
|
|
6
|
+
};
|
|
7
|
+
const mockPdfDocument = {
|
|
8
|
+
numPages: 5,
|
|
9
|
+
data: new TextEncoder().encode("hello world"),
|
|
10
|
+
};
|
|
11
|
+
const mockPages = [
|
|
12
|
+
{
|
|
13
|
+
pageNum: 1,
|
|
14
|
+
width: 612,
|
|
15
|
+
height: 792,
|
|
16
|
+
textItems: [],
|
|
17
|
+
paths: [],
|
|
18
|
+
images: [],
|
|
19
|
+
annotations: [],
|
|
20
|
+
},
|
|
21
|
+
{
|
|
22
|
+
pageNum: 2,
|
|
23
|
+
width: 612,
|
|
24
|
+
height: 792,
|
|
25
|
+
textItems: [],
|
|
26
|
+
paths: [],
|
|
27
|
+
images: [],
|
|
28
|
+
},
|
|
29
|
+
{
|
|
30
|
+
pageNum: 3,
|
|
31
|
+
width: 612,
|
|
32
|
+
height: 792,
|
|
33
|
+
textItems: [],
|
|
34
|
+
paths: [],
|
|
35
|
+
images: [],
|
|
36
|
+
annotations: [],
|
|
37
|
+
},
|
|
38
|
+
{
|
|
39
|
+
pageNum: 4,
|
|
40
|
+
width: 612,
|
|
41
|
+
height: 792,
|
|
42
|
+
textItems: [],
|
|
43
|
+
paths: [],
|
|
44
|
+
images: [],
|
|
45
|
+
},
|
|
46
|
+
{
|
|
47
|
+
pageNum: 5,
|
|
48
|
+
width: 612,
|
|
49
|
+
height: 792,
|
|
50
|
+
textItems: [],
|
|
51
|
+
paths: [],
|
|
52
|
+
images: [],
|
|
53
|
+
annotations: [],
|
|
54
|
+
},
|
|
55
|
+
];
|
|
56
|
+
const mockBoundingBoxes = [{ x1: 0, y1: 0, x2: 300, y2: 400 }];
|
|
57
|
+
const mockParsedPages = [
|
|
58
|
+
{
|
|
59
|
+
pageNum: 1,
|
|
60
|
+
width: 612,
|
|
61
|
+
height: 792,
|
|
62
|
+
text: "Sample text for page 1",
|
|
63
|
+
textItems: [],
|
|
64
|
+
boundingBoxes: undefined,
|
|
65
|
+
},
|
|
66
|
+
{
|
|
67
|
+
pageNum: 2,
|
|
68
|
+
width: 612,
|
|
69
|
+
height: 792,
|
|
70
|
+
text: "Sample text for page 2",
|
|
71
|
+
textItems: [],
|
|
72
|
+
},
|
|
73
|
+
{
|
|
74
|
+
pageNum: 3,
|
|
75
|
+
width: 612,
|
|
76
|
+
height: 792,
|
|
77
|
+
text: "Sample text for page 3",
|
|
78
|
+
textItems: [],
|
|
79
|
+
boundingBoxes: undefined,
|
|
80
|
+
},
|
|
81
|
+
{
|
|
82
|
+
pageNum: 4,
|
|
83
|
+
width: 612,
|
|
84
|
+
height: 792,
|
|
85
|
+
text: "Sample text for page 4",
|
|
86
|
+
textItems: [],
|
|
87
|
+
},
|
|
88
|
+
{
|
|
89
|
+
pageNum: 5,
|
|
90
|
+
width: 612,
|
|
91
|
+
height: 792,
|
|
92
|
+
text: "Sample text for page 5",
|
|
93
|
+
textItems: [],
|
|
94
|
+
boundingBoxes: undefined,
|
|
95
|
+
},
|
|
96
|
+
];
|
|
97
|
+
const mockParsedPagesWithBoundingBoxes = mockParsedPages.map((page) => ({
|
|
98
|
+
...page,
|
|
99
|
+
boundingBoxes: mockBoundingBoxes,
|
|
100
|
+
}));
|
|
101
|
+
const mockOcrResults = [
|
|
102
|
+
{ text: "Hello World", bbox: [10, 20, 200, 40], confidence: 0.98 },
|
|
103
|
+
{ text: "Sample text", bbox: [10, 50, 180, 70], confidence: 0.85 },
|
|
104
|
+
{ text: "Page footer", bbox: [10, 750, 300, 770], confidence: 0.76 },
|
|
105
|
+
];
|
|
106
|
+
const mockTextItems = mockOcrResults
|
|
107
|
+
.filter((r) => r.confidence > 0.1) // Filter low confidence
|
|
108
|
+
.filter((r) => {
|
|
109
|
+
// Filter out OCR text that already exists in native PDF text
|
|
110
|
+
const ocrText = r.text.trim().toLowerCase();
|
|
111
|
+
return ocrText.length > 0;
|
|
112
|
+
})
|
|
113
|
+
.map((r) => ({
|
|
114
|
+
str: r.text,
|
|
115
|
+
x: r.bbox[0],
|
|
116
|
+
y: r.bbox[1],
|
|
117
|
+
width: r.bbox[2] - r.bbox[0],
|
|
118
|
+
height: r.bbox[3] - r.bbox[1],
|
|
119
|
+
w: r.bbox[2] - r.bbox[0],
|
|
120
|
+
h: r.bbox[3] - r.bbox[1],
|
|
121
|
+
fontName: "OCR",
|
|
122
|
+
fontSize: r.bbox[3] - r.bbox[1],
|
|
123
|
+
}));
|
|
124
|
+
const mockTextItemsJson = mockOcrResults.map((r) => ({
|
|
125
|
+
text: r.text,
|
|
126
|
+
x: r.bbox[0],
|
|
127
|
+
y: r.bbox[1],
|
|
128
|
+
width: r.bbox[2] - r.bbox[0],
|
|
129
|
+
height: r.bbox[3] - r.bbox[1],
|
|
130
|
+
fontName: "OCR",
|
|
131
|
+
fontSize: r.bbox[3] - r.bbox[1],
|
|
132
|
+
}));
|
|
133
|
+
const mockParsedPagesOcr = [
|
|
134
|
+
{
|
|
135
|
+
pageNum: 1,
|
|
136
|
+
width: 612,
|
|
137
|
+
height: 792,
|
|
138
|
+
text: "Sample text for page 1",
|
|
139
|
+
textItems: mockTextItems,
|
|
140
|
+
boundingBoxes: undefined,
|
|
141
|
+
},
|
|
142
|
+
{
|
|
143
|
+
pageNum: 2,
|
|
144
|
+
width: 612,
|
|
145
|
+
height: 792,
|
|
146
|
+
text: "Sample text for page 2",
|
|
147
|
+
textItems: mockTextItems,
|
|
148
|
+
},
|
|
149
|
+
{
|
|
150
|
+
pageNum: 3,
|
|
151
|
+
width: 612,
|
|
152
|
+
height: 792,
|
|
153
|
+
text: "Sample text for page 3",
|
|
154
|
+
textItems: mockTextItems,
|
|
155
|
+
boundingBoxes: undefined,
|
|
156
|
+
},
|
|
157
|
+
{
|
|
158
|
+
pageNum: 4,
|
|
159
|
+
width: 612,
|
|
160
|
+
height: 792,
|
|
161
|
+
text: "Sample text for page 4",
|
|
162
|
+
textItems: mockTextItems,
|
|
163
|
+
},
|
|
164
|
+
{
|
|
165
|
+
pageNum: 5,
|
|
166
|
+
width: 612,
|
|
167
|
+
height: 792,
|
|
168
|
+
text: "Sample text for page 5",
|
|
169
|
+
textItems: mockTextItems,
|
|
170
|
+
boundingBoxes: undefined,
|
|
171
|
+
},
|
|
172
|
+
];
|
|
173
|
+
const mockParsedPagesWithBoundingBoxesOcr = mockParsedPagesOcr.map((page) => ({
|
|
174
|
+
...page,
|
|
175
|
+
boundingBoxes: mockBoundingBoxes,
|
|
176
|
+
}));
|
|
177
|
+
const mockParseResultJson = {
|
|
178
|
+
pages: [
|
|
179
|
+
{
|
|
180
|
+
page: 1,
|
|
181
|
+
width: 612,
|
|
182
|
+
height: 792,
|
|
183
|
+
text: "Sample text for page 1",
|
|
184
|
+
textItems: [],
|
|
185
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
186
|
+
},
|
|
187
|
+
{
|
|
188
|
+
page: 2,
|
|
189
|
+
width: 612,
|
|
190
|
+
height: 792,
|
|
191
|
+
text: "Sample text for page 2",
|
|
192
|
+
textItems: [],
|
|
193
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
194
|
+
},
|
|
195
|
+
{
|
|
196
|
+
page: 3,
|
|
197
|
+
width: 612,
|
|
198
|
+
height: 792,
|
|
199
|
+
text: "Sample text for page 3",
|
|
200
|
+
textItems: [],
|
|
201
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
202
|
+
},
|
|
203
|
+
{
|
|
204
|
+
page: 4,
|
|
205
|
+
width: 612,
|
|
206
|
+
height: 792,
|
|
207
|
+
text: "Sample text for page 4",
|
|
208
|
+
textItems: [],
|
|
209
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
210
|
+
},
|
|
211
|
+
{
|
|
212
|
+
page: 5,
|
|
213
|
+
width: 612,
|
|
214
|
+
height: 792,
|
|
215
|
+
text: "Sample text for page 5",
|
|
216
|
+
textItems: [],
|
|
217
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
218
|
+
},
|
|
219
|
+
],
|
|
220
|
+
};
|
|
221
|
+
const mockParseResultJsonOcr = {
|
|
222
|
+
pages: [
|
|
223
|
+
{
|
|
224
|
+
page: 1,
|
|
225
|
+
width: 612,
|
|
226
|
+
height: 792,
|
|
227
|
+
text: "Sample text for page 1",
|
|
228
|
+
textItems: mockTextItemsJson,
|
|
229
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
230
|
+
},
|
|
231
|
+
{
|
|
232
|
+
page: 2,
|
|
233
|
+
width: 612,
|
|
234
|
+
height: 792,
|
|
235
|
+
text: "Sample text for page 2",
|
|
236
|
+
textItems: mockTextItemsJson,
|
|
237
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
238
|
+
},
|
|
239
|
+
{
|
|
240
|
+
page: 3,
|
|
241
|
+
width: 612,
|
|
242
|
+
height: 792,
|
|
243
|
+
text: "Sample text for page 3",
|
|
244
|
+
textItems: mockTextItemsJson,
|
|
245
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
246
|
+
},
|
|
247
|
+
{
|
|
248
|
+
page: 4,
|
|
249
|
+
width: 612,
|
|
250
|
+
height: 792,
|
|
251
|
+
text: "Sample text for page 4",
|
|
252
|
+
textItems: mockTextItemsJson,
|
|
253
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
254
|
+
},
|
|
255
|
+
{
|
|
256
|
+
page: 5,
|
|
257
|
+
width: 612,
|
|
258
|
+
height: 792,
|
|
259
|
+
text: "Sample text for page 5",
|
|
260
|
+
textItems: mockTextItemsJson,
|
|
261
|
+
boundingBoxes: [{ x1: 0, y1: 0, x2: 300, y2: 400 }],
|
|
262
|
+
},
|
|
263
|
+
],
|
|
264
|
+
};
|
|
265
|
+
const mockImageBuffer = Buffer.from(new Uint8Array([1, 2, 3, 4, 5]));
|
|
266
|
+
const mockScreenshotResults = [
|
|
267
|
+
{
|
|
268
|
+
pageNum: 1,
|
|
269
|
+
width: 612,
|
|
270
|
+
height: 792,
|
|
271
|
+
imageBuffer: mockImageBuffer,
|
|
272
|
+
},
|
|
273
|
+
{
|
|
274
|
+
pageNum: 2,
|
|
275
|
+
width: 612,
|
|
276
|
+
height: 792,
|
|
277
|
+
imageBuffer: mockImageBuffer,
|
|
278
|
+
},
|
|
279
|
+
{
|
|
280
|
+
pageNum: 3,
|
|
281
|
+
width: 612,
|
|
282
|
+
height: 792,
|
|
283
|
+
imageBuffer: mockImageBuffer,
|
|
284
|
+
},
|
|
285
|
+
{
|
|
286
|
+
pageNum: 4,
|
|
287
|
+
width: 612,
|
|
288
|
+
height: 792,
|
|
289
|
+
imageBuffer: mockImageBuffer,
|
|
290
|
+
},
|
|
291
|
+
{
|
|
292
|
+
pageNum: 5,
|
|
293
|
+
width: 612,
|
|
294
|
+
height: 792,
|
|
295
|
+
imageBuffer: mockImageBuffer,
|
|
296
|
+
},
|
|
297
|
+
];
|
|
298
|
+
return {
|
|
299
|
+
mockPdfConvertedResult,
|
|
300
|
+
mockPdfDocument,
|
|
301
|
+
mockPages,
|
|
302
|
+
mockParsedPages,
|
|
303
|
+
mockBoundingBoxes,
|
|
304
|
+
mockOcrResults,
|
|
305
|
+
mockParseResultJson,
|
|
306
|
+
mockParsedPagesOcr,
|
|
307
|
+
mockParseResultJsonOcr,
|
|
308
|
+
mockTextItems,
|
|
309
|
+
mockParsedPagesWithBoundingBoxes,
|
|
310
|
+
mockParsedPagesWithBoundingBoxesOcr,
|
|
311
|
+
mockScreenshotResults,
|
|
312
|
+
};
|
|
313
|
+
});
|
|
314
|
+
import { LiteParse } from "./parser";
|
|
315
|
+
vi.mock("../conversion/convertToPdf.js", async () => {
|
|
316
|
+
const actual = await vi.importActual("../conversion/convertToPdf.js");
|
|
317
|
+
return {
|
|
318
|
+
...actual,
|
|
319
|
+
convertToPdf: vi.fn(async () => {
|
|
320
|
+
return mockPdfConvertedResult;
|
|
321
|
+
}),
|
|
322
|
+
cleanupConversionFiles: vi.fn(async () => { }),
|
|
323
|
+
};
|
|
324
|
+
});
|
|
325
|
+
vi.mock("../engines/pdf/pdfjs.js", async () => {
|
|
326
|
+
const actual = await vi.importActual("../engines/pdf/pdfjs.js");
|
|
327
|
+
return {
|
|
328
|
+
...actual,
|
|
329
|
+
PdfJsEngine: vi.fn(class {
|
|
330
|
+
constructor() { }
|
|
331
|
+
loadDocument = vi.fn().mockResolvedValue(mockPdfDocument);
|
|
332
|
+
extractAllPages = vi.fn().mockResolvedValue(mockPages);
|
|
333
|
+
extractPage = vi.fn().mockResolvedValue(mockPages[0]);
|
|
334
|
+
renderPageImage = vi.fn(async () => Buffer.from(new Uint8Array([1, 2, 3, 4, 5])));
|
|
335
|
+
close = vi.fn(async () => { });
|
|
336
|
+
}),
|
|
337
|
+
};
|
|
338
|
+
});
|
|
339
|
+
vi.mock("../engines/ocr/tesseract.js", async () => {
|
|
340
|
+
const actual = await vi.importActual("../engines/ocr/tesseract.js");
|
|
341
|
+
return {
|
|
342
|
+
...actual,
|
|
343
|
+
TesseractEngine: vi.fn(class {
|
|
344
|
+
constructor() { }
|
|
345
|
+
recognize = vi.fn().mockResolvedValue(mockOcrResults);
|
|
346
|
+
}),
|
|
347
|
+
};
|
|
348
|
+
});
|
|
349
|
+
vi.mock("../engines/ocr/http-simple.js", async () => {
|
|
350
|
+
const actual = await vi.importActual("../engines/ocr/http-simple.js");
|
|
351
|
+
return {
|
|
352
|
+
...actual,
|
|
353
|
+
HttpOcrEngine: vi.fn(class {
|
|
354
|
+
url;
|
|
355
|
+
constructor(url) {
|
|
356
|
+
this.url = url;
|
|
357
|
+
if (this.url != "http://localhost:8000") {
|
|
358
|
+
// works as an assertion that the URL is being passed to the class
|
|
359
|
+
// at init time
|
|
360
|
+
throw new Error("cannot accept a URL that is not http://localhost:8000");
|
|
361
|
+
}
|
|
362
|
+
}
|
|
363
|
+
recognize = vi.fn().mockResolvedValue(mockOcrResults);
|
|
364
|
+
}),
|
|
365
|
+
};
|
|
366
|
+
});
|
|
367
|
+
vi.mock("../processing/grid.js", async () => {
|
|
368
|
+
const actual = await vi.importActual("../processing/grid.js");
|
|
369
|
+
return {
|
|
370
|
+
...actual,
|
|
371
|
+
projectPagesToGrid: vi
|
|
372
|
+
.fn()
|
|
373
|
+
// these get modified, so we need to pass a deep copy of the array instead of the original one
|
|
374
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPages)) // no ocr - text
|
|
375
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPages)) // no ocr - json
|
|
376
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPagesOcr)) // ocr (tesseract) - text
|
|
377
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPagesOcr)) // ocr (tesseract) - json
|
|
378
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPagesOcr)) // ocr (http) - text
|
|
379
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPagesOcr)) // ocr (http) - json
|
|
380
|
+
.mockImplementationOnce(() => structuredClone(mockParsedPages)), //no bounding boxes
|
|
381
|
+
};
|
|
382
|
+
});
|
|
383
|
+
vi.mock("../processing/bbox.js", async () => {
|
|
384
|
+
const actual = await vi.importActual("../processing/bbox.js");
|
|
385
|
+
return {
|
|
386
|
+
...actual,
|
|
387
|
+
buildBoundingBoxes: vi.fn().mockReturnValue(mockBoundingBoxes),
|
|
388
|
+
};
|
|
389
|
+
});
|
|
390
|
+
vi.mock("../engines/pdf/pdfium-renderer.js", async () => {
|
|
391
|
+
const actual = await vi.importActual("../engines/pdf/pdfium-renderer.js");
|
|
392
|
+
return {
|
|
393
|
+
...actual,
|
|
394
|
+
PdfiumRenderer: vi.fn(class {
|
|
395
|
+
constructor() { }
|
|
396
|
+
renderPageToBuffer = vi.fn(async () => Buffer.from(new Uint8Array([1, 2, 3, 4, 5])));
|
|
397
|
+
close = vi.fn(async () => { });
|
|
398
|
+
}),
|
|
399
|
+
};
|
|
400
|
+
});
|
|
401
|
+
describe("Parse tests", () => {
|
|
402
|
+
it("test parse without OCR and text format", async () => {
|
|
403
|
+
const config = { ocrEnabled: false, outputFormat: "text" };
|
|
404
|
+
const liteparse = new LiteParse(config);
|
|
405
|
+
expect(liteparse.getConfig().ocrEnabled).toBeFalsy();
|
|
406
|
+
expect(liteparse.getConfig().outputFormat).toBe("text");
|
|
407
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
408
|
+
expect(result.json).toBeUndefined();
|
|
409
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
410
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxes);
|
|
411
|
+
});
|
|
412
|
+
it("test parse without OCR and json format", async () => {
|
|
413
|
+
const config = { ocrEnabled: false, outputFormat: "json" };
|
|
414
|
+
const liteparse = new LiteParse(config);
|
|
415
|
+
expect(liteparse.getConfig().ocrEnabled).toBeFalsy();
|
|
416
|
+
expect(liteparse.getConfig().outputFormat).toBe("json");
|
|
417
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
418
|
+
expect(result.json).toBeDefined();
|
|
419
|
+
expect(result.json).toStrictEqual(mockParseResultJson);
|
|
420
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
421
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxes);
|
|
422
|
+
});
|
|
423
|
+
it("test parse with OCR (tesseract) and text format", async () => {
|
|
424
|
+
const config = {
|
|
425
|
+
ocrEnabled: true,
|
|
426
|
+
outputFormat: "text",
|
|
427
|
+
ocrServerUrl: undefined,
|
|
428
|
+
};
|
|
429
|
+
const liteparse = new LiteParse(config);
|
|
430
|
+
expect(liteparse.getConfig().ocrEnabled).toBeTruthy();
|
|
431
|
+
expect(liteparse.getConfig().outputFormat).toBe("text");
|
|
432
|
+
expect(liteparse.getConfig().ocrServerUrl).toBeUndefined();
|
|
433
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
434
|
+
console.log(result.pages.at(0)?.textItems);
|
|
435
|
+
expect(result.pages.filter((page) => {
|
|
436
|
+
return page.textItems.length > 0;
|
|
437
|
+
}).length).toBeGreaterThan(0);
|
|
438
|
+
expect(result.json).toBeUndefined();
|
|
439
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
440
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxesOcr);
|
|
441
|
+
});
|
|
442
|
+
it("test parse with OCR (tesseract) and json format", async () => {
|
|
443
|
+
const config = {
|
|
444
|
+
ocrEnabled: true,
|
|
445
|
+
outputFormat: "json",
|
|
446
|
+
ocrServerUrl: undefined,
|
|
447
|
+
};
|
|
448
|
+
const liteparse = new LiteParse(config);
|
|
449
|
+
expect(liteparse.getConfig().ocrEnabled).toBeTruthy();
|
|
450
|
+
expect(liteparse.getConfig().outputFormat).toBe("json");
|
|
451
|
+
expect(liteparse.getConfig().ocrServerUrl).toBeUndefined();
|
|
452
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
453
|
+
expect(result.pages.filter((page) => {
|
|
454
|
+
return page.textItems.length > 0;
|
|
455
|
+
}).length).toBe(result.pages.length);
|
|
456
|
+
expect(result.pages.at(0).textItems).toStrictEqual(mockTextItems);
|
|
457
|
+
expect(result.json).toBeDefined();
|
|
458
|
+
expect(result.json).toStrictEqual(mockParseResultJsonOcr);
|
|
459
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
460
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxesOcr);
|
|
461
|
+
});
|
|
462
|
+
it("test parse with OCR (http) and text format", async () => {
|
|
463
|
+
const config = {
|
|
464
|
+
ocrEnabled: true,
|
|
465
|
+
outputFormat: "text",
|
|
466
|
+
ocrServerUrl: "http://localhost:8000",
|
|
467
|
+
};
|
|
468
|
+
const liteparse = new LiteParse(config);
|
|
469
|
+
expect(liteparse.getConfig().ocrEnabled).toBeTruthy();
|
|
470
|
+
expect(liteparse.getConfig().outputFormat).toBe("text");
|
|
471
|
+
expect(liteparse.getConfig().ocrServerUrl).toBe("http://localhost:8000");
|
|
472
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
473
|
+
console.log(result.pages.at(0)?.textItems);
|
|
474
|
+
expect(result.pages.filter((page) => {
|
|
475
|
+
return page.textItems.length > 0;
|
|
476
|
+
}).length).toBe(result.pages.length);
|
|
477
|
+
expect(result.pages.at(0).textItems).toStrictEqual(mockTextItems);
|
|
478
|
+
expect(result.json).toBeUndefined();
|
|
479
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
480
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxesOcr);
|
|
481
|
+
});
|
|
482
|
+
it("test parse with OCR (http) and json format", async () => {
|
|
483
|
+
const config = {
|
|
484
|
+
ocrEnabled: true,
|
|
485
|
+
outputFormat: "json",
|
|
486
|
+
ocrServerUrl: "http://localhost:8000",
|
|
487
|
+
};
|
|
488
|
+
const liteparse = new LiteParse(config);
|
|
489
|
+
expect(liteparse.getConfig().ocrEnabled).toBeTruthy();
|
|
490
|
+
expect(liteparse.getConfig().outputFormat).toBe("json");
|
|
491
|
+
expect(liteparse.getConfig().ocrServerUrl).toBe("http://localhost:8000");
|
|
492
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
493
|
+
expect(result.pages.filter((page) => {
|
|
494
|
+
return page.textItems.length > 0;
|
|
495
|
+
}).length).toBe(result.pages.length);
|
|
496
|
+
expect(result.pages.at(0).textItems).toStrictEqual(mockTextItems);
|
|
497
|
+
expect(result.json).toBeDefined();
|
|
498
|
+
expect(result.json).toStrictEqual(mockParseResultJsonOcr);
|
|
499
|
+
expect(result.text).toBe(mockParsedPages.map((p) => p.text).join("\n\n"));
|
|
500
|
+
expect(result.pages).toStrictEqual(mockParsedPagesWithBoundingBoxesOcr);
|
|
501
|
+
});
|
|
502
|
+
it("test without bounding boxes", async () => {
|
|
503
|
+
const config = {
|
|
504
|
+
ocrEnabled: false,
|
|
505
|
+
outputFormat: "text",
|
|
506
|
+
preciseBoundingBox: false,
|
|
507
|
+
};
|
|
508
|
+
const liteparse = new LiteParse(config);
|
|
509
|
+
expect(liteparse.getConfig().preciseBoundingBox).toBeFalsy();
|
|
510
|
+
const result = await liteparse.parse("/tmp/test.docx");
|
|
511
|
+
console.log(result.pages.at(0).boundingBoxes);
|
|
512
|
+
expect(result.pages.filter((page) => {
|
|
513
|
+
return typeof page.boundingBoxes != "undefined";
|
|
514
|
+
}).length).toBe(0);
|
|
515
|
+
});
|
|
516
|
+
});
|
|
517
|
+
describe("test screenshot", () => {
|
|
518
|
+
it("test screenshot with all pages", async () => {
|
|
519
|
+
const config = { ocrEnabled: false, outputFormat: "text" };
|
|
520
|
+
const liteparse = new LiteParse(config);
|
|
521
|
+
const results = await liteparse.screenshot("/tmp/test.docx");
|
|
522
|
+
expect(results.length).toBe(5);
|
|
523
|
+
expect(results).toStrictEqual(mockScreenshotResults);
|
|
524
|
+
});
|
|
525
|
+
it("test screenshot with selected pages", async () => {
|
|
526
|
+
const config = { ocrEnabled: false, outputFormat: "text" };
|
|
527
|
+
const liteparse = new LiteParse(config);
|
|
528
|
+
const results = await liteparse.screenshot("/tmp/test.docx", [1, 3, 4]);
|
|
529
|
+
expect(results.length).toBe(3);
|
|
530
|
+
expect(results).toStrictEqual([
|
|
531
|
+
mockScreenshotResults[0],
|
|
532
|
+
mockScreenshotResults[2],
|
|
533
|
+
mockScreenshotResults[3],
|
|
534
|
+
]);
|
|
535
|
+
});
|
|
536
|
+
});
|
|
537
|
+
//# sourceMappingURL=parser.test.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"parser.test.js","sourceRoot":"","sources":["../../../src/core/parser.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,EAAE,EAAE,QAAQ,EAAE,EAAE,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAElD,MAAM,EACJ,SAAS,EACT,eAAe,EACf,eAAe,EACf,iBAAiB,EACjB,sBAAsB,EACtB,cAAc,EACd,mBAAmB,EACnB,kBAAkB,EAClB,sBAAsB,EACtB,aAAa,EACb,gCAAgC,EAChC,mCAAmC,EACnC,qBAAqB,GACtB,GAAG,EAAE,CAAC,OAAO,CAAC,GAAG,EAAE;IAClB,MAAM,sBAAsB,GAAG;QAC7B,OAAO,EAAE,oBAAoB;QAC7B,iBAAiB,EAAE,OAAO;KAC3B,CAAC;IAEF,MAAM,eAAe,GAAG;QACtB,QAAQ,EAAE,CAAC;QACX,IAAI,EAAE,IAAI,WAAW,EAAE,CAAC,MAAM,CAAC,aAAa,CAAC;KAC9C,CAAC;IAEF,MAAM,SAAS,GAAG;QAChB;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE,EAAE;YACb,KAAK,EAAE,EAAE;YACT,MAAM,EAAE,EAAE;YACV,WAAW,EAAE,EAAE;SAChB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE,EAAE;YACb,KAAK,EAAE,EAAE;YACT,MAAM,EAAE,EAAE;SACX;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE,EAAE;YACb,KAAK,EAAE,EAAE;YACT,MAAM,EAAE,EAAE;YACV,WAAW,EAAE,EAAE;SAChB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE,EAAE;YACb,KAAK,EAAE,EAAE;YACT,MAAM,EAAE,EAAE;SACX;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,SAAS,EAAE,EAAE;YACb,KAAK,EAAE,EAAE;YACT,MAAM,EAAE,EAAE;YACV,WAAW,EAAE,EAAE;SAChB;KACF,CAAC;IAEF,MAAM,iBAAiB,GAAG,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;IAE/D,MAAM,eAAe,GAAG;QACtB;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,EAAE;YACb,aAAa,EAAE,SAAS;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,EAAE;SACd;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,EAAE;YACb,aAAa,EAAE,SAAS;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,EAAE;SACd;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,EAAE;YACb,aAAa,EAAE,SAAS;SACzB;KACF,CAAC;IAEF,MAAM,gCAAgC,GAAG,eAAe,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QACtE,GAAG,IAAI;QACP,aAAa,EAAE,iBAAiB;KACjC,CAAC,CAAC,CAAC;IAEJ,MAAM,cAAc,GAAG;QACrB,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;QAClE,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;QAClE,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,EAAE,UAAU,EAAE,IAAI,EAAE;KACrE,CAAC;IAEF,MAAM,aAAa,GAAG,cAAc;SACjC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,GAAG,GAAG,CAAC,CAAC,wBAAwB;SAC1D,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE;QACZ,6DAA6D;QAC7D,MAAM,OAAO,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;QAC5C,OAAO,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;IAC5B,CAAC,CAAC;SACD,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACX,GAAG,EAAE,CAAC,CAAC,IAAI;QACX,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACZ,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACZ,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QAC5B,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QAC7B,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACxB,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACxB,QAAQ,EAAE,KAAK;QACf,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;KAChC,CAAC,CAAC,CAAC;IAEN,MAAM,iBAAiB,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QACnD,IAAI,EAAE,CAAC,CAAC,IAAI;QACZ,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACZ,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACZ,KAAK,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QAC5B,MAAM,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QAC7B,QAAQ,EAAE,KAAK;QACf,QAAQ,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;KAChC,CAAC,CAAC,CAAC;IAEJ,MAAM,kBAAkB,GAAG;QACzB;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,aAAa;YACxB,aAAa,EAAE,SAAS;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,aAAa;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,aAAa;YACxB,aAAa,EAAE,SAAS;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,aAAa;SACzB;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,IAAI,EAAE,wBAAwB;YAC9B,SAAS,EAAE,aAAa;YACxB,aAAa,EAAE,SAAS;SACzB;KACF,CAAC;IAEF,MAAM,mCAAmC,GAAG,kBAAkB,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC;QAC5E,GAAG,IAAI;QACP,aAAa,EAAE,iBAAiB;KACjC,CAAC,CAAC,CAAC;IAEJ,MAAM,mBAAmB,GAAG;QAC1B,KAAK,EAAE;YACL;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,EAAE;gBACb,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,EAAE;gBACb,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,EAAE;gBACb,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,EAAE;gBACb,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,EAAE;gBACb,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;SACF;KACF,CAAC;IAEF,MAAM,sBAAsB,GAAG;QAC7B,KAAK,EAAE;YACL;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,iBAAiB;gBAC5B,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,iBAAiB;gBAC5B,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,iBAAiB;gBAC5B,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,iBAAiB;gBAC5B,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;YACD;gBACE,IAAI,EAAE,CAAC;gBACP,KAAK,EAAE,GAAG;gBACV,MAAM,EAAE,GAAG;gBACX,IAAI,EAAE,wBAAwB;gBAC9B,SAAS,EAAE,iBAAiB;gBAC5B,aAAa,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,EAAE,EAAE,GAAG,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC;aACpD;SACF;KACF,CAAC;IAEF,MAAM,eAAe,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;IACrE,MAAM,qBAAqB,GAAuB;QAChD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,WAAW,EAAE,eAAe;SAC7B;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,WAAW,EAAE,eAAe;SAC7B;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,WAAW,EAAE,eAAe;SAC7B;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,WAAW,EAAE,eAAe;SAC7B;QACD;YACE,OAAO,EAAE,CAAC;YACV,KAAK,EAAE,GAAG;YACV,MAAM,EAAE,GAAG;YACX,WAAW,EAAE,eAAe;SAC7B;KACF,CAAC;IAEF,OAAO;QACL,sBAAsB;QACtB,eAAe;QACf,SAAS;QACT,eAAe;QACf,iBAAiB;QACjB,cAAc;QACd,mBAAmB;QACnB,kBAAkB;QAClB,sBAAsB;QACtB,aAAa;QACb,gCAAgC;QAChC,mCAAmC;QACnC,qBAAqB;KACtB,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,OAAO,EAAE,SAAS,EAAE,MAAM,UAAU,CAAC;AAGrC,EAAE,CAAC,IAAI,CAAC,+BAA+B,EAAE,KAAK,IAAI,EAAE;IAClD,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC,YAAY,CAClC,+BAA+B,CAChC,CAAC;IACF,OAAO;QACL,GAAG,MAAM;QACT,YAAY,EAAE,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE;YAC7B,OAAO,sBAAsB,CAAC;QAChC,CAAC,CAAC;QACF,sBAAsB,EAAE,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE,GAAE,CAAC,CAAC;KAC9C,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,yBAAyB,EAAE,KAAK,IAAI,EAAE;IAC5C,MAAM,MAAM,GACV,MAAM,EAAE,CAAC,YAAY,CAA2C,yBAAyB,CAAC,CAAC;IAC7F,OAAO;QACL,GAAG,MAAM;QACT,WAAW,EAAE,EAAE,CAAC,EAAE,CAChB;YACE,gBAAe,CAAC;YAEhB,YAAY,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,eAAe,CAAC,CAAC;YAC1D,eAAe,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,SAAS,CAAC,CAAC;YACvD,WAAW,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;YACtD,eAAe,GAAG,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAClF,KAAK,GAAG,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE,GAAE,CAAC,CAAC,CAAC;SAC/B,CACF;KACF,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,6BAA6B,EAAE,KAAK,IAAI,EAAE;IAChD,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC,YAAY,CAClC,6BAA6B,CAC9B,CAAC;IACF,OAAO;QACL,GAAG,MAAM;QACT,eAAe,EAAE,EAAE,CAAC,EAAE,CACpB;YACE,gBAAe,CAAC;YAEhB,SAAS,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,cAAc,CAAC,CAAC;SACvD,CACF;KACF,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,+BAA+B,EAAE,KAAK,IAAI,EAAE;IAClD,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC,YAAY,CAClC,+BAA+B,CAChC,CAAC;IACF,OAAO;QACL,GAAG,MAAM;QACT,aAAa,EAAE,EAAE,CAAC,EAAE,CAClB;YACU,GAAG,CAAC;YAEZ,YAAY,GAAW;gBACrB,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC;gBACf,IAAI,IAAI,CAAC,GAAG,IAAK,uBAAkC,EAAE,CAAC;oBACpD,kEAAkE;oBAClE,eAAe;oBACf,MAAM,IAAI,KAAK,CAAC,uDAAuD,CAAC,CAAC;gBAC3E,CAAC;YACH,CAAC;YAED,SAAS,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,iBAAiB,CAAC,cAAc,CAAC,CAAC;SACvD,CACF;KACF,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;IAC1C,MAAM,MAAM,GACV,MAAM,EAAE,CAAC,YAAY,CAAyC,uBAAuB,CAAC,CAAC;IACzF,OAAO;QACL,GAAG,MAAM;QACT,kBAAkB,EAAE,EAAE;aACnB,EAAE,EAAE;YACL,8FAA8F;aAC7F,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,eAAe,CAAC,CAAC,CAAC,gBAAgB;aAC/E,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,eAAe,CAAC,CAAC,CAAC,gBAAgB;aAC/E,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,yBAAyB;aAC3F,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,yBAAyB;aAC3F,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,oBAAoB;aACtF,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,kBAAkB,CAAC,CAAC,CAAC,oBAAoB;aACtF,sBAAsB,CAAC,GAAG,EAAE,CAAC,eAAe,CAAC,eAAe,CAAC,CAAC,EAAE,mBAAmB;KACvF,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,uBAAuB,EAAE,KAAK,IAAI,EAAE;IAC1C,MAAM,MAAM,GACV,MAAM,EAAE,CAAC,YAAY,CAAyC,uBAAuB,CAAC,CAAC;IACzF,OAAO;QACL,GAAG,MAAM;QACT,kBAAkB,EAAE,EAAE,CAAC,EAAE,EAAE,CAAC,eAAe,CAAC,iBAAiB,CAAC;KAC/D,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,EAAE,CAAC,IAAI,CAAC,mCAAmC,EAAE,KAAK,IAAI,EAAE;IACtD,MAAM,MAAM,GAAG,MAAM,EAAE,CAAC,YAAY,CAClC,mCAAmC,CACpC,CAAC;IACF,OAAO;QACL,GAAG,MAAM;QACT,cAAc,EAAE,EAAE,CAAC,EAAE,CACnB;YACE,gBAAe,CAAC;YAEhB,kBAAkB,GAAG,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,UAAU,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YACrF,KAAK,GAAG,EAAE,CAAC,EAAE,CAAC,KAAK,IAAI,EAAE,GAAE,CAAC,CAAC,CAAC;SAC/B,CACF;KACF,CAAC;AACJ,CAAC,CAAC,CAAC;AAEH,QAAQ,CAAC,aAAa,EAAE,GAAG,EAAE;IAC3B,EAAE,CAAC,wCAAwC,EAAE,KAAK,IAAI,EAAE;QACtD,MAAM,MAAM,GAA6B,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,EAAE,MAAM,EAAE,CAAC;QACrF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,EAAE,CAAC;QACpC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,gCAAgC,CAAC,CAAC;IACvE,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,wCAAwC,EAAE,KAAK,IAAI,EAAE;QACtD,MAAM,MAAM,GAA6B,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,EAAE,MAAM,EAAE,CAAC;QACrF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;QAClC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,CAAC,mBAAmB,CAAC,CAAC;QACvD,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,gCAAgC,CAAC,CAAC;IACvE,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,iDAAiD,EAAE,KAAK,IAAI,EAAE;QAC/D,MAAM,MAAM,GAA6B;YACvC,UAAU,EAAE,IAAI;YAChB,YAAY,EAAE,MAAM;YACpB,YAAY,EAAE,SAAS;SACxB,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,UAAU,EAAE,CAAC;QACtD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,aAAa,EAAE,CAAC;QAC3D,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;QAC3C,MAAM,CACJ,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;YAC3B,OAAO,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC;QACnC,CAAC,CAAC,CAAC,MAAM,CACV,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC;QACrB,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,EAAE,CAAC;QACpC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,mCAAmC,CAAC,CAAC;IAC1E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,iDAAiD,EAAE,KAAK,IAAI,EAAE;QAC/D,MAAM,MAAM,GAA6B;YACvC,UAAU,EAAE,IAAI;YAChB,YAAY,EAAE,MAAM;YACpB,YAAY,EAAE,SAAS;SACxB,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,UAAU,EAAE,CAAC;QACtD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,aAAa,EAAE,CAAC;QAC3D,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,MAAM,CACJ,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;YAC3B,OAAO,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC;QACnC,CAAC,CAAC,CAAC,MAAM,CACV,CAAC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAC5B,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAE,CAAC,SAAS,CAAC,CAAC,aAAa,CAAC,aAAa,CAAC,CAAC;QACnE,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;QAClC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,CAAC,sBAAsB,CAAC,CAAC;QAC1D,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,mCAAmC,CAAC,CAAC;IAC1E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4CAA4C,EAAE,KAAK,IAAI,EAAE;QAC1D,MAAM,MAAM,GAA6B;YACvC,UAAU,EAAE,IAAI;YAChB,YAAY,EAAE,MAAM;YACpB,YAAY,EAAE,uBAAuB;SACtC,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,UAAU,EAAE,CAAC;QACtD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;QACzE,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,SAAS,CAAC,CAAC;QAC3C,MAAM,CACJ,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;YAC3B,OAAO,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC;QACnC,CAAC,CAAC,CAAC,MAAM,CACV,CAAC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAC5B,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAE,CAAC,SAAS,CAAC,CAAC,aAAa,CAAC,aAAa,CAAC,CAAC;QACnE,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,EAAE,CAAC;QACpC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,mCAAmC,CAAC,CAAC;IAC1E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,4CAA4C,EAAE,KAAK,IAAI,EAAE;QAC1D,MAAM,MAAM,GAA6B;YACvC,UAAU,EAAE,IAAI;YAChB,YAAY,EAAE,MAAM;YACpB,YAAY,EAAE,uBAAuB;SACtC,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,UAAU,CAAC,CAAC,UAAU,EAAE,CAAC;QACtD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACxD,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,uBAAuB,CAAC,CAAC;QACzE,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,MAAM,CACJ,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;YAC3B,OAAO,IAAI,CAAC,SAAS,CAAC,MAAM,GAAG,CAAC,CAAC;QACnC,CAAC,CAAC,CAAC,MAAM,CACV,CAAC,IAAI,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAC5B,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAE,CAAC,SAAS,CAAC,CAAC,aAAa,CAAC,aAAa,CAAC,CAAC;QACnE,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC;QAClC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,aAAa,CAAC,sBAAsB,CAAC,CAAC;QAC1D,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC;QAC1E,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,aAAa,CAAC,mCAAmC,CAAC,CAAC;IAC1E,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,6BAA6B,EAAE,KAAK,IAAI,EAAE;QAC3C,MAAM,MAAM,GAA6B;YACvC,UAAU,EAAE,KAAK;YACjB,YAAY,EAAE,MAAM;YACpB,kBAAkB,EAAE,KAAK;SAC1B,CAAC;QACF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,CAAC,SAAS,CAAC,SAAS,EAAE,CAAC,kBAAkB,CAAC,CAAC,SAAS,EAAE,CAAC;QAC7D,MAAM,MAAM,GAAG,MAAM,SAAS,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC;QACvD,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,CAAC,CAAE,CAAC,aAAa,CAAC,CAAC;QAC/C,MAAM,CACJ,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;YAC3B,OAAO,OAAO,IAAI,CAAC,aAAa,IAAI,WAAW,CAAC;QAClD,CAAC,CAAC,CAAC,MAAM,CACV,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IACZ,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC;AAEH,QAAQ,CAAC,iBAAiB,EAAE,GAAG,EAAE;IAC/B,EAAE,CAAC,gCAAgC,EAAE,KAAK,IAAI,EAAE;QAC9C,MAAM,MAAM,GAA6B,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,EAAE,MAAM,EAAE,CAAC;QACrF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,OAAO,GAAG,MAAM,SAAS,CAAC,UAAU,CAAC,gBAAgB,CAAC,CAAC;QAC7D,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC/B,MAAM,CAAC,OAAO,CAAC,CAAC,aAAa,CAAC,qBAAqB,CAAC,CAAC;IACvD,CAAC,CAAC,CAAC;IACH,EAAE,CAAC,qCAAqC,EAAE,KAAK,IAAI,EAAE;QACnD,MAAM,MAAM,GAA6B,EAAE,UAAU,EAAE,KAAK,EAAE,YAAY,EAAE,MAAM,EAAE,CAAC;QACrF,MAAM,SAAS,GAAG,IAAI,SAAS,CAAC,MAAM,CAAC,CAAC;QACxC,MAAM,OAAO,GAAG,MAAM,SAAS,CAAC,UAAU,CAAC,gBAAgB,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;QACxE,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAC/B,MAAM,CAAC,OAAO,CAAC,CAAC,aAAa,CAAC;YAC5B,qBAAqB,CAAC,CAAC,CAAC;YACxB,qBAAqB,CAAC,CAAC,CAAC;YACxB,qBAAqB,CAAC,CAAC,CAAC;SACzB,CAAC,CAAC;IACL,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|