@llamaindex/liteparse 1.5.3 → 2.0.0-beta.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/README.md +49 -448
- package/dist/cli.d.ts +3 -0
- package/dist/cli.d.ts.map +1 -0
- package/dist/cli.js +87 -0
- package/dist/cli.js.map +1 -0
- package/dist/lib.d.ts +58 -0
- package/dist/lib.d.ts.map +1 -0
- package/dist/lib.js +88 -0
- package/dist/lib.js.map +1 -0
- package/dist/native.d.ts +54 -0
- package/dist/native.d.ts.map +1 -0
- package/dist/native.js +70 -0
- package/dist/native.js.map +1 -0
- package/libpdfium.so +0 -0
- package/liteparse.linux-x64-gnu.node +0 -0
- package/package.json +36 -50
- package/LICENSE +0 -201
- package/dist/cli/parse.d.ts +0 -4
- package/dist/cli/parse.d.ts.map +0 -1
- package/dist/cli/parse.js +0 -450
- package/dist/cli/parse.js.map +0 -1
- package/dist/package.json +0 -90
- package/dist/src/conversion/convertToPdf.d.ts +0 -65
- package/dist/src/conversion/convertToPdf.d.ts.map +0 -1
- package/dist/src/conversion/convertToPdf.js +0 -405
- package/dist/src/conversion/convertToPdf.js.map +0 -1
- package/dist/src/conversion/convertToPdf.test.d.ts +0 -2
- package/dist/src/conversion/convertToPdf.test.d.ts.map +0 -1
- package/dist/src/conversion/convertToPdf.test.js +0 -327
- package/dist/src/conversion/convertToPdf.test.js.map +0 -1
- package/dist/src/core/config.d.ts +0 -4
- package/dist/src/core/config.d.ts.map +0 -1
- package/dist/src/core/config.js +0 -26
- package/dist/src/core/config.js.map +0 -1
- package/dist/src/core/config.test.d.ts +0 -2
- package/dist/src/core/config.test.d.ts.map +0 -1
- package/dist/src/core/config.test.js +0 -21
- package/dist/src/core/config.test.js.map +0 -1
- package/dist/src/core/parser.d.ts +0 -92
- package/dist/src/core/parser.d.ts.map +0 -1
- package/dist/src/core/parser.js +0 -401
- package/dist/src/core/parser.js.map +0 -1
- package/dist/src/core/parser.test.d.ts +0 -2
- package/dist/src/core/parser.test.d.ts.map +0 -1
- package/dist/src/core/parser.test.js +0 -541
- package/dist/src/core/parser.test.js.map +0 -1
- package/dist/src/core/types.d.ts +0 -370
- package/dist/src/core/types.d.ts.map +0 -1
- package/dist/src/core/types.js +0 -2
- package/dist/src/core/types.js.map +0 -1
- package/dist/src/engines/ocr/http-simple.d.ts +0 -19
- package/dist/src/engines/ocr/http-simple.d.ts.map +0 -1
- package/dist/src/engines/ocr/http-simple.js +0 -69
- package/dist/src/engines/ocr/http-simple.js.map +0 -1
- package/dist/src/engines/ocr/http-simple.test.d.ts +0 -2
- package/dist/src/engines/ocr/http-simple.test.d.ts.map +0 -1
- package/dist/src/engines/ocr/http-simple.test.js +0 -108
- package/dist/src/engines/ocr/http-simple.test.js.map +0 -1
- package/dist/src/engines/ocr/interface.d.ts +0 -15
- package/dist/src/engines/ocr/interface.d.ts.map +0 -1
- package/dist/src/engines/ocr/interface.js +0 -2
- package/dist/src/engines/ocr/interface.js.map +0 -1
- package/dist/src/engines/ocr/tesseract.d.ts +0 -20
- package/dist/src/engines/ocr/tesseract.d.ts.map +0 -1
- package/dist/src/engines/ocr/tesseract.js +0 -162
- package/dist/src/engines/ocr/tesseract.js.map +0 -1
- package/dist/src/engines/ocr/tesseract.test.d.ts +0 -2
- package/dist/src/engines/ocr/tesseract.test.d.ts.map +0 -1
- package/dist/src/engines/ocr/tesseract.test.js +0 -94
- package/dist/src/engines/ocr/tesseract.test.js.map +0 -1
- package/dist/src/engines/pdf/interface.d.ts +0 -84
- package/dist/src/engines/pdf/interface.d.ts.map +0 -1
- package/dist/src/engines/pdf/interface.js +0 -2
- package/dist/src/engines/pdf/interface.js.map +0 -1
- package/dist/src/engines/pdf/pdfium-renderer.d.ts +0 -31
- package/dist/src/engines/pdf/pdfium-renderer.d.ts.map +0 -1
- package/dist/src/engines/pdf/pdfium-renderer.js +0 -145
- package/dist/src/engines/pdf/pdfium-renderer.js.map +0 -1
- package/dist/src/engines/pdf/pdfium-renderer.test.d.ts +0 -2
- package/dist/src/engines/pdf/pdfium-renderer.test.d.ts.map +0 -1
- package/dist/src/engines/pdf/pdfium-renderer.test.js +0 -109
- package/dist/src/engines/pdf/pdfium-renderer.test.js.map +0 -1
- package/dist/src/engines/pdf/pdfjs.d.ts +0 -14
- package/dist/src/engines/pdf/pdfjs.d.ts.map +0 -1
- package/dist/src/engines/pdf/pdfjs.js +0 -804
- package/dist/src/engines/pdf/pdfjs.js.map +0 -1
- package/dist/src/engines/pdf/pdfjs.test.d.ts +0 -2
- package/dist/src/engines/pdf/pdfjs.test.d.ts.map +0 -1
- package/dist/src/engines/pdf/pdfjs.test.js +0 -225
- package/dist/src/engines/pdf/pdfjs.test.js.map +0 -1
- package/dist/src/engines/pdf/pdfjsImporter.d.ts +0 -5
- package/dist/src/engines/pdf/pdfjsImporter.d.ts.map +0 -1
- package/dist/src/engines/pdf/pdfjsImporter.js +0 -45
- package/dist/src/engines/pdf/pdfjsImporter.js.map +0 -1
- package/dist/src/index.d.ts +0 -3
- package/dist/src/index.d.ts.map +0 -1
- package/dist/src/index.js +0 -5
- package/dist/src/index.js.map +0 -1
- package/dist/src/lib.d.ts +0 -19
- package/dist/src/lib.d.ts.map +0 -1
- package/dist/src/lib.js +0 -17
- package/dist/src/lib.js.map +0 -1
- package/dist/src/output/json.d.ts +0 -10
- package/dist/src/output/json.d.ts.map +0 -1
- package/dist/src/output/json.js +0 -32
- package/dist/src/output/json.js.map +0 -1
- package/dist/src/output/json.test.d.ts +0 -2
- package/dist/src/output/json.test.d.ts.map +0 -1
- package/dist/src/output/json.test.js +0 -199
- package/dist/src/output/json.test.js.map +0 -1
- package/dist/src/output/text.d.ts +0 -10
- package/dist/src/output/text.d.ts.map +0 -1
- package/dist/src/output/text.js +0 -17
- package/dist/src/output/text.js.map +0 -1
- package/dist/src/output/text.test.d.ts +0 -2
- package/dist/src/output/text.test.d.ts.map +0 -1
- package/dist/src/output/text.test.js +0 -65
- package/dist/src/output/text.test.js.map +0 -1
- package/dist/src/processing/bbox.d.ts +0 -20
- package/dist/src/processing/bbox.d.ts.map +0 -1
- package/dist/src/processing/bbox.js +0 -258
- package/dist/src/processing/bbox.js.map +0 -1
- package/dist/src/processing/bbox.test.d.ts +0 -2
- package/dist/src/processing/bbox.test.d.ts.map +0 -1
- package/dist/src/processing/bbox.test.js +0 -334
- package/dist/src/processing/bbox.test.js.map +0 -1
- package/dist/src/processing/cleanText.d.ts +0 -6
- package/dist/src/processing/cleanText.d.ts.map +0 -1
- package/dist/src/processing/cleanText.js +0 -73
- package/dist/src/processing/cleanText.js.map +0 -1
- package/dist/src/processing/cleanText.test.d.ts +0 -2
- package/dist/src/processing/cleanText.test.d.ts.map +0 -1
- package/dist/src/processing/cleanText.test.js +0 -46
- package/dist/src/processing/cleanText.test.js.map +0 -1
- package/dist/src/processing/grid.d.ts +0 -7
- package/dist/src/processing/grid.d.ts.map +0 -1
- package/dist/src/processing/grid.js +0 -13
- package/dist/src/processing/grid.js.map +0 -1
- package/dist/src/processing/gridDebugLogger.d.ts +0 -206
- package/dist/src/processing/gridDebugLogger.d.ts.map +0 -1
- package/dist/src/processing/gridDebugLogger.js +0 -446
- package/dist/src/processing/gridDebugLogger.js.map +0 -1
- package/dist/src/processing/gridProjection.d.ts +0 -19
- package/dist/src/processing/gridProjection.d.ts.map +0 -1
- package/dist/src/processing/gridProjection.js +0 -1813
- package/dist/src/processing/gridProjection.js.map +0 -1
- package/dist/src/processing/gridProjection.test.d.ts +0 -2
- package/dist/src/processing/gridProjection.test.d.ts.map +0 -1
- package/dist/src/processing/gridProjection.test.js +0 -495
- package/dist/src/processing/gridProjection.test.js.map +0 -1
- package/dist/src/processing/gridVisualizer.d.ts +0 -14
- package/dist/src/processing/gridVisualizer.d.ts.map +0 -1
- package/dist/src/processing/gridVisualizer.js +0 -166
- package/dist/src/processing/gridVisualizer.js.map +0 -1
- package/dist/src/processing/markupUtils.d.ts +0 -7
- package/dist/src/processing/markupUtils.d.ts.map +0 -1
- package/dist/src/processing/markupUtils.js +0 -25
- package/dist/src/processing/markupUtils.js.map +0 -1
- package/dist/src/processing/markupUtils.test.d.ts +0 -2
- package/dist/src/processing/markupUtils.test.d.ts.map +0 -1
- package/dist/src/processing/markupUtils.test.js +0 -26
- package/dist/src/processing/markupUtils.test.js.map +0 -1
- package/dist/src/processing/ocrUtils.d.ts +0 -24
- package/dist/src/processing/ocrUtils.d.ts.map +0 -1
- package/dist/src/processing/ocrUtils.js +0 -79
- package/dist/src/processing/ocrUtils.js.map +0 -1
- package/dist/src/processing/octUtils.test.d.ts +0 -2
- package/dist/src/processing/octUtils.test.d.ts.map +0 -1
- package/dist/src/processing/octUtils.test.js +0 -72
- package/dist/src/processing/octUtils.test.js.map +0 -1
- package/dist/src/processing/searchItems.d.ts +0 -26
- package/dist/src/processing/searchItems.d.ts.map +0 -1
- package/dist/src/processing/searchItems.js +0 -93
- package/dist/src/processing/searchItems.js.map +0 -1
- package/dist/src/processing/searchItems.test.d.ts +0 -2
- package/dist/src/processing/searchItems.test.d.ts.map +0 -1
- package/dist/src/processing/searchItems.test.js +0 -84
- package/dist/src/processing/searchItems.test.js.map +0 -1
- package/dist/src/processing/textUtils.d.ts +0 -20
- package/dist/src/processing/textUtils.d.ts.map +0 -1
- package/dist/src/processing/textUtils.js +0 -142
- package/dist/src/processing/textUtils.js.map +0 -1
- package/dist/src/processing/textUtils.test.d.ts +0 -2
- package/dist/src/processing/textUtils.test.d.ts.map +0 -1
- package/dist/src/processing/textUtils.test.js +0 -45
- package/dist/src/processing/textUtils.test.js.map +0 -1
- package/dist/src/vendor/pdfjs/LICENSE +0 -177
- 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 +0 -3
- 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 +0 -3
- 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 +0 -4
- 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 +0 -36
- 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/jbig2.wasm +0 -0
- package/dist/src/vendor/pdfjs/openjpeg.wasm +0 -0
- package/dist/src/vendor/pdfjs/pdf.mjs +0 -33603
- package/dist/src/vendor/pdfjs/pdf.mjs.map +0 -1
- package/dist/src/vendor/pdfjs/pdf.sandbox.mjs +0 -4936
- package/dist/src/vendor/pdfjs/pdf.sandbox.mjs.map +0 -1
- package/dist/src/vendor/pdfjs/pdf.worker.mjs +0 -70100
- package/dist/src/vendor/pdfjs/pdf.worker.mjs.map +0 -1
- package/dist/src/vendor/pdfjs/qcms_bg.wasm +0 -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 +0 -27
- package/dist/src/vendor/pdfjs/standard_fonts/LICENSE_LIBERATION +0 -102
- 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/src/vendor/pdfjs/LICENSE +0 -177
- 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 +0 -3
- 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 +0 -3
- 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 +0 -4
- 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 +0 -36
- 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/jbig2.wasm +0 -0
- package/src/vendor/pdfjs/openjpeg.wasm +0 -0
- package/src/vendor/pdfjs/pdf.mjs +0 -33603
- package/src/vendor/pdfjs/pdf.mjs.map +0 -1
- package/src/vendor/pdfjs/pdf.sandbox.mjs +0 -4936
- package/src/vendor/pdfjs/pdf.sandbox.mjs.map +0 -1
- package/src/vendor/pdfjs/pdf.worker.mjs +0 -70100
- package/src/vendor/pdfjs/pdf.worker.mjs.map +0 -1
- package/src/vendor/pdfjs/qcms_bg.wasm +0 -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 +0 -27
- package/src/vendor/pdfjs/standard_fonts/LICENSE_LIBERATION +0 -102
- 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
|
@@ -1,334 +0,0 @@
|
|
|
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
|
|
@@ -1 +0,0 @@
|
|
|
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"}
|
|
@@ -1,6 +0,0 @@
|
|
|
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
|
|
@@ -1 +0,0 @@
|
|
|
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"}
|
|
@@ -1,73 +0,0 @@
|
|
|
1
|
-
/**
|
|
2
|
-
* Detect and remove margins from a single page.
|
|
3
|
-
* Removes:
|
|
4
|
-
* - Left margin (consistent leading whitespace)
|
|
5
|
-
* - Top margin (empty lines at start)
|
|
6
|
-
* - Bottom margin (empty lines at end)
|
|
7
|
-
* - Right margin (trailing whitespace on each line)
|
|
8
|
-
*/
|
|
9
|
-
function detectAndRemoveMarginOnPage(page) {
|
|
10
|
-
const lines = page.text.split("\n");
|
|
11
|
-
let minX = undefined;
|
|
12
|
-
let minY = undefined;
|
|
13
|
-
let maxY = undefined;
|
|
14
|
-
// Find margins
|
|
15
|
-
for (let index = 0; index < lines.length; index++) {
|
|
16
|
-
const line = lines[index];
|
|
17
|
-
if (line.trim().length === 0) {
|
|
18
|
-
continue;
|
|
19
|
-
}
|
|
20
|
-
// Find first non-whitespace character position (left margin)
|
|
21
|
-
const x = line.search(/\S/);
|
|
22
|
-
if (minX === undefined || x < minX) {
|
|
23
|
-
minX = x;
|
|
24
|
-
}
|
|
25
|
-
// First non-empty line (top margin)
|
|
26
|
-
if (minY === undefined) {
|
|
27
|
-
minY = index;
|
|
28
|
-
}
|
|
29
|
-
// Last non-empty line (bottom margin)
|
|
30
|
-
maxY = index;
|
|
31
|
-
}
|
|
32
|
-
// If page is entirely empty, just return
|
|
33
|
-
if (minX === undefined || minY === undefined || maxY === undefined) {
|
|
34
|
-
page.text = "";
|
|
35
|
-
return;
|
|
36
|
-
}
|
|
37
|
-
// Remove margins
|
|
38
|
-
const newLines = [];
|
|
39
|
-
for (let index = 0; index < lines.length; index++) {
|
|
40
|
-
// Skip lines before first content (top margin) or after last content (bottom margin)
|
|
41
|
-
if (index < minY || index > maxY) {
|
|
42
|
-
continue;
|
|
43
|
-
}
|
|
44
|
-
let line = lines[index];
|
|
45
|
-
// Remove trailing whitespace (right margin)
|
|
46
|
-
line = line.trimEnd();
|
|
47
|
-
// Remove left margin
|
|
48
|
-
newLines.push(line.slice(minX));
|
|
49
|
-
}
|
|
50
|
-
page.text = newLines.join("\n");
|
|
51
|
-
}
|
|
52
|
-
/**
|
|
53
|
-
* Detect and remove margins from all pages.
|
|
54
|
-
* Processes each page independently.
|
|
55
|
-
*/
|
|
56
|
-
function detectAndRemoveMargin(pages) {
|
|
57
|
-
for (const page of pages) {
|
|
58
|
-
detectAndRemoveMarginOnPage(page);
|
|
59
|
-
}
|
|
60
|
-
}
|
|
61
|
-
/**
|
|
62
|
-
* Clean raw text output - removes margins, null characters
|
|
63
|
-
*/
|
|
64
|
-
export function cleanRawText(pages, _config) {
|
|
65
|
-
// Remove margins (per-page)
|
|
66
|
-
detectAndRemoveMargin(pages);
|
|
67
|
-
// Remove null characters
|
|
68
|
-
for (const page of pages) {
|
|
69
|
-
// eslint-disable-next-line no-control-regex
|
|
70
|
-
page.text = page.text.replace(/\u0000/g, " ");
|
|
71
|
-
}
|
|
72
|
-
}
|
|
73
|
-
//# sourceMappingURL=cleanText.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"cleanText.js","sourceRoot":"","sources":["../../../src/processing/cleanText.ts"],"names":[],"mappings":"AAEA;;;;;;;GAOG;AACH,SAAS,2BAA2B,CAAC,IAAgB;IACnD,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAEpC,IAAI,IAAI,GAAuB,SAAS,CAAC;IACzC,IAAI,IAAI,GAAuB,SAAS,CAAC;IACzC,IAAI,IAAI,GAAuB,SAAS,CAAC;IAEzC,eAAe;IACf,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAClD,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;QAC1B,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAC7B,SAAS;QACX,CAAC;QAED,6DAA6D;QAC7D,MAAM,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;QAC5B,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC;YACnC,IAAI,GAAG,CAAC,CAAC;QACX,CAAC;QAED,oCAAoC;QACpC,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACvB,IAAI,GAAG,KAAK,CAAC;QACf,CAAC;QAED,sCAAsC;QACtC,IAAI,GAAG,KAAK,CAAC;IACf,CAAC;IAED,yCAAyC;IACzC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;QACnE,IAAI,CAAC,IAAI,GAAG,EAAE,CAAC;QACf,OAAO;IACT,CAAC;IAED,iBAAiB;IACjB,MAAM,QAAQ,GAAa,EAAE,CAAC;IAC9B,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QAClD,qFAAqF;QACrF,IAAI,KAAK,GAAG,IAAI,IAAI,KAAK,GAAG,IAAI,EAAE,CAAC;YACjC,SAAS;QACX,CAAC;QAED,IAAI,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;QAExB,4CAA4C;QAC5C,IAAI,GAAG,IAAI,CAAC,OAAO,EAAE,CAAC;QAEtB,qBAAqB;QACrB,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC;IAClC,CAAC;IAED,IAAI,CAAC,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAClC,CAAC;AAED;;;GAGG;AACH,SAAS,qBAAqB,CAAC,KAAmB;IAChD,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,2BAA2B,CAAC,IAAI,CAAC,CAAC;IACpC,CAAC;AACH,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,CAAC,KAAmB,EAAE,OAAwB;IACxE,4BAA4B;IAC5B,qBAAqB,CAAC,KAAK,CAAC,CAAC;IAE7B,yBAAyB;IACzB,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,4CAA4C;QAC5C,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC;IAChD,CAAC;AACH,CAAC"}
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"cleanText.test.d.ts","sourceRoot":"","sources":["../../../src/processing/cleanText.test.ts"],"names":[],"mappings":""}
|
|
@@ -1,46 +0,0 @@
|
|
|
1
|
-
import { describe, expect, it } from "vitest";
|
|
2
|
-
import { cleanRawText } from "./cleanText.js";
|
|
3
|
-
import { DEFAULT_CONFIG } from "../core/config.js";
|
|
4
|
-
const mockPages = [
|
|
5
|
-
{
|
|
6
|
-
// Normal page with margins
|
|
7
|
-
pageNum: 1,
|
|
8
|
-
width: 612,
|
|
9
|
-
height: 792,
|
|
10
|
-
text: " Hello World \n Foo Bar \n",
|
|
11
|
-
textItems: [],
|
|
12
|
-
},
|
|
13
|
-
{
|
|
14
|
-
// Empty page
|
|
15
|
-
pageNum: 2,
|
|
16
|
-
width: 612,
|
|
17
|
-
height: 792,
|
|
18
|
-
text: " \n \n",
|
|
19
|
-
textItems: [],
|
|
20
|
-
},
|
|
21
|
-
{
|
|
22
|
-
// Single line
|
|
23
|
-
pageNum: 3,
|
|
24
|
-
width: 612,
|
|
25
|
-
height: 792,
|
|
26
|
-
text: " Hello ",
|
|
27
|
-
textItems: [],
|
|
28
|
-
},
|
|
29
|
-
];
|
|
30
|
-
const expectedTexts = [
|
|
31
|
-
// minX=3, minY=0, maxY=1 → slice(3) + trimEnd
|
|
32
|
-
"Hello World\nFoo Bar",
|
|
33
|
-
// entirely empty → ""
|
|
34
|
-
"",
|
|
35
|
-
// minX=2, minY=0, maxY=0 → slice(2) + trimEnd
|
|
36
|
-
"Hello",
|
|
37
|
-
];
|
|
38
|
-
describe("test cleanText", () => {
|
|
39
|
-
it("test cleanRawText", () => {
|
|
40
|
-
cleanRawText(mockPages, DEFAULT_CONFIG);
|
|
41
|
-
for (let i = 0; i < mockPages.length; i++) {
|
|
42
|
-
expect(mockPages[i].text).toBe(expectedTexts[i]);
|
|
43
|
-
}
|
|
44
|
-
});
|
|
45
|
-
});
|
|
46
|
-
//# sourceMappingURL=cleanText.test.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"cleanText.test.js","sourceRoot":"","sources":["../../../src/processing/cleanText.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,EAAE,EAAE,EAAE,MAAM,QAAQ,CAAC;AAE9C,OAAO,EAAE,YAAY,EAAE,MAAM,gBAAgB,CAAC;AAC9C,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AAEnD,MAAM,SAAS,GAAiB;IAC9B;QACE,2BAA2B;QAC3B,OAAO,EAAE,CAAC;QACV,KAAK,EAAE,GAAG;QACV,MAAM,EAAE,GAAG;QACX,IAAI,EAAE,oCAAoC;QAC1C,SAAS,EAAE,EAAE;KACd;IACD;QACE,aAAa;QACb,OAAO,EAAE,CAAC;QACV,KAAK,EAAE,GAAG;QACV,MAAM,EAAE,GAAG;QACX,IAAI,EAAE,YAAY;QAClB,SAAS,EAAE,EAAE;KACd;IACD;QACE,cAAc;QACd,OAAO,EAAE,CAAC;QACV,KAAK,EAAE,GAAG;QACV,MAAM,EAAE,GAAG;QACX,IAAI,EAAE,WAAW;QACjB,SAAS,EAAE,EAAE;KACd;CACF,CAAC;AAEF,MAAM,aAAa,GAAa;IAC9B,8CAA8C;IAC9C,sBAAsB;IACtB,sBAAsB;IACtB,EAAE;IACF,8CAA8C;IAC9C,OAAO;CACR,CAAC;AAEF,QAAQ,CAAC,gBAAgB,EAAE,GAAG,EAAE;IAC9B,EAAE,CAAC,mBAAmB,EAAE,GAAG,EAAE;QAC3B,YAAY,CAAC,SAAS,EAAE,cAAc,CAAC,CAAC;QACxC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,SAAS,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;YAC1C,MAAM,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC;QACnD,CAAC;IACH,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}
|
|
@@ -1,7 +0,0 @@
|
|
|
1
|
-
import { PageData } from "../engines/pdf/interface.js";
|
|
2
|
-
import { ParsedPage, LiteParseConfig } from "../core/types.js";
|
|
3
|
-
/**
|
|
4
|
-
* Projects text items onto a grid for spatial text extraction
|
|
5
|
-
*/
|
|
6
|
-
export declare function projectPagesToGrid(pages: PageData[], config?: Partial<LiteParseConfig>): Promise<ParsedPage[]>;
|
|
7
|
-
//# sourceMappingURL=grid.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"grid.d.ts","sourceRoot":"","sources":["../../../src/processing/grid.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,6BAA6B,CAAC;AACvD,OAAO,EAAE,UAAU,EAAE,eAAe,EAAE,MAAM,kBAAkB,CAAC;AAI/D;;GAEG;AACH,wBAAsB,kBAAkB,CACtC,KAAK,EAAE,QAAQ,EAAE,EACjB,MAAM,CAAC,EAAE,OAAO,CAAC,eAAe,CAAC,GAChC,OAAO,CAAC,UAAU,EAAE,CAAC,CAOvB"}
|
|
@@ -1,13 +0,0 @@
|
|
|
1
|
-
import { projectPagesToGrid as projectPagesToGridComplete } from "./gridProjection.js";
|
|
2
|
-
import { DEFAULT_CONFIG } from "../core/config.js";
|
|
3
|
-
/**
|
|
4
|
-
* Projects text items onto a grid for spatial text extraction
|
|
5
|
-
*/
|
|
6
|
-
export async function projectPagesToGrid(pages, config) {
|
|
7
|
-
const fullConfig = {
|
|
8
|
-
...DEFAULT_CONFIG,
|
|
9
|
-
...config,
|
|
10
|
-
};
|
|
11
|
-
return projectPagesToGridComplete(pages, fullConfig);
|
|
12
|
-
}
|
|
13
|
-
//# sourceMappingURL=grid.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"grid.js","sourceRoot":"","sources":["../../../src/processing/grid.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,kBAAkB,IAAI,0BAA0B,EAAE,MAAM,qBAAqB,CAAC;AACvF,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AAEnD;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,KAAiB,EACjB,MAAiC;IAEjC,MAAM,UAAU,GAAG;QACjB,GAAG,cAAc;QACjB,GAAG,MAAM;KACV,CAAC;IAEF,OAAO,0BAA0B,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;AACvD,CAAC"}
|