UNPKG

pomljs

Version:

Prompt Orchestration Markup Language

1 lines 4.1 kB
{"version":3,"file":"pdf.cjs","sources":["../../.build/util/pdf.js"],"sourcesContent":["const log = console.log;\nconsole.log = (m, ...a) => /Cannot polyfill `(DOMMatrix|Path2D)`/.test(m) ? null : log(m, ...a);\n// import * as PDFJS from 'pdfjs-dist';\n// import PDFJS from 'pdfjs-dist/legacy/build/pdf.js';\nimport * as PDFJS from 'pdfjs-dist/legacy/build/pdf.js';\nlet pdfjs = PDFJS;\nif (PDFJS.GlobalWorkerOptions === undefined) {\n // in esm\n pdfjs = PDFJS.default;\n}\nelse {\n // in commonjs\n pdfjs = PDFJS;\n}\npdfjs.GlobalWorkerOptions.workerSrc = 'pdfjs-dist/legacy/build/pdf.worker.js';\nconsole.log = log; // restore original console.log\nexport async function getNumPages(pdfBuffer) {\n const uint8Array = pdfBuffer instanceof ArrayBuffer ? new Uint8Array(pdfBuffer) : new Uint8Array(pdfBuffer);\n const loadingTask = pdfjs.getDocument({ data: uint8Array });\n const pdfDocument = await loadingTask.promise;\n return pdfDocument.numPages;\n}\nexport async function pdfParse(pdfBuffer, maxPages) {\n const uint8Array = pdfBuffer instanceof ArrayBuffer ? new Uint8Array(pdfBuffer) : new Uint8Array(pdfBuffer);\n const loadingTask = pdfjs.getDocument({ data: uint8Array });\n const pdfDocument = await loadingTask.promise;\n let fullTexts = [];\n if (maxPages == undefined) {\n maxPages = pdfDocument.numPages;\n }\n else {\n maxPages = Math.min(maxPages, pdfDocument.numPages);\n }\n for (let pageNum = 1; pageNum <= maxPages; pageNum++) {\n const page = await pdfDocument.getPage(pageNum);\n const pageText = await extractTextFromPage(page);\n fullTexts.push(pageText);\n }\n return fullTexts.join('\\n\\n');\n}\nasync function extractTextFromPage(page) {\n const textContent = await page.getTextContent();\n let lastY, text = '';\n for (let item of textContent.items) {\n if (lastY == item.transform[5] || !lastY) {\n text += item.str;\n }\n else {\n text += '\\n' + item.str;\n }\n lastY = item.transform[5];\n }\n return text;\n}\n//# sourceMappingURL=pdf.js.map"],"names":["PDFJS"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;AAAA,MAAM,GAAG,GAAG,OAAO,CAAC,GAAG;AACvB,OAAO,CAAC,GAAG,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,KAAK,sCAAsC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,GAAG,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC;AAI/F,IAAI,KAAK,GAAGA,gBAAK;AACjB,IAAIA,gBAAK,CAAC,mBAAmB,KAAK,SAAS,EAAE;AAC7C;AACA,IAAI,KAAK,GAAGA,gBAAK,CAAC,OAAO;AACzB;AACA,KAAK;AACL;AACA,IAAI,KAAK,GAAGA,gBAAK;AACjB;AACA,KAAK,CAAC,mBAAmB,CAAC,SAAS,GAAG,uCAAuC;AAC7E,OAAO,CAAC,GAAG,GAAG,GAAG,CAAC;AACX,eAAe,WAAW,CAAC,SAAS,EAAE;AAC7C,IAAI,MAAM,UAAU,GAAG,SAAS,YAAY,WAAW,GAAG,IAAI,UAAU,CAAC,SAAS,CAAC,GAAG,IAAI,UAAU,CAAC,SAAS,CAAC;AAC/G,IAAI,MAAM,WAAW,GAAG,KAAK,CAAC,WAAW,CAAC,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC;AAC/D,IAAI,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO;AACjD,IAAI,OAAO,WAAW,CAAC,QAAQ;AAC/B;AACO,eAAe,QAAQ,CAAC,SAAS,EAAE,QAAQ,EAAE;AACpD,IAAI,MAAM,UAAU,GAAG,SAAS,YAAY,WAAW,GAAG,IAAI,UAAU,CAAC,SAAS,CAAC,GAAG,IAAI,UAAU,CAAC,SAAS,CAAC;AAC/G,IAAI,MAAM,WAAW,GAAG,KAAK,CAAC,WAAW,CAAC,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC;AAC/D,IAAI,MAAM,WAAW,GAAG,MAAM,WAAW,CAAC,OAAO;AACjD,IAAI,IAAI,SAAS,GAAG,EAAE;AACtB,IAAI,IAAI,QAAQ,IAAI,SAAS,EAAE;AAC/B,QAAQ,QAAQ,GAAG,WAAW,CAAC,QAAQ;AACvC;AACA,SAAS;AACT,QAAQ,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,EAAE,WAAW,CAAC,QAAQ,CAAC;AAC3D;AACA,IAAI,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,EAAE,EAAE;AAC1D,QAAQ,MAAM,IAAI,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC,OAAO,CAAC;AACvD,QAAQ,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,IAAI,CAAC;AACxD,QAAQ,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC;AAChC;AACA,IAAI,OAAO,SAAS,CAAC,IAAI,CAAC,MAAM,CAAC;AACjC;AACA,eAAe,mBAAmB,CAAC,IAAI,EAAE;AACzC,IAAI,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE;AACnD,IAAI,IAAI,KAAK,EAAE,IAAI,GAAG,EAAE;AACxB,IAAI,KAAK,IAAI,IAAI,IAAI,WAAW,CAAC,KAAK,EAAE;AACxC,QAAQ,IAAI,KAAK,IAAI,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,EAAE;AAClD,YAAY,IAAI,IAAI,IAAI,CAAC,GAAG;AAC5B;AACA,aAAa;AACb,YAAY,IAAI,IAAI,IAAI,GAAG,IAAI,CAAC,GAAG;AACnC;AACA,QAAQ,KAAK,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC;AACjC;AACA,IAAI,OAAO,IAAI;AACf;;;;;"}