unpdf 0.5.4 → 0.6.1

package/dist/index.worker.d.ts

@@ -1,10 +0,0 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.js';
-export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.js';
-import 'pdfjs-dist/types/src/display/api';
-import 'pdfjs-dist';
-
-declare const getPDFMeta: typeof getPDFMeta$1;
-declare const extractPDFText: typeof extractPDFText$1;
-declare const getImagesFromPage: typeof getImagesFromPage$1;
-
-export { extractPDFText, getImagesFromPage, getPDFMeta };

package/dist/index.worker.mjs

@@ -1,13 +0,0 @@
-export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.ab6a2535.mjs';
-
-const getPDFMeta = async () => {
-  throw new Error("Not implemented in worker context yet");
-};
-const extractPDFText = async () => {
-  throw new Error("Not implemented in worker context yet");
-};
-const getImagesFromPage = async () => {
-  throw new Error("Not implemented in worker context yet");
-};
-
-export { extractPDFText, getImagesFromPage, getPDFMeta };

package/dist/shared/unpdf.3dcacb97.mjs

@@ -1,49 +0,0 @@
-import { i as isPDFDocumentProxy, g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.ab6a2535.mjs';
-
-async function getPDFMeta(data) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const meta = await pdf.getMetadata().catch(() => null);
-  return {
-    info: meta?.info ?? {},
-    metadata: meta?.metadata?.getAll() ?? {}
-  };
-}
-
-async function extractPDFText(data, options = {}) {
-  const { mergePages = false } = { ...options };
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const texts = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
-  );
-  const filteredTexts = texts.filter(Boolean);
-  return {
-    totalPages: pdf.numPages,
-    text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
-  };
-}
-async function getPageText(pdf, pageNumber) {
-  const page = await pdf.getPage(pageNumber);
-  const content = await page.getTextContent();
-  const items = content.items;
-  if (items.length === 0)
-    return;
-  return items.map(({ str }) => str).join("\n");
-}
-
-async function getImagesFromPage(data, pageNumber) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const page = await pdf.getPage(pageNumber);
-  const operatorList = await page.getOperatorList();
-  const { OPS } = await getResolvedPDFJS();
-  const images = [];
-  for (const op of operatorList.fnArray) {
-    if (op !== OPS.paintImageXObject)
-      continue;
-    const image = await page.objs.get(operatorList.argsArray[op][0]);
-    if (image.data)
-      images.push(image.data.buffer);
-  }
-  return images;
-}
-
-export { getImagesFromPage as a, extractPDFText as e, getPDFMeta as g };

package/dist/shared/unpdf.42987ff1.cjs

@@ -1,52 +0,0 @@
-'use strict';
-
-let resolvedModule;
-async function getDocumentProxy(data, options = {}) {
-  const { getDocument } = await getResolvedPDFJS();
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    isEvalSupported: false,
-    useSystemFonts: true,
-    ...options
-  }).promise;
-  return pdf;
-}
-async function defineUnPDFConfig(options) {
-  const { pdfjs } = { ...options };
-  if (pdfjs)
-    await resolvePDFJSImports(pdfjs);
-}
-async function getResolvedPDFJS() {
-  if (!resolvedModule)
-    await resolvePDFJSImports();
-  return resolvedModule;
-}
-async function resolvePDFJSImports(pdfjs) {
-  if (resolvedModule)
-    return;
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-      return;
-    } catch (error) {
-      throw new Error("Resolving PDF.js failed. Please check the provided configuration.");
-    }
-  }
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    resolvedModule = mod;
-  } catch (error) {
-    throw new Error("PDF.js is not available. Please add the package as a dependency.");
-  }
-}
-function isPDFDocumentProxy(data) {
-  return typeof data === "object" && data !== null && "_pdfInfo" in data;
-}
-
-exports.defineUnPDFConfig = defineUnPDFConfig;
-exports.getDocumentProxy = getDocumentProxy;
-exports.getResolvedPDFJS = getResolvedPDFJS;
-exports.isPDFDocumentProxy = isPDFDocumentProxy;
-exports.resolvePDFJSImports = resolvePDFJSImports;

package/dist/shared/unpdf.6363738c.d.cts

@@ -1,43 +0,0 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
-    info: Record<string, any>;
-    metadata: Record<string, any>;
-}>;
-
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
-
-interface UnPDFConfiguration {
-    /**
-     * By default, UnPDF will use the latest version of PDF.js. If you want to
-     * use an older version or the legacy build, set a promise that resolves to
-     * the PDF.js module.
-     *
-     * @example
-     * () => import('pdfjs-dist/legacy/build/pdf.js')
-     */
-    pdfjs?: () => Promise<typeof PDFJS>;
-}
-
-/**
- * Returns a PDFDocumentProxy instance from a given binary data.
- *
- * Applies the following defaults:
- * - `useWorkerFetch: false`
- * - `isEvalSupported: false`
- * - `useSystemFonts: true`
- */
-declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
-
-export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/unpdf.6363738c.d.mts

@@ -1,43 +0,0 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
-    info: Record<string, any>;
-    metadata: Record<string, any>;
-}>;
-
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
-
-interface UnPDFConfiguration {
-    /**
-     * By default, UnPDF will use the latest version of PDF.js. If you want to
-     * use an older version or the legacy build, set a promise that resolves to
-     * the PDF.js module.
-     *
-     * @example
-     * () => import('pdfjs-dist/legacy/build/pdf.js')
-     */
-    pdfjs?: () => Promise<typeof PDFJS>;
-}
-
-/**
- * Returns a PDFDocumentProxy instance from a given binary data.
- *
- * Applies the following defaults:
- * - `useWorkerFetch: false`
- * - `isEvalSupported: false`
- * - `useSystemFonts: true`
- */
-declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
-
-export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/unpdf.6363738c.d.ts

@@ -1,43 +0,0 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
-    info: Record<string, any>;
-    metadata: Record<string, any>;
-}>;
-
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
-
-interface UnPDFConfiguration {
-    /**
-     * By default, UnPDF will use the latest version of PDF.js. If you want to
-     * use an older version or the legacy build, set a promise that resolves to
-     * the PDF.js module.
-     *
-     * @example
-     * () => import('pdfjs-dist/legacy/build/pdf.js')
-     */
-    pdfjs?: () => Promise<typeof PDFJS>;
-}
-
-/**
- * Returns a PDFDocumentProxy instance from a given binary data.
- *
- * Applies the following defaults:
- * - `useWorkerFetch: false`
- * - `isEvalSupported: false`
- * - `useSystemFonts: true`
- */
-declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
-
-export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/unpdf.ab6a2535.mjs

@@ -1,46 +0,0 @@
-let resolvedModule;
-async function getDocumentProxy(data, options = {}) {
-  const { getDocument } = await getResolvedPDFJS();
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    isEvalSupported: false,
-    useSystemFonts: true,
-    ...options
-  }).promise;
-  return pdf;
-}
-async function defineUnPDFConfig(options) {
-  const { pdfjs } = { ...options };
-  if (pdfjs)
-    await resolvePDFJSImports(pdfjs);
-}
-async function getResolvedPDFJS() {
-  if (!resolvedModule)
-    await resolvePDFJSImports();
-  return resolvedModule;
-}
-async function resolvePDFJSImports(pdfjs) {
-  if (resolvedModule)
-    return;
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-      return;
-    } catch (error) {
-      throw new Error("Resolving PDF.js failed. Please check the provided configuration.");
-    }
-  }
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    resolvedModule = mod;
-  } catch (error) {
-    throw new Error("PDF.js is not available. Please add the package as a dependency.");
-  }
-}
-function isPDFDocumentProxy(data) {
-  return typeof data === "object" && data !== null && "_pdfInfo" in data;
-}
-
-export { getResolvedPDFJS as a, defineUnPDFConfig as d, getDocumentProxy as g, isPDFDocumentProxy as i, resolvePDFJSImports as r };

package/dist/shared/unpdf.eba9a3dc.cjs

@@ -1,53 +0,0 @@
-'use strict';
-
-const utils = require('./unpdf.42987ff1.cjs');
-
-async function getPDFMeta(data) {
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const meta = await pdf.getMetadata().catch(() => null);
-  return {
-    info: meta?.info ?? {},
-    metadata: meta?.metadata?.getAll() ?? {}
-  };
-}
-
-async function extractPDFText(data, options = {}) {
-  const { mergePages = false } = { ...options };
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const texts = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
-  );
-  const filteredTexts = texts.filter(Boolean);
-  return {
-    totalPages: pdf.numPages,
-    text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
-  };
-}
-async function getPageText(pdf, pageNumber) {
-  const page = await pdf.getPage(pageNumber);
-  const content = await page.getTextContent();
-  const items = content.items;
-  if (items.length === 0)
-    return;
-  return items.map(({ str }) => str).join("\n");
-}
-
-async function getImagesFromPage(data, pageNumber) {
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const page = await pdf.getPage(pageNumber);
-  const operatorList = await page.getOperatorList();
-  const { OPS } = await utils.getResolvedPDFJS();
-  const images = [];
-  for (const op of operatorList.fnArray) {
-    if (op !== OPS.paintImageXObject)
-      continue;
-    const image = await page.objs.get(operatorList.argsArray[op][0]);
-    if (image.data)
-      images.push(image.data.buffer);
-  }
-  return images;
-}
-
-exports.extractPDFText = extractPDFText;
-exports.getImagesFromPage = getImagesFromPage;
-exports.getPDFMeta = getPDFMeta;