unpdf 0.3.0 → 0.4.1

package/README.md

@@ -10,6 +10,7 @@ This library is also intended as a modern alternative to the unmaintained [`pdf-
 
 - 🏗️ Conditional exports for Browser, Node and worker environments
 - 💬 Extract text from PDFs
+- 🧱 Use custom PDF.js build
 
 ## Installation
 
@@ -38,8 +39,30 @@ const { totalPages, info, metadata, text } = await decodePDFText(
 )
 ```
 
+## Config
+
+```ts
+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?: () => typeof PDFJS
+}
+```
+
 ## Methods
 
+### `defineUnPDFConfig`
+
+```ts
+function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>
+```
+
 ### `decodePDFText`
 
 ```ts

package/dist/index.node.cjs

@@ -1,15 +1,17 @@
 'use strict';
 
-const image = require('./shared/unpdf.970262b7.cjs');
+const image = require('./shared/unpdf.ff73f2a9.cjs');
+const utils = require('./shared/unpdf.54007d6f.cjs');
 
 const decodePDFText = async (...args) => {
-  await image.resolvePDFJSNodeImports();
+  await utils.resolvePDFJSImports();
   return await image.decodePDFText(...args);
 };
 const getImagesFromPage = async (...args) => {
-  await image.resolvePDFJSNodeImports();
+  await utils.resolvePDFJSImports();
   return await image.getImagesFromPage(...args);
 };
 
+exports.defineUnPDFConfig = utils.defineUnPDFConfig;
 exports.decodePDFText = decodePDFText;
 exports.getImagesFromPage = getImagesFromPage;

package/dist/index.node.d.cts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.cjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.cjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.cjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.node.d.mts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.mjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.mjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.mjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.node.d.ts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.js';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.js';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.js';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.node.mjs

@@ -1,11 +1,13 @@
-import { a as resolvePDFJSNodeImports, d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72a42ec0.mjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.0f3aa1eb.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.5b4a24ed.mjs';
+export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
 
 const decodePDFText = async (...args) => {
-  await resolvePDFJSNodeImports();
+  await resolvePDFJSImports();
   return await decodePDFText$1(...args);
 };
 const getImagesFromPage = async (...args) => {
-  await resolvePDFJSNodeImports();
+  await resolvePDFJSImports();
   return await getImagesFromPage$1(...args);
 };
 

package/dist/index.web.cjs

@@ -1,15 +1,17 @@
 'use strict';
 
-const image = require('./shared/unpdf.970262b7.cjs');
+const image = require('./shared/unpdf.ff73f2a9.cjs');
+const utils = require('./shared/unpdf.54007d6f.cjs');
 
 const decodePDFText = async (...args) => {
-  await image.resolvePDFJSWebImports();
+  await utils.resolvePDFJSImports();
   return await image.decodePDFText(...args);
 };
 const getImagesFromPage = async (...args) => {
-  await image.resolvePDFJSWebImports();
+  await utils.resolvePDFJSImports();
   return await image.getImagesFromPage(...args);
 };
 
+exports.defineUnPDFConfig = utils.defineUnPDFConfig;
 exports.decodePDFText = decodePDFText;
 exports.getImagesFromPage = getImagesFromPage;

package/dist/index.web.d.cts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.cjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.cjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.cjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.web.d.mts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.mjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.mjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.mjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.web.d.ts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.js';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.js';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.js';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.web.mjs

@@ -1,11 +1,13 @@
-import { r as resolvePDFJSWebImports, d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72a42ec0.mjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.0f3aa1eb.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.5b4a24ed.mjs';
+export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
 
 const decodePDFText = async (...args) => {
-  await resolvePDFJSWebImports();
+  await resolvePDFJSImports();
   return await decodePDFText$1(...args);
 };
 const getImagesFromPage = async (...args) => {
-  await resolvePDFJSWebImports();
+  await resolvePDFJSImports();
   return await getImagesFromPage$1(...args);
 };
 

package/dist/index.worker.cjs

@@ -1,5 +1,7 @@
 'use strict';
 
+const utils = require('./shared/unpdf.54007d6f.cjs');
+
 const decodePDFText = async () => {
   throw new Error("Not implemented in worker context yet");
 };
@@ -7,5 +9,6 @@ const getImagesFromPage = async () => {
   throw new Error("Not implemented in worker context yet");
 };
 
+exports.defineUnPDFConfig = utils.defineUnPDFConfig;
 exports.decodePDFText = decodePDFText;
 exports.getImagesFromPage = getImagesFromPage;

package/dist/index.worker.d.cts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.cjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.cjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.cjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.worker.d.mts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.mjs';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.mjs';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.mjs';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.worker.d.ts

@@ -1,4 +1,6 @@
-import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.5c39ffe2.js';
+import { d as decodePDFText$1, g as getImagesFromPage$1 } from './shared/unpdf.72be7352.js';
+export { a as defineUnPDFConfig } from './shared/unpdf.72be7352.js';
+import 'pdfjs-dist';
 
 declare const decodePDFText: typeof decodePDFText$1;
 declare const getImagesFromPage: typeof getImagesFromPage$1;

package/dist/index.worker.mjs

@@ -1,3 +1,5 @@
+export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
+
 const decodePDFText = async () => {
   throw new Error("Not implemented in worker context yet");
 };

package/dist/shared/{unpdf.72a42ec0.mjs → unpdf.0f3aa1eb.mjs}

@@ -1,41 +1,4 @@
-let instance;
-async function getDocumentProxy(data) {
-  const { getDocument } = instance;
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    useSystemFonts: true,
-    isEvalSupported: false
-  }).promise;
-  return pdf;
-}
-function getResolvedPDFJSInstance() {
-  return instance;
-}
-async function resolvePDFJSWebImports() {
-  if (instance)
-    return;
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    instance = mod;
-  } catch (error) {
-    throw new Error(
-      "PDF.js is not available. Please add the package as a dependency."
-    );
-  }
-}
-async function resolvePDFJSNodeImports() {
-  if (instance)
-    return;
-  try {
-    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
-    instance = mod;
-  } catch (error) {
-    throw new Error(
-      "PDF.js is not available. Please add the package as a dependency."
-    );
-  }
-}
+import { g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.5b4a24ed.mjs';
 
 async function decodePDFText(data, { mergePages = false } = {}) {
   const pdf = await getDocumentProxy(data);
@@ -64,7 +27,7 @@ async function getImagesFromPage(data, pageNumber) {
   const pdf = await getDocumentProxy(data);
   const page = await pdf.getPage(pageNumber);
   const operatorList = await page.getOperatorList();
-  const { OPS } = getResolvedPDFJSInstance();
+  const { OPS } = getResolvedPDFJS();
   const images = [];
   for (const op of operatorList.fnArray) {
     if (op !== OPS.paintImageXObject)
@@ -76,4 +39,4 @@ async function getImagesFromPage(data, pageNumber) {
   return images;
 }
 
-export { resolvePDFJSNodeImports as a, decodePDFText as d, getImagesFromPage as g, resolvePDFJSWebImports as r };
+export { decodePDFText as d, getImagesFromPage as g };

package/dist/shared/unpdf.54007d6f.cjs

@@ -0,0 +1,41 @@
+'use strict';
+
+let resolvedModule;
+async function getDocumentProxy(data) {
+  const { getDocument } = getResolvedPDFJS();
+  const pdf = await getDocument({
+    data,
+    useWorkerFetch: false,
+    useSystemFonts: true,
+    isEvalSupported: false
+  }).promise;
+  return pdf;
+}
+async function defineUnPDFConfig({ pdfjs }) {
+  if (pdfjs) {
+    try {
+      const { default: mod } = await pdfjs();
+      resolvedModule = mod;
+    } catch (error) {
+      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
+    }
+  }
+}
+function getResolvedPDFJS() {
+  return resolvedModule;
+}
+async function resolvePDFJSImports() {
+  if (resolvedModule)
+    return;
+  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.");
+  }
+}
+
+exports.defineUnPDFConfig = defineUnPDFConfig;
+exports.getDocumentProxy = getDocumentProxy;
+exports.getResolvedPDFJS = getResolvedPDFJS;
+exports.resolvePDFJSImports = resolvePDFJSImports;

package/dist/shared/unpdf.5b4a24ed.mjs

@@ -0,0 +1,36 @@
+let resolvedModule;
+async function getDocumentProxy(data) {
+  const { getDocument } = getResolvedPDFJS();
+  const pdf = await getDocument({
+    data,
+    useWorkerFetch: false,
+    useSystemFonts: true,
+    isEvalSupported: false
+  }).promise;
+  return pdf;
+}
+async function defineUnPDFConfig({ pdfjs }) {
+  if (pdfjs) {
+    try {
+      const { default: mod } = await pdfjs();
+      resolvedModule = mod;
+    } catch (error) {
+      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
+    }
+  }
+}
+function getResolvedPDFJS() {
+  return resolvedModule;
+}
+async function resolvePDFJSImports() {
+  if (resolvedModule)
+    return;
+  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.");
+  }
+}
+
+export { getResolvedPDFJS as a, defineUnPDFConfig as d, getDocumentProxy as g, resolvePDFJSImports as r };

package/dist/shared/unpdf.72be7352.d.cts

@@ -0,0 +1,29 @@
+import PDFJS from 'pdfjs-dist';
+
+interface PDFContent {
+    totalPages: number;
+    info?: Record<string, any>;
+    metadata?: any;
+    text: string | string[];
+}
+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>;
+}
+
+declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
+    mergePages?: boolean;
+}): Promise<PDFContent>;
+
+declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
+
+declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+
+export { defineUnPDFConfig as a, decodePDFText as d, getImagesFromPage as g };

package/dist/shared/unpdf.72be7352.d.mts

@@ -0,0 +1,29 @@
+import PDFJS from 'pdfjs-dist';
+
+interface PDFContent {
+    totalPages: number;
+    info?: Record<string, any>;
+    metadata?: any;
+    text: string | string[];
+}
+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>;
+}
+
+declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
+    mergePages?: boolean;
+}): Promise<PDFContent>;
+
+declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
+
+declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+
+export { defineUnPDFConfig as a, decodePDFText as d, getImagesFromPage as g };

package/dist/shared/unpdf.72be7352.d.ts

@@ -0,0 +1,29 @@
+import PDFJS from 'pdfjs-dist';
+
+interface PDFContent {
+    totalPages: number;
+    info?: Record<string, any>;
+    metadata?: any;
+    text: string | string[];
+}
+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>;
+}
+
+declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
+    mergePages?: boolean;
+}): Promise<PDFContent>;
+
+declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
+
+declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+
+export { defineUnPDFConfig as a, decodePDFText as d, getImagesFromPage as g };

package/dist/shared/{unpdf.970262b7.cjs → unpdf.ff73f2a9.cjs}

@@ -1,46 +1,9 @@
 'use strict';
 
-let instance;
-async function getDocumentProxy(data) {
-  const { getDocument } = instance;
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    useSystemFonts: true,
-    isEvalSupported: false
-  }).promise;
-  return pdf;
-}
-function getResolvedPDFJSInstance() {
-  return instance;
-}
-async function resolvePDFJSWebImports() {
-  if (instance)
-    return;
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    instance = mod;
-  } catch (error) {
-    throw new Error(
-      "PDF.js is not available. Please add the package as a dependency."
-    );
-  }
-}
-async function resolvePDFJSNodeImports() {
-  if (instance)
-    return;
-  try {
-    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
-    instance = mod;
-  } catch (error) {
-    throw new Error(
-      "PDF.js is not available. Please add the package as a dependency."
-    );
-  }
-}
+const utils = require('./unpdf.54007d6f.cjs');
 
 async function decodePDFText(data, { mergePages = false } = {}) {
-  const pdf = await getDocumentProxy(data);
+  const pdf = await utils.getDocumentProxy(data);
   const meta = await pdf.getMetadata().catch(() => null);
   const texts = await Promise.all(
     Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
@@ -63,10 +26,10 @@ async function getPageText(pdf, pageNumber) {
 }
 
 async function getImagesFromPage(data, pageNumber) {
-  const pdf = await getDocumentProxy(data);
+  const pdf = await utils.getDocumentProxy(data);
   const page = await pdf.getPage(pageNumber);
   const operatorList = await page.getOperatorList();
-  const { OPS } = getResolvedPDFJSInstance();
+  const { OPS } = utils.getResolvedPDFJS();
   const images = [];
   for (const op of operatorList.fnArray) {
     if (op !== OPS.paintImageXObject)
@@ -80,5 +43,3 @@ async function getImagesFromPage(data, pageNumber) {
 
 exports.decodePDFText = decodePDFText;
 exports.getImagesFromPage = getImagesFromPage;
-exports.resolvePDFJSNodeImports = resolvePDFJSNodeImports;
-exports.resolvePDFJSWebImports = resolvePDFJSWebImports;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "unpdf",
   "type": "module",
-  "version": "0.3.0",
+  "version": "0.4.1",
   "packageManager": "pnpm@8.6.12",
   "description": "Utilities to work with PDFs, like extracting text",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",
@@ -53,6 +53,7 @@
     "lint": "eslint .",
     "lint:fix": "eslint . --fix",
     "release": "bumpp --commit --push --tag",
+    "test": "tsx test/index.ts",
     "test:types": "tsc --noEmit"
   },
   "peerDependencies": {
@@ -60,9 +61,11 @@
   },
   "devDependencies": {
     "@antfu/eslint-config": "^0.40.2",
+    "@types/node": "^20.4.10",
     "bumpp": "^9.1.1",
     "eslint": "^8.47.0",
     "pdfjs-dist": "^3.9.179",
+    "tsx": "^3.12.7",
     "typescript": "^5.1.6",
     "unbuild": "^2.0.0-rc.0"
   }

package/dist/shared/unpdf.5c39ffe2.d.cts

@@ -1,14 +0,0 @@
-interface PDFContent {
-    totalPages: number;
-    info?: Record<string, any>;
-    metadata?: any;
-    text: string | string[];
-}
-
-declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<PDFContent>;
-
-declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
-
-export { decodePDFText as d, getImagesFromPage as g };

package/dist/shared/unpdf.5c39ffe2.d.mts

@@ -1,14 +0,0 @@
-interface PDFContent {
-    totalPages: number;
-    info?: Record<string, any>;
-    metadata?: any;
-    text: string | string[];
-}
-
-declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<PDFContent>;
-
-declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
-
-export { decodePDFText as d, getImagesFromPage as g };

package/dist/shared/unpdf.5c39ffe2.d.ts

@@ -1,14 +0,0 @@
-interface PDFContent {
-    totalPages: number;
-    info?: Record<string, any>;
-    metadata?: any;
-    text: string | string[];
-}
-
-declare function decodePDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<PDFContent>;
-
-declare function getImagesFromPage(data: ArrayBuffer, pageNumber: number): Promise<ArrayBuffer[]>;
-
-export { decodePDFText as d, getImagesFromPage as g };