unpdf 0.5.0 → 0.5.2

package/README.md

@@ -1,15 +1,15 @@
 # unpdf
 
-A collection of utilities to work with PDFs. Uses Mozilla's [PDF.js](https://github.com/mozilla/pdf.js) under the hood.
+A collection of utilities to work with PDFs. Uses Mozilla's [PDF.js](https://github.com/mozilla/pdf.js) under the hood and lazily initializes the library.
 
-`unpdf` takes advantage of [export conditions](https://nodejs.org/api/packages.html#packages_conditional_exports) to circumvent build issues in serverless environments. PDF.js depends on the optional `canvas` module, which [doesn't work inside worker threads](https://github.com/Automattic/node-canvas/issues/1394).
+`unpdf` takes advantage of [export conditions](https://nodejs.org/api/packages.html#packages_conditional_exports) to circumvent build issues in serverless environments. For example, PDF.js depends on the optional `canvas` module, which [doesn't work inside worker threads](https://github.com/Automattic/node-canvas/issues/1394).
 
-This library is also intended as a modern alternative to the unmaintained [`pdf-parse`](https://www.npmjs.com/package/pdf-parse).
+This library is also intended as a modern alternative to the unmaintained but still popular [`pdf-parse`](https://www.npmjs.com/package/pdf-parse).
 
 ## Features
 
 - 🏗️ Conditional exports for Browser, Node and worker environments
-- 💬 Extract text from PDFs
+- 💬 Extract text and images from PDFs
 - 🧱 Opt-in to legacy PDF.js build
 
 ## Installation
@@ -18,13 +18,13 @@ Run the following command to add `unpdf` to your project.
 
 ```bash
 # pnpm
-pnpm add -D unpdf
+pnpm add unpdf
 
 # npm
-npm install -D unpdf
+npm install unpdf
 
 # yarn
-yarn add -D unpdf
+yarn add unpdf
 ```
 
 ## Usage
@@ -55,6 +55,14 @@ defineUnPDFConfig({
 // …
 ```
 
+### Access the PDF.js Module
+
+```ts
+import { getResolvedPDFJS } from 'unpdf'
+
+const { version } = await getResolvedPDFJS()
+```
+
 ## Config
 
 ```ts
@@ -75,16 +83,28 @@ interface UnPDFConfiguration {
 
 ### `defineUnPDFConfig`
 
+Define a custom PDF.js module, like the legacy build.
+
 ```ts
-function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>
+function defineUnPDFConfig(config: UnPDFConfiguration): Promise<void>
+```
+
+### `getResolvedPDFJS`
+
+Returns the resolved PDF.js module.
+
+```ts
+function getResolvedPDFJS(): Promise<typeof import('pdfjs-dist')>
 ```
 
 ### `getPDFMeta`
 
 ```ts
-function getPDFMeta(data: ArrayBuffer): Promise<{
+function getPDFMeta(
+  data: BinaryData | PDFDocumentProxy
+): Promise<{
   info: Record<string, any>
-  metadata: any
+  metadata: Record<string, any>
 }>
 ```
 
@@ -92,7 +112,7 @@ function getPDFMeta(data: ArrayBuffer): Promise<{
 
 ```ts
 function extractPDFText(
-  data: ArrayBuffer,
+  data: BinaryData | PDFDocumentProxy,
   { mergePages }?: { mergePages?: boolean }
 ): Promise<{
   totalPages: number
@@ -104,7 +124,7 @@ function extractPDFText(
 
 ```ts
 function getImagesFromPage(
-  data: ArrayBuffer,
+  data: BinaryData | PDFDocumentProxy,
   pageNumber: number
 ): Promise<ArrayBuffer[]>
 ```

package/dist/index.node.cjs

@@ -1,7 +1,7 @@
 'use strict';
 
-const image = require('./shared/unpdf.e023f326.cjs');
-const utils = require('./shared/unpdf.54007d6f.cjs');
+const image = require('./shared/unpdf.6502e2c8.cjs');
+const utils = require('./shared/unpdf.9b5cc790.cjs');
 
 const getPDFMeta = async (...args) => {
   await utils.resolvePDFJSImports();
@@ -17,6 +17,9 @@ const getImagesFromPage = async (...args) => {
 };
 
 exports.defineUnPDFConfig = utils.defineUnPDFConfig;
+exports.getDocumentProxy = utils.getDocumentProxy;
+exports.getResolvedPDFJS = utils.getResolvedPDFJS;
+exports.resolvePDFJSImports = utils.resolvePDFJSImports;
 exports.extractPDFText = extractPDFText;
 exports.getImagesFromPage = getImagesFromPage;
 exports.getPDFMeta = getPDFMeta;

package/dist/index.node.d.cts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.cjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.cjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.cjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.node.d.mts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.mjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.mjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.mjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.node.d.ts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.js';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.js';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.js';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.js';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.node.mjs

@@ -1,6 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0db47641.mjs';
-import { r as resolvePDFJSImports } from './shared/unpdf.5b4a24ed.mjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.e3e60ef4.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.f801f444.mjs';
+export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS } from './shared/unpdf.f801f444.mjs';
 
 const getPDFMeta = async (...args) => {
   await resolvePDFJSImports();
@@ -15,4 +15,4 @@ const getImagesFromPage = async (...args) => {
   return await getImagesFromPage$1(...args);
 };
 
-export { extractPDFText, getImagesFromPage, getPDFMeta };
+export { extractPDFText, getImagesFromPage, getPDFMeta, resolvePDFJSImports };

package/dist/index.web.cjs

@@ -1,7 +1,7 @@
 'use strict';
 
-const image = require('./shared/unpdf.e023f326.cjs');
-const utils = require('./shared/unpdf.54007d6f.cjs');
+const image = require('./shared/unpdf.6502e2c8.cjs');
+const utils = require('./shared/unpdf.9b5cc790.cjs');
 
 const getPDFMeta = async (...args) => {
   await utils.resolvePDFJSImports();
@@ -17,6 +17,9 @@ const getImagesFromPage = async (...args) => {
 };
 
 exports.defineUnPDFConfig = utils.defineUnPDFConfig;
+exports.getDocumentProxy = utils.getDocumentProxy;
+exports.getResolvedPDFJS = utils.getResolvedPDFJS;
+exports.resolvePDFJSImports = utils.resolvePDFJSImports;
 exports.extractPDFText = extractPDFText;
 exports.getImagesFromPage = getImagesFromPage;
 exports.getPDFMeta = getPDFMeta;

package/dist/index.web.d.cts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.cjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.cjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.cjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.web.d.mts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.mjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.mjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.mjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.web.d.ts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.js';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.js';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.js';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.js';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.web.mjs

@@ -1,6 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0db47641.mjs';
-import { r as resolvePDFJSImports } from './shared/unpdf.5b4a24ed.mjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.e3e60ef4.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.f801f444.mjs';
+export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS } from './shared/unpdf.f801f444.mjs';
 
 const getPDFMeta = async (...args) => {
   await resolvePDFJSImports();
@@ -15,4 +15,4 @@ const getImagesFromPage = async (...args) => {
   return await getImagesFromPage$1(...args);
 };
 
-export { extractPDFText, getImagesFromPage, getPDFMeta };
+export { extractPDFText, getImagesFromPage, getPDFMeta, resolvePDFJSImports };

package/dist/index.worker.cjs

@@ -1,6 +1,6 @@
 'use strict';
 
-const utils = require('./shared/unpdf.54007d6f.cjs');
+const utils = require('./shared/unpdf.9b5cc790.cjs');
 
 const getPDFMeta = async () => {
   throw new Error("Not implemented in worker context yet");
@@ -13,6 +13,9 @@ const getImagesFromPage = async () => {
 };
 
 exports.defineUnPDFConfig = utils.defineUnPDFConfig;
+exports.getDocumentProxy = utils.getDocumentProxy;
+exports.getResolvedPDFJS = utils.getResolvedPDFJS;
+exports.resolvePDFJSImports = utils.resolvePDFJSImports;
 exports.extractPDFText = extractPDFText;
 exports.getImagesFromPage = getImagesFromPage;
 exports.getPDFMeta = getPDFMeta;

package/dist/index.worker.d.cts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.cjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.cjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.cjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.worker.d.mts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.mjs';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.mjs';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.mjs';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.worker.d.ts

@@ -1,5 +1,6 @@
-import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.842960e8.js';
-export { d as defineUnPDFConfig } from './shared/unpdf.842960e8.js';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.0c92a4fe.js';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.0c92a4fe.js';
+import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 
 declare const getPDFMeta: typeof getPDFMeta$1;

package/dist/index.worker.mjs

@@ -1,4 +1,4 @@
-export { d as defineUnPDFConfig } from './shared/unpdf.5b4a24ed.mjs';
+export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.f801f444.mjs';
 
 const getPDFMeta = async () => {
   throw new Error("Not implemented in worker context yet");

package/dist/shared/unpdf.0c92a4fe.d.cts

@@ -0,0 +1,43 @@
+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, { mergePages }?: {
+    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({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
+declare function resolvePDFJSImports(): 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.0c92a4fe.d.mts

@@ -0,0 +1,43 @@
+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, { mergePages }?: {
+    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({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
+declare function resolvePDFJSImports(): 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.0c92a4fe.d.ts

@@ -0,0 +1,43 @@
+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, { mergePages }?: {
+    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({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
+declare function resolvePDFJSImports(): 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.e023f326.cjs → unpdf.6502e2c8.cjs}

@@ -1,18 +1,18 @@
 'use strict';
 
-const utils = require('./unpdf.54007d6f.cjs');
+const utils = require('./unpdf.9b5cc790.cjs');
 
 async function getPDFMeta(data) {
-  const pdf = await utils.getDocumentProxy(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()
+    info: meta?.info ?? {},
+    metadata: meta?.metadata?.getAll() ?? {}
   };
 }
 
 async function extractPDFText(data, { mergePages = false } = {}) {
-  const pdf = await utils.getDocumentProxy(data);
+  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))
   );
@@ -32,10 +32,10 @@ async function getPageText(pdf, pageNumber) {
 }
 
 async function getImagesFromPage(data, pageNumber) {
-  const pdf = await utils.getDocumentProxy(data);
+  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
   const page = await pdf.getPage(pageNumber);
   const operatorList = await page.getOperatorList();
-  const { OPS } = utils.getResolvedPDFJS();
+  const { OPS } = await utils.getResolvedPDFJS();
   const images = [];
   for (const op of operatorList.fnArray) {
     if (op !== OPS.paintImageXObject)

package/dist/shared/{unpdf.54007d6f.cjs → unpdf.9b5cc790.cjs}

@@ -1,13 +1,20 @@
 'use strict';
 
 let resolvedModule;
-async function getDocumentProxy(data) {
-  const { getDocument } = getResolvedPDFJS();
+async function getDocumentProxy(data, options = {}) {
+  const {
+    useWorkerFetch = false,
+    isEvalSupported = false,
+    useSystemFonts = true,
+    ...rest
+  } = options;
+  const { getDocument } = await getResolvedPDFJS();
   const pdf = await getDocument({
     data,
-    useWorkerFetch: false,
-    useSystemFonts: true,
-    isEvalSupported: false
+    useWorkerFetch,
+    useSystemFonts,
+    isEvalSupported,
+    ...rest
   }).promise;
   return pdf;
 }
@@ -21,7 +28,9 @@ async function defineUnPDFConfig({ pdfjs }) {
     }
   }
 }
-function getResolvedPDFJS() {
+async function getResolvedPDFJS() {
+  if (!resolvedModule)
+    await resolvePDFJSImports();
   return resolvedModule;
 }
 async function resolvePDFJSImports() {
@@ -34,8 +43,12 @@ async function resolvePDFJSImports() {
     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.0db47641.mjs → unpdf.e3e60ef4.mjs}

@@ -1,16 +1,16 @@
-import { g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.5b4a24ed.mjs';
+import { i as isPDFDocumentProxy, g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.f801f444.mjs';
 
 async function getPDFMeta(data) {
-  const pdf = await getDocumentProxy(data);
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   const meta = await pdf.getMetadata().catch(() => null);
   return {
-    info: meta?.info,
-    metadata: meta?.metadata?.getAll()
+    info: meta?.info ?? {},
+    metadata: meta?.metadata?.getAll() ?? {}
   };
 }
 
 async function extractPDFText(data, { mergePages = false } = {}) {
-  const pdf = await getDocumentProxy(data);
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   const texts = await Promise.all(
     Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
   );
@@ -30,10 +30,10 @@ async function getPageText(pdf, pageNumber) {
 }
 
 async function getImagesFromPage(data, pageNumber) {
-  const pdf = await getDocumentProxy(data);
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
   const page = await pdf.getPage(pageNumber);
   const operatorList = await page.getOperatorList();
-  const { OPS } = getResolvedPDFJS();
+  const { OPS } = await getResolvedPDFJS();
   const images = [];
   for (const op of operatorList.fnArray) {
     if (op !== OPS.paintImageXObject)

package/dist/shared/{unpdf.5b4a24ed.mjs → unpdf.f801f444.mjs}

@@ -1,11 +1,18 @@
 let resolvedModule;
-async function getDocumentProxy(data) {
-  const { getDocument } = getResolvedPDFJS();
+async function getDocumentProxy(data, options = {}) {
+  const {
+    useWorkerFetch = false,
+    isEvalSupported = false,
+    useSystemFonts = true,
+    ...rest
+  } = options;
+  const { getDocument } = await getResolvedPDFJS();
   const pdf = await getDocument({
     data,
-    useWorkerFetch: false,
-    useSystemFonts: true,
-    isEvalSupported: false
+    useWorkerFetch,
+    useSystemFonts,
+    isEvalSupported,
+    ...rest
   }).promise;
   return pdf;
 }
@@ -19,7 +26,9 @@ async function defineUnPDFConfig({ pdfjs }) {
     }
   }
 }
-function getResolvedPDFJS() {
+async function getResolvedPDFJS() {
+  if (!resolvedModule)
+    await resolvePDFJSImports();
   return resolvedModule;
 }
 async function resolvePDFJSImports() {
@@ -32,5 +41,8 @@ async function resolvePDFJSImports() {
     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, resolvePDFJSImports as r };
+export { getResolvedPDFJS as a, defineUnPDFConfig as d, getDocumentProxy as g, isPDFDocumentProxy as i, resolvePDFJSImports as r };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "unpdf",
   "type": "module",
-  "version": "0.5.0",
+  "version": "0.5.2",
   "packageManager": "pnpm@8.6.12",
   "description": "Utilities to work with PDFs, like extracting text",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",

package/dist/shared/unpdf.842960e8.d.cts

@@ -1,31 +0,0 @@
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: ArrayBuffer): Promise<{
-    info: Record<string, any>;
-    metadata: any;
-}>;
-
-declare function extractPDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: ArrayBuffer, 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>;
-}
-
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
-
-export { getImagesFromPage as a, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g };

package/dist/shared/unpdf.842960e8.d.mts

@@ -1,31 +0,0 @@
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: ArrayBuffer): Promise<{
-    info: Record<string, any>;
-    metadata: any;
-}>;
-
-declare function extractPDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: ArrayBuffer, 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>;
-}
-
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
-
-export { getImagesFromPage as a, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g };

package/dist/shared/unpdf.842960e8.d.ts

@@ -1,31 +0,0 @@
-import PDFJS from 'pdfjs-dist';
-
-declare function getPDFMeta(data: ArrayBuffer): Promise<{
-    info: Record<string, any>;
-    metadata: any;
-}>;
-
-declare function extractPDFText(data: ArrayBuffer, { mergePages }?: {
-    mergePages?: boolean;
-}): Promise<{
-    totalPages: number;
-    text: string | string[];
-}>;
-
-declare function getImagesFromPage(data: ArrayBuffer, 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>;
-}
-
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
-
-export { getImagesFromPage as a, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g };