unpdf 0.5.2 → 0.5.4

package/README.md

@@ -32,11 +32,16 @@ yarn add unpdf
 ```ts
 import { extractPDFText } from 'unpdf'
 
-const pdfBuffer = await fetch('https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf')
+// Fetch a PDF file from the web
+const pdf = await fetch('https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf')
   .then(res => res.arrayBuffer())
 
+// Or load it from the filesystem
+const pdf = await readFile('./dummy.pdf')
+
+// Pass the PDF buffer to the relevant method
 const { totalPages, text } = await extractPDFText(
-  new Uint8Array(pdfBuffer), { mergePages: true }
+  new Uint8Array(pdf), { mergePages: true }
 )
 ```
 
@@ -75,7 +80,7 @@ interface UnPDFConfiguration {
    * @example
    * () => import('pdfjs-dist/legacy/build/pdf.js')
    */
-  pdfjs?: () => typeof PDFJS
+  pdfjs?: () => Promise<typeof PDFJS>
 }
 ```
 
@@ -83,7 +88,7 @@ interface UnPDFConfiguration {
 
 ### `defineUnPDFConfig`
 
-Define a custom PDF.js module, like the legacy build.
+Define a custom PDF.js module, like the legacy build. Make sure to call this method before using any other methods.
 
 ```ts
 function defineUnPDFConfig(config: UnPDFConfiguration): Promise<void>
@@ -91,7 +96,7 @@ function defineUnPDFConfig(config: UnPDFConfiguration): Promise<void>
 
 ### `getResolvedPDFJS`
 
-Returns the resolved PDF.js module.
+Returns the resolved PDF.js module. If no build is defined, the latest version will be initialized.
 
 ```ts
 function getResolvedPDFJS(): Promise<typeof import('pdfjs-dist')>

package/dist/index.node.cjs

@@ -1,7 +1,7 @@
 'use strict';
 
-const image = require('./shared/unpdf.6502e2c8.cjs');
-const utils = require('./shared/unpdf.9b5cc790.cjs');
+const image = require('./shared/unpdf.eba9a3dc.cjs');
+const utils = require('./shared/unpdf.42987ff1.cjs');
 
 const getPDFMeta = async (...args) => {
   await utils.resolvePDFJSImports();

package/dist/index.node.d.cts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.cjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.node.d.mts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.mjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.node.d.ts

@@ -1,5 +1,5 @@
-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 { 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';
 

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.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';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.3dcacb97.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.ab6a2535.mjs';
+export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS } from './shared/unpdf.ab6a2535.mjs';
 
 const getPDFMeta = async (...args) => {
   await resolvePDFJSImports();

package/dist/index.web.cjs

@@ -1,7 +1,7 @@
 'use strict';
 
-const image = require('./shared/unpdf.6502e2c8.cjs');
-const utils = require('./shared/unpdf.9b5cc790.cjs');
+const image = require('./shared/unpdf.eba9a3dc.cjs');
+const utils = require('./shared/unpdf.42987ff1.cjs');
 
 const getPDFMeta = async (...args) => {
   await utils.resolvePDFJSImports();

package/dist/index.web.d.cts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.cjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.web.d.mts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.mjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.web.d.ts

@@ -1,5 +1,5 @@
-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 { 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';
 

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.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';
+import { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.3dcacb97.mjs';
+import { r as resolvePDFJSImports } from './shared/unpdf.ab6a2535.mjs';
+export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS } from './shared/unpdf.ab6a2535.mjs';
 
 const getPDFMeta = async (...args) => {
   await resolvePDFJSImports();

package/dist/index.worker.cjs

@@ -1,6 +1,6 @@
 'use strict';
 
-const utils = require('./shared/unpdf.9b5cc790.cjs');
+const utils = require('./shared/unpdf.42987ff1.cjs');
 
 const getPDFMeta = async () => {
   throw new Error("Not implemented in worker context yet");

package/dist/index.worker.d.cts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.cjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.cjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.worker.d.mts

@@ -1,5 +1,5 @@
-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 { g as getPDFMeta$1, e as extractPDFText$1, a as getImagesFromPage$1 } from './shared/unpdf.6363738c.mjs';
+export { d as defineUnPDFConfig, b as getDocumentProxy, c as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.6363738c.mjs';
 import 'pdfjs-dist/types/src/display/api';
 import 'pdfjs-dist';
 

package/dist/index.worker.d.ts

@@ -1,5 +1,5 @@
-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 { 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';
 

package/dist/index.worker.mjs

@@ -1,4 +1,4 @@
-export { d as defineUnPDFConfig, g as getDocumentProxy, a as getResolvedPDFJS, r as resolvePDFJSImports } from './shared/unpdf.f801f444.mjs';
+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");

package/dist/shared/{unpdf.e3e60ef4.mjs → unpdf.3dcacb97.mjs}

@@ -1,4 +1,4 @@
-import { i as isPDFDocumentProxy, g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.f801f444.mjs';
+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);
@@ -9,7 +9,8 @@ async function getPDFMeta(data) {
   };
 }
 
-async function extractPDFText(data, { mergePages = false } = {}) {
+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))

package/dist/shared/{unpdf.9b5cc790.cjs → unpdf.42987ff1.cjs}

@@ -2,40 +2,38 @@
 
 let resolvedModule;
 async function getDocumentProxy(data, options = {}) {
-  const {
-    useWorkerFetch = false,
-    isEvalSupported = false,
-    useSystemFonts = true,
-    ...rest
-  } = options;
   const { getDocument } = await getResolvedPDFJS();
   const pdf = await getDocument({
     data,
-    useWorkerFetch,
-    useSystemFonts,
-    isEvalSupported,
-    ...rest
+    useWorkerFetch: false,
+    isEvalSupported: false,
+    useSystemFonts: true,
+    ...options
   }).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.");
-    }
-  }
+async function defineUnPDFConfig(options) {
+  const { pdfjs } = { ...options };
+  if (pdfjs)
+    await resolvePDFJSImports(pdfjs);
 }
 async function getResolvedPDFJS() {
   if (!resolvedModule)
     await resolvePDFJSImports();
   return resolvedModule;
 }
-async function resolvePDFJSImports() {
+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;

package/dist/shared/{unpdf.0c92a4fe.d.cts → unpdf.6363738c.d.cts}

@@ -6,7 +6,7 @@ declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     metadata: Record<string, any>;
 }>;
 
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, { mergePages }?: {
+declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
     mergePages?: boolean;
 }): Promise<{
     totalPages: number;
@@ -36,8 +36,8 @@ interface UnPDFConfiguration {
  * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
 declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(): Promise<void>;
+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.0c92a4fe.d.mts → unpdf.6363738c.d.mts}

@@ -6,7 +6,7 @@ declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     metadata: Record<string, any>;
 }>;
 
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, { mergePages }?: {
+declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
     mergePages?: boolean;
 }): Promise<{
     totalPages: number;
@@ -36,8 +36,8 @@ interface UnPDFConfiguration {
  * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
 declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(): Promise<void>;
+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.0c92a4fe.d.ts → unpdf.6363738c.d.ts}

@@ -6,7 +6,7 @@ declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     metadata: Record<string, any>;
 }>;
 
-declare function extractPDFText(data: BinaryData | PDFDocumentProxy, { mergePages }?: {
+declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
     mergePages?: boolean;
 }): Promise<{
     totalPages: number;
@@ -36,8 +36,8 @@ interface UnPDFConfiguration {
  * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
-declare function defineUnPDFConfig({ pdfjs }: UnPDFConfiguration): Promise<void>;
+declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
 declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(): Promise<void>;
+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.f801f444.mjs → unpdf.ab6a2535.mjs}

@@ -1,39 +1,37 @@
 let resolvedModule;
 async function getDocumentProxy(data, options = {}) {
-  const {
-    useWorkerFetch = false,
-    isEvalSupported = false,
-    useSystemFonts = true,
-    ...rest
-  } = options;
   const { getDocument } = await getResolvedPDFJS();
   const pdf = await getDocument({
     data,
-    useWorkerFetch,
-    useSystemFonts,
-    isEvalSupported,
-    ...rest
+    useWorkerFetch: false,
+    isEvalSupported: false,
+    useSystemFonts: true,
+    ...options
   }).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.");
-    }
-  }
+async function defineUnPDFConfig(options) {
+  const { pdfjs } = { ...options };
+  if (pdfjs)
+    await resolvePDFJSImports(pdfjs);
 }
 async function getResolvedPDFJS() {
   if (!resolvedModule)
     await resolvePDFJSImports();
   return resolvedModule;
 }
-async function resolvePDFJSImports() {
+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;

package/dist/shared/{unpdf.6502e2c8.cjs → unpdf.eba9a3dc.cjs}

@@ -1,6 +1,6 @@
 'use strict';
 
-const utils = require('./unpdf.9b5cc790.cjs');
+const utils = require('./unpdf.42987ff1.cjs');
 
 async function getPDFMeta(data) {
   const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
@@ -11,7 +11,8 @@ async function getPDFMeta(data) {
   };
 }
 
-async function extractPDFText(data, { mergePages = false } = {}) {
+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))

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "unpdf",
   "type": "module",
-  "version": "0.5.2",
+  "version": "0.5.4",
   "packageManager": "pnpm@8.6.12",
   "description": "Utilities to work with PDFs, like extracting text",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",