unpdf 0.1.0 → 0.2.0

package/README.md

@@ -14,7 +14,14 @@ A collection of utilities to work with PDFs.
 Run the following command to add `unpdf` to your project.
 
 ```bash
-pnpm install unpdf # or npm or yarn
+# pnpm
+pnpm add -D unpdf
+
+# npm
+npm install -D unpdf
+
+# yarn
+yarn add -D unpdf
 ```
 
 ## Usage

package/dist/index.node.cjs

@@ -1,6 +1,8 @@
 'use strict';
 
-async function decodePDFText(data, { mergePages = false } = {}) {
+const pdfjsDist = require('pdfjs-dist');
+
+async function getDocumentProxy(data) {
   const { getDocument } = await getPDFJSImports();
   const pdf = await getDocument({
     data,
@@ -8,6 +10,23 @@ async function decodePDFText(data, { mergePages = false } = {}) {
     useSystemFonts: true,
     isEvalSupported: false
   }).promise;
+  return pdf;
+}
+async function getPDFJSImports() {
+  try {
+    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
+    const { getDocument, version } = mod;
+    return { getDocument, version };
+  } catch (e) {
+    console.error(e);
+    throw new Error(
+      "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
+    );
+  }
+}
+
+async function decodePDFText(data, { mergePages = false } = {}) {
+  const pdf = await getDocumentProxy(data);
   const meta = await pdf.getMetadata().catch(() => null);
   const texts = await Promise.all(
     Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
@@ -28,17 +47,21 @@ async function getPageText(pdf, pageNumber) {
     return;
   return items.map(({ str }) => str).join("\n");
 }
-async function getPDFJSImports() {
-  try {
-    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
-    const { getDocument, version } = mod;
-    return { getDocument, version };
-  } catch (e) {
-    console.error(e);
-    throw new Error(
-      "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
-    );
+
+async function getImagesFromPage(data, pageNumber) {
+  const pdf = await getDocumentProxy(data);
+  const page = await pdf.getPage(pageNumber);
+  const operatorList = await page.getOperatorList();
+  const images = [];
+  for (const op of operatorList.fnArray) {
+    if (op !== pdfjsDist.OPS.paintImageXObject)
+      continue;
+    const image = await page.objs.get(operatorList.argsArray[op][0]);
+    if (image.data)
+      images.push(image.data.buffer);
   }
+  return images;
 }
 
 exports.decodePDFText = decodePDFText;
+exports.getImagesFromPage = getImagesFromPage;

package/dist/index.node.d.cts

@@ -1 +1,14 @@
-export { d as decodePDFText } from './shared/unpdf.98b795c6.cjs';
+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, getImagesFromPage };

package/dist/index.node.d.mts

@@ -1 +1,14 @@
-export { d as decodePDFText } from './shared/unpdf.98b795c6.mjs';
+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, getImagesFromPage };

package/dist/index.node.d.ts

@@ -1 +1,14 @@
-export { d as decodePDFText } from './shared/unpdf.98b795c6.js';
+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, getImagesFromPage };

package/dist/index.node.mjs

@@ -1,4 +1,6 @@
-async function decodePDFText(data, { mergePages = false } = {}) {
+import { OPS } from 'pdfjs-dist';
+
+async function getDocumentProxy(data) {
   const { getDocument } = await getPDFJSImports();
   const pdf = await getDocument({
     data,
@@ -6,6 +8,23 @@ async function decodePDFText(data, { mergePages = false } = {}) {
     useSystemFonts: true,
     isEvalSupported: false
   }).promise;
+  return pdf;
+}
+async function getPDFJSImports() {
+  try {
+    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
+    const { getDocument, version } = mod;
+    return { getDocument, version };
+  } catch (e) {
+    console.error(e);
+    throw new Error(
+      "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
+    );
+  }
+}
+
+async function decodePDFText(data, { mergePages = false } = {}) {
+  const pdf = await getDocumentProxy(data);
   const meta = await pdf.getMetadata().catch(() => null);
   const texts = await Promise.all(
     Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
@@ -26,17 +45,20 @@ async function getPageText(pdf, pageNumber) {
     return;
   return items.map(({ str }) => str).join("\n");
 }
-async function getPDFJSImports() {
-  try {
-    const { default: mod } = await import('pdfjs-dist/legacy/build/pdf');
-    const { getDocument, version } = mod;
-    return { getDocument, version };
-  } catch (e) {
-    console.error(e);
-    throw new Error(
-      "PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again."
-    );
+
+async function getImagesFromPage(data, pageNumber) {
+  const pdf = await getDocumentProxy(data);
+  const page = await pdf.getPage(pageNumber);
+  const operatorList = await page.getOperatorList();
+  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 { decodePDFText };
+export { decodePDFText, getImagesFromPage };

package/dist/index.web.cjs

@@ -1,7 +1,11 @@
 'use strict';
 
-async function decodePDFText(..._args) {
+const decodePDFText = async () => {
   throw new Error("Not supported in browser context yet");
-}
+};
+const getImagesFromPage = async () => {
+  throw new Error("Not supported in browser context yet");
+};
 
 exports.decodePDFText = decodePDFText;
+exports.getImagesFromPage = getImagesFromPage;

package/dist/index.web.d.cts

@@ -1,5 +1,6 @@
-import { d as decodePDFText$1, P as PDFContent } from './shared/unpdf.98b795c6.cjs';
+import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.cjs';
 
-declare function decodePDFText(..._args: Parameters<typeof decodePDFText$1>): Promise<PDFContent>;
+declare const decodePDFText: typeof decodePDFText$1;
+declare const getImagesFromPage: typeof getImagesFromPage$1;
 
-export { decodePDFText };
+export { decodePDFText, getImagesFromPage };

package/dist/index.web.d.mts

@@ -1,5 +1,6 @@
-import { d as decodePDFText$1, P as PDFContent } from './shared/unpdf.98b795c6.mjs';
+import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.mjs';
 
-declare function decodePDFText(..._args: Parameters<typeof decodePDFText$1>): Promise<PDFContent>;
+declare const decodePDFText: typeof decodePDFText$1;
+declare const getImagesFromPage: typeof getImagesFromPage$1;
 
-export { decodePDFText };
+export { decodePDFText, getImagesFromPage };

package/dist/index.web.d.ts

@@ -1,5 +1,6 @@
-import { d as decodePDFText$1, P as PDFContent } from './shared/unpdf.98b795c6.js';
+import { decodePDFText as decodePDFText$1, getImagesFromPage as getImagesFromPage$1 } from './index.node.js';
 
-declare function decodePDFText(..._args: Parameters<typeof decodePDFText$1>): Promise<PDFContent>;
+declare const decodePDFText: typeof decodePDFText$1;
+declare const getImagesFromPage: typeof getImagesFromPage$1;
 
-export { decodePDFText };
+export { decodePDFText, getImagesFromPage };

package/dist/index.web.mjs

@@ -1,5 +1,8 @@
-async function decodePDFText(..._args) {
+const decodePDFText = async () => {
   throw new Error("Not supported in browser context yet");
-}
+};
+const getImagesFromPage = async () => {
+  throw new Error("Not supported in browser context yet");
+};
 
-export { decodePDFText };
+export { decodePDFText, getImagesFromPage };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "unpdf",
   "type": "module",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "packageManager": "pnpm@8.6.12",
   "description": "Utilities to work with PDFs",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",

package/dist/shared/unpdf.98b795c6.d.cts

@@ -1,12 +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>;
-
-export { PDFContent as P, decodePDFText as d };

package/dist/shared/unpdf.98b795c6.d.mts

@@ -1,12 +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>;
-
-export { PDFContent as P, decodePDFText as d };

package/dist/shared/unpdf.98b795c6.d.ts

@@ -1,12 +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>;
-
-export { PDFContent as P, decodePDFText as d };