feat: resolve PDF.js build for browser context

unjs · Aug 12, 2023 · 325d450 · 325d450
1 parent 66a2fc7
commit 325d450
Show file tree

Hide file tree

Showing 9 changed files with 88 additions and 35 deletions.
diff --git a/README.md b/README.md
@@ -1,13 +1,15 @@
 # unpdf
 
-A collection of utilities to work with PDFs.
+A collection of utilities to work with PDFs. Uses Mozilla's [PDF.js](https://github.com/mozilla/pdf.js) under the hood.
 
-`unpdf` takes advantage of [export conditions](https://nodejs.org/api/packages.html#packages_conditional_exports) to provide a minimal bundle size for the browser. As of now, the available methods are only supported in Node contexts.
+`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).
 
-**Why this package then?**
+This library is also intended as a modern alternative to the unmaintained [`pdf-parse`](https://www.npmjs.com/package/pdf-parse).
 
-- To circumvent build issues in serverless environments, where the `canvas` package used by `PDF.js` is not supported.
-- WIP and more to come.
+## Features
+
+- 🏗️ Conditional exports for Browser, Node and worker environments
+- 💬 Extract text from PDFs
 
 ## Installation
 

diff --git a/build.config.ts b/build.config.ts
@@ -1,7 +1,7 @@
 import { defineBuildConfig } from 'unbuild'
 
 export default defineBuildConfig({
-  entries: ['src/index.node', 'src/index.web.ts'],
+  entries: ['src/index.web.ts', 'src/index.worker', 'src/index.node'],
   clean: true,
   declaration: true,
   externals: ['pdfjs-dist'],

diff --git a/package.json b/package.json
@@ -3,7 +3,7 @@
   "type": "module",
   "version": "0.2.2",
   "packageManager": "pnpm@8.6.12",
-  "description": "Utilities to work with PDFs",
+  "description": "Utilities to work with PDFs, like extracting text",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",
   "license": "MIT",
   "homepage": "https://github.com/johannschopplich/unpdf#readme",
@@ -21,16 +21,16 @@
   "sideEffects": false,
   "exports": {
     "browser": "./dist/index.web.mjs",
-    "bun": "./dist/index.web.mjs",
-    "deno": "./dist/index.web.mjs",
-    "edge-light": "./dist/index.web.mjs",
-    "edge-routine": "./dist/index.web.mjs",
-    "lagon": "./dist/index.web.mjs",
-    "netlify": "./dist/index.web.mjs",
-    "react-native": "./dist/index.web.mjs",
-    "wintercg": "./dist/index.web.mjs",
-    "worker": "./dist/index.web.mjs",
-    "workerd": "./dist/index.web.mjs",
+    "bun": "./dist/index.worker.mjs",
+    "deno": "./dist/index.worker.mjs",
+    "edge-light": "./dist/index.worker.mjs",
+    "edge-routine": "./dist/index.worker.mjs",
+    "lagon": "./dist/index.worker.mjs",
+    "netlify": "./dist/index.worker.mjs",
+    "react-native": "./dist/index.worker.mjs",
+    "wintercg": "./dist/index.worker.mjs",
+    "worker": "./dist/index.worker.mjs",
+    "workerd": "./dist/index.worker.mjs",
     "node": {
       "types": "./dist/index.node.d.ts",
       "import": "./dist/index.node.mjs",

diff --git a/src/image.ts b/src/image.ts
@@ -1,4 +1,4 @@
-import { getDocumentProxy, getPDFJSImports } from './utils'
+import { getDocumentProxy, getResolvedPDFJSInstance } from './utils'
 
 export async function getImagesFromPage(
   data: ArrayBuffer,
@@ -7,7 +7,7 @@ export async function getImagesFromPage(
   const pdf = await getDocumentProxy(data)
   const page = await pdf.getPage(pageNumber)
   const operatorList = await page.getOperatorList()
-  const { OPS } = await getPDFJSImports()
+  const { OPS } = getResolvedPDFJSInstance()
 
   const images: ArrayBuffer[] = []
   for (const op of operatorList.fnArray) {

diff --git a/src/index.node.ts b/src/index.node.ts
@@ -1,2 +1,13 @@
-export * from './text'
-export * from './image'
+import { decodePDFText as _decodePDFText } from './text'
+import { getImagesFromPage as _getImagesFromPage } from './image'
+import { resolvePDFJSNodeImports } from './utils'
+
+export const decodePDFText: typeof _decodePDFText = async (...args) => {
+  await resolvePDFJSNodeImports()
+  return await _decodePDFText(...args)
+}
+
+export const getImagesFromPage: typeof _getImagesFromPage = async (...args) => {
+  await resolvePDFJSNodeImports()
+  return await _getImagesFromPage(...args)
+}
diff --git a/src/index.web.ts b/src/index.web.ts
@@ -1,12 +1,13 @@
-import type {
-  decodePDFText as _decodePDFText,
-  getImagesFromPage as _getImagesFromPage,
-} from './index.node'
+import { decodePDFText as _decodePDFText } from './text'
+import { getImagesFromPage as _getImagesFromPage } from './image'
+import { resolvePDFJSWebImports } from './utils'
 
-export const decodePDFText: typeof _decodePDFText = async () => {
-  throw new Error('Not supported in browser context yet')
+export const decodePDFText: typeof _decodePDFText = async (...args) => {
+  await resolvePDFJSWebImports()
+  return await _decodePDFText(...args)
 }
 
-export const getImagesFromPage: typeof _getImagesFromPage = async () => {
-  throw new Error('Not supported in browser context yet')
+export const getImagesFromPage: typeof _getImagesFromPage = async (...args) => {
+  await resolvePDFJSWebImports()
+  return await _getImagesFromPage(...args)
 }
diff --git a/src/index.worker.ts b/src/index.worker.ts
@@ -0,0 +1,10 @@
+import type { decodePDFText as _decodePDFText } from './text'
+import type { getImagesFromPage as _getImagesFromPage } from './image'
+
+export const decodePDFText: typeof _decodePDFText = async () => {
+  throw new Error('Not implemented in worker context yet')
+}
+
+export const getImagesFromPage: typeof _getImagesFromPage = async () => {
+  throw new Error('Not implemented in worker context yet')
+}
diff --git a/src/types.ts b/src/types.ts
@@ -4,3 +4,7 @@ export interface PDFContent {
   metadata?: any
   text: string | string[]
 }
+
+export type Prettify<T> = {
+  [K in keyof T]: T[K];
+} & Record<never, never>
diff --git a/src/utils.ts b/src/utils.ts
@@ -1,5 +1,9 @@
+import type * as PDFJS from 'pdfjs-dist'
+
+let instance: typeof PDFJS | undefined
+
 export async function getDocumentProxy(data: ArrayBuffer) {
-  const { getDocument } = await getPDFJSImports()
+  const { getDocument } = instance!
   const pdf = await getDocument({
     data,
     useWorkerFetch: false,
@@ -10,15 +14,36 @@ export async function getDocumentProxy(data: ArrayBuffer) {
   return pdf
 }
 
-export async function getPDFJSImports() {
+export function getResolvedPDFJSInstance() {
+  return instance!
+}
+
+export 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.',
+    )
+  }
+}
+
+export async function resolvePDFJSNodeImports() {
+  if (instance)
+    return
+
   try {
     const { default: mod } = await import('pdfjs-dist/legacy/build/pdf')
-    return mod
+    instance = mod
   }
-  catch (e) {
-    console.error(e)
+  catch (error) {
     throw new Error(
-      'PDF.js is not available. Please run `pnpm add -D pdfjs-dist` and try again.',
+      'PDF.js is not available. Please add the package as a dependency.',
     )
   }
 }