Chore: Remove node-canvas dependency (#12238)

Co-authored-by: Laurent Cozic <laurent22@users.noreply.github.com>

.eslintignore

@@ -1516,6 +1516,7 @@ packages/lib/utils/replaceUnsupportedCharacters.test.js
 packages/lib/utils/replaceUnsupportedCharacters.js
 packages/lib/utils/resolvePathWithinDir.test.js
 packages/lib/utils/resolvePathWithinDir.js
+packages/lib/utils/types/pdfJs.js
 packages/lib/utils/userFetcher.js
 packages/lib/utils/webDAVUtils.test.js
 packages/lib/utils/webDAVUtils.js

.gitignore

@@ -1490,6 +1490,7 @@ packages/lib/utils/replaceUnsupportedCharacters.test.js
 packages/lib/utils/replaceUnsupportedCharacters.js
 packages/lib/utils/resolvePathWithinDir.test.js
 packages/lib/utils/resolvePathWithinDir.js
+packages/lib/utils/types/pdfJs.js
 packages/lib/utils/userFetcher.js
 packages/lib/utils/webDAVUtils.test.js
 packages/lib/utils/webDAVUtils.js

devbox.json

@@ -11,9 +11,6 @@
     },
     "nodejs":     "latest",
     "pkg-config": "latest",
-    "pixman":     "latest",
-    "cairo.dev":  "",
-    "pango.dev":  "",
     "darwin.apple_sdk.frameworks.Foundation": { // satisfies missing CoreText/CoreText.h
       // https://github.com/NixOS/nixpkgs/blob/master/pkgs/os-specific/darwin/apple-sdk/default.nix
       "version":   "",
@@ -26,7 +23,6 @@
       "excluded_platforms": ["aarch64-darwin", "x86_64-darwin"],
     },
     "git":    "latest",
-    "giflib": "latest",
   },
   "shell": {
     "init_hook": [

packages/lib/jest.setup.js

@@ -10,7 +10,7 @@ const React = require('react');
 
 require('../../jest.base-setup.js')();
 
-shimInit({ sharp, nodeSqlite, pdfJs, React, appVersion: () => packageInfo.version });
+shimInit({ pdfJs, sharp, nodeSqlite, React, appVersion: () => packageInfo.version });
 
 global.afterEach(async () => {
 	await afterEachCleanUp();

packages/lib/services/ocr/OcrService.test.ts

@@ -88,7 +88,6 @@ describe('OcrService', () => {
 		// Use embedded text (skip OCR)
 		['dummy.pdf', 'Dummy PDF file'],
 		['multi_page__embedded_text.pdf', 'This is a test.\nTesting...\nThis PDF has 3 pages.\nThis is page 3.'],
-		['multi_page__no_embedded_text.pdf', 'This is a multi-page PDF\nwith no embedded text.\nPage 2: more text.\nThe third page.'],
 	])('should process PDF resources', async (samplePath: string, expectedText: string) => {
 		const { resource } = await createNoteAndResource({ path: `${ocrSampleDir}/${samplePath}` });
 

packages/lib/shim-init-node.ts

@@ -6,10 +6,8 @@ import Note from './models/Note';
 import Resource from './models/Resource';
 import { basename, fileExtension, safeFileExtension } from './path-utils';
 import * as fs from 'fs-extra';
-import * as pdfJsNamespace from 'pdfjs-dist';
 import { writeFile } from 'fs/promises';
 import { ResourceEntity } from './services/database/types';
-import { TextItem } from 'pdfjs-dist/types/src/display/api';
 import replaceUnsupportedCharacters from './utils/replaceUnsupportedCharacters';
 import { FetchBlobOptions } from './types';
 import { fromFile as fileTypeFromFile } from 'file-type';
@@ -20,6 +18,7 @@ import * as mimeUtils from './mime-utils';
 import BaseItem from './models/BaseItem';
 import { Size } from '@joplin/utils/types';
 import { cpus } from 'os';
+import type PdfJs from './utils/types/pdfJs';
 const { _ } = require('./locale');
 const http = require('http');
 const https = require('https');
@@ -111,7 +110,7 @@ interface ShimInitOptions {
 	electronBridge: any;
 	// eslint-disable-next-line @typescript-eslint/no-explicit-any -- Old code before rule was applied
 	nodeSqlite: any;
-	pdfJs: typeof pdfJsNamespace;
+	pdfJs: PdfJs;
 	isAppleSilicon?: ()=> boolean;
 }
 
@@ -823,7 +822,7 @@ function shimInit(options: ShimInitOptions = null) {
 				const textContent = await page.getTextContent();
 
 				const strings = textContent.items.map(item => {
-					const text = (item as TextItem).str ?? '';
+					const text = item.str ?? '';
 					return text;
 				}).join('\n');
 
@@ -839,32 +838,24 @@ function shimInit(options: ShimInitOptions = null) {
 	};
 
 	shim.pdfToImages = async (pdfPath: string, outputDirectoryPath: string, options?: CreatePdfFromImagesOptions): Promise<string[]> => {
-		// We handle both the Electron app and testing framework. Potentially
-		// the same code could be use to support the CLI app.
-		const isTesting = !shim.isElectron();
+		if (typeof HTMLCanvasElement === 'undefined') {
+			throw new Error('Unsupported -- the Canvas element is required.');
+		}
 
 		const createCanvas = () => {
-			if (isTesting) {
-				return require('canvas').createCanvas();
-			}
 			return document.createElement('canvas');
 		};
 
-		// eslint-disable-next-line @typescript-eslint/no-explicit-any -- Old code before rule was applied
-		const canvasToBuffer = async (canvas: any): Promise<Buffer> => {
+		const canvasToBuffer = async (canvas: HTMLCanvasElement): Promise<Buffer> => {
 			const quality = 0.8;
-			if (isTesting) {
-				return canvas.toBuffer('image/jpeg', { quality });
-			} else {
-				const canvasToBlob = async (canvas: HTMLCanvasElement): Promise<Blob> => {
-					return new Promise(resolve => {
-						canvas.toBlob(blob => resolve(blob), 'image/jpg', quality);
-					});
-				};
+			const canvasToBlob = async (canvas: HTMLCanvasElement): Promise<Blob> => {
+				return new Promise(resolve => {
+					canvas.toBlob(blob => resolve(blob), 'image/jpg', quality);
+				});
+			};
 
-				const blob = await canvasToBlob(canvas);
-				return Buffer.from(await blob.arrayBuffer());
-			}
+			const blob = await canvasToBlob(canvas);
+			return Buffer.from(await blob.arrayBuffer());
 		};
 
 		const filePrefix = `page_${Date.now()}`;
@@ -879,6 +870,9 @@ function shimInit(options: ShimInitOptions = null) {
 				const viewport = page.getViewport({ scale: options?.scaleFactor ?? 2 });
 				const canvas = createCanvas();
 				const ctx = canvas.getContext('2d');
+				if (!ctx) {
+					throw new Error('Unable to get 2D rendering context from canvas.');
+				}
 
 				canvas.height = viewport.height;
 				canvas.width = viewport.width;

packages/lib/utils/types/pdfJs.js

@@ -0,0 +1,6 @@
+"use strict";
+// Custom type definitions for pdfjs-dist. packages/lib should avoid depending on
+// pdfjs-dist directly, since an unpatched version of pdfjs-dist depends on node-canvas,
+// which makes it more difficult to build Joplin.
+Object.defineProperty(exports, "__esModule", { value: true });
+//# sourceMappingURL=pdfJs.js.map

packages/lib/utils/types/pdfJs.ts

@@ -0,0 +1,48 @@
+
+// Custom type definitions for pdfjs-dist. packages/lib should avoid depending on
+// pdfjs-dist directly, since an unpatched version of pdfjs-dist depends on node-canvas,
+// which makes it more difficult to build Joplin.
+
+interface PdfTextContent {
+	items: { str?: string }[];
+}
+
+interface Viewport {
+	width: number;
+	height: number;
+}
+
+interface RenderOptions {
+	canvasContext: CanvasRenderingContext2D|OffscreenCanvasRenderingContext2D;
+	viewport: Viewport;
+}
+
+interface PdfPage {
+	getViewport(options: { scale: number }): Viewport;
+	render(options: RenderOptions): { promise: Promise<void> };
+	getTextContent(): Promise<PdfTextContent>;
+}
+
+interface PdfDocument {
+	numPages: number;
+	getPage(n: number): Promise<PdfPage>;
+	destroy(): Promise<void>;
+}
+
+interface GetDocumentTask {
+	promise: Promise<PdfDocument>;
+}
+
+interface GetDocumentOptions {
+	url: string;
+	useSystemFonts: boolean;
+	// IMPORTANT: Set to false to mitigate CVE-2024-4367.
+	isEvalSupported: false;
+}
+
+interface PdfJs {
+	getDocument(options: GetDocumentOptions): GetDocumentTask;
+}
+
+export default PdfJs;
+