fix(editor): add footnote url markdown preprocessor to avoid link node parsing (#11888)

Closes: [BS-3282](https://linear.app/affine-design/issue/BS-3282/预处理-footnote-definition-中的-url-避免-markdown-link-node-parsing)

blocksuite/affine/all/src/__tests__/adapters/markdown.unit.spec.ts

@@ -4028,11 +4028,8 @@ hhh
     expect(nanoidReplacement(rawBlockSnapshot)).toEqual(blockSnapshot);
   });
 
-  test('without footnote middleware', async () => {
+  describe('footnote', () => {
-    const markdown =
+    const createFootnoteBlockSnapshot = (url: string): BlockSnapshot => ({
-      'aaa[^1][^2][^3]\n\n[^1]: {"type":"url","url":"https%3A%2F%2Fwww.example.com"}\n\n[^2]: {"type":"doc","docId":"deadbeef"}\n\n[^3]: {"type":"attachment","blobId":"abcdefg","fileName":"test.txt","fileType":"text/plain"}\n';
-
-    const blockSnapshot: BlockSnapshot = {
       type: 'block',
       id: 'matchesReplaceMap[0]',
       flavour: 'affine:note',
@@ -4063,7 +4060,7 @@ hhh
                       label: '1',
                       reference: {
                         type: 'url',
-                        url: 'https://www.example.com',
+                        url,
                       },
                     },
                   },
@@ -4100,13 +4097,37 @@ hhh
           children: [],
         },
       ],
-    };
-
-    const mdAdapter = new MarkdownAdapter(createJob(), provider);
-    const rawBlockSnapshot = await mdAdapter.toBlockSnapshot({
-      file: markdown,
     });
-    expect(nanoidReplacement(rawBlockSnapshot)).toEqual(blockSnapshot);
+
+    test('with encoded url', async () => {
+      const markdown =
+        'aaa[^1][^2][^3]\n\n[^1]: {"type":"url","url":"https%3A%2F%2Fwww.example.com"}\n\n[^2]: {"type":"doc","docId":"deadbeef"}\n\n[^3]: {"type":"attachment","blobId":"abcdefg","fileName":"test.txt","fileType":"text/plain"}\n';
+
+      const blockSnapshot = createFootnoteBlockSnapshot(
+        'https://www.example.com'
+      );
+
+      const mdAdapter = new MarkdownAdapter(createJob(), provider);
+      const rawBlockSnapshot = await mdAdapter.toBlockSnapshot({
+        file: markdown,
+      });
+      expect(nanoidReplacement(rawBlockSnapshot)).toEqual(blockSnapshot);
+    });
+
+    test('with unencoded url', async () => {
+      const markdown =
+        'aaa[^1][^2][^3]\n\n[^1]: {"type":"url","url":"https://www.example.com"}\n\n[^2]: {"type":"doc","docId":"deadbeef"}\n\n[^3]: {"type":"attachment","blobId":"abcdefg","fileName":"test.txt","fileType":"text/plain"}\n';
+
+      const blockSnapshot = createFootnoteBlockSnapshot(
+        'https://www.example.com'
+      );
+
+      const mdAdapter = new MarkdownAdapter(createJob(), provider);
+      const rawBlockSnapshot = await mdAdapter.toBlockSnapshot({
+        file: markdown,
+      });
+      expect(nanoidReplacement(rawBlockSnapshot)).toEqual(blockSnapshot);
+    });
   });
 
   test('should not wrap url with angle brackets if it is not a url', async () => {

blocksuite/affine/all/src/adapters/markdown/preprocessor.ts

@@ -1,7 +1,9 @@
+import { BookmarkBlockMarkdownPreprocessorExtension } from '@blocksuite/affine-block-bookmark';
 import { CodeMarkdownPreprocessorExtension } from '@blocksuite/affine-block-code';
 import { LatexMarkdownPreprocessorExtension } from '@blocksuite/affine-block-latex';
 
 export const defaultMarkdownPreprocessors = [
   LatexMarkdownPreprocessorExtension,
   CodeMarkdownPreprocessorExtension,
+  BookmarkBlockMarkdownPreprocessorExtension,
 ];

blocksuite/affine/blocks/bookmark/package.json

@@ -30,6 +30,9 @@
     "yjs": "^13.6.23",
     "zod": "^3.23.8"
   },
+  "devDependencies": {
+    "vitest": "3.1.1"
+  },
   "exports": {
     ".": "./src/index.ts",
     "./effects": "./src/effects.ts",

blocksuite/affine/blocks/bookmark/src/__tests__/adapters/preprocessor.unit.spec.ts

@@ -0,0 +1,53 @@
+import { describe, expect, it } from 'vitest';
+
+import { footnoteUrlPreprocessor } from '../../adapters/markdown/preprocessor';
+
+describe('footnoteUrlPreprocessor', () => {
+  it('should encode unencoded URLs in footnote definitions', () => {
+    const input =
+      '[^ref]: {"type":"url","url":"https://example.com?param=value"}';
+    const expected =
+      '[^ref]: {"type":"url","url":"https%3A%2F%2Fexample.com%3Fparam%3Dvalue"}';
+    expect(footnoteUrlPreprocessor(input)).toBe(expected);
+  });
+
+  it('should not encode already encoded URLs', () => {
+    const input = '[^ref]: {"type":"url","url":"https%3A%2F%2Fexample.com"}';
+    expect(footnoteUrlPreprocessor(input)).toBe(input);
+  });
+
+  it('should handle invalid JSON content', () => {
+    const input = '[^ref]: {"invalid json"}';
+    expect(footnoteUrlPreprocessor(input)).toBe(input);
+  });
+
+  it('should handle non-object footnote data', () => {
+    const input = '[^ref]: "not an object"';
+    expect(footnoteUrlPreprocessor(input)).toBe(input);
+  });
+
+  it('should handle footnote data without url property', () => {
+    const input = '[^ref]: {"type":"url"}';
+    expect(footnoteUrlPreprocessor(input)).toBe(input);
+  });
+
+  it('should handle multiple footnote definitions', () => {
+    const input = `
+[^ref1]: {"type":"url","url":"https://example1.com"}
+[^ref2]: {"type":"url","url":"https://example2.com"}
+    `.trim();
+    const expected = `
+[^ref1]: {"type":"url","url":"https%3A%2F%2Fexample1.com"}
+[^ref2]: {"type":"url","url":"https%3A%2F%2Fexample2.com"}
+    `.trim();
+    expect(footnoteUrlPreprocessor(input)).toBe(expected);
+  });
+
+  it('should handle special characters in URLs', () => {
+    const input =
+      '[^ref]: {"type":"url","url":"https://example.com/path with spaces?param=value&another=param"}';
+    const expected =
+      '[^ref]: {"type":"url","url":"https%3A%2F%2Fexample.com%2Fpath%20with%20spaces%3Fparam%3Dvalue%26another%3Dparam"}';
+    expect(footnoteUrlPreprocessor(input)).toBe(expected);
+  });
+});

blocksuite/affine/blocks/bookmark/src/adapters/extension.ts

@@ -1,13 +1,13 @@
 import type { ExtensionType } from '@blocksuite/store';
 
 import { BookmarkBlockHtmlAdapterExtension } from './html.js';
-import { BookmarkBlockMarkdownAdapterExtension } from './markdown.js';
+import { BookmarkBlockMarkdownAdapterExtensions } from './markdown/index.js';
 import { BookmarkBlockNotionHtmlAdapterExtension } from './notion-html.js';
 import { BookmarkBlockPlainTextAdapterExtension } from './plain-text.js';
 
 export const BookmarkBlockAdapterExtensions: ExtensionType[] = [
   BookmarkBlockHtmlAdapterExtension,
-  BookmarkBlockMarkdownAdapterExtension,
+  BookmarkBlockMarkdownAdapterExtensions,
   BookmarkBlockNotionHtmlAdapterExtension,
   BookmarkBlockPlainTextAdapterExtension,
-];
+].flat();

blocksuite/affine/blocks/bookmark/src/adapters/index.ts

@@ -1,4 +1,4 @@
 export * from './html.js';
-export * from './markdown.js';
+export * from './markdown/index.js';
 export * from './notion-html.js';
 export * from './plain-text.js';

blocksuite/affine/blocks/bookmark/src/adapters/markdown/index.ts

@@ -0,0 +1,12 @@
+import type { ExtensionType } from '@blocksuite/store';
+
+import { BookmarkBlockMarkdownAdapterExtension } from './markdown.js';
+import { BookmarkBlockMarkdownPreprocessorExtension } from './preprocessor.js';
+
+export * from './markdown.js';
+export * from './preprocessor.js';
+
+export const BookmarkBlockMarkdownAdapterExtensions: ExtensionType[] = [
+  BookmarkBlockMarkdownPreprocessorExtension,
+  BookmarkBlockMarkdownAdapterExtension,
+];

blocksuite/affine/blocks/bookmark/src/adapters/markdown/preprocessor.ts

@@ -0,0 +1,67 @@
+import {
+  type MarkdownAdapterPreprocessor,
+  MarkdownPreprocessorExtension,
+} from '@blocksuite/affine-shared/adapters';
+
+// Check if a URL is already encoded with encodeURIComponent
+function isEncoded(uri: string): boolean {
+  try {
+    // If decoding produces a different result than the original,
+    // then the URI contains encoded characters
+    return uri !== decodeURIComponent(uri);
+  } catch {
+    // If decoding fails, the URI contains invalid percent-encoding
+    return true;
+  }
+}
+
+// Format footnote definition with consistent spacing
+function formatFootnoteDefinition(reference: string, data: object): string {
+  return `[^${reference}]: ${JSON.stringify(data)}`;
+}
+
+/**
+ * Preprocessor for footnote url
+ * We should encode url in footnote definition to avoid markdown link parsing
+ *
+ * Example of footnote definition:
+ * [^ref]: {"type":"url","url":"https://example.com"}
+ */
+export function footnoteUrlPreprocessor(content: string): string {
+  // Match footnote definitions with any reference (not just numbers)
+  // Format: [^reference]: {json_content}
+  return content.replace(
+    /\[\^([^\]]+)\]:\s*({[\s\S]*?})/g,
+    (match, reference, jsonContent) => {
+      try {
+        const footnoteData = JSON.parse(jsonContent.trim());
+        // If footnoteData is not an object or doesn't have url, return original content
+        // If the url is already encoded, return original content
+        if (
+          typeof footnoteData !== 'object' ||
+          !footnoteData.url ||
+          isEncoded(footnoteData.url)
+        ) {
+          return match;
+        }
+
+        return formatFootnoteDefinition(reference, {
+          ...footnoteData,
+          url: encodeURIComponent(footnoteData.url),
+        });
+      } catch {
+        // Keep original content if JSON parsing fails
+        return match;
+      }
+    }
+  );
+}
+
+const bookmarkBlockPreprocessor: MarkdownAdapterPreprocessor = {
+  name: 'bookmark-block',
+  levels: ['block', 'slice', 'doc'],
+  preprocess: footnoteUrlPreprocessor,
+};
+
+export const BookmarkBlockMarkdownPreprocessorExtension =
+  MarkdownPreprocessorExtension(bookmarkBlockPreprocessor);

blocksuite/affine/blocks/bookmark/vitest.config.ts

@@ -0,0 +1,25 @@
+import { defineConfig } from 'vitest/config';
+
+export default defineConfig({
+  esbuild: {
+    target: 'es2018',
+  },
+  test: {
+    browser: {
+      enabled: true,
+      headless: true,
+      name: 'chromium',
+      provider: 'playwright',
+      isolate: false,
+      providerOptions: {},
+    },
+    include: ['src/__tests__/**/*.unit.spec.ts'],
+    testTimeout: 500,
+    coverage: {
+      provider: 'istanbul',
+      reporter: ['lcov'],
+      reportsDirectory: '../../../.coverage/bookmark',
+    },
+    restoreMocks: true,
+  },
+});

yarn.lock

@@ -2417,6 +2417,7 @@ __metadata:
     lit: "npm:^3.2.0"
     minimatch: "npm:^10.0.1"
     rxjs: "npm:^7.8.1"
+    vitest: "npm:3.1.1"
     yjs: "npm:^13.6.23"
     zod: "npm:^3.23.8"
   languageName: unknown