feat(parser): Integrate linkify into parser

.github/workflows/ci.yml

@@ -7,6 +7,7 @@ on:
     # Only trigger when relevant code changes to save resources
     paths:
       - '.github/workflows/ci.yml'
+      - 'packages/linkify/**'
       - 'packages/markdown-parser/**'
       - 'packages/markdown-parser-lit/**'
       - 'packages/markdown-parser-vue/**'
@@ -18,6 +19,7 @@ on:
   pull_request:
     paths:
       - '.github/workflows/ci.yml'
+      - 'packages/linkify/**'
       - 'packages/markdown-parser/**'
       - 'packages/markdown-parser-lit/**'
       - 'packages/markdown-parser-vue/**'
@@ -53,6 +55,8 @@ jobs:
           cache: pnpm
       - name: Install dependencies
         run: pnpm install --frozen-lockfile --ignore-scripts
+      - name: Build linkify package
+        run: pnpm --filter @fuyeor/linkify build
       - name: Typecheck parser package
         run: pnpm --filter @fuyeor/markdown-parser typecheck
       - name: Run parser unit tests

Dockerfile

@@ -14,6 +14,7 @@ WORKDIR /app
 # prepare depends
 COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 # copy shared packages
+COPY packages/linkify/package.json packages/linkify/
 COPY packages/markdown-parser/package.json packages/markdown-parser/
 COPY packages/markdown-parser-lit/package.json packages/markdown-parser-lit/
 COPY packages/playground/package.json packages/playground/
@@ -23,6 +24,8 @@ RUN --mount=type=cache,id=pnpm,target=/pnpm/store pnpm install --frozen-lockfile
 # copy source code
 COPY . .
 
+# build linkify before parser consumers resolve its package exports
+RUN pnpm --filter @fuyeor/linkify build
 # build parser package before workspace consumers resolve its package exports
 RUN pnpm --filter @fuyeor/markdown-parser build
 

packages/markdown-parser/package.json

@@ -18,6 +18,9 @@
   "exports": {
     ".": "./src/index.ts"
   },
+  "dependencies": {
+    "@fuyeor/linkify": "workspace:*"
+  },
   "devDependencies": {
     "@types/node": "^25.6.0",
     "typescript": "^6.0.3",

packages/markdown-parser/src/core/parser.ts

@@ -1,31 +1,40 @@
 // @fuyeor/markdown-parser/src/core/parser.ts
 import { BlockState, InlineState } from './state';
+import { linkify } from '@fuyeor/linkify';
 import type {
   ASTNode,
   BlockRule,
   InlineRule,
+  Linkifier,
   MarkdownPlugin,
   MarkdownParserOptions,
   ParserContext,
 } from '#/types';
 import { isSafeLinkUrl } from '#/core/url';
 
-const LINKIFY_REGEX =
-  /(https?:\/\/[^\s]+|(?<![@\w])(?:[a-zA-Z0-9\-]+\.)+[a-zA-Z]{2,}(?:\/[^\s]*)?)/gv;
+const LINKIFY_CANDIDATE_REGEX = /(?:https?:\/\/|[A-Za-z0-9-]+\.)/u;
+
+/** Skip linkify scans for text without a possible URL candidate. */
+function linkifyWithCandidateCheck(text: string): ReturnType<typeof linkify> {
+  if (!LINKIFY_CANDIDATE_REGEX.test(text)) return [];
+  return linkify(text);
+}
 
 export class MarkdownParser {
   #blockRuleMap = new Map<string, BlockRule[]>();
   #inlineRuleMap = new Map<string, InlineRule[]>();
   #idSequence = 0;
   #isPreflight = false;
   readonly #maxNestingDepth: number;
+  readonly #linkifier: Linkifier;
 
   constructor(options: MarkdownParserOptions = {}) {
     const maxNestingDepth = options.maxNestingDepth ?? 64;
     if (!Number.isInteger(maxNestingDepth) || maxNestingDepth < 1)
       throw new RangeError('maxNestingDepth must be a positive integer');
 
     this.#maxNestingDepth = maxNestingDepth;
+    this.#linkifier = options.linkifier ?? linkifyWithCandidateCheck;
   }
 
   // construct a context for recursive rule calls
@@ -227,44 +236,20 @@ export class MarkdownParser {
         const textBuffer = state.content.slice(textStart, endPos);
         let lastIdx = 0;
 
-        for (const match of textBuffer.matchAll(LINKIFY_REGEX)) {
-          let urlStr = match[0];
-          let matchIdx = match.index!;
-
-          // remove the punctuation marks at the end.
-          const punctuation = '.,:;?!';
-          while (
-            urlStr.length > 0 &&
-            punctuation.includes(urlStr[urlStr.length - 1])
-          ) {
-            urlStr = urlStr.slice(0, -1);
-          }
-
-          // handle bracket matching
-          // If a link ends with `)` and the inner brackets are unbalanced, then strip `)`
-          if (urlStr.endsWith(')')) {
-            const openCount = (urlStr.match(/\(/g) || []).length;
-            const closeCount = (urlStr.match(/\)/g) || []).length;
-            if (closeCount > openCount) urlStr = urlStr.slice(0, -1);
-          }
-
-          const fullUrl = urlStr.startsWith('http')
-            ? urlStr
-            : `https://${urlStr}`;
-
-          if (isSafeLinkUrl(fullUrl)) {
-            if (matchIdx > lastIdx) {
+        for (const match of this.#linkifier(textBuffer)) {
+          if (isSafeLinkUrl(match.url)) {
+            if (match.index > lastIdx) {
               nodes.push({
                 type: 'text',
-                content: textBuffer.slice(lastIdx, matchIdx),
+                content: textBuffer.slice(lastIdx, match.index),
               });
             }
             nodes.push({
               type: 'link',
-              url: fullUrl,
-              children: [{ type: 'text', content: urlStr }],
+              url: match.url,
+              children: [{ type: 'text', content: match.text }],
             });
-            lastIdx = matchIdx + urlStr.length;
+            lastIdx = match.lastIndex;
           }
         }
 

packages/markdown-parser/src/index.spec.ts

@@ -56,6 +56,16 @@ describe('test @fuyeor/markdown-parser', () => {
     expect(children[3].url).toBe('https://fuyeor.com');
   });
 
+  it('uses linkify for internationalized fuzzy domains', () => {
+    const ast = createFuyeorMarkdownParser()('Visit fuyeor.xn--p1ai');
+    const link = ast[0].children?.find((node) => node.type === 'link');
+
+    expect(link).toMatchObject({
+      url: 'https://fuyeor.рф',
+      children: [{ type: 'text', content: 'fuyeor.рф' }],
+    });
+  });
+
   it('parse table', () => {
     const content = `
 | title 1 | title 2 |

packages/markdown-parser/src/index.ts

@@ -27,6 +27,7 @@ export type {
   ParserContext,
   BlockRule,
   InlineRule,
+  Linkifier,
   MarkdownParserOptions,
   MarkdownPlugin,
 } from './types';

packages/markdown-parser/src/types.ts

@@ -1,5 +1,6 @@
 // @fuyeor/markdown-parser/src/types.ts
 
+import type { LinkMatch } from '@fuyeor/linkify';
 import type { MarkdownParser } from './core/parser';
 import type { BlockState, InlineState } from './core/state';
 
@@ -73,8 +74,11 @@ export interface InlineRule {
   ) => { node: ASTNode; consumedChars: number } | null;
 }
 
+export type Linkifier = (text: string) => readonly LinkMatch[];
+
 export interface MarkdownParserOptions {
   maxNestingDepth?: number;
+  linkifier?: Linkifier;
 }
 
 export type MarkdownPlugin = (parser: MarkdownParser) => void;