Prevent link highlight in markdown code blocks and spans (#140816)

* Prevent link highlight in markdown codeblocks (#139770) * Handle inline codespan variants for markdown link provider (#139770) * Refactor codespan detection in markdown link provider (#139770)
2025-12-24 12:19:20 +00:00 · 2022-01-26 04:38:50 +08:00
parent 283f478fc3
commit a514979b02
4 changed files with 153 additions and 39 deletions
--- a/extensions/markdown-language-features/src/extension.ts
+++ b/extensions/markdown-language-features/src/extension.ts
@@ -55,7 +55,7 @@ function registerMarkdownLanguageFeatures(

 	return vscode.Disposable.from(
 		vscode.languages.registerDocumentSymbolProvider(selector, symbolProvider),
-		vscode.languages.registerDocumentLinkProvider(selector, new LinkProvider()),
+		vscode.languages.registerDocumentLinkProvider(selector, new LinkProvider(engine)),
 		vscode.languages.registerFoldingRangeProvider(selector, new MarkdownFoldingProvider(engine)),
 		vscode.languages.registerSelectionRangeProvider(selector, new MarkdownSmartSelect(engine)),
 		vscode.languages.registerWorkspaceSymbolProvider(new MarkdownWorkspaceSymbolProvider(symbolProvider)),
--- a/extensions/markdown-language-features/src/features/documentLinkProvider.ts
+++ b/extensions/markdown-language-features/src/features/documentLinkProvider.ts
@@ -6,6 +6,7 @@
 import * as vscode from 'vscode';
 import * as nls from 'vscode-nls';
 import { OpenDocumentLinkCommand } from '../commands/openDocumentLink';
+import { MarkdownEngine } from '../markdownEngine';
 import { getUriForLinkWithKnownExternalScheme, isOfScheme, Schemes } from '../util/links';
 import { dirname } from '../util/path';

@@ -105,33 +106,66 @@ export function stripAngleBrackets(link: string) {
 const linkPattern = /(\[((!\[[^\]]*?\]\(\s*)([^\s\(\)]+?)\s*\)\]|(?:\\\]|[^\]])*\])\(\s*)(([^\s\(\)]|\([^\s\(\)]*?\))+)\s*(".*?")?\)/g;
 const referenceLinkPattern = /(\[((?:\\\]|[^\]])+)\]\[\s*?)([^\s\]]*?)\]/g;
 const definitionPattern = /^([\t ]*\[(?!\^)((?:\\\]|[^\]])+)\]:\s*)([^<]\S*|<[^>]+>)/gm;
+const inlineCodePattern = /(?:(?<!`)(`+)(?!`)(?:.+?|.*?(?:(?:\r?\n).+?)*?)(?:\r?\n)?(?<!`)\1(?!`))/g;
+
+type CodeInDocument = {
+	/**
+	 * code blocks and fences each represented by [line_start,line_end).
+	 */
+	multiline: [number, number][];
+	/**
+	 * inline code spans each represented by {@link vscode.Range}.
+	 */
+	inline: vscode.Range[];
+};
+
+async function findCode(document: vscode.TextDocument, engine: MarkdownEngine): Promise<CodeInDocument> {
+	const tokens = await engine.parse(document);
+	const multiline = tokens.filter(t => (t.type === 'code_block' || t.type === 'fence') && !!t.map).map(t => t.map) as [number, number][];
+
+	const text = document.getText();
+	const inline = [...text.matchAll(inlineCodePattern)].map(match => {
+		const start = match.index || 0;
+		return new vscode.Range(document.positionAt(start), document.positionAt(start + match[0].length));
+	});
+
+	return { multiline, inline };
+}
+
+function isLinkInsideCode(code: CodeInDocument, link: vscode.DocumentLink) {
+	return code.multiline.some(interval => link.range.start.line >= interval[0] && link.range.start.line < interval[1]) ||
+		code.inline.some(position => position.intersection(link.range));
+}

 export default class LinkProvider implements vscode.DocumentLinkProvider {
+	constructor(
+		private readonly engine: MarkdownEngine
+	) { }

-	public provideDocumentLinks(
+	public async provideDocumentLinks(
 		document: vscode.TextDocument,
 		_token: vscode.CancellationToken
-	): vscode.DocumentLink[] {
+	): Promise<vscode.DocumentLink[]> {
 		const text = document.getText();
-
 		return [
-			...this.providerInlineLinks(text, document),
+			...(await this.providerInlineLinks(text, document)),
 			...this.provideReferenceLinks(text, document)
 		];
 	}

-	private providerInlineLinks(
+	private async providerInlineLinks(
 		text: string,
 		document: vscode.TextDocument,
-	): vscode.DocumentLink[] {
+	): Promise<vscode.DocumentLink[]> {
 		const results: vscode.DocumentLink[] = [];
+		const codeInDocument = await findCode(document, this.engine);
 		for (const match of text.matchAll(linkPattern)) {
 			const matchImage = match[4] && extractDocumentLink(document, match[3].length + 1, match[4], match.index);
-			if (matchImage) {
+			if (matchImage && !isLinkInsideCode(codeInDocument, matchImage)) {
 				results.push(matchImage);
 			}
 			const matchLink = extractDocumentLink(document, match[1].length, match[5], match.index);
-			if (matchLink) {
+			if (matchLink && !isLinkInsideCode(codeInDocument, matchLink)) {
 				results.push(matchLink);
 			}
 		}
--- a/extensions/markdown-language-features/src/test/documentLinkProvider.test.ts
+++ b/extensions/markdown-language-features/src/test/documentLinkProvider.test.ts
@@ -7,15 +7,16 @@ import * as assert from 'assert';
 import 'mocha';
 import * as vscode from 'vscode';
 import LinkProvider from '../features/documentLinkProvider';
+import { createNewMarkdownEngine } from './engine';
 import { InMemoryDocument } from './inMemoryDocument';
-import { noopToken } from './util';
+import { joinLines, noopToken } from './util';


 const testFile = vscode.Uri.joinPath(vscode.workspace.workspaceFolders![0].uri, 'x.md');

 function getLinksForFile(fileContents: string) {
 	const doc = new InMemoryDocument(testFile, fileContents);
-	const provider = new LinkProvider();
+	const provider = new LinkProvider(createNewMarkdownEngine());
 	return provider.provideDocumentLinks(doc, noopToken);
 }

@@ -27,63 +28,63 @@ function assertRangeEqual(expected: vscode.Range, actual: vscode.Range) {
 }

 suite('markdown.DocumentLinkProvider', () => {
-	test('Should not return anything for empty document', () => {
-		const links = getLinksForFile('');
+	test('Should not return anything for empty document', async () => {
+		const links = await getLinksForFile('');
 		assert.strictEqual(links.length, 0);
 	});

-	test('Should not return anything for simple document without links', () => {
-		const links = getLinksForFile('# a\nfdasfdfsafsa');
+	test('Should not return anything for simple document without links', async () => {
+		const links = await getLinksForFile('# a\nfdasfdfsafsa');
 		assert.strictEqual(links.length, 0);
 	});

-	test('Should detect basic http links', () => {
-		const links = getLinksForFile('a [b](https://example.com) c');
+	test('Should detect basic http links', async () => {
+		const links = await getLinksForFile('a [b](https://example.com) c');
 		assert.strictEqual(links.length, 1);
 		const [link] = links;
 		assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 25));
 	});

-	test('Should detect basic workspace links', () => {
+	test('Should detect basic workspace links', async () => {
 		{
-			const links = getLinksForFile('a [b](./file) c');
+			const links = await getLinksForFile('a [b](./file) c');
 			assert.strictEqual(links.length, 1);
 			const [link] = links;
 			assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 12));
 		}
 		{
-			const links = getLinksForFile('a [b](file.png) c');
+			const links = await getLinksForFile('a [b](file.png) c');
 			assert.strictEqual(links.length, 1);
 			const [link] = links;
 			assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 14));
 		}
 	});

-	test('Should detect links with title', () => {
-		const links = getLinksForFile('a [b](https://example.com "abc") c');
+	test('Should detect links with title', async () => {
+		const links = await getLinksForFile('a [b](https://example.com "abc") c');
 		assert.strictEqual(links.length, 1);
 		const [link] = links;
 		assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 25));
 	});

 	// #35245
-	test('Should handle links with escaped characters in name', () => {
-		const links = getLinksForFile('a [b\\]](./file)');
+	test('Should handle links with escaped characters in name', async () => {
+		const links = await getLinksForFile('a [b\\]](./file)');
 		assert.strictEqual(links.length, 1);
 		const [link] = links;
 		assertRangeEqual(link.range, new vscode.Range(0, 8, 0, 14));
 	});


-	test('Should handle links with balanced parens', () => {
+	test('Should handle links with balanced parens', async () => {
 		{
-			const links = getLinksForFile('a [b](https://example.com/a()c) c');
+			const links = await getLinksForFile('a [b](https://example.com/a()c) c');
 			assert.strictEqual(links.length, 1);
 			const [link] = links;
 			assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 30));
 		}
 		{
-			const links = getLinksForFile('a [b](https://example.com/a(b)c) c');
+			const links = await getLinksForFile('a [b](https://example.com/a(b)c) c');
 			assert.strictEqual(links.length, 1);
 			const [link] = links;
 			assertRangeEqual(link.range, new vscode.Range(0, 6, 0, 31));
@@ -91,15 +92,15 @@ suite('markdown.DocumentLinkProvider', () => {
 		}
 		{
 			// #49011
-			const links = getLinksForFile('[A link](http://ThisUrlhasParens/A_link(in_parens))');
+			const links = await getLinksForFile('[A link](http://ThisUrlhasParens/A_link(in_parens))');
 			assert.strictEqual(links.length, 1);
 			const [link] = links;
 			assertRangeEqual(link.range, new vscode.Range(0, 9, 0, 50));
 		}
 	});

-	test('Should handle two links without space', () => {
-		const links = getLinksForFile('a ([test](test)[test2](test2)) c');
+	test('Should handle two links without space', async () => {
+		const links = await getLinksForFile('a ([test](test)[test2](test2)) c');
 		assert.strictEqual(links.length, 2);
 		const [link1, link2] = links;
 		assertRangeEqual(link1.range, new vscode.Range(0, 10, 0, 14));
@@ -107,23 +108,23 @@ suite('markdown.DocumentLinkProvider', () => {
 	});

 	// #49238
-	test('should handle hyperlinked images', () => {
+	test('should handle hyperlinked images', async () => {
 		{
-			const links = getLinksForFile('[![alt text](image.jpg)](https://example.com)');
+			const links = await getLinksForFile('[![alt text](image.jpg)](https://example.com)');
 			assert.strictEqual(links.length, 2);
 			const [link1, link2] = links;
 			assertRangeEqual(link1.range, new vscode.Range(0, 13, 0, 22));
 			assertRangeEqual(link2.range, new vscode.Range(0, 25, 0, 44));
 		}
 		{
-			const links = getLinksForFile('[![a]( whitespace.jpg )]( https://whitespace.com )');
+			const links = await getLinksForFile('[![a]( whitespace.jpg )]( https://whitespace.com )');
 			assert.strictEqual(links.length, 2);
 			const [link1, link2] = links;
 			assertRangeEqual(link1.range, new vscode.Range(0, 7, 0, 21));
 			assertRangeEqual(link2.range, new vscode.Range(0, 26, 0, 48));
 		}
 		{
-			const links = getLinksForFile('[![a](img1.jpg)](file1.txt) text [![a](img2.jpg)](file2.txt)');
+			const links = await getLinksForFile('[![a](img1.jpg)](file1.txt) text [![a](img2.jpg)](file2.txt)');
 			assert.strictEqual(links.length, 4);
 			const [link1, link2, link3, link4] = links;
 			assertRangeEqual(link1.range, new vscode.Range(0, 6, 0, 14));
@@ -133,13 +134,13 @@ suite('markdown.DocumentLinkProvider', () => {
 		}
 	});

-	test('Should not consider link references starting with ^ character valid (#107471)', () => {
-		const links = getLinksForFile('[^reference]: https://example.com');
+	test('Should not consider link references starting with ^ character valid (#107471)', async () => {
+		const links = await getLinksForFile('[^reference]: https://example.com');
 		assert.strictEqual(links.length, 0);
 	});

-	test('Should find definitions links with spaces in angle brackets (#136073)', () => {
-		const links = getLinksForFile([
+	test('Should find definitions links with spaces in angle brackets (#136073)', async () => {
+		const links = await getLinksForFile([
 			'[a]: <b c>',
 			'[b]: <cd>',
 		].join('\n'));
@@ -149,6 +150,75 @@ suite('markdown.DocumentLinkProvider', () => {
 		assertRangeEqual(link1.range, new vscode.Range(0, 6, 0, 9));
 		assertRangeEqual(link2.range, new vscode.Range(1, 6, 1, 8));
 	});
+
+	test('Should not consider links in code fenced with backticks', async () => {
+		const text = joinLines(
+			'```',
+			'[b](https://example.com)',
+			'```');
+		const links = await getLinksForFile(text);
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links in code fenced with tilda', async () => {
+		const text = joinLines(
+			'~~~',
+			'[b](https://example.com)',
+			'~~~');
+		const links = await getLinksForFile(text);
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links in indented code', async () => {
+		const links = await getLinksForFile('    [b](https://example.com)');
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links in inline code span', async () => {
+		const links = await getLinksForFile('`[b](https://example.com)`');
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links with code span inside', async () => {
+		const links = await getLinksForFile('[li`nk](https://example.com`)');
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links in multiline inline code span', async () => {
+		const text = joinLines(
+			'`` ',
+			'[b](https://example.com)',
+			'``');
+		const links = await getLinksForFile(text);
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not consider links in multiline inline code span between between text', async () => {
+		const text = joinLines(
+			'[b](https://1.com) `[b](https://2.com)',
+			'` [b](https://3.com)');
+		const links = await getLinksForFile(text);
+		assert.deepStrictEqual(links.map(l => l.target?.authority), ['1.com', '3.com'])
+	});
+
+	test('Should not consider links in multiline inline code span with new line after the first backtick', async () => {
+		const text = joinLines(
+			'`',
+			'[b](https://example.com)`');
+		const links = await getLinksForFile(text);
+		assert.strictEqual(links.length, 0);
+	});
+
+	test('Should not miss links in invalid multiline inline code span', async () => {
+		const text = joinLines(
+			'`` ',
+			'',
+			'[b](https://example.com)',
+			'',
+			'``');
+		const links = await getLinksForFile(text);
+		assert.strictEqual(links.length, 1);
+	});
 });