perf(fmt): optimize Yuku parser selection (#312)

Author: chenjiahan

packages/rstack/src/fmt/yukuPlugin.ts

@@ -1,17 +1,16 @@
 import * as prettierEstreePlugin from 'prettier/plugins/estree';
 import type { Parser, ParserOptions, Plugin, SupportLanguage } from 'prettier';
 import {
+  langFromPath,
   parse as parseWithYuku,
   type Comment,
   type Diagnostic,
   type ParseOptions,
   type ParseResult,
-  type SourceLang,
   type SourceType,
 } from 'yuku-parser';
 
 const AST_FORMAT = 'estree-yuku';
-const JSX_REGEXP = /^[^"'`]*<\/|^[^/]{2}.*\/>/m;
 const SOURCE_TYPE_COMBINATIONS: SourceType[] = ['module', 'commonjs'];
 
 type Range = [start: number, end: number];
@@ -200,15 +199,24 @@ const mergeNestedJsdocComments = (comments: PrettierComment[]): void => {
 };
 
 const stripComments = (originalText: string, comments: PrettierComment[]): string => {
-  let text = originalText;
+  if (comments.length === 0) {
+    return originalText;
+  }
 
+  const chunks: string[] = [];
+  let cursor = 0;
+
+  // Yuku returns comments in source order, so mask each range while copying the source only once.
   for (const comment of comments) {
     const start = locStart(comment);
     const end = locEnd(comment);
-    text = text.slice(0, start) + text.slice(start, end).replace(/[^\n]/g, ' ') + text.slice(end);
+    chunks.push(originalText.slice(cursor, start));
+    chunks.push(originalText.slice(start, end).replace(/[^\n]/g, ' '));
+    cursor = end;
   }
 
-  return text;
+  chunks.push(originalText.slice(cursor));
+  return chunks.join('');
 };
 
 const setContentEnd = (
@@ -441,11 +449,7 @@ const parseWithOptions = (text: string, options: ParseOptions): ParseResult => {
   return result;
 };
 
-const getSourceType = (filepath: unknown): SourceType | undefined => {
-  if (typeof filepath !== 'string') {
-    return undefined;
-  }
-
+const getSourceType = (filepath: string): SourceType | undefined => {
   if (/\.(?:mjs|mts)$/i.test(filepath)) {
     return 'module';
   }
@@ -457,20 +461,6 @@ const getSourceType = (filepath: unknown): SourceType | undefined => {
   return undefined;
 };
 
-const getLanguageCombinations = (text: string, filepath: unknown): SourceLang[] => {
-  if (typeof filepath === 'string') {
-    if (/\.(?:jsx|tsx)$/i.test(filepath)) {
-      return ['tsx'];
-    }
-
-    if (filepath.toLowerCase().endsWith('.d.ts')) {
-      return ['dts'];
-    }
-  }
-
-  return JSX_REGEXP.test(text) ? ['tsx', 'ts', 'dts'] : ['ts', 'tsx', 'dts'];
-};
-
 const tryCombinations = (combinations: (() => ParseResult)[]): ParseResult => {
   let firstError: unknown;
   let hasError = false;
@@ -505,9 +495,9 @@ const parseJavaScript = (text: string, options: ParserOptions<AstNode>): AstNode
 
 const parseTypeScript = (text: string, options: ParserOptions<AstNode>): AstNode => {
   const sourceType = getSourceType(options.filepath);
-  const languages = getLanguageCombinations(text, options.filepath);
-  const combinations = (sourceType ? [sourceType] : SOURCE_TYPE_COMBINATIONS).flatMap((candidate) =>
-    languages.map((lang) => () => parseWithOptions(text, { sourceType: candidate, lang })),
+  const lang = langFromPath(options.filepath.toLowerCase());
+  const combinations = (sourceType ? [sourceType] : SOURCE_TYPE_COMBINATIONS).map(
+    (candidate) => () => parseWithOptions(text, { sourceType: candidate, lang }),
   );
   const { program, comments } = tryCombinations(combinations);
 

packages/rstack/tests/fmt/yukuPlugin.test.ts

@@ -7,9 +7,9 @@ const formatWithYuku = (
   options: Options & { parser: 'yuku' | 'yuku-ts' },
 ): Promise<string> =>
   format(source, {
-    filepath: `example.${options.parser === 'yuku' ? 'js' : 'ts'}`,
     plugins: [yukuPlugin],
     ...options,
+    filepath: options.filepath ?? `example.${options.parser === 'yuku' ? 'js' : 'ts'}`,
   });
 
 test('exposes the same JavaScript and TypeScript language mappings as the official plugin', async () => {
@@ -34,6 +34,39 @@ test('exposes the same JavaScript and TypeScript language mappings as the offici
   ]);
 });
 
+test('parses JSX in JavaScript files', async () => {
+  await expect(
+    formatWithYuku('const view=<Component/>', {
+      filepath: 'example.js',
+      parser: 'yuku',
+    }),
+  ).resolves.toBe('const view = <Component />;\n');
+});
+
+test.each(['example.ts', 'example.mts', 'example.cts'])(
+  'rejects JSX syntax in %s',
+  async (filepath) => {
+    await expect(
+      formatWithYuku('const view=<Component/>', {
+        filepath,
+        parser: 'yuku-ts',
+      }),
+    ).rejects.toThrow();
+  },
+);
+
+test.each(['example.d.ts', 'example.d.mts', 'example.d.cts'])(
+  'rejects function implementations in %s',
+  async (filepath) => {
+    await expect(
+      formatWithYuku('export function value() { return 1; }', {
+        filepath,
+        parser: 'yuku-ts',
+      }),
+    ).rejects.toThrow('An implementation cannot be declared in ambient contexts');
+  },
+);
+
 test.each([
   {
     name: 'hashbangs and unicode locations',