fix(astro-markflow): handle single-line getHeadings in RE_SCAN_NOISE regex

packages/astro-markflow/src/transforms/inject-components.test.ts

@@ -133,6 +133,34 @@ export default function Content() {
     expect(result).toContain('import { Aside }');
   });
 
+  it('should strip single-line getHeadings before scanning', () => {
+    const code = `
+export function getHeadings() { return [{"depth":1,"slug":"aside","text":"Aside"}]; }
+
+export default function Content() {
+  return <Aside>Content</Aside>;
+}`;
+
+    const result = injectComponentImports(code, ['Aside'], '@astrojs/starlight/components');
+
+    expect(result).toContain('import { Aside }');
+  });
+
+  it('should not false-positive when heading text contains semicolons before component names', () => {
+    const code = `
+export const headings = [
+  { depth: 2, slug: 'use-aside', text: 'Use; <Aside> wisely' }
+];
+
+export default function Content() {
+  return <div>No components here</div>;
+}`;
+
+    const result = injectComponentImports(code, ['Aside'], '@astrojs/starlight/components');
+
+    expect(result).not.toContain('import');
+  });
+
   it('should handle components in nested structures', () => {
     const code = `
 export default function Content() {

packages/astro-markflow/src/transforms/inject-components.ts

@@ -7,11 +7,29 @@ import type { Registry } from 'markflow/registry';
 import { astroLibrary } from 'markflow/registry';
 import { collectImportedNames, insertAfterImports } from '../utils/imports.js';
 import { resolveStarlightConfig, type StarlightUserConfig } from '../utils/config.js';
-import { stripHeadingsMeta } from '../utils/validation.js';
 
-/** Strip set:html={...} string content to avoid false component matches in code blocks */
-function stripSetHtmlContent(code: string): string {
-  return code.replace(/set:html=\{("(?:[^"\\]|\\.)*")\}/g, 'set:html={""}');
+
+/** Combined regex to strip set:html content and headings metadata in a single pass */
+const RE_SCAN_NOISE = /set:html=\{"(?:[^"\\]|\\.)*"\}|export const headings[\s\S]*?];\n?|export function getHeadings\(\) \{(?:[^\n]+\n|[\s\S]*?\n\}\n?)/g;
+
+/** Strip set:html content and headings metadata in a single pass to avoid false component matches */
+function stripScanNoise(code: string): string {
+  return code.replace(RE_SCAN_NOISE, (match) =>
+    match.startsWith('set:html=') ? 'set:html={""}' : ''
+  );
+}
+
+/** Cache compiled regex per registry instance (registry is immutable after init) */
+const registryRegexCache = new WeakMap<Registry, RegExp>();
+
+function getComponentPattern(registry: Registry): RegExp | null {
+  let cached = registryRegexCache.get(registry);
+  if (cached) return cached;
+  const all = registry.getAllComponents();
+  if (all.length === 0) return null;
+  cached = new RegExp(`<(${all.map(c => c.name).join('|')})\\b`, 'g');
+  registryRegexCache.set(registry, cached);
+  return cached;
 }
 
 /**
@@ -35,7 +53,7 @@ export function injectComponentImports(
   if (!code || typeof code !== 'string' || components.length === 0) {
     return code;
   }
-  const scanTarget = stripSetHtmlContent(stripHeadingsMeta(code));
+  const scanTarget = stripScanNoise(code);
 
   // PERF: Use single combined regex instead of per-component regex
   // This reduces from O(n) regex compilations to O(1)
@@ -52,7 +70,7 @@ export function injectComponentImports(
   const used = components.filter((name) => usedSet.has(name));
   if (used.length === 0) return code;
 
-  const imported = collectImportedNames(code);
+  const imported = collectImportedNames(code, { skipCodeFences: true });
   const missing = used.filter((name) => !imported.has(name));
   if (missing.length === 0) return code;
 
@@ -123,18 +141,13 @@ export function injectComponentImportsFromRegistry(
     return code;
   }
 
-  const allComponents = registry.getAllComponents();
-  if (allComponents.length === 0) {
-    return code;
-  }
+  const combinedPattern = getComponentPattern(registry);
+  if (!combinedPattern) return code;
 
-  const scanTarget = stripSetHtmlContent(stripHeadingsMeta(code));
-  const imported = collectImportedNames(code);
+  const scanTarget = stripScanNoise(code);
+  const imported = collectImportedNames(code, { skipCodeFences: true });
 
-  // PERF: Use single combined regex instead of per-component regex
-  // This reduces from O(n) regex compilations to O(1)
-  const componentNames = allComponents.map((c) => c.name);
-  const combinedPattern = new RegExp(`<(${componentNames.join('|')})\\b`, 'g');
+  combinedPattern.lastIndex = 0;
   const matches = scanTarget.match(combinedPattern);
   if (!matches) return code;
 
@@ -148,6 +161,7 @@ export function injectComponentImportsFromRegistry(
   // Find used components that are missing imports
   const missingByModule = new Map<string, Array<{ name: string; exportType: string }>>();
 
+  const allComponents = registry.getAllComponents();
   for (const comp of allComponents) {
     if (usedNames.has(comp.name) && !imported.has(comp.name)) {
       const modulePath = comp.modulePath;