Fix translation corruption patterns causing search scraping timeouts (#60544)

Author: Ebonsignori

src/languages/lib/correct-translation-content.ts

@@ -188,6 +188,13 @@ export function correctTranslatedContentStrings(
     content = content.replaceAll('{% dados variables', '{% data variables')
     content = content.replaceAll('{% de dados variables', '{% data variables')
     content = content.replaceAll('{% dados reusables', '{% data reusables')
+    // Fully translated reusables path: `{% dados reutilizáveis.X.Y %}` → `{% data reusables.X.Y %}`
+    content = content.replaceAll('{% dados reutilizáveis.', '{% data reusables.')
+    // Translated path segment inside reusables path: `repositórios` → `repositories`
+    content = content.replaceAll(
+      '{% data reusables.repositórios.',
+      '{% data reusables.repositories.',
+    )
     content = content.replaceAll('{{% dados ', '{% data ')
     content = content.replaceAll('{{% datas ', '{% data ')
     content = content.replaceAll('{% senão %}', '{% else %}')
@@ -375,6 +382,18 @@ export function correctTranslatedContentStrings(
     content = content.replaceAll('{% conseil %}', '{% tip %}')
     content = content.replaceAll('{%- conseil %}', '{%- tip %}')
     content = content.replaceAll('{%- conseil -%}', '{%- tip -%}')
+    // Remove orphaned {% endif %} tags when no ifversion/elsif opener exists in the content.
+    // Caused by translations where only the closing tag survived (e.g. user-api.md reusable).
+    if (
+      !content.includes('{% ifversion ') &&
+      !content.includes('{%- ifversion ') &&
+      !content.includes('{% elsif ') &&
+      !content.includes('{%- elsif ')
+    ) {
+      content = content.replaceAll('{% endif %}', '')
+      content = content.replaceAll('{%- endif %}', '')
+      content = content.replaceAll('{%- endif -%}', '')
+    }
   }
 
   if (context.code === 'ko') {
@@ -403,6 +422,11 @@ export function correctTranslatedContentStrings(
     content = content.replaceAll('{% daten variables', '{% data variables')
     content = content.replaceAll('{% Data variables', '{% data variables')
     content = content.replaceAll('{% Daten reusables', '{% data reusables')
+    content = content.replaceAll('{% Data reusables', '{% data reusables')
+    // `wiederverwendbare` is German for "reusables" — fix translated reusables paths
+    content = content.replaceAll('{% data wiederverwendbare.', '{% data reusables.')
+    content = content.replaceAll('{% Daten wiederverwendbare.', '{% data reusables.')
+    content = content.replaceAll('{% Data wiederverwendbare.', '{% data reusables.')
     content = content.replaceAll('{%-Daten variables', '{%- data variables')
     content = content.replaceAll('{%-Daten-variables', '{%- data variables')
     content = content.replaceAll('{%- ifversion fpt oder ghec %}', '{%- ifversion fpt or ghec %}')
@@ -474,6 +498,8 @@ export function correctTranslatedContentStrings(
 
   // Corrupted `{ endif %}%` → `{% endif %}` (delimiters shuffled)
   content = content.replaceAll('{ endif %}%', '{% endif %}')
+  // Corrupted `{ endif% %}` → `{% endif %}` (percent placed after keyword instead of after brace)
+  content = content.replaceAll('{ endif% %}', '{% endif %}')
   // Empty tag `{%}` (no space, no name) — typically `{% else %}`
   content = content.replace(/\{%\}(?!})/g, '{% else %}')
   // `{% }` or `{%  }` (tag with just `}` or spaces as name) — almost always `{% endif %}`

src/languages/tests/correct-translation-content.ts

@@ -211,6 +211,11 @@ describe('correctTranslatedContentStrings', () => {
         '{%- assign supportLevel = entry.support -%}',
       )
     })
+
+    test('fixes garbled endif with percent placed after keyword', () => {
+      // `{ endif% %}` — percent appears after "endif" instead of after the opening brace
+      expect(fix('some content\n{ endif% %}\nmore', 'ja')).toBe('some content\n{% endif %}\nmore')
+    })
   })
 
   // ─── PORTUGUESE (pt) ───────────────────────────────────────────────
@@ -264,6 +269,21 @@ describe('correctTranslatedContentStrings', () => {
       expect(fix('{% ifversion fpt ou ghec %}', 'pt')).toBe('{% ifversion fpt or ghec %}')
       expect(fix('{%- elsif fpt ou ghec %}', 'pt')).toBe('{%- elsif fpt or ghec %}')
     })
+
+    test('fixes fully translated reutilizáveis reusables path', () => {
+      // `reutilizáveis` is Portuguese for "reusables"
+      expect(fix('{% dados reutilizáveis.repositórios.reaction_list %}', 'pt')).toBe(
+        '{% data reusables.repositories.reaction_list %}',
+      )
+      expect(fix('{% dados reutilizáveis.foo.bar %}', 'pt')).toBe('{% data reusables.foo.bar %}')
+    })
+
+    test('fixes translated repositórios path segment', () => {
+      // `repositórios` is Portuguese for "repositories"
+      expect(fix('{% data reusables.repositórios.reaction_list %}', 'pt')).toBe(
+        '{% data reusables.repositories.reaction_list %}',
+      )
+    })
   })
 
   // ─── CHINESE (zh) ──────────────────────────────────────────────────
@@ -492,6 +512,25 @@ describe('correctTranslatedContentStrings', () => {
       expect(fix('{%- conseil %}', 'fr')).toBe('{%- tip %}')
       expect(fix('{%- conseil -%}', 'fr')).toBe('{%- tip -%}')
     })
+
+    test('removes orphaned endif when no matching ifversion/elsif opener exists', () => {
+      // Caused by translations where only the closing tag survived (e.g. user-api.md reusable)
+      expect(fix('Some content\n{% endif %}\nMore content', 'fr')).toBe(
+        'Some content\n\nMore content',
+      )
+      expect(fix('Line one\n{%- endif %}\nLine two', 'fr')).toBe('Line one\n\nLine two')
+      expect(fix('Text {%- endif -%} more', 'fr')).toBe('Text  more')
+    })
+
+    test('preserves endif when matching ifversion opener is present', () => {
+      const input = '{% ifversion ghec %}content{% endif %}'
+      expect(fix(input, 'fr')).toBe(input)
+    })
+
+    test('preserves endif when elsif opener is present', () => {
+      const input = '{% ifversion fpt %}a{% elsif ghec %}b{% endif %}'
+      expect(fix(input, 'fr')).toBe(input)
+    })
   })
 
   // ─── KOREAN (ko) ──────────────────────────────────────────────────
@@ -586,6 +625,21 @@ describe('correctTranslatedContentStrings', () => {
       )
       expect(fix('{% für entry in list %}', 'de')).toBe('{% for entry in list %}')
     })
+
+    test('fixes wiederverwendbare reusables path', () => {
+      // `wiederverwendbare` is German for "reusables"
+      expect(fix('{% data wiederverwendbare.audit_log.reference %}', 'de')).toBe(
+        '{% data reusables.audit_log.reference %}',
+      )
+      expect(fix('{% Daten wiederverwendbare.audit_log.reference %}', 'de')).toBe(
+        '{% data reusables.audit_log.reference %}',
+      )
+      // Full real-world example: `{% Data wiederverwendbare.audit_log.referenz-nach-kategorie-gruppiert %}`
+      // The `{% Data ` → `{% data ` fix runs before this, so by the time we check:
+      expect(
+        fix('{% Data wiederverwendbare.audit_log.referenz-nach-kategorie-gruppiert %}', 'de'),
+      ).toBe('{% data reusables.audit_log.referenz-nach-kategorie-gruppiert %}')
+    })
   })
 
   // ─── GENERIC FIXES ────────────────────────────────────────────────