Add multilingual audit CI pipeline + extract mandelblog_content_guard

2026-03-29 20:49:42 +02:00
parent 2a51989fa4
commit 1f05011a63
104 changed files with 3372 additions and 6 deletions
--- a/mandelblog_content_guard/validators/rules/cta.py
+++ b/mandelblog_content_guard/validators/rules/cta.py
@@ -0,0 +1,146 @@
+from __future__ import annotations
+
+import re
+
+from ...types import make_issue
+
+CTA_RULES = {
+    "nl": (
+        r"^Plan ",
+        r"^Bekijk ",
+        r"^Vraag ",
+        r"^Bespreek ",
+        r"^Contact$",
+        r"^Start ",
+        r"^Meer ",
+        r"^Verstuur ",
+        r"^Neem ",
+    ),
+    "en": (
+        r"^Book ",
+        r"^View ",
+        r"^Schedule ",
+        r"^Start ",
+        r"^Talk ",
+        r"^Discuss ",
+        r"^Contact$",
+        r"^Explore ",
+        r"^Learn ",
+        r"^Request ",
+        r"^Send ",
+    ),
+    "de": (
+        r"^Plan",
+        r"^Mehr",
+        r"^Support",
+        r"^Start",
+        r"^Kontakt",
+        r"^Gespr",
+        r"^Kostenlose",
+        r"^Anfrage",
+        r"^Projekte",
+        r"^Verein",
+        r"^Besprech",
+        r"^Anzeig",
+        r"^Ansehen",
+        r"^Technisch",
+        r"^Unterst",
+        r"^Unsere",
+        r"^Service",
+        r"^Dienstleistungen",
+        r"^Erstgespräch",
+        r"^Einführ",
+        r"^Anpassung",
+        r"^Ansichts",
+        r"^Prozess",
+        r"^Pakete",
+        r"^Demo",
+        r"^Alle ",
+        r"^Ein ",
+        r"^Webshop",
+    ),
+    "fr": (
+        r"^Planifier",
+        r"^Voir",
+        r"^Découvrir",
+        r"^Demander",
+        r"^Lancer",
+        r"^Démarrer",
+        r"^Contacter",
+        r"^Contact$",
+        r"^Parler",
+        r"^Lancez",
+        r"^Prendre",
+        r"^Envoyer",
+        r"^Afficher",
+    ),
+    "es": (
+        r"^Reservar",
+        r"^Ver",
+        r"^Solicitar",
+        r"^Inicia",
+        r"^Hablar",
+        r"^Descubrir",
+        r"^Contactar",
+        r"^Planificar",
+        r"^Programe",
+        r"^Concertar",
+        r"^Enviar",
+        r"^Mostrar",
+        r"^Comenta",
+    ),
+    "it": (
+        r"^Prenota",
+        r"^Vedi",
+        r"^Avvia",
+        r"^Richiedi",
+        r"^Contatta",
+        r"^Contatto$",
+        r"^Scopri",
+        r"^Pianifica",
+        r"^Invia",
+        r"^Mostra",
+        r"^Parla",
+        r"^Parliamo",
+    ),
+    "pt": (
+        r"^Agendar",
+        r"^Ver",
+        r"^Iniciar",
+        r"^Pedir",
+        r"^Contactar",
+        r"^Falar",
+        r"^Explorar",
+        r"^Marcar",
+        r"^Solicitar",
+        r"^Enviar",
+        r"^Mostrar",
+    ),
+    "ru": (
+        r"^Заплан",
+        r"^Посмотр",
+        r"^Запуст",
+        r"^Связ",
+        r"^Подробнее",
+        r"^Показать",
+        r"^Отправ",
+        r"^Получ",
+        r"^Запрос",
+    ),
+}
+
+CTA_FIELDS = {
+    "cta_text",
+    "primary_cta_text",
+    "secondary_cta_text",
+    "submit_button_text",
+}
+
+
+def validate_cta(locale_code: str, field_path: str, normalized: str):
+    last_segment = field_path.split(".")[-1]
+    if last_segment not in CTA_FIELDS:
+        return []
+    if any(re.search(pattern, normalized) for pattern in CTA_RULES.get(locale_code, ())):
+        return []
+    return [make_issue("cta_language_mismatch", field_path, normalized)]