Add multilingual audit CI pipeline + extract mandelblog_content_guard

This commit is contained in:
2026-03-29 20:49:42 +02:00
parent 643aca26d0
commit e3bafd3a73
104 changed files with 3372 additions and 6 deletions

View File

@@ -0,0 +1,269 @@
from __future__ import annotations
import re
from ...types import make_issue
from ...system_strings import (
build_system_rewrite_candidates,
is_canonical_system_string,
system_string_replacement,
)
GLOBAL_BAD_PATTERNS = (
"The Spanish translation",
"The Spanish translation of",
"As the input",
"The input",
"Poiché l'input",
'Unternehmen" è tedesco',
"Support anzeigen",
"Starter intake",
"Business intake",
"Plan Starter intake",
"Plan Business intake",
"Plan de admisión",
"None",
)
LOCALE_FORBIDDEN = {
"nl": ("Starter intake", "Business intake", "Poiché", "Correo electrónico", "Mostrar los servicios", "Plan de admisión"),
"en": ("Starter intake", "Business intake", "Poiché", "Correo electrónico", "Mostrar los servicios", "Questions fréquemment posées", "Plan de admisión"),
"de": ("Starter intake", "Business intake", "Poiché", "Correo electrónico", "Mostrar los servicios", "Questions fréquemment posées", "Plan de admisión"),
"fr": ("Starter intake", "Business intake", "Poiché", "Correo electrónico", "Mostrar los servicios", "Plan de admisión", "Support anzeigen"),
"es": ("Poiché", 'Unternehmen" è tedesco', "Support anzeigen", "Questions fréquemment posées"),
"it": ("Poiché l'input", "Consulta inicial sin compromiso", "Mostrar los servicios", "Questions fréquentes", "Plan de admisión", "Correo electrónico"),
"pt": ("Poiché l'input", "Consulta inicial sin compromiso", "Mostrar los servicios", "Correo electrónico", 'Unternehmen" è tedesco', "Questions fréquemment posées"),
"ru": ("Poiché l'input", "Consulta inicial sin compromiso", "Correo electrónico", 'Unternehmen" è tedesco', "Mostrar los servicios"),
}
PLACEHOLDER_VALUES = {"None", "-", "N/A", "null"}
GENERIC_BADGE_LABELS = {
"New",
"Popular",
"PLAN",
"PIANO",
"SERVICES",
}
GLOBAL_REWRITE_CANDIDATES = {
**build_system_rewrite_candidates(
(
"days_label",
"average_delivery",
"response_time",
"without_commitment",
"transparent_label",
"weeks_1_2",
"customer_reviews",
"editable_label",
"core_pages_label",
"detailed_page_structure",
"business_process_cta",
"multilingual_rollout",
"customization_integrations",
"transparent_investment",
)
),
}
LOCALE_REWRITE_CANDIDATES = {
"en": {
"Service packages (from) Transparent starting points.": "foreign_ui_label",
"Frequently Asked Questions Transparent about planning, approach, and management.": "foreign_ui_label",
"Transparent investment": "foreign_ui_label",
},
"de": {
"New": "weak_marketing_copy",
"Intakegespräch": "weak_marketing_copy",
"SEO-ready basis": "foreign_ui_label",
"Sales-ready mit skalierbarem Stack": "foreign_ui_label",
"Continuous Verbesserung": "foreign_ui_label",
"Was du bekommst": "weak_marketing_copy",
"Einführungsmeeting": "weak_marketing_copy",
"Starter Website": "weak_marketing_copy",
"Business Website": "weak_marketing_copy",
"Häufig gestellte Fragen Transparent über Planung, Vorgehensweise und Management.": "foreign_ui_label",
},
"es": {
"Preguntas frecuentes Transparente sobre la planificación, el proceso y la gestión.": "foreign_ui_label",
"Unverbindliches Gespräch, klares Angebot": "foreign_ui_label",
},
"pt": {
"Siti web e negozi online": "mixed_locale_heading",
"Caso de cliente en directo": "weak_marketing_copy",
"El primer proyecto de producción finalizado con éxito.": "weak_marketing_copy",
"Más sobre el proceso": "foreign_ui_label",
"Modifiez simplement vous-même.": "foreign_ui_label",
"Opciones de la tienda web": "foreign_ui_label",
"Planes de soporte": "foreign_ui_label",
"Multilingüe": "foreign_ui_label",
"Unsere Serviços": "mixed_locale_heading",
"Elija el camino": "mixed_locale_heading",
"Début en direct": "foreign_ui_label",
"Demande d'admission initiale": "foreign_ui_label",
"Site Web d'Entreprise": "foreign_ui_label",
"Hablar sobre el proceso empresarial": "foreign_ui_label",
"Mise en place de boutique en ligne": "foreign_ui_label",
"Maintenance & gestion": "foreign_ui_label",
"Afficher le plan de soutien": "foreign_ui_label",
"Introducción multilingüe": "foreign_ui_label",
"Forfaits de services (à partir de)": "mixed_locale_heading",
"Kundenschätzung": "foreign_ui_label",
"Gestisca lei stesso il contenuto": "foreign_ui_label",
"Optimizado para móviles": "foreign_ui_label",
"Schnell online mit einer starken Basis": "weak_marketing_copy",
"La entrada \"Unterstützung oder Erweiterung\"": "foreign_ui_label",
"Suivi + corrections": "foreign_ui_label",
"Mejoras mensuales": "foreign_ui_label",
"¿A qué velocidad puede comenzar?": "foreign_ui_label",
"¿Puedo editar textos e imágenes yo mismo?": "foreign_ui_label",
"Transparente sobre o planejamento, o processo e a gestão.": "foreign_ui_label",
"Ab 2.250 €": "foreign_ui_label",
"Boutique en ligne": "foreign_ui_label",
"Sales-ready mit skalierbarem Stack": "foreign_ui_label",
},
"fr": {
"Erstes Produktionsprojekt erfolgreich abgeschlossen.": "weak_marketing_copy",
"Von Kickoff bis zum Launch mit einem klaren Umfang.": "foreign_ui_label",
"Demande d'admission initiale": "weak_marketing_copy",
"Entretien d'accueil": "weak_marketing_copy",
"Vraag over diensten": "foreign_ui_label",
"Konkrete erste Schätzung": "foreign_ui_label",
"Ansatz, der zu Ihrem Budget passt": "foreign_ui_label",
**build_system_rewrite_candidates(("weeks_2_4",)),
"Bereit, mit der Business-Website zu starten?": "foreign_ui_label",
},
"it": {
"Planificación clara": "foreign_ui_label",
"Mehrsprachiger Rollout-Plan": "foreign_ui_label",
"Unverbindliches Gespräch, klares Angebot": "foreign_ui_label",
},
"ru": {
"Base prête pour le SEO": "foreign_ui_label",
"Unverbindliches Gespräch, klares Angebot": "foreign_ui_label",
},
}
KNOWN_REPLACEMENTS = {
"Starter intake": {
"nl": "Plan startergesprek",
"en": "Book starter call",
"de": "Starter-Gespräch planen",
"fr": "Planifier lentretien de départ",
"es": "Reservar llamada inicial",
"it": "Prenota una chiamata iniziale",
"pt": "Agendar chamada inicial",
"ru": "Запланировать стартовый звонок",
},
"Business intake": {
"nl": "Plan zakelijk gesprek",
"en": "Book business call",
"de": "Beratungsgespräch planen",
"fr": "Planifier lentretien commercial",
"es": "Reservar llamada comercial",
"it": "Prenota una chiamata commerciale",
"pt": "Agendar chamada comercial",
"ru": "Запланировать деловой звонок",
},
"Plan Starter intake": {
"nl": "Plan startergesprek",
"en": "Book starter call",
"de": "Starter-Gespräch planen",
"fr": "Planifier lentretien de départ",
"es": "Reservar llamada inicial",
"it": "Prenota una chiamata iniziale",
"pt": "Agendar chamada inicial",
"ru": "Запланировать стартовый звонок",
},
"Plan Business intake": {
"nl": "Plan zakelijk gesprek",
"en": "Book business call",
"de": "Beratungsgespräch planen",
"fr": "Planifier lentretien commercial",
"es": "Reservar llamada comercial",
"it": "Prenota una chiamata commerciale",
"pt": "Agendar chamada comercial",
"ru": "Запланировать деловой звонок",
},
"Mostrar los servicios": {
"es": "Mostrar los servicios",
"it": "Vedi servizi",
"pt": "Ver serviços",
"ru": "Показать услуги",
},
"Correo electrónico": {"pt": "E-mail", "ru": "Электронная почта"},
'Unternehmen" è tedesco, non olandese. La traduzione spagnola di "Unternehmen" è "empresa".': {
"pt": "Empresa",
"ru": "Компания",
},
'Poiché l\'input "Unverbindliche Erstberatung" è in tedesco (non in olandese), la traduzione in spagnolo è: "Consulta inicial sin compromiso".': {
"it": "Senza impegno",
"pt": "Sem compromisso",
"ru": "Без обязательств",
"es": "Consulta inicial sin compromiso",
},
}
def _contains_fragment(text: str, fragment: str) -> bool:
if re.fullmatch(r"[\wÀ-ÿ-]+", fragment, flags=re.UNICODE):
pattern = re.compile(rf"(?<![\wÀ-ÿ-]){re.escape(fragment)}(?![\wÀ-ÿ-])", re.UNICODE)
return bool(pattern.search(text))
return fragment in text
def validate_patterns(locale_code: str, field_path: str, normalized: str):
issues = []
for fragment in GLOBAL_BAD_PATTERNS:
if _contains_fragment(normalized, fragment):
issues.append(
make_issue(
"known_bad_pattern",
field_path,
fragment,
KNOWN_REPLACEMENTS.get(fragment, {}).get(locale_code, ""),
)
)
for fragment in LOCALE_FORBIDDEN.get(locale_code, ()):
if _contains_fragment(normalized, fragment):
issues.append(
make_issue(
"wrong_language_fragment",
field_path,
fragment,
KNOWN_REPLACEMENTS.get(fragment, {}).get(locale_code, ""),
)
)
if normalized in GENERIC_BADGE_LABELS and not is_canonical_system_string(locale_code, normalized):
issues.append(
make_issue(
"generic_badge_label",
field_path,
normalized,
system_string_replacement(locale_code, normalized),
)
)
for fragment, issue_type in GLOBAL_REWRITE_CANDIDATES.items():
if _contains_fragment(normalized, fragment):
if is_canonical_system_string(locale_code, fragment):
continue
issues.append(
make_issue(
issue_type,
field_path,
fragment,
system_string_replacement(locale_code, fragment),
)
)
for fragment, issue_type in LOCALE_REWRITE_CANDIDATES.get(locale_code, {}).items():
if _contains_fragment(normalized, fragment):
issues.append(
make_issue(
issue_type,
field_path,
fragment,
system_string_replacement(locale_code, fragment),
)
)
return issues