airflow-translations

Ajouter ou mettre à jour les traductions pour l'interface utilisateur d'Apache Airflow. Guide pour configurer les locales, générer les fichiers de traduction, traduire avec des paramètres régionaux spécifiques…

npx skills add https://github.com/astronomer/airflow --skill airflow-translations

Airflow Translations

Determining the Task

Translation work falls into one of two categories depending on whether the target locale already exists. Check if a directory for the locale exists under airflow-core/src/airflow/ui/public/i18n/locales/<locale>/. If it does, skip ahead to Updating an Existing Translation. If not, start with Adding a Translation below.


Adding a Translation

When adding a translation, some configuration files need to be updated before translation work can begin.

Setting up the locale

First, create the locale directory:

mkdir -p airflow-core/src/airflow/ui/public/i18n/locales/<locale>/

Then update the following configuration files, keeping the existing alphabetical ordering in each file:

airflow-core/src/airflow/ui/src/i18n/config.ts: add the locale to the supportedLanguages array:

{ code: "<locale>", name: "<native name>" },

dev/breeze/src/airflow_breeze/commands/ui_commands.py: add the plural suffixes for the language to the PLURAL_SUFFIXES dict. Check the i18next plural rules for the language at https://jsfiddle.net/6bpxsgd4 to determine which suffixes are needed:

"<locale>": ["<suffixes>"],

.github/boring-cyborg.yml: under labelPRBasedOnFilePath, add:

translation:<locale>:
  - airflow-core/src/airflow/ui/public/i18n/locales/<locale>/*

Scaffolding the translation files

Once the configuration is in place, run the breeze command to copy every English namespace file into the new locale directory. This populates each key with a TODO: translate: stub:

breeze ui check-translation-completeness --language <locale> --add-missing

The generated files will look like this:

{
  "allRuns": "TODO: translate: All Runs",
  "blockingDeps": {
    "dependency": "TODO: translate: Dependency",
    "reason": "TODO: translate: Reason"
  }
}

Translating

With the scaffolded files in place, read the locale-specific guideline for the target language (see the table under Locale-Specific Guidelines below). If one exists, it contains the glossary, tone rules, and formatting conventions that must be followed. If no locale-specific guideline exists yet, follow the translation rules described later in this document.

Replace every TODO: translate: <English terminology> entry, including the prefix, with the translated string.

After all entries are translated, continue to Validation below.


Updating an Existing Translation

When a locale already exists and you need to fill translation gaps, revise existing translations, or remove stale keys, start by reading the locale-specific guideline for the language (see the table under Locale-Specific Guidelines below). This establishes the glossary and formatting rules to follow.

Next, read the locale's existing JSON files under airflow-core/src/airflow/ui/public/i18n/locales/<locale>/ to learn the terminology already in use. Consistency with established translations is critical. If a term has been translated a certain way, reuse that exact translation.

Then check the current state of completeness:

breeze ui check-translation-completeness --language <locale>

If there are missing keys, scaffold them with TODO: translate: stubs:

breeze ui check-translation-completeness --language <locale> --add-missing

If there are unused keys — keys that are not required, meaning they are absent from the English locale or are plural suffixes this language does not need — remove them:

breeze ui check-translation-completeness --language <locale> --remove-unused

Now translate the TODO: translate: entries following the locale-specific guideline, then continue to Validation below.


Validation

After completing translations, run these checks:

Check completeness. The output table should show 0 missing, 0 TODOs, 0 unused, and 100% coverage:

breeze ui check-translation-completeness --language <locale>

Run pre-commit hooks to fix formatting, licenses, and linting issues:

prek run --from-ref main --hook-stage pre-commit

General Translation Rules

The following rules apply globally. If the locale-specific guideline for a language states differently, follow the locale-specific guideline.

Terms Kept in English

The terms below should remain in English by default. Locale-specific guidelines may override individual entries where an established local convention exists:

TermReason
AirflowProduct name
Dag / DagsAirflow convention; always Dag, never DAG
XCom / XComsAirflow cross-communication mechanism name
Provider / ProvidersAirflow extension package name
REST APIStandard technical term
JSONStandard technical format name
IDUniversal abbreviation
PIDUnix process identifier
UTCTime standard
SchemaDatabase term

Variables and Placeholders

Translation strings use {{variable}} interpolation (i18next format). Never translate or remove variable names inside {{…}}. Placeholders may be reordered as needed for natural word order, but the exact variable casing must be preserved (e.g., {{dagDisplayName}}).

Plural Forms

Airflow uses i18next plural suffixes (_one, _other, and optionally _zero, _two, _few, _many). Provide translations for all plural suffixes that the language requires — the locale-specific guideline specifies which ones. If no locale guideline exists, check the i18next plural rules at https://jsfiddle.net/6bpxsgd4 and provide at minimum _one and _other.

Hotkeys

Hotkey values (e.g., "hotkey": "e") are literal key bindings and should not be translated unless the locale-specific guideline says otherwise.


Translation File Structure

All translation files are JSON files located at:

airflow-core/src/airflow/ui/public/i18n/locales/<locale-name>/

Each locale directory contains namespace JSON files that mirror the English locale (en/). The English locale is the default locale and the primary source for all translations. The current namespace files are:

admin.json, assets.json, browse.json, common.json, components.json, dag.json, dags.json, dashboard.json, hitl.json, tasks.json


Locale-Specific Guidelines

Before translating, read the locale-specific guideline file for the target language. These contain glossaries, tone rules, and formatting conventions tailored to each language. If a locale-specific guideline states differently from a global rule in this document, follow the locale-specific guideline.

Locale CodeLanguageGuideline File
arArabiclocales/ar.md
caCatalanlocales/ca.md
deGermanlocales/de.md
elGreeklocales/el.md
esSpanishlocales/es.md
frFrenchlocales/fr.md
heHebrewlocales/he.md
hiHindilocales/hi.md
huHungarianlocales/hu.md
itItalianlocales/it.md
jaJapaneselocales/ja.md
koKoreanlocales/ko.md
nlDutchlocales/nl.md
plPolishlocales/pl.md
ptPortugueselocales/pt.md
thThailocales/th.md
trTurkishlocales/tr.md
zh-CNSimplified Chineselocales/zh-CN.md
zh-TWTraditional Chineselocales/zh-TW.md

If the target locale file does not yet exist, follow only the global rules in this document.

Plus de skills de astronomer

airflow-state-store
astronomer
Persists task and asset state across retries and DAG runs using Airflow 3.3's AIP-103 key/value stores (`task_state_store`, `asset_state_store`) and the…
creating-openlineage-extractors
astronomer
Extracteurs OpenLineage personnalisés pour les opérateurs Airflow non pris en charge et les scénarios de lignage complexes. Deux approches : ajouter des méthodes OpenLineage directement aux opérateurs que vous possédez (recommandé), ou créer des extracteurs personnalisés pour les opérateurs tiers que vous ne pouvez pas modifier. Les extracteurs interceptent l'exécution des opérateurs à trois moments : avant l'exécution pour le lignage statique, après le succès pour les sorties déterminées à l'exécution, et optionnellement après l'échec pour un lignage partiel. Enregistrez les extracteurs via airflow.cfg ou l'environnement...
debugging-dags
astronomer
Analyse systématique des causes profondes et correction des DAG Airflow défaillants, avec des flux d'investigation structurés. Guide à travers un processus de diagnostic en quatre étapes : identifier l'échec, extraire les détails de l'erreur, rassembler les informations contextuelles et fournir des étapes de correction exploitables. Classe les échecs en quatre types (données, code, infrastructure, dépendance) pour cibler l'investigation et suggérer les correctifs appropriés. Fournit des commandes CLI prêtes à l'emploi pour la récupération des logs, la comparaison des exécutions, l'effacement des tâches et des DAG...
delegating-to-otto
astronomer
Drives Astronomer's Otto agent (`astro otto`) as a delegated sub-agent for Airflow, dbt, and data-engineering work. Use when the user explicitly asks to "use…
deploying-airflow
astronomer
Déployer des DAGs et projets Airflow. Utiliser lorsque l'utilisateur souhaite déployer du code, pousser des DAGs, configurer CI/CD, déployer en production, ou demande des stratégies de déploiement…
deploying-go-sdk-bundles
astronomer
Construit, empaquette et déploie des bundles compilés du SDK Go Airflow afin que l'ExecutableCoordinator puisse les exécuter. À utiliser lorsque l'utilisateur souhaite compiler un bundle de tâches Go, demande…
testing-dags
astronomer
Cycles itératifs de test-débogage-correction pour les DAGs Airflow avec diagnostic complet des échecs. Commencez par af runs trigger-wait <dag_id> pour exécuter un DAG et attendre son achèvement ; aucune vérification préalable nécessaire. En cas d'échec, utilisez af runs diagnose pour un résumé complet des échecs et af tasks logs pour inspecter les détails des erreurs de tâches spécifiques. Prend en charge la configuration personnalisée, les délais d'attente et les tentatives de réessai ; gère les scénarios de succès, d'échec et de dépassement de délai avec une interprétation claire des réponses. Validation rapide disponible...
tracing-downstream-lineage
astronomer
Tracer la lignée des données en aval pour évaluer l'impact des modifications avant de modifier des tables ou des DAG. Identifie les consommateurs directs d'une table ou d'un DAG cible via la recherche dans le code source, les dépendances de vues et les connexions aux outils BI. Construit un arbre de dépendances complet cartographiant tous les impacts en aval, des tables aux tableaux de bord en passant par les modèles ML. Catégorise les dépendances par criticité (critique, élevée, moyenne, faible) pour prioriser la communication avec les parties prenantes et les tests. Génère un rapport d'impact avec évaluation des risques, éléments affectés...