Skill: translate

Translate missing strings key-by-key across all locale files (or a specific one).

Usage

/translate [locale-file]

• /translate — processes all locale files
• /translate it.json — processes Italian only

How translations work in this project

• All locale files live in assets/locales/.
• assets/locales/en-US.json is the source of truth.
• A string is untranslated when its value equals its key in a locale file.
• en-GB.json is always skipped (near-identical to US English).

Workflow

This skill uses _automated_translation.json at the project root to track context for every string. Always check this file before translating.

Step 0 — Sync keys and ensure tracking file is current

python3 scripts/update_en_strings.py

Then verify tracking file is in sync:

# If keys were added/removed, regenerate the tracking file (see CLAUDE.md for script)

Step 1 — Discover what needs translating

Run the discovery script to get a structured list of every key that is missing in one or more locale files:

# All locales:
python3 scripts/find_missing_translations.py

# Specific locale(s):
python3 scripts/find_missing_translations.py --locale it.json
python3 scripts/find_missing_translations.py --locale it.json --locale de.json

Output is a JSON array:

[
  { "key": "Apply", "missing_locales": ["de.json", "fr.json", "it.json"] },
  { "key": "Amount (Ascending)", "missing_locales": ["it.json"] }
]

If the array is empty, tell the user everything is translated and stop.

Step 2 — Process key by key

For each entry in the discovery output:

2a. Check _automated_translation.json for context

Look up the key in _automated_translation.json:

python3 -c "import json; d=json.load(open('_automated_translation.json')); print(json.dumps(d['keys']['your_key'], indent=2))"

Three possibilities:

Status = "verified" — Context is documented, proceed to translation (2c)

Status = "pending" — Context not yet researched, do research first (2b)

Status = "skip" — Don't translate (brand names, technical terms, universal)

2b. If pending: Research and document context

Search the Dart source to understand where the key is used:

grep -rn "Key text here" lib/

Read the surrounding code (5-10 lines before/after) to understand:

• Page context: Settings, Wallets, Records, Currencies, Categories, etc.?
• Component context: Button label, dropdown option, form field, error message, dialog title, description text?
• User perspective: What would the user see on screen? What does this word mean in that UI context?

Example:

• "Left" in a currency symbol position dropdown = positional (left of amount), not directional
• Context changes translation from simple directional to positional description

Update _automated_translation.json with findings:

{
  "keys": {
    "Left": {
      "key": "Left",
      "context": "Currency symbol position option",
      "file": "lib/settings/constants/preferences-options.dart:136",
      "page": "Settings > Currency Formatting",
      "component": "currencySymbolPosition dropdown option",
      "meaning": "Position the currency symbol to the LEFT of the amount (e.g., $100 vs 100$)",
      "notes": "Positional context - pair with 'Default' and 'Right' options",
      "status": "verified"
    }
  }
}

2c. Translate into all missing locales at once

With verified context in mind, produce the most natural translation for every locale in missing_locales:

• Context is the source of truth: Use documented meaning, not literal word
• Keep placeholders exactly as-is: %s, %d, %1$s, etc.
• Match tone/terminology of already-translated strings in the same locale file
• Strings with status="skip" should be kept as English (e.g., "Oinkoin Pro", "PIN", "CSV")
• If the word is identical in English and target language, verify it's correct in context (often fine for tech terms)

2d. Write translations immediately

Apply all translations for this key to affected locale files in one batch:

python3 scripts/write_translations.py <<'EOF'
{
  "de.json":  { "Apply": "Anwenden" },
  "fr.json":  { "Apply": "Appliquer" },
  "it.json":  { "Apply": "Applica" }
}
EOF

The script only overwrites values that still equal their key (untranslated). Already-translated strings are never touched.

Step 3 — Continue with the next key

Repeat Step 2 for every entry in the discovery output.

Step 4 — Report

After processing all keys, print a compact summary table:

Helper scripts reference

Important rules

• Never modify already-translated strings (value ≠ key).
• Never change keys — only values.
• Process the entire discovery list in one invocation; do not stop to ask for confirmation.
• en-GB.json is always skipped.