config.example.yaml

# Configuration for Translate Java Property Files

# Absolute path to the Git repository root
target_project_root: "/path/to/your/git/repo"

# Directory containing all .properties files
input_folder: "/path/to/properties/files"

# Path to glossary
glossary_file_path: "glossary.json"

# OpenAI model to use
model_name: "gpt-4o-mini"

# (Optional) OpenAI model for the holistic review step.
# If not provided, defaults to the 'model_name' above.
# Use a more powerful model like "gpt-4o" for higher quality review.
review_model_name: "gpt-4o"
# Tip: You can override this at runtime with the REVIEW_MODEL_NAME environment variable.
# Example: export REVIEW_MODEL_NAME="gpt-4o"

# (Optional) Number of keys to include in a single holistic review API call.
# Lower this value if you encounter "request too large" errors from the API.
holistic_review_chunk_size: 75

# (Optional) If specified, only files matching this glob pattern will be sent for AI translation.
# All other changed .properties files will still be included in the final PR.
# This is useful for targeting specific sets of files, like 'mobile_*.properties'.
# If commented out or empty, all changed .properties files will be translated.
# Set to null or remove to translate all changed files.
# translation_file_filter_glob: null

# Optional: Process all translation files under input_folder instead of only git-changed files.
# Useful for one-time ledger bootstrap or forced full retranslation runs.
# Keep false for normal incremental operation.
process_all_files: false

# Optional: Pull source files (*_en.properties) from Transifex as well as translations.
# Set to `true` if the canonical source of truth for English strings is Transifex.
# The default behavior (`false`) is to only pull translation files and assume
# the source files in the Git repository are the source of truth.
pull_source_files_from_transifex: false

# Translation queue folders
translation_queue_folder: "translation_queue"
translated_queue_folder: "translated_queue"

# If true, the script logs actions without altering files
dry_run: false

# If true, existing keys whose target text is identical to source text
# are retranslated on every run (legacy behavior). Keep this false to
# only translate newly added keys by default and reduce translation churn.
retranslate_identical_source_strings: false

# Optional persistent per-key hash ledger used to detect source changes
# without reprocessing already-stable translations.
translation_key_ledger_file_path: "logs/translation_key_ledger.json"

# Concurrency setting for OpenAI API calls.
# This controls how many API requests can be active at the same time.
# A lower value (e.g., 1 or 2) is safer to avoid API rate limits.
# A higher value can speed up translation but may lead to "429 Too Many Requests" errors.
# Default value if not specified is 1.
max_concurrent_api_calls: 1

# Each locale has a 'code' and a human-readable 'name'
supported_locales:
  - code: "cs"
    name: "Czech"
  - code: "de"
    name: "German"
  - code: "es"
    name: "Spanish"
  - code: "it"
    name: "Italian"
  - code: "pt_BR"
    name: "Brazilian Portuguese"
  - code: "pcm"
    name: "Nigerian Pidgin"
  - code: "ru"
    name: "Russian"
  - code: "af_ZA"
    name: "Afrikaans"

# (Optional) Language-specific style and formatting rules to be injected into the AI prompt.
# Use this to guide the AI on tone, casing, terminology, and other stylistic conventions.
# NOTE: Keys here must be language codes present in supported_locales (e.g., de, pt_BR).
style_rules:
  de:
    - "Use the formal 'Sie' form of address."
    - "For quantities, prefer 'ermitteln' or 'bestimmen' over 'bewerten'."
    - "Ensure compound nouns are correctly formed (e.g., 'Sicherheitsschritt')."
  cs:
    - "Use impersonal and neutral phrasing for system messages; avoid the first-person plural."
    - "For camera-related nouns, use the correct preposition 'ke kameře'."
  es:
    - "When referring to 'billetera' (wallet), use feminine pronouns (e.g., 'Configúrala', 'restáurala')."
    - "Ensure questions that open with '¿' are properly closed with '?'."
  pcm:
    - "Maintain consistency with terms like 'connection' and 'camera' (do not use 'kamera')."
    - "For negative outcomes, prefer the 'no gree work' pattern where appropriate."
  ru:
    - "In a wallet context, prefer hyphenated terms like 'seed-фраза' and 'биткойн-кошелька'."
  af_ZA: []
  it: []
  pt_BR: []

# (Optional) A list of brand names and technical terms that should never be translated.
brand_technical_glossary:
  - ACH
  - ACH Transfer
  - Advanced Cash
  - AliPay
  - Amazon eGift Card
  - Bisq
  - Bisq Easy
  - Bitcoin
  - Bizum
  - BSQ Swaps
  - BTC
  - Capitual
  - Cash App
  - CelPay
  - Chase QuickPay
  - CIPS
  - FPS (UK)
  - Faster Payments
  - HalCash
  - I2P
  - IMPS
  - Interac e-Transfer
  - Japan Furikomi
  - Lightning
  - Lightning Escrow
  - Liquid
  - Liquid Assets
  - Monese
  - MoneyBeam (N26)
  - Monero Swaps
  - MuSig
  - NEFT
  - Nequi
  - Paxum
  - PayTM
  - Paysera
  - Perfect Money
  - Pix
  - Popmoney
  - PromptPay
  - RSK
  - RTGS
  - Revolut
  - SEPA
  - SEPA Instant
  - Satispay
  - Strike
  - Submarine Swaps
  - SWIFT
  - SWIFT International Wire Transfer
  - Swish
  - Tikkie
  - Tor
  - UPI
  - US Money Order
  - Uphold
  - Verse
  - WeChat Pay
  - Wire Transfer
  - Wise
  - Wise-USD
  - Zelle