lang.py

"""Language configuration and path helpers for multi-language support."""

import json
from pathlib import Path

_ROOT = Path(__file__).parent
_current_lang: str | None = None
_config_cache: dict[str, dict] = {}


def available_languages() -> list[str]:
    """Return language codes with a config.json, ordered by the config's
    "sort" rank — web-content share per
    https://en.wikipedia.org/wiki/Languages_used_on_the_Internet — then code.
    """
    langs_dir = _ROOT / "languages"
    ranked = []
    for d in langs_dir.iterdir():
        cfg_path = d / "config.json"
        if not (d.is_dir() and cfg_path.exists()):
            continue
        with open(cfg_path, encoding="utf-8") as f:
            rank = json.load(f).get("sort", 999)
        ranked.append((rank, d.name))
    return [code for _, code in sorted(ranked)]


def set_lang(lang: str) -> None:
    """Set the active language. Must be called before loading language data."""
    global _current_lang
    ld = _ROOT / "languages" / lang
    if not ld.exists():
        raise ValueError(f"Unknown language: {lang!r} (no directory {ld})")
    _current_lang = lang


def get_lang() -> str:
    """Return the active language code. Raises if set_lang() was not called."""
    if _current_lang is None:
        raise RuntimeError("Language not set. Call lang.set_lang() first.")
    return _current_lang


def config() -> dict:
    """Load and cache the config.json for the active language."""
    lang = get_lang()
    if lang not in _config_cache:
        path = _ROOT / "languages" / lang / "config.json"
        with open(path, encoding="utf-8") as f:
            _config_cache[lang] = json.load(f)
    return _config_cache[lang]


def lang_dir() -> Path:
    """Path to languages/<lang>/ (features, corpus config)."""
    return _ROOT / "languages" / get_lang()


def data_dir() -> Path:
    """Path to data/<lang>/ (generated wordlists and transition tables)."""
    return _ROOT / "data" / get_lang()


def features_path() -> Path:
    return lang_dir() / "letter_features.json"


def corpus_dir() -> Path:
    return lang_dir() / "corpus"


def wordlist_path() -> Path:
    return data_dir() / "wordlist.json"


def bigram_transitions_path() -> Path:
    return data_dir() / "bigram_transitions.json"


def trigram_transitions_path() -> Path:
    return data_dir() / "trigram_transitions.json"


def alphabet() -> str:
    """Return the ordered alphabet string for the current language."""
    return config()["alphabet"]


def alphabet_regex() -> str:
    """Return a regex character class body for the current language."""
    return config()["alphabet_regex"]