schon/core/management/commands/deepl_translate.py

import os
import re
from tempfile import NamedTemporaryFile

import polib
import requests
from django.apps import apps
from django.core.management.base import BaseCommand, CommandError

from core.management.commands import RootDirectory

# Mapping from Django locale codes to DeepL API codes
DEEPL_TARGET_LANGUAGES_MAPPING = {
    "en-gb": "EN-GB",
    "ar-ar": "AR",
    "cs-cz": "CS",
    "da-dk": "DA",
    "de-de": "DE",
    "en-us": "EN-US",
    "es-es": "ES",
    "fr-fr": "FR",
    "hi-in": "unsupported",
    "it-it": "IT",
    "ja-jp": "JA",
    "kk-kz": "unsupported",
    "nl-nl": "NL",
    "pl-pl": "PL",
    "pt-br": "PT-BR",
    "ro-ro": "RO",
    "ru-ru": "RU",
    "zh-hans": "ZH-HANS",
}

# Patterns to identify placeholders
PLACEHOLDER_REGEXES = [
    re.compile(r"\{[^}]+"),  # {name}, {product_uuid}
    re.compile(r"%\([^)]+\)[sd]"),  # %(name)s, %(count)d
]


def placeholderize(text: str) -> tuple[str, list[str]]:
    """
    Replace placeholders with tokens and collect originals.
    Returns (protected_text, placeholders_list).
    """
    placeholders: list[str] = []

    def _repl(match: re.Match) -> str:
        idx = len(placeholders)
        placeholders.append(match.group(0))
        return f"__PH_{idx}__"

    for rx in PLACEHOLDER_REGEXES:
        text = rx.sub(_repl, text)
    return text, placeholders


def deplaceholderize(text: str, placeholders: list[str]) -> str:
    """
    Restore tokens back to original placeholders.
    """
    for idx, ph in enumerate(placeholders):
        text = text.replace(f"__PH_{idx}__", ph)
    return text


def load_po_sanitized(path: str) -> polib.POFile | None:
    """
    Load a .po file, sanitizing common issues if necessary.
    """
    try:
        return polib.pofile(path)
    except OSError:
        with open(path, encoding="utf-8") as f:
            text = f.read()
        # fix malformed fuzzy flags
        text = re.sub(r"^#,(?!\s)", "#, ", text, flags=re.MULTILINE)
        # remove empty header entries
        parts = text.split("\n\n", 1)
        header = parts[0]
        rest = parts[1] if len(parts) > 1 else ""
        rest_clean = re.sub(r"^msgid \"\"\s*\nmsgstr \"\"\s*\n?", "", rest, flags=re.MULTILINE)
        sanitized = header + "\n\n" + rest_clean
        tmp = NamedTemporaryFile(  # noqa: SIM115
            mode="w+", delete=False, suffix=".po", encoding="utf-8"
        )
        try:
            tmp.write(sanitized)
            tmp.flush()
            tmp.close()
            return polib.pofile(tmp.name)
        finally:
            try:
                os.unlink(tmp.name)
            except OSError as e:
                raise CommandError("Failed to write sanitized .po file") from e


class Command(BaseCommand):
    help = "Merge msgid/msgstr from en_GB PO into target-language POs via DeepL, preserving placeholders."

    def add_arguments(self, parser):
        parser.add_argument(
            "-l",
            "--language",
            dest="target_languages",
            action="append",
            required=True,
            metavar="LANG",
            help="Locale code for translation, e.g. de-DE, fr-FR.",
        )
        parser.add_argument(
            "-a",
            "--app",
            dest="target_apps",
            action="append",
            required=True,
            metavar="APP",
            help="App label for translation, e.g. core, payments.",
        )

    def handle(self, *args, **options) -> None:
        target_langs = options["target_languages"]
        target_apps = set(options["target_apps"])
        auth_key = os.environ.get("DEEPL_AUTH_KEY")
        if not auth_key:
            raise CommandError("DEEPL_AUTH_KEY not set")

        # attempt to import readline for interactive prefill
        try:
            import readline
        except ImportError:
            readline = None  # fallback

        for target_lang in target_langs:
            api_code = DEEPL_TARGET_LANGUAGES_MAPPING.get(target_lang)
            if not api_code:
                self.stdout.write(self.style.WARNING(f"Unknown language '{target_lang}'"))
                continue
            if api_code == "unsupported":
                self.stdout.write(self.style.WARNING(f"Unsupported language '{target_lang}'"))
                continue

            self.stdout.write(self.style.MIGRATE_HEADING(f"→ Translating into {target_lang}"))

            configs = list(apps.get_app_configs()) + [RootDirectory()]

            for app_conf in configs:
                if app_conf.label not in target_apps:
                    continue

                en_path = os.path.join(app_conf.path, "locale", "en_GB", "LC_MESSAGES", "django.po")
                if not os.path.isfile(en_path):
                    self.stdout.write(self.style.WARNING(f"• {app_conf.label}: no en_GB PO"))
                    continue

                self.stdout.write(f"• {app_conf.label}: loading English PO…")
                en_po = load_po_sanitized(en_path)
                if not en_po:
                    raise CommandError(f"Failed to load en_GB PO for {app_conf.label}")

                # gather entries with missing translations
                missing = [e for e in en_po if e.msgid and not e.msgstr and not e.obsolete]
                if missing:
                    self.stdout.write(self.style.NOTICE(f"⚠️ {len(missing)} missing in en_GB"))
                    for e in missing:
                        default = e.msgid
                        if readline:

                            def hook():
                                readline.insert_text(default)
                                readline.redisplay()

                            readline.set_pre_input_hook(hook)

                        prompt = f"Enter translation for '{e.msgid}': "
                        user_in = input(prompt).strip()

                        if readline:
                            readline.set_pre_input_hook(None)

                        if user_in:
                            e.msgstr = user_in
                        else:
                            e.msgstr = e.msgid

                    en_po.save(en_path)
                    self.stdout.write(self.style.SUCCESS("Updated en_GB PO"))

                # … rest of your DeepL logic unchanged …
                # build new_po, translate missing entries, save target PO, etc.

        self.stdout.write(self.style.SUCCESS("Done."))