Blaise commited on
Commit ·
8f90c2f
1
Parent(s): bfac87c
Add initial implementation of XLIFF translation script and Gradio interface
Browse files- app.py +14 -0
- requirements.txt +3 -0
- translate_xlf.py +140 -0
app.py
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr, pathlib, tempfile
|
| 2 |
+
from translate_xlf import translate_xlf
|
| 3 |
+
|
| 4 |
+
def translate(file, target):
|
| 5 |
+
tmp_in = tempfile.NamedTemporaryFile(delete=False, suffix=".xlf")
|
| 6 |
+
tmp_in.write(file.read()); tmp_in.flush()
|
| 7 |
+
out = pathlib.Path(tmp_in.name).with_stem(f"{tmp_in.name}_{target}")
|
| 8 |
+
translate_xlf(tmp_in.name, target, out)
|
| 9 |
+
return out
|
| 10 |
+
|
| 11 |
+
demo = gr.Interface(fn=translate,
|
| 12 |
+
inputs=[gr.File(label="Fichier XLF"), gr.Textbox(label="Langue cible", value="fr")],
|
| 13 |
+
outputs=gr.File())
|
| 14 |
+
demo.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
gradio
|
| 2 |
+
googletrans==4.0.0-rc1
|
| 3 |
+
# deepl # décommentez si vous utilisez une clé DeepL
|
translate_xlf.py
ADDED
|
@@ -0,0 +1,140 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
"""translate_xlf.py – Script de traduction XLIFF
|
| 3 |
+
|
| 4 |
+
Ce script lit un fichier XLF (XLIFF 1.2), traduit chaque entrée <source> dans la
|
| 5 |
+
langue cible désirée et écrit le résultat dans un nouveau fichier.
|
| 6 |
+
|
| 7 |
+
Dépendances :
|
| 8 |
+
pip install googletrans==4.0.0-rc1
|
| 9 |
+
|
| 10 |
+
Optionnel : si vous disposez d’une clef API DeepL dans la variable d’environnement
|
| 11 |
+
"DEEPL_AUTH_KEY", le script utilisera automatiquement DeepL pour une meilleure
|
| 12 |
+
qualité de traduction :
|
| 13 |
+
pip install deepl
|
| 14 |
+
|
| 15 |
+
Usage :
|
| 16 |
+
python translate_xlf.py chemin/vers/fichier.xlf fr -o chemin/sortie_fr.xlf
|
| 17 |
+
|
| 18 |
+
Auteur : ChatGPT – 23/05/2025
|
| 19 |
+
"""
|
| 20 |
+
from __future__ import annotations
|
| 21 |
+
|
| 22 |
+
import argparse
|
| 23 |
+
import os
|
| 24 |
+
import sys
|
| 25 |
+
import xml.etree.ElementTree as ET
|
| 26 |
+
from pathlib import Path
|
| 27 |
+
from typing import Optional
|
| 28 |
+
|
| 29 |
+
# ---------------------------------------------------------------------------
|
| 30 |
+
# Choix du moteur de traduction
|
| 31 |
+
# ---------------------------------------------------------------------------
|
| 32 |
+
TRANSLATOR = None
|
| 33 |
+
|
| 34 |
+
if os.getenv("DEEPL_AUTH_KEY"):
|
| 35 |
+
try:
|
| 36 |
+
import deepl # type: ignore
|
| 37 |
+
|
| 38 |
+
_deepl_translator = deepl.Translator(os.environ["DEEPL_AUTH_KEY"])
|
| 39 |
+
|
| 40 |
+
def _translate(text: str, dest: str, src: Optional[str] = None) -> str: # noqa: D401
|
| 41 |
+
"""Utilise l’API DeepL."""
|
| 42 |
+
|
| 43 |
+
result = _deepl_translator.translate_text(text, target_lang=dest.upper())
|
| 44 |
+
return result.text
|
| 45 |
+
|
| 46 |
+
TRANSLATOR = _translate
|
| 47 |
+
except ModuleNotFoundError:
|
| 48 |
+
print("Le module 'deepl' n’est pas installé. Installation recommandée : 'pip install deepl'", file=sys.stderr)
|
| 49 |
+
|
| 50 |
+
if TRANSLATOR is None:
|
| 51 |
+
try:
|
| 52 |
+
from googletrans import Translator # type: ignore
|
| 53 |
+
|
| 54 |
+
_gt = Translator()
|
| 55 |
+
|
| 56 |
+
def _translate(text: str, dest: str, src: Optional[str] = None) -> str: # noqa: D401
|
| 57 |
+
"""Fallback sur Google Translate."""
|
| 58 |
+
|
| 59 |
+
return _gt.translate(text, dest=dest, src=src).text
|
| 60 |
+
|
| 61 |
+
TRANSLATOR = _translate
|
| 62 |
+
except ModuleNotFoundError:
|
| 63 |
+
print("Aucun moteur de traduction disponible. Installez 'googletrans==4.0.0-rc1' ou configurez DEEPL_AUTH_KEY.", file=sys.stderr)
|
| 64 |
+
sys.exit(1)
|
| 65 |
+
|
| 66 |
+
# ---------------------------------------------------------------------------
|
| 67 |
+
# Fonctions principales
|
| 68 |
+
# ---------------------------------------------------------------------------
|
| 69 |
+
|
| 70 |
+
def translate_xlf(input_path: Path, target_lang: str, output_path: Path, source_lang: Optional[str] = None, overwrite: bool = True) -> None:
|
| 71 |
+
"""Traduit toutes les <source> d’un fichier XLF et écrit <target>."""
|
| 72 |
+
|
| 73 |
+
tree = ET.parse(input_path)
|
| 74 |
+
root = tree.getroot()
|
| 75 |
+
|
| 76 |
+
# Namespace éventuel
|
| 77 |
+
nsmap = {}
|
| 78 |
+
if root.tag.startswith("{"):
|
| 79 |
+
uri, _, _ = root.tag[1:].partition("}")
|
| 80 |
+
nsmap = {"ns": uri}
|
| 81 |
+
trans_unit_xpath = ".//ns:trans-unit"
|
| 82 |
+
else:
|
| 83 |
+
trans_unit_xpath = ".//trans-unit"
|
| 84 |
+
|
| 85 |
+
units = root.findall(trans_unit_xpath, namespaces=nsmap)
|
| 86 |
+
if not units:
|
| 87 |
+
print("Aucune balise <trans-unit> trouvée. Vérifiez le format du fichier.", file=sys.stderr)
|
| 88 |
+
sys.exit(1)
|
| 89 |
+
|
| 90 |
+
for tu in units:
|
| 91 |
+
source_elem = tu.find("ns:source" if nsmap else "source", namespaces=nsmap)
|
| 92 |
+
if source_elem is None or not (src_text := source_elem.text):
|
| 93 |
+
continue # nothing to translate
|
| 94 |
+
|
| 95 |
+
target_elem = tu.find("ns:target" if nsmap else "target", namespaces=nsmap)
|
| 96 |
+
if target_elem is None:
|
| 97 |
+
target_elem = ET.SubElement(tu, f"{{{nsmap.get('ns','')}}}target" if nsmap else "target")
|
| 98 |
+
|
| 99 |
+
if target_elem.text and not overwrite:
|
| 100 |
+
# Ne pas réécrire si déjà rempli et overwrite désactivé
|
| 101 |
+
continue
|
| 102 |
+
|
| 103 |
+
translated_text = TRANSLATOR(src_text, dest=target_lang, src=source_lang)
|
| 104 |
+
target_elem.text = translated_text
|
| 105 |
+
|
| 106 |
+
tree.write(output_path, encoding="utf-8", xml_declaration=True)
|
| 107 |
+
print(f"Fichier traduit écrit : {output_path}")
|
| 108 |
+
|
| 109 |
+
|
| 110 |
+
def build_argparser() -> argparse.ArgumentParser:
|
| 111 |
+
parser = argparse.ArgumentParser(description="Traduire un fichier XLF (XLIFF) dans la langue cible.")
|
| 112 |
+
parser.add_argument("input", type=Path, help="Fichier .xlf d’entrée")
|
| 113 |
+
parser.add_argument("target_lang", help="Langue cible (code ISO, p. ex. fr, de, es)")
|
| 114 |
+
parser.add_argument("-s", "--source-lang", dest="source_lang", help="Langue source (code ISO). Laisser vide pour détection auto.")
|
| 115 |
+
parser.add_argument("-o", "--output", type=Path, help="Chemin du fichier de sortie. Défaut : <input>_<lang>.xlf")
|
| 116 |
+
parser.add_argument("--no-overwrite", action="store_true", help="Ne pas écraser les balises <target> déjà existantes")
|
| 117 |
+
return parser
|
| 118 |
+
|
| 119 |
+
|
| 120 |
+
def main() -> None:
|
| 121 |
+
parser = build_argparser()
|
| 122 |
+
args = parser.parse_args()
|
| 123 |
+
|
| 124 |
+
input_path: Path = args.input
|
| 125 |
+
if not input_path.is_file():
|
| 126 |
+
parser.error("Le fichier d’entrée n’existe pas : %s" % input_path)
|
| 127 |
+
|
| 128 |
+
output_path: Path = args.output or input_path.with_stem(f"{input_path.stem}_{args.target_lang}")
|
| 129 |
+
|
| 130 |
+
translate_xlf(
|
| 131 |
+
input_path=input_path,
|
| 132 |
+
target_lang=args.target_lang,
|
| 133 |
+
output_path=output_path,
|
| 134 |
+
source_lang=args.source_lang,
|
| 135 |
+
overwrite=not args.no_overwrite,
|
| 136 |
+
)
|
| 137 |
+
|
| 138 |
+
|
| 139 |
+
if __name__ == "__main__":
|
| 140 |
+
main()
|