Support for ISO 639-1 - 15924 codes

This commit is contained in:
Piero Toffanin 2025-04-07 23:28:34 -04:00
parent 4fe84200fb
commit 4e86bfa991
4 changed files with 48 additions and 19 deletions

View file

@ -1 +1 @@
1.6.5
1.7.0

View file

@ -22,7 +22,7 @@ from werkzeug.http import http_date
from werkzeug.utils import secure_filename
from libretranslate import flood, remove_translated_files, scheduler, secret, security, storage
from libretranslate.language import detect_languages, improve_translation_formatting
from libretranslate.language import model2iso, iso2model, detect_languages, improve_translation_formatting
from libretranslate.locales import (
_,
_lazy,
@ -499,7 +499,10 @@ def create_app(args):
type: string
description: Supported target language codes
"""
return jsonify([{"code": l.code, "name": _lazy(l.name), "targets": language_pairs.get(l.code, [])} for l in languages])
return jsonify([{"code": model2iso(l.code),
"name": _lazy(l.name),
"targets": model2iso(language_pairs.get(l.code, []))
} for l in languages])
# Add cors
@bp.after_request
@ -626,14 +629,14 @@ def create_app(args):
if request.is_json:
json = get_json_dict(request)
q = json.get("q")
source_lang = json.get("source")
target_lang = json.get("target")
source_lang = iso2model(json.get("source"))
target_lang = iso2model(json.get("target"))
text_format = json.get("format")
num_alternatives = int(json.get("alternatives", 0))
else:
q = request.values.get("q")
source_lang = request.values.get("source")
target_lang = request.values.get("target")
source_lang = iso2model(request.values.get("source"))
target_lang = iso2model(request.values.get("target"))
text_format = request.values.get("format")
num_alternatives = request.values.get("alternatives", 0)
@ -689,7 +692,7 @@ def create_app(args):
candidate_langs = detect_languages(src_texts)
detected_src_lang = candidate_langs[0]
else:
detected_src_lang = {"confidence": 100.0, "language": source_lang}
detected_src_lang = {"confidence": 100.0, "language": model2iso(source_lang)}
else:
detected_src_lang = {"confidence": 0.0, "language": "en"}
@ -736,7 +739,7 @@ def create_app(args):
result = {"translatedText": batch_results}
if source_lang == "auto":
result["detectedLanguage"] = [detected_src_lang] * len(q)
result["detectedLanguage"] = [model2iso(detected_src_lang)] * len(q)
if num_alternatives > 0:
result["alternatives"] = batch_alternatives
@ -761,7 +764,7 @@ def create_app(args):
result = {"translatedText": translated_text}
if source_lang == "auto":
result["detectedLanguage"] = detected_src_lang
result["detectedLanguage"] = model2iso(detected_src_lang)
if num_alternatives > 0:
result["alternatives"] = alternatives
@ -857,8 +860,8 @@ def create_app(args):
if args.disable_files_translation:
abort(403, description=_("Files translation are disabled on this server."))
source_lang = request.form.get("source")
target_lang = request.form.get("target")
source_lang = iso2model(request.form.get("source"))
target_lang = iso2model(request.form.get("target"))
file = request.files['file']
char_limit = get_char_limit(args.char_limit, api_keys_db)
@ -1034,7 +1037,7 @@ def create_app(args):
if not q:
abort(400, description=_("Invalid request: missing %(name)s parameter", name='q'))
return jsonify(detect_languages(q))
return jsonify(model2iso(detect_languages(q)))
@bp.route("/frontend/settings")
@limiter.exempt
@ -1180,13 +1183,13 @@ def create_app(args):
json = get_json_dict(request)
q = json.get("q")
s = json.get("s")
source_lang = json.get("source")
target_lang = json.get("target")
source_lang = iso2model(json.get("source"))
target_lang = iso2model(json.get("target"))
else:
q = request.values.get("q")
s = request.values.get("s")
source_lang = request.values.get("source")
target_lang = request.values.get("target")
source_lang = iso2model(request.values.get("source"))
target_lang = iso2model(request.values.get("target"))
if not q:
abort(400, description=_("Invalid request: missing %(name)s parameter", name='q'))

View file

@ -21,9 +21,11 @@ def check_lang(langcodes, lang):
def normalized_lang_code(lang):
code = lang.lang
# Handle zh-cn
if code.startswith("zh"):
# Handle Chinese
if code == "zh-cn":
code = "zh"
elif code == "zh-tw":
code = "zt"
return code
class Detector:

View file

@ -6,6 +6,30 @@ from argostranslate import translate
from libretranslate.detect import Detector
__languages = None
aliases = {
'pb': 'pt-BR',
'zh': 'zh-Hans',
'zt': 'zh-Hant',
}
rev_aliases = {v.lower(): k for k, v in aliases.items()}
def iso2model(lang):
if not isinstance(lang, str):
return lang
lang = lang.lower()
return rev_aliases.get(lang, lang)
def model2iso(lang):
if isinstance(lang, dict) and 'language' in lang:
d = dict(lang)
d['language'] = model2iso(d['language'])
return d
elif isinstance(lang, list):
return [model2iso(l) for l in lang]
lang = lang.lower()
return aliases.get(lang, lang)
def load_languages():
global __languages