mirrors/LibreTranslate
1
1
Fork 1
mirror of https://github.com/LibreTranslate/LibreTranslate.git synced 2024-11-25 01:11:00 +00:00
Code Issues Projects Releases Packages Wiki Activity

move and improve_translation in language.py, use it for transliteration

Browse source
This commit is contained in:
dingedi 2022-09-23 13:59:13 +02:00
parent 5aa3f1eee8
commit 077d6e9efa
No known key found for this signature in database
GPG key ID: EE23A12D271A3CED
2 changed files with 40 additions and 34 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

37
app/app.py
View file

@ -15,7 +15,7 @@ from translatehtml import translate_html
from werkzeug.utils import secure_filename
from app import flood, remove_translated_files, security
from app.language import detect_languages, transliterate
from app.language import detect_languages, transliterate, improve_translation_formatting
from .api_keys import Database, RemoteDatabase
from .suggestions import Database as SuggestionsDatabase
@ -483,46 +483,15 @@ def create_app(args):
if text_format not in ["text", "html"]:
abort(400, description="%s format is not supported" % text_format)
def improve_translation(source, translation):
source = source.strip()
source_last_char = source[len(source) - 1]
translation_last_char = translation[len(translation) - 1]
punctuation_chars = ['!', '?', '.', ',', ';']
if source_last_char in punctuation_chars:
if translation_last_char != source_last_char:
if translation_last_char in punctuation_chars:
translation = translation[:-1]
translation += source_last_char
elif translation_last_char in punctuation_chars:
translation = translation[:-1]
if source.islower():
return translation.lower()
if source.isupper():
return translation.upper()
if source[0].islower():
return translation[0].lower() + translation[1:]
if source[0].isupper():
return translation[0].upper() + translation[1:]
return translation
try:
if batch:
results = []
for idx, text in enumerate(q):
translator = src_langs[idx].get_translation(tgt_lang)
if text_format == "html":
translated_text = str(translate_html(translator, text))
else:
translated_text = improve_translation(text, translator.translate(
translated_text = improve_translation_formatting(text, translator.translate(
transliterate(text, target_lang=source_langs[idx]["language"])))
results.append(unescape(translated_text))
@ -545,7 +514,7 @@ def create_app(args):
if text_format == "html":
translated_text = str(translate_html(translator, q))
else:
translated_text = improve_translation(q, translator.translate(
translated_text = improve_translation_formatting(q, translator.translate(
transliterate(q, target_lang=source_langs[0]["language"])))
if source_lang == "auto":

37
app/language.py
View file

@ -79,6 +79,41 @@ def detect_languages(text):
return [{"confidence": l.confidence, "language": l.code} for l in candidate_langs]
def improve_translation_formatting(source, translation, improve_punctuation=True):
source = source.strip()
if not len(source):
return ""
if improve_punctuation:
source_last_char = source[len(source) - 1]
translation_last_char = translation[len(translation) - 1]
punctuation_chars = ['!', '?', '.', ',', ';']
if source_last_char in punctuation_chars:
if translation_last_char != source_last_char:
if translation_last_char in punctuation_chars:
translation = translation[:-1]
translation += source_last_char
elif translation_last_char in punctuation_chars:
translation = translation[:-1]
if source.islower():
return translation.lower()
if source.isupper():
return translation.upper()
if source[0].islower():
return translation[0].lower() + translation[1:]
if source[0].isupper():
return translation[0].upper() + translation[1:]
return translation
def __transliterate_line(transliterator, line_text):
new_text = []
@ -94,6 +129,8 @@ def __transliterate_line(transliterator, line_text):
# the actual transliteration of the word
t_word = transliterator.transliterate(orig_word.strip(string.punctuation))
t_word = improve_translation_formatting(orig_word.strip(string.punctuation), t_word, improve_punctuation=False)
# if transliteration fails, default back to the original word
if not t_word:
t_word = orig_word