1
0
mirror of https://github.com/chubin/cheat.sh.git synced 2026-06-20 13:16:44 +02:00

Disable language detection

This commit is contained in:
Igor Chubin
2021-01-31 20:23:09 +01:00
parent dabff40376
commit e70b815306
2 changed files with 29 additions and 23 deletions
+23 -20
View File
@@ -12,8 +12,9 @@ import os
import re
from subprocess import Popen, PIPE
from polyglot.detect import Detector
from polyglot.detect.base import UnknownLanguage
#Language detetction disabled
#from polyglot.detect import Detector
#from polyglot.detect.base import UnknownLanguage
from config import CONFIG
from languages_data import SO_NAME
@@ -83,25 +84,27 @@ class Question(UpstreamAdapter):
topic = " ".join(topic_words)
lang = 'en'
try:
query_text = topic # " ".join(topic)
query_text = re.sub('^[^/]*/+', '', query_text.rstrip('/'))
query_text = re.sub('/[0-9]+$', '', query_text)
query_text = re.sub('/[0-9]+$', '', query_text)
detector = Detector(query_text)
supposed_lang = detector.languages[0].code
if len(topic_words) > 2 \
or supposed_lang in ['az', 'ru', 'uk', 'de', 'fr', 'es', 'it', 'nl']:
lang = supposed_lang
if supposed_lang.startswith('zh_') or supposed_lang == 'zh':
lang = 'zh'
elif supposed_lang.startswith('pt_'):
lang = 'pt'
if supposed_lang in ['ja', 'ko']:
lang = supposed_lang
except UnknownLanguage:
print("Unknown language (%s)" % query_text)
# Question language detection disabled
# try:
# query_text = topic # " ".join(topic)
# query_text = re.sub('^[^/]*/+', '', query_text.rstrip('/'))
# query_text = re.sub('/[0-9]+$', '', query_text)
# query_text = re.sub('/[0-9]+$', '', query_text)
# detector = Detector(query_text)
# supposed_lang = detector.languages[0].code
# if len(topic_words) > 2 \
# or supposed_lang in ['az', 'ru', 'uk', 'de', 'fr', 'es', 'it', 'nl']:
# lang = supposed_lang
# if supposed_lang.startswith('zh_') or supposed_lang == 'zh':
# lang = 'zh'
# elif supposed_lang.startswith('pt_'):
# lang = 'pt'
# if supposed_lang in ['ja', 'ko']:
# lang = supposed_lang
# except UnknownLanguage:
# print("Unknown language (%s)" % query_text)
if lang != 'en':
topic = ['--human-language', lang, topic]
+6 -3
View File
@@ -9,10 +9,13 @@ redis
colored
langdetect
cffi
polyglot
PyICU
pycld2
colorama
pyyaml
python-Levenshtein
pylint
#Disable language detection feature
#
#polyglot
#PyICU
#pycld2