commented code
This commit is contained in:
+15
-45
@@ -1,15 +1,11 @@
|
|||||||
|
# public packages
|
||||||
from langdetect import detect
|
from langdetect import detect
|
||||||
from google_trans_new import google_translator
|
from google_trans_new import google_translator
|
||||||
import pandas as pd
|
# database-related
|
||||||
import multiprocessing as mp
|
from database import Data_table, get_untranslated, insert_translated
|
||||||
import numpy as np
|
|
||||||
from dotenv import load_dotenv
|
from dotenv import load_dotenv
|
||||||
from tqdm import tqdm
|
# own scripts
|
||||||
import logging
|
|
||||||
|
|
||||||
from database import Data_table
|
|
||||||
from clean_data import clean_content
|
from clean_data import clean_content
|
||||||
from setup_logging import setup_logging
|
|
||||||
|
|
||||||
def translate(text, target_lang='en'):
|
def translate(text, target_lang='en'):
|
||||||
try:
|
try:
|
||||||
@@ -26,65 +22,39 @@ def translate(text, target_lang='en'):
|
|||||||
return ''
|
return ''
|
||||||
return text_trans
|
return text_trans
|
||||||
|
|
||||||
def get_untranslated(dt):
|
|
||||||
query = (
|
|
||||||
f'SELECT * FROM {dt._name} '
|
|
||||||
'WHERE content_en IS NULL '
|
|
||||||
'ORDER BY random() '
|
|
||||||
'LIMIT 1 '
|
|
||||||
)
|
|
||||||
df = pd.read_sql_query(query, dt.con)
|
|
||||||
if len(df)==0: return None
|
|
||||||
val_dict = {
|
|
||||||
'id': int(df['id'][0]),
|
|
||||||
'content': str(df['content'][0])
|
|
||||||
}
|
|
||||||
return val_dict
|
|
||||||
|
|
||||||
def insert_translated(dt, val_dict):
|
|
||||||
assert 'id' in val_dict.keys()
|
|
||||||
assert 'content_en' in val_dict.keys()
|
|
||||||
query = (
|
|
||||||
f'UPDATE {dt._name} '
|
|
||||||
' SET content_en = (%(content_en)s) '
|
|
||||||
'WHERE id = (%(id)s) '
|
|
||||||
)
|
|
||||||
dt.execute(query, val_dict)
|
|
||||||
|
|
||||||
def continously_translate():
|
def continously_translate():
|
||||||
with Data_table() as dt:
|
with Data_table() as dt:
|
||||||
logging.info('began translating')
|
print('began translating')
|
||||||
while True:
|
while True: # keep going forever
|
||||||
try:
|
try:
|
||||||
val_dict = get_untranslated(dt)
|
val_dict = get_untranslated(dt)
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
logging.warning(f'error when getting untranslated content from database: {e}')
|
print(f'error when getting untranslated content from database: {e}')
|
||||||
continue
|
continue
|
||||||
|
# if nother left untranslated, stop
|
||||||
if val_dict is None: break
|
if val_dict is None: break
|
||||||
# clean
|
# clean text
|
||||||
try:
|
try:
|
||||||
content_clean = clean_content(val_dict['content'])
|
content_clean = clean_content(val_dict['content'])
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
logging.warning(f'error when cleaning content: {e}')
|
print(f'error when cleaning content: {e}')
|
||||||
continue
|
continue
|
||||||
# translate
|
# translate text
|
||||||
try:
|
try:
|
||||||
content_en = translate(content_clean)
|
content_en = translate(content_clean)
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
logging.warning(f'error when translating content: {e}')
|
print(f'error when translating content: {e}')
|
||||||
continue
|
continue
|
||||||
# upload to database
|
# upload to database
|
||||||
val_dict['content_en'] = content_en
|
val_dict['content_en'] = content_en
|
||||||
try:
|
try:
|
||||||
insert_translated(dt, val_dict)
|
insert_translated(dt, val_dict)
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
logging.warning(f'error when sending translated content to database: {e}')
|
print(f'error when sending translated content to database: {e}')
|
||||||
continue
|
continue
|
||||||
logging.info('translated content')
|
print('translated content')
|
||||||
logging.warning('stopped translating')
|
print('stopped translating')
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
load_dotenv()
|
load_dotenv()
|
||||||
setup_logging()
|
|
||||||
continously_translate()
|
continously_translate()
|
||||||
# later try to set all empty strings back to null and retranslate failed...
|
|
||||||
|
|||||||
Reference in New Issue
Block a user