在平时使用谷歌翻译的过程中，经常会遇到需要批量翻译大量文本的情景，这种时候需要调用谷歌翻译的API
首先可以使用python库googletrans
pip install googletrans
#使用方法
from googletrans import Translator
translator = Translator(service_urls=[''])
source = '我还是不开心！'
text = translator.translate(source,src='zh-cn',dest='en').text
print(text)
"i'm still not happy!"
但是在面对大规模需要翻译的句子时就会很慢，所以可以使用协程的方法。
这里我们使用了基于gevents库的grequests库。
仔细看了下googletrans库的核心代码，发现主要是构造一个url，然后发起get请求，得到一个json的结果，从中提取出翻译结果。
构造url的过程需要一个token，根据某些规则生成，所以为了方便还是调用googletrans的部分函数。
具体的参考代码如下：
这里是将德语(de)翻译成英语(en)，代码中需要相应修改。
import grequests
import logging
import json
from googletrans import Translator
from googletrans.utils import format_json
headers = {'User-Agent':'Mozilla/5.0 (Windows NT 10.0; Win64; x64)'}
translator = Translator(service_urls=[''])
logging.basicConfig(level = logging.INFO,format = '%(asctime)s - %(name)s - %(levelname)s - %(message)s',filename='log.txt')
logger = logging.getLogger()
def exception_handler(request, exception):
logger.warning('exception when at %s :%s',request.url,exception)
def work(urls):
reqs = (grequests.get(u,verify=True, allow_redirects=True, timeout=4) for u in urls)
res = grequests.map(reqs, exception_handler=exception_handler,size=20)
return res
def totaltranslate():
file2 = open('de2en_en.txt',mode='a',encoding='utf-8')
with open('de.txt',mode='r',encoding='utf-8') as f:
urls = []
num = 0
for line in f:
num+=1
line = line.strip()
token = translator.token_acquirer.do(line)
url="/translate_a/single?client=t&sl=de&tl=en&hl=en&dt=at&dt=bd&dt=ex&dt=ld&dt=md&dt=qca&dt=rw&dt=rm&dt=ss&dt=t&ie=UTF-8&oe=UTF-8&otf=1&ssel=3&tsel=0&kc=1&tk={0}&q={1}".format(token,line)
urls.append(url)
if len(urls) >= 50:
res = work(urls)
for r in res:
if hasattr(r,'status_code'):
if r.status_code == 200:
try:
a=format_json(r.text)
target = ''.join([d[0] if d[0] else '' for d in a[0]])
source = ''.join([d[1] if d[1] else '' for d in a[0]])
except Exception as e:
logger.error('when format:%s',e)
logger.error('%s\n%s',r.text)
source = ''
target = ''
if len(source) != 0 and len(target) != 0:
file2.write(target+'\n')
else:
file2.write('\n')
else:
file2.write('\n')
urls = []
logger.info('finish 50 sentence, now at %s',num)
file2.close()
def sentencetranslate(line):
line = line.strip()
text = translator.translate(line,src='de',dest='en').text
return text
def completetranslate():
file1 = open('de2en_en.txt',mode='r',encoding='utf-8')
file2 = open('new_de2en_en.txt',mode='a',encoding='utf-8')
i = 1
with open('de.txt',mode='r',encoding='utf-8') as f:
for line in f:
t = file1.readline()
if len(t) == 1:#'only \n'
text = sentencetranslate(line)
file2.write(text+'\n')
else:
file2.write(t)
i += 1
if i%100 == 0:
print(i)
file1.close()
file2.close()
if __name__ == "__main__":
totaltranslate()
completetranslate()
totaltranslate()以及翻译了大部分的句子，但是可能因为某些原因有一些句子没有翻译，所以还需要completetranslate()来补全翻译结果。
在grequests.map()中size设置为20的情况下，平均每秒翻译20句。
