gcolle.py:Add try block
This commit is contained in:
lededev
@@ -6,65 +6,68 @@ from ADC_function import *
|
|||||||
from lxml import etree
|
from lxml import etree
|
||||||
|
|
||||||
def main(number):
|
def main(number):
|
||||||
config_file = config.getInstance()
|
try:
|
||||||
|
number = number.upper().replace('GCOLLE-','')
|
||||||
|
session = get_html_session()
|
||||||
|
|
||||||
number = number.upper().replace('GCOLLE-','')
|
htmlcode = session.get('https://gcolle.net/product_info.php/products_id/' + number).text
|
||||||
session = get_html_session()
|
html = etree.HTML(htmlcode)
|
||||||
|
# R18 countinue
|
||||||
|
htmlcode = session.get(html.xpath('//*[@id="main_content"]/table[1]/tbody/tr/td[2]/table/tbody/tr/td/h4/a[2]/@href')[0]).text
|
||||||
|
gcolle_crawler = Crawler(htmlcode)
|
||||||
|
|
||||||
htmlcode = session.get('https://gcolle.net/product_info.php/products_id/' + number).text
|
number_html = gcolle_crawler.getString('//td[contains(text(),"商品番号")]/../td[2]/text()')
|
||||||
html = etree.HTML(htmlcode)
|
if number != number_html:
|
||||||
# R18 countinue
|
raise Exception('[-]gcolle.py: number not match')
|
||||||
htmlcode = session.get(html.xpath('//*[@id="main_content"]/table[1]/tbody/tr/td[2]/table/tbody/tr/td/h4/a[2]/@href')[0]).text
|
|
||||||
gcolle_crawler = Crawler(htmlcode)
|
|
||||||
|
|
||||||
number_html = gcolle_crawler.getString('//td[contains(text(),"商品番号")]/../td[2]/text()')
|
# get extrafanart url
|
||||||
if number != number_html:
|
if len(gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/img/@src')) == 0:
|
||||||
if config_file.debug():
|
extrafanart = gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/a/img/@src')
|
||||||
print('[!]gcolle.py: number not match')
|
else:
|
||||||
return {'title':''}
|
extrafanart = gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/img/@src')
|
||||||
|
# Add "https:" in each extrafanart url
|
||||||
|
for i in range(len(extrafanart)):
|
||||||
|
extrafanart[i] = 'https:' + extrafanart[i]
|
||||||
|
|
||||||
# get extrafanart url
|
dic = {
|
||||||
if len(gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/img/@src')) == 0:
|
"title": gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[1]/td/h1/text()'),
|
||||||
extrafanart = gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/a/img/@src')
|
"studio": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
||||||
else:
|
"year": re.findall('\d{4}',gcolle_crawler.getString('//td[contains(text(),"商品登録日")]/../td[2]/time/@datetime'))[0],
|
||||||
extrafanart = gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[3]/td/div/img/@src')
|
"outline": gcolle_crawler.getOutline('//*[@id="cart_quantity"]/table/tr[3]/td/p/text()'),
|
||||||
# Add "https:" in each extrafanart url
|
"runtime": '',
|
||||||
for i in range(len(extrafanart)):
|
"director": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
||||||
extrafanart[i] = 'https:' + extrafanart[i]
|
"actor": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
||||||
|
"release": re.findall('\d{4}-\d{2}-\d{2}',gcolle_crawler.getString('//td[contains(text(),"商品登録日")]/../td[2]/time/@datetime'))[0],
|
||||||
|
"number": "GCOLLE-" + str(number_html),
|
||||||
|
"cover": "https:" + gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[3]/td/table/tr/td/a/@href'),
|
||||||
|
"thumb": "https:" + gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[3]/td/table/tr/td/a/@href'),
|
||||||
|
"trailer": '',
|
||||||
|
"actor_photo":'',
|
||||||
|
"imagecut": 4, # 该值为4时同时也是有码影片 也用人脸识别裁剪封面
|
||||||
|
"tag": gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[4]/td/a/text()'),
|
||||||
|
"extrafanart":extrafanart,
|
||||||
|
"label": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
||||||
|
"website": 'https://gcolle.net/product_info.php/products_id/' + number,
|
||||||
|
"source": 'gcolle.py',
|
||||||
|
"series": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
||||||
|
'无码': False,
|
||||||
|
}
|
||||||
|
# for k,v in dic.items():
|
||||||
|
# if k == 'outline':
|
||||||
|
# print(k,len(v))
|
||||||
|
# else:
|
||||||
|
# print(k,v)
|
||||||
|
# print('===============================================================')
|
||||||
|
except Exception as e:
|
||||||
|
dic = {'title':''}
|
||||||
|
if config.getInstance().debug():
|
||||||
|
print(e)
|
||||||
|
|
||||||
dic = {
|
|
||||||
"title": gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[1]/td/h1/text()'),
|
|
||||||
"studio": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
|
||||||
"year": re.findall('\d{4}',gcolle_crawler.getString('//td[contains(text(),"商品登録日")]/../td[2]/time/@datetime'))[0],
|
|
||||||
"outline": gcolle_crawler.getOutline('//*[@id="cart_quantity"]/table/tr[3]/td/p/text()'),
|
|
||||||
"runtime": '',
|
|
||||||
"director": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
|
||||||
"actor": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
|
||||||
"release": re.findall('\d{4}-\d{2}-\d{2}',gcolle_crawler.getString('//td[contains(text(),"商品登録日")]/../td[2]/time/@datetime'))[0],
|
|
||||||
"number": "GCOLLE-" + str(number_html),
|
|
||||||
"cover": "https:" + gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[3]/td/table/tr/td/a/@href'),
|
|
||||||
"thumb": "https:" + gcolle_crawler.getString('//*[@id="cart_quantity"]/table/tr[3]/td/table/tr/td/a/@href'),
|
|
||||||
"trailer": '',
|
|
||||||
"actor_photo":'',
|
|
||||||
"imagecut": 4, # 该值为4时同时也是有码影片 也用人脸识别裁剪封面
|
|
||||||
"tag": gcolle_crawler.getStrings('//*[@id="cart_quantity"]/table/tr[4]/td/a/text()'),
|
|
||||||
"extrafanart":extrafanart,
|
|
||||||
"label": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
|
||||||
"website": 'https://gcolle.net/product_info.php/products_id/' + number,
|
|
||||||
"source": 'gcolle.py',
|
|
||||||
"series": gcolle_crawler.getString('//td[contains(text(),"アップロード会員名")]/b/text()'),
|
|
||||||
'无码': False,
|
|
||||||
}
|
|
||||||
# for k,v in dic.items():
|
|
||||||
# if k == 'outline':
|
|
||||||
# print(k,len(v))
|
|
||||||
# else:
|
|
||||||
# print(k,v)
|
|
||||||
# print('===============================================================')
|
|
||||||
return dic
|
return dic
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
from pprint import pprint
|
from pprint import pprint
|
||||||
|
config.getInstance().set_override("debug_mode:switch=1")
|
||||||
pprint(main('840724'))
|
pprint(main('840724'))
|
||||||
pprint(main('840386'))
|
pprint(main('840386'))
|
||||||
pprint(main('838671'))
|
pprint(main('838671'))
|
||||||
|
|||||||
Reference in New Issue
Block a user