Update 1.8

This commit is contained in:
Yoshiko 2019-12-15 20:30:20 +08:00 committed by GitHub
parent ec9957c75c
commit 3b84877a53
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
4 changed files with 17 additions and 21 deletions

View File

@ -8,6 +8,9 @@ import re
import time
import sys
from lxml import etree
import sys
import io
sys.stdout = io.TextIOWrapper(sys.stdout.buffer, errors = 'replace', line_buffering = True)
config_file='config.ini'
config = ConfigParser()
@ -103,16 +106,7 @@ def get_html(url,cookies = None):#网页请求核心
getweb = requests.get(str(url), headers=headers, timeout=timeout, cookies=cookies)
getweb.encoding = 'utf-8'
return getweb.text
except requests.exceptions.RequestException:
i += 1
print('[-]Connect retry '+str(i)+'/'+str(retry_count))
except requests.exceptions.ConnectionError:
i += 1
print('[-]Connect retry '+str(i)+'/'+str(retry_count))
except requests.exceptions.ProxyError:
i += 1
print('[-]Connect retry '+str(i)+'/'+str(retry_count))
except requests.exceptions.ConnectTimeout:
except:
i += 1
print('[-]Connect retry '+str(i)+'/'+str(retry_count))
print('[-]Connect Failed! Please check your Proxy or Network!')

View File

@ -51,6 +51,8 @@ def getYear(release):
def main(number):
number=number.replace('PPV','').replace('ppv','').strip('fc2_').strip('fc2-').strip('ppv-').strip('PPV-').strip('FC2_').strip('FC2-').strip('ppv-').strip('PPV-').replace('fc2ppv-','').replace('FC2PPV-','')
htmlcode2 = ADC_function.get_html('http://adult.contents.fc2.com/article_search.php?id='+str(number).lstrip("FC2-").lstrip("fc2-").lstrip("fc2_").lstrip("fc2-")+'')
print('http://adult.contents.fc2.com/article_search.php?id='+str(number).lstrip("FC2-").lstrip("fc2-").lstrip("fc2_").lstrip("fc2-")+'')
print(htmlcode2)
#htmlcode = ADC_function.get_html('http://fc2fans.club/html/FC2-' + number + '.html')
dic = {
'title': getTitle(htmlcode2),
@ -70,6 +72,4 @@ def main(number):
'source': 'fc2fans_club.py',
}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'),)#.encode('UTF-8')
return js
#print(main('1145465'))
return js

View File

@ -134,4 +134,5 @@ def main_uncensored(number):
'source': 'javbus.py',
}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) # .encode('UTF-8')
return js
return js

15
siro.py
View File

@ -7,7 +7,7 @@ from ADC_function import *
def getTitle(a):
try:
html = etree.fromstring(a, etree.HTMLParser())
result = str(html.xpath('//*[@id="center_column"]/div[2]/h1/text()')).strip(" ['']")
result = str(html.xpath('//*[@id="center_column"]/div[1]/h1/text()')).strip(" ['']")
return result.replace('/', ',')
except:
return ''
@ -62,7 +62,8 @@ def getTag(a):
return str(result1 + result2).strip('+').replace("', '\\n",",").replace("', '","").replace('"','')
def getCover(htmlcode):
html = etree.fromstring(htmlcode, etree.HTMLParser())
result = str(html.xpath('//*[@id="center_column"]/div[2]/div[1]/div/div/h2/img/@src')).strip(" ['']")
result = str(html.xpath('//*[@id="center_column"]/div[1]/div[1]/div/div/h2/img/@src')).strip(" ['']")
# /html/body/div[2]/article[2]/div[1]/div[1]/div/div/h2/img/@src
return result
def getDirector(a):
html = etree.fromstring(a, etree.HTMLParser()) # //table/tr[1]/td[1]/text()
@ -77,7 +78,7 @@ def getOutline(htmlcode):
return result
def main(number2):
number=number2.upper()
htmlcode=get_html('https://www.mgstage.com/product/product_detail/'+str(number)+'/',cookies={'adc':'1'})
htmlcode=str(get_html('https://www.mgstage.com/product/product_detail/'+str(number)+'/',cookies={'adc':'1'}))
soup = BeautifulSoup(htmlcode, 'lxml')
a = str(soup.find(attrs={'class': 'detail_data'})).replace('\n ','').replace(' ','').replace('\n ','').replace('\n ','')
dic = {
@ -98,7 +99,7 @@ def main(number2):
'website':'https://www.mgstage.com/product/product_detail/'+str(number)+'/',
'source': 'siro.py',
}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'),)#.encode('UTF-8')
return js
#print(main('300maan-373'))
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) # .encode('UTF-8')
print(js)
#print(htmlcode)
return js