bacow/AV_Data_Capture
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #499 from lededev/fc2-nbr

Browse Source 
javdb:FC2 PPV number precise matching
This commit is contained in:
Yoshiko2
2021-05-31 21:41:48 +08:00
committed by GitHub
parent a3ceaf42b2 a5363430b1
commit 98f50e76cc
1 changed files with 12 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
WebCrawler/javdb.py
View File

@@ -212,6 +212,7 @@ def main(number):
# else: # else:
# number = number.upper() # number = number.upper()
number = number.upper() number = number.upper()
isFC2PPV = bool(re.search(r'^FC2-\d+', number))
cookie_json = './' + javdb_site + '.json' cookie_json = './' + javdb_site + '.json'
javdb_cookies = None javdb_cookies = None
# 不加载过期的cookie，javdb登录界面显示为7天免登录，故假定cookie有效期为7天 # 不加载过期的cookie，javdb登录界面显示为7天免登录，故假定cookie有效期为7天
@@ -240,6 +241,9 @@ def main(number):
try: try:
correct_url = urls[ids.index(number)] correct_url = urls[ids.index(number)]
except: except:
# 为避免获得错误番号，FC2 PPV 只要精确对应的结果
if isFC2PPV and ids[0] != number:
raise ValueError("number not found")
# if input number is "STAR438" not "STAR-438", use first search result. # if input number is "STAR438" not "STAR-438", use first search result.
correct_url = urls[0] correct_url = urls[0]
try: try:
@@ -264,9 +268,12 @@ def main(number):
imagecut = 1 imagecut = 1
cover_small = getCover(detail_page) cover_small = getCover(detail_page)
number = getNum(detail_page) dp_number = getNum(detail_page)
if isFC2PPV and dp_number != number:
raise ValueError("number not found")
title = getTitle(detail_page) title = getTitle(detail_page)
if title and number: if title and dp_number:
number = dp_number
# remove duplicate title # remove duplicate title
title = title.replace(number, '').strip() title = title.replace(number, '').strip()
@@ -305,6 +312,7 @@ def main(number):
if __name__ == "__main__": if __name__ == "__main__":
# print(main('blacked.20.05.30')) # print(main('blacked.20.05.30'))
# print(main('AGAV-042')) # print(main('AGAV-042'))
print(main('BANK-022')) # print(main('BANK-022'))
print(main('FC2-735670')) print(main('FC2-735670'))
print(main('MVSD-439')) print(main('FC2-1174949'))
print(main('MVSD-439'))