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

[WebCrawler/javdb] refine title value

Browse Source
This commit is contained in:
Max Zhao
2020-09-06 18:09:49 +08:00
parent 3d9c92aac5
commit 42e646e92c
1 changed files with 9 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
WebCrawler/javdb.py
View File

@@ -127,15 +127,21 @@ def main(number):
imagecut = 1 imagecut = 1
cover_small = getCover(detail_page) cover_small = getCover(detail_page)
number = getNum(detail_page)
title = getTitle(detail_page)
if title and number:
# remove duplicate title
title = title.replace(number, '').strip()
dic = { dic = {
'actor': getActor(detail_page), 'actor': getActor(detail_page),
'title': getTitle(detail_page), 'title': title,
'studio': getStudio(detail_page), 'studio': getStudio(detail_page),
'outline': getOutline(detail_page), 'outline': getOutline(detail_page),
'runtime': getRuntime(detail_page), 'runtime': getRuntime(detail_page),
'director': getDirector(detail_page), 'director': getDirector(detail_page),
'release': getRelease(detail_page), 'release': getRelease(detail_page),
'number': getNum(detail_page), 'number': number,
'cover': getCover(detail_page), 'cover': getCover(detail_page),
'cover_small': cover_small, 'cover_small': cover_small,
'imagecut': imagecut, 'imagecut': imagecut,
@@ -156,4 +162,4 @@ def main(number):
# main('DV-1562') # main('DV-1562')
# input("[+][+]Press enter key exit, you can check the error messge before you exit.\n[+][+]按回车键结束，你可以在结束之前查看和错误信息。") # input("[+][+]Press enter key exit, you can check the error messge before you exit.\n[+][+]按回车键结束，你可以在结束之前查看和错误信息。")
if __name__ == "__main__": if __name__ == "__main__":
print(main('snyz-007')) print(main('GS-351'))