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

xcity.py: Mode 3 requires the file name to remain unchanged

Browse Source
This commit is contained in:
lededev
2021-09-25 06:45:08 +08:00
parent 50574a705b
commit 6c990e8482
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
WebCrawler/xcity.py
View File

@@ -182,9 +182,10 @@ def getExtrafanart(htmlcode): # 获取剧照
def main(number): def main(number):
try: try:
xcity_number = number.replace('-','')
query_result, browser = get_html_by_form( query_result, browser = get_html_by_form(
'https://xcity.jp/about/', 'https://xcity.jp/about/',
fields = {'q' : number.replace('-','').lower()}, fields = {'q' : xcity_number.lower()},
return_type = 'browser') return_type = 'browser')
if not query_result or not query_result.ok: if not query_result or not query_result.ok:
raise ValueError("xcity.py: page not found") raise ValueError("xcity.py: page not found")
@@ -193,6 +194,10 @@ def main(number):
raise ValueError("xcity.py: detail page not found") raise ValueError("xcity.py: detail page not found")
detail_page = str(browser.page) detail_page = str(browser.page)
url = browser.url url = browser.url
newnum = getNum(detail_page).upper()
number_up = number.upper()
if newnum != number_up and newnum == xcity_number.upper():
newnum = number_up
dic = { dic = {
'actor': getActor(browser), 'actor': getActor(browser),
'title': getTitle(detail_page), 'title': getTitle(detail_page),
@@ -201,7 +206,7 @@ def main(number):
'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': newnum,
'cover': getCover(detail_page), 'cover': getCover(detail_page),
'cover_small': '', 'cover_small': '',
'extrafanart': getExtrafanart(detail_page), 'extrafanart': getExtrafanart(detail_page),