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

Fix mgstage source error Exception handling

Browse Source
This commit is contained in:
yoshiko2
2021-05-03 22:14:22 +08:00
parent ef87a626da
commit fceed1d04e
1 changed files with 24 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
WebCrawler/mgstage.py
View File

@@ -114,8 +114,9 @@ def main(number2):
a = str(soup.find(attrs={'class': 'detail_data'})).replace('\n ','').replace(' ','').replace('\n ','').replace('\n ','') a = str(soup.find(attrs={'class': 'detail_data'})).replace('\n ','').replace(' ','').replace('\n ','').replace('\n ','')
b = str(soup.find(attrs={'id': 'introduction'})).replace('\n ','').replace(' ','').replace('\n ','').replace('\n ','') b = str(soup.find(attrs={'id': 'introduction'})).replace('\n ','').replace(' ','').replace('\n ','').replace('\n ','')
#print(b) #print(b)
try:
dic = { dic = {
'title': getTitle(htmlcode).replace("\\n",'').replace(' ',''), 'title': getTitle(htmlcode).replace("\\n", '').replace(' ', ''),
'studio': getStudio(a), 'studio': getStudio(a),
'outline': getOutline(b), 'outline': getOutline(b),
'runtime': getRuntime(a), 'runtime': getRuntime(a),
@@ -126,17 +127,19 @@ def main(number2):
'cover': getCover(htmlcode), 'cover': getCover(htmlcode),
'imagecut': 0, 'imagecut': 0,
'tag': getTag(a), 'tag': getTag(a),
'label':getLabel(a), 'label': getLabel(a),
'extrafanart': getExtrafanart(htmlcode), 'extrafanart': getExtrafanart(htmlcode),
'year': getYear(getRelease(a)), # str(re.search('\d{4}',getRelease(a)).group()), 'year': getYear(getRelease(a)), # str(re.search('\d{4}',getRelease(a)).group()),
'actor_photo': '', 'actor_photo': '',
'website':'https://www.mgstage.com/product/product_detail/'+str(number)+'/', 'website': 'https://www.mgstage.com/product/product_detail/' + str(number) + '/',
'source': 'mgstage.py', 'source': 'mgstage.py',
'series': getSeries(a), 'series': getSeries(a),
} }
except:
dic = {"title": ""}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) # .encode('UTF-8') js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) # .encode('UTF-8')
return js return js
#print(htmlcode)
if __name__ == '__main__': if __name__ == '__main__':
print(main('SIRO-4149')) print(main('SIRO-4149'))