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

45
WebCrawler/mgstage.py
View File

@@ -114,29 +114,32 @@ def main(number2):
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 ','')
#print(b)
dic = {
'title': getTitle(htmlcode).replace("\\n",'').replace(' ',''),
'studio': getStudio(a),
'outline': getOutline(b),
'runtime': getRuntime(a),
'director': getDirector(a),
'actor': getActor(a),
'release': getRelease(a),
'number': getNum(a),
'cover': getCover(htmlcode),
'imagecut': 0,
'tag': getTag(a),
'label':getLabel(a),
'extrafanart': getExtrafanart(htmlcode),
'year': getYear(getRelease(a)), # str(re.search('\d{4}',getRelease(a)).group()),
'actor_photo': '',
'website':'https://www.mgstage.com/product/product_detail/'+str(number)+'/',
'source': 'mgstage.py',
'series': getSeries(a),
}
try:
dic = {
'title': getTitle(htmlcode).replace("\\n", '').replace(' ', ''),
'studio': getStudio(a),
'outline': getOutline(b),
'runtime': getRuntime(a),
'director': getDirector(a),
'actor': getActor(a),
'release': getRelease(a),
'number': getNum(a),
'cover': getCover(htmlcode),
'imagecut': 0,
'tag': getTag(a),
'label': getLabel(a),
'extrafanart': getExtrafanart(htmlcode),
'year': getYear(getRelease(a)), # str(re.search('\d{4}',getRelease(a)).group()),
'actor_photo': '',
'website': 'https://www.mgstage.com/product/product_detail/' + str(number) + '/',
'source': 'mgstage.py',
'series': getSeries(a),
}
except:
dic = {"title": ""}
js = json.dumps(dic, ensure_ascii=False, sort_keys=True, indent=4, separators=(',', ':'), ) # .encode('UTF-8')
return js
#print(htmlcode)
if __name__ == '__main__':
print(main('SIRO-4149'))