s434776/PRI_2020-FE
2
0
Fork 0
Code Issues Pull Requests Releases Activity

remove HTML tags from messages

Browse Source
This commit is contained in:
Karolina Boczoń 2020-05-10 14:52:34 +00:00
parent ecb4b1148b
commit 1735d9ddb8
1 changed files with 10 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
backend/xmlParser.py
View File

@ -20,17 +20,20 @@ out['id'] = soup.forum.get('id')
out['name'] = soup.forum.find('name').text
out['discussions'] = []
for d in soup.forum.find_all('discussion'):
posts = []
for p in d.find_all('post'):
message_soup = BeautifulSoup(p.message.get_text(), "xml")
posts.append({
'id': p.get('id'),
'parent': p.find('parent').text,
'author': p.userid.text,
'message': message_soup.get_text()
})
out['discussions'].append({
'id': d.get('id'),
'title': d.find('name').text,
'first_post': d.firstpost.text,
'posts': [
{
'id': p.get('id'),
'parent': p.find('parent').text,
'author': p.userid.text,
'message': p.message.get_text()
} for p in d.find_all('post')]
'posts': posts
})
with open('parsed.json', 'w') as outfile: