def post(self):
new_post = Post()
if len(self.request.get('caption').strip()) > 0:
new_post.text_caption = self.request.get('caption')
try:
upload = self.get_uploads()[0]
new_post.image_blob = upload.key()
except:
pass
user = users.get_current_user()
myuser_key = ndb.Key('User', user.user_id())
myuser = myuser_key.get()
new_post.owner_user = myuser_key
new_post.date = datetime.now()
new_key = new_post.put()
myuser.posts.insert(0, new_key)
myuser.put()
self.redirect('/')
def _parse(self):
for element in self.root.iter():
if element.tag == 'ID':
self.id = element.text
elif element.tag == 'WRITING':
post = Post()
elif element.tag == 'TITLE':
post.title = element.text
elif element.tag == 'DATE':
d = Date(element.text)
post.date = d
elif element.tag == 'TEXT':
#self._find_emojis2(element.text)
post.text = element.text
elif element.tag == 'INFO':
post.info = element.text
self.posts.append(post)
def parse_all_posts(self, author=True, title=True, date=True, contents=True, messages=True,
reply=True, images=True, ip=False):
try:
if self.post_urls is None:
raise Exception("You must run get_posts_url first")
except Exception as err:
print(err)
sys.exit(1)
post_objects = list()
for url in self.post_urls:
post_url = "http://www.ptt.cc/bbs/{}/{}".format(self.board_name, url)
soup = self._get_soup(post_url)
post = Post(post_url)
try:
article_meta = soup.find_all('span', class_="article-meta-value")
if author:
post.author = article_meta[0].contents[0]
if title:
post.title = article_meta[2].contents[0]
if date:
post.date = article_meta[3].contents[0]
# TODO: ip
if ip:
pass
if contents:
a = str(soup.find(id="main-container").contents[1])
a = a.split("</div>")
a = a[4].split("<span class=\"f2\">※ 發信站: 批踢踢實業坊(ptt.cc),")
post.contents = a[0].replace(' ', '').replace('\n', '').replace('\t', '')
if images:
a = soup.find_all('img')
imgs = list()
for i in a:
imgs.append(i.attrs.get('src'))
post.images = imgs
# TODO: There are some push_content lost due to hyperlink in the content
if messages or reply:
# messages = list()
for tag in soup.find_all("div", "push"):
# d = dict()
push_tag = tag.find("span", "push-tag").\
string.replace(' ', '')
# d.setdefault('狀態', push_tag)
# d.setdefault('留言者', push_userid)
# d.setdefault('留言內容', push_content)
# d.setdefault('留言時間', push_ipdatetime)
# messages.append(d)
if push_tag == '推':
post.good += 1
elif push_tag == '噓':
post.bad += 1
else:
post.normal += 1
# post.messages = messages
except:
pass
post_objects.append(post)
return post_objects
