def get_image_url(soup, url):
og_image = get_og_property(soup, 'image')
if og_image:
return urljoin(url, og_image)
tw_image = get_twitter_property(soup, 'image')
if tw_image:
return urljoin(url, tw_image)
def get_preview_video(soup):
#TODO: find using property="twitter:player:url" and "twitter:player:stream:url"
# (ref: view-source:https://www.kickstarter.com/projects/1087256999/odiun-a-web-site)
if soup:
og_vids = get_og_property(soup, 'video', deep=True)
for og_vid in og_vids:
if og_vid.endswith('.mp4'):
return og_vid
if len(og_vids) > 0:
return og_vids[0]
tw_vid = get_twitter_property(soup, 'player')
if tw_vid:
return tw_vid
og_vid_url = get_og_property(soup, 'video:url')
if og_vid_url:
return og_vid_url
return ''
def get_preview_picture(soup):
if soup:
og_pic = get_og_property(soup, 'image')
if og_pic:
return og_pic
tw_pic = get_twitter_property(soup, 'image')
if tw_pic:
return tw_pic
return ''
def get_type(url, soup):
if 'imgur.com' in url:
type_str = 'image'
else:
type_str = get_og_property(soup, 'type')
if type_str:
type_str = type_str.split('.')[0]
return ContentType.get_content_type(type_str)
return None
def get_description(soup, feed):
og_desc = get_og_property(soup, 'description')
if og_desc:
return og_desc
tw_desc = get_twitter_property(soup, 'description')
if tw_desc:
return tw_desc
meta_desc = get_meta_property(soup, 'description')
if meta_desc:
return meta_desc
return clean_html(feed.description)
def get_title(soup, feed):
og_title = get_og_property(soup, 'title')
if og_title:
return og_title
tw_title = get_twitter_property(soup, 'title')
if tw_title:
return tw_title
meta_title = get_meta_property(soup, 'title')
if meta_title:
return meta_title
return clean_html(feed.title)
def get_metadata_description(soup):
og_desc = get_og_property(soup, 'description')
if og_desc:
return og_desc
tw_desc = get_twitter_property(soup, 'description')
if tw_desc:
return tw_desc
meta_desc = get_meta_property(soup, 'description')
if not meta_desc:
meta_desc = get_meta_property(soup, 'Description')
if meta_desc:
return meta_desc
return ''
def get_metadata_title(soup):
og_title = get_og_property(soup, 'title')
if og_title:
return og_title
tw_title = get_twitter_property(soup, 'title')
if tw_title:
return tw_title
meta_title = get_meta_property(soup, 'title')
if meta_title:
return meta_title
if soup:
if soup.find('title'):
return soup.find('title').get_text()
return ''
def get_site_name(soup, session):
site_name = get_og_property(soup, 'site_name')
if site_name:
return SiteName.get_or_create_site_name(session, site_name)
return None
def get_url(page_request, soup):
url = get_og_property(soup, 'url')
if not url or not is_valid_url(url):
url = page_request.url
return url