def run(self):
"""
Fetch homepage URLs, lookup content item IDS, and set number of minutes
it's been on the homepage.
"""
p = self.options.pop('page')
for link in pageone.get(p, **self.options):
u = link.get('url')
# smartly handle urls
u = url.prepare(u, canonicalize=False)
if u and not u in self.url_lookup:
u = url.prepare(u, canonicalize=True)
# yield metrics
if u and u in self.url_lookup:
cids = self.url_lookup[u]
for cid in cids:
yield {
'datetime': dates.now(),
'content_item_id': cid,
'metrics': {
'time_on_homepage': self.recipe.get('minutes', 60)
}
}
def run(self):
"""
Fetch homepage URLs, lookup content item IDS, and set number of minutes
it's been on the homepage.
"""
p = self.options.pop('page')
for link in pageone.get(p, **self.options):
u = link.get('url')
# smartly handle urls
u = url.prepare(u, canonicalize=False)
if u and not u in self.url_lookup:
u = url.prepare(u, canonicalize=True)
# yield metrics
if u and u in self.url_lookup:
cids = self.url_lookup[u]
for cid in cids:
yield {
'datetime': dates.now(),
'content_item_id': cid,
'metrics': {
'time_on_homepage': self.recipe.get('minutes', 60)
}
}
def _prepare_url(o, field, source=None, **kw):
"""
Prepare a url
"""
if field not in o:
return None
if o[field] is None:
return None
if kw.get('canonicalize', False):
return url.prepare(o[field], source=source, **kw)
# prepare urls before attempting cached request.
u = url.prepare(o[field], source=source, expand=False, canonicalize=False)
cache_response = url_cache.get(u)
return cache_response.value
def _prepare_url(o, field, source=None, **kw):
"""
Prepare a url
"""
if field not in o:
return None
if o[field] is None:
return None
if kw.get('canonicalize', False):
return url.prepare(o[field], source=source, **kw)
# prepare urls before attempting cached request.
u = url.prepare(o[field], source=source, expand=False, canonicalize=False)
cache_response = url_cache.get(u)
return cache_response.value
def format(self, obj):
"""
For now all of these options are standard to twitter events.
"""
# set the status.
obj['status'] = self.options.get('event_status', 'pending')
# prepare url (these are formatted as redirects).
obj['url'] = url.prepare(obj['url'], expand=False, canonicalize=False)
# ignore bad domains / org's own domains.
if self._is_bad_domain(obj['url']):
return
# extract and merge article data.
if url.is_article(obj['url']):
data = article.extract(obj['url'], type=None)
if data:
obj.update(data)
obj.pop('type', None)
obj.pop('site_name', None)
obj.pop('favicon', None)
# set source id:
_id = obj.pop('id', obj.get('url', gen_uuid()))
if ":" in _id:
_id = _id.split(':')[-1]
obj['source_id'] = _id
# TODO: Make formatting more elegant.
if self.options.get('set_event_title', None):
obj['title'] = self.options.get(
'set_event_title').format(**self._fmt(obj))
if self.options.get('set_event_description', None):
obj['description'] = self.options.get(
'set_event_description').format(**self._fmt(obj))
if self.options.get('set_event_tag_ids', None) and \
len(self.options.get('set_event_tag_ids')):
obj['tag_ids'] = self.options.get('set_event_tag_ids')
# hack because the app cant handle this field being a list.
if self.options.get('set_event_content_items', None):
if 'content_item_ids' not in obj:
obj['content_item_ids'] = []
for c in self.options.get('set_event_content_items', []):
if isinstance(c, dict):
if c.get('id', None):
obj['content_item_ids'].append(c.get('id'))
elif isinstance(c, int):
obj['content_item_ids'].append(c)
# filter links.
if self.options.get('must_link', False) \
and not len(obj.get('links', [])):
return None
return obj
def work(self, raw_url):
"""
Standardize + cache a raw url
returning it's standardized url + global bitly url.
"""
# standradize the url
if url.is_abs(raw_url):
source = raw_url
else:
source = None
return url.prepare(raw_url, source=source, canonicalize=True, expand=True)
def _gen_lookups(self):
"""
Create a tree of url > content item ids.
"""
# create containers
self.url_lookup = defaultdict(list)
# populate with ALL content items.
for c in self.api.orgs.simple_content():
u = c.pop('url', None)
if u:
self.url_lookup[url.prepare(u)].append(c['id'])
def prepare_url(o, field, source=None):
"""
Prepare a url
"""
if field not in o:
return None
if o[field] is None:
return None
# prepare it first before the sending to the canoncilation cache
u = url.prepare(o[field], source=source, canonicalize=False, expand=False)
cache_response = url_cache.get(u)
return cache_response.value
def _gen_lookups(self):
"""
Create a tree of url > content item ids.
"""
# create containers
self.url_lookup = defaultdict(list)
# populate with ALL content items.
for c in self.api.orgs.simple_content():
u = c.pop('url', None)
if u:
self.url_lookup[url.prepare(u)].append(c['id'])
def prepare_url(o, field, source=None):
"""
Prepare a url
"""
if field not in o:
return None
if o[field] is None:
return None
# prepare it first before the sending to the canoncilation cache
u = url.prepare(o[field], source=source, canonicalize=False, expand=False)
cache_response = url_cache.get(u)
return cache_response.value
def work(self, raw_url):
"""
Standardize + cache a raw url
returning it's standardized url + global bitly url.
"""
# standradize the url
if url.is_abs(raw_url):
source = raw_url
else:
source = None
return url.prepare(raw_url,
source=source,
canonicalize=True,
expand=True)
def test_unshorten_url(self):
cases = [
('http://nyti.ms/1oxYm3e',
'http://www.nytimes.com/video/movies/100000002920951/anatomy-8216the-fault-in-our-stars8217.html'),
('nyti.ms/1oxYm3e',
'http://www.nytimes.com/video/movies/100000002920951/anatomy-8216the-fault-in-our-stars8217.html'),
('http://bit.ly/1kzIQWw',
'http://www.fromscratchradio.com/show/marc-dacosta'),
('bit.ly/aaaaaa', 'http://bit.ly/aaaaaa'),
('http://ow.ly/i/5OTms', 'http://ow.ly/i/5OTms'),
# ('http://j.mp/1jBOKo1', 'http://earthfix.info/portables')
]
for c in cases:
test, truth = c
try:
test = url.prepare(test)
assert(test == truth)
except AssertionError:
print "failed on %s" % test
raise
def test_unshorten_url(self):
cases = [
('http://nyti.ms/1oxYm3e',
'http://www.nytimes.com/video/movies/100000002920951/anatomy-8216the-fault-in-our-stars8217.html'
),
('nyti.ms/1oxYm3e',
'http://www.nytimes.com/video/movies/100000002920951/anatomy-8216the-fault-in-our-stars8217.html'
),
('http://bit.ly/1kzIQWw',
'http://www.fromscratchradio.com/show/marc-dacosta'),
('bit.ly/aaaaaa', 'http://bit.ly/aaaaaa'),
('http://ow.ly/i/5OTms', 'http://ow.ly/i/5OTms'),
# ('http://j.mp/1jBOKo1', 'http://earthfix.info/portables')
]
for c in cases:
test, truth = c
try:
test = url.prepare(test)
assert (test == truth)
except AssertionError:
print "failed on %s" % test
raise
def extract(source_url):
"""
Article extraction. Method is as follows:
1. Get html from url.
2. Canonicalize URL.
3. If not canonical, prepare the url.
4. Extract meta tags.
5. If embedly is active, use it for content extraction.
6. If embedly doesnt return content or is not active, use readability
7. If readability doesnt return content, use article tag.
8. If authors aren't detcted from meta tags, detect them in article body.
"""
# fetch page
page_html = network.get(source_url)
# something failed.
if not page_html:
log.warning("Failed to extract html from {}".format(source_url))
return None
soup = BeautifulSoup(page_html)
# get canonical url
canonical_url = meta.canonical_url(soup)
if not canonical_url:
canonical_url = url.prepare(
source_url, source=source_url, canonicalize=False)
# domain
domain = url.get_domain(canonical_url)
# get meta tags + other data
data = {
'url': canonical_url,
'domain': domain,
'title': meta.title(soup, canonical_url),
'description': meta.description(soup, canonical_url),
'img_url': meta.img_url(soup, canonical_url),
'created': meta.publish_date(soup, canonical_url),
'favicon': meta.favicon(soup, canonical_url),
'site_name': meta.site_name(soup, canonical_url),
'page_type': meta.page_type(soup, canonical_url),
'authors': author.extract(soup),
'body': None
}
# embed videos
if url.is_video(canonical_url):
data['body'] = embed.video(canonical_url)
return data
# extract article body
if settings.EMBEDLY_ENABLED:
data['body'] = body_via_embedly(canonical_url)
if not data['body']:
data['body'] = body_via_readability(page_html, canonical_url)
# # extract body from article tag
body, raw_html = body_via_article_tag(soup, canonical_url)
# merge body
if not data['body']:
data['body'] = body
# get creators from raw article html
if not len(data['authors']) and raw_html:
data['authors'] = author.extract(raw_html, tags=author.OPTIMISTIC_TAGS)
# remove site name from authors
if data.get('site_name'):
data['authors'] = [
a.replace(data['site_name'].upper(), "").strip()
for a in data['authors']
]
# # get links from raw_html + content
# links = [u for u in url.from_any(data['body']) if source_url not in u]
# for u in url.from_any(raw_html, source=source_url):
# if u not in links and (u != source_url or not u.startswith(source_url)):
# links.append(u)
# # split out internal / external links / article links
# data['links'] = url.categorize_links(links, domain)
return data
def test_prepare_with_redirect_back(self):
source = 'https://www.revealnews.org/article/a-brief-history-of-the-modern-strawberry/'
u = '//cdn.embedly.com/widgets/media.html?url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DfPxUIz5GHAE&src=http%3A%2F%2Fwww.youtube.com%2Fembed%2FfPxUIz5GHAE&type=text%2Fhtml&key=1b74e47c9db441f8a998fb6138abca72&schema=youtube'
out = url.prepare(u, source)
assert(out == 'http://www.youtube.com/watch?v=fPxUIz5GHAE')
def extract(source_url):
"""
Article extraction. Method is as follows:
1. Get html from url.
2. Canonicalize URL.
3. If not canonical, prepare the url.
4. Extract meta tags.
5. If embedly is active, use it for content extraction.
6. If embedly doesnt return content or is not active, use readability
7. If readability doesnt return content, use article tag.
8. If authors aren't detcted from meta tags, detect them in article body.
"""
# fetch page
page_html = network.get(source_url)
# something failed.
if not page_html:
log.warning("Failed to extract html from {}".format(source_url))
return None
soup = BeautifulSoup(page_html)
# get canonical url
canonical_url = meta.canonical_url(soup)
if not canonical_url:
canonical_url = url.prepare(source_url,
source=source_url,
canonicalize=False)
# domain
domain = url.get_domain(canonical_url)
# get meta tags + other data
data = {
'url': canonical_url,
'domain': domain,
'title': meta.title(soup, canonical_url),
'description': meta.description(soup, canonical_url),
'img_url': meta.img_url(soup, canonical_url),
'created': meta.publish_date(soup, canonical_url),
'favicon': meta.favicon(soup, canonical_url),
'site_name': meta.site_name(soup, canonical_url),
'page_type': meta.page_type(soup, canonical_url),
'authors': author.extract(soup),
'body': None
}
# extract body from embedly + readability
if settings.EMBEDLY_ENABLED:
data['body'] = body_via_embedly(canonical_url)
if not data['body']:
data['body'] = body_via_readability(page_html, canonical_url)
# # extract body from article tag
body, raw_html = body_via_article_tag(soup, canonical_url)
# merge body
if not data['body']:
data['body'] = body
# get creators from raw article html
if not len(data['authors']) and raw_html:
data['authors'] = author.extract(raw_html, tags=author.OPTIMISTIC_TAGS)
# remove site name from authors
if data.get('site_name'):
data['authors'] = [
a.replace(data['site_name'].upper(), "").strip()
for a in data['authors']
]
# # get links from raw_html + content
# links = [u for u in url.from_any(data['body']) if source_url not in u]
# for u in url.from_any(raw_html, source=source_url):
# if u not in links and (u != source_url or not u.startswith(source_url)):
# links.append(u)
# # split out internal / external links / article links
# data['links'] = url.categorize_links(links, domain)
return data
def test_prepare_with_redirect_back(self):
source = 'https://www.revealnews.org/article/a-brief-history-of-the-modern-strawberry/'
u = '//cdn.embedly.com/widgets/media.html?url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DfPxUIz5GHAE&src=http%3A%2F%2Fwww.youtube.com%2Fembed%2FfPxUIz5GHAE&type=text%2Fhtml&key=1b74e47c9db441f8a998fb6138abca72&schema=youtube'
out = url.prepare(u, source)
assert (out == 'http://www.youtube.com/watch?v=fPxUIz5GHAE')
