def test_link_stats(self):
"""
TK: better tests, I'm just checking it runs various ways. for now:
"""
for link in pageone.get('https://www.propublica.org/', pattern=p):
assert('url' in link)
for link in pageone.get('https://www.npr.org', pattern=p, visible_only=True):
assert(link['visible'])
def test_link_stats(self):
"""
TK: better tests, I'm just checking it runs various ways. for now:
"""
for link in pageone.get('https://www.propublica.org/', pattern=p):
assert ('url' in link)
for link in pageone.get('https://www.npr.org',
pattern=p,
visible_only=True):
assert (link['visible'])
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)
}
}
