def callback(self, response):
"""Will process the result from the challenge, creating missing article
and updating the feed"""
article_created = False
if response.result().status_code != 204:
results = response.result().json()
logger.debug('%r %r - %d entries were not matched '
'and will be created',
self.feed['id'], self.feed['title'], len(results))
for id_to_create in results:
article_created = True
entry = construct_article(
self.entries[tuple(sorted(id_to_create.items()))],
self.feed)
logger.info('%r %r - creating %r for %r - %r', self.feed['id'],
self.feed['title'], entry['title'],
entry['user_id'], id_to_create)
self.query_pyagg('post', 'article', entry)
logger.debug('%r %r - updating feed etag %r last_mod %r',
self.feed['id'], self.feed['title'],
self.headers.get('etag', ''),
self.headers.get('last-modified', ''))
up_feed = {'error_count': 0, 'last_error': None,
'etag': self.headers.get('etag', ''),
'last_modified': self.headers.get('last-modified',
strftime('%a, %d %b %Y %X %Z', gmtime()))}
fresh_feed = construct_feed_from(url=self.feed['link'],
fp_parsed=self.parsed_feed)
for key in ('description', 'site_link', 'icon_url'):
if fresh_feed.get(key) and fresh_feed[key] != self.feed.get(key):
up_feed[key] = fresh_feed[key]
if not self.feed.get('title'):
up_feed['title'] = fresh_feed.get('title', '')
up_feed['user_id'] = self.feed['user_id']
# re-getting that feed earlier since new entries appeared
if article_created:
up_feed['last_retrieved'] \
= (datetime.now() - timedelta(minutes=45)).isoformat()
diff_keys = {key for key in up_feed
if up_feed[key] != self.feed.get(key)}
if not diff_keys:
return # no change in the feed, no update
if not article_created and diff_keys == {'last_modified', 'etag'}:
return # meaningless if no new article has been published
logger.info('%r %r - pushing feed attrs %r',
self.feed['id'], self.feed['title'],
{key: "%s -> %s" % (up_feed[key], self.feed.get(key))
for key in up_feed if up_feed[key] != self.feed.get(key)})
self.query_pyagg('put', 'feed/%d' % self.feed['id'], up_feed)
async def insert_database(user, feed):
articles = await parse_feed(user, feed)
if None is articles:
return []
logger.debug('inserting articles for {}'.format(feed.title))
logger.info("Database insertion...")
new_articles = []
art_contr = ArticleController(user.id)
for article in articles:
existing_article_req = art_contr.read(feed_id=feed.id,
**extract_id(article))
exist = existing_article_req.count() != 0
if exist:
# if the article has been already retrieved, we only update
# the content or the title
logger.debug("Article %r (%r) already in the database.",
article['title'], article['link'])
existing_article = existing_article_req.first()
new_updated_date = None
try:
new_updated_date = dateutil.parser.parse(article['updated'])
except Exception as e:
print(e) #new_updated_date = existing_article.date
if None is existing_article.updated_date:
existing_article.updated_date = new_updated_date.replace(
tzinfo=None)
if existing_article.updated_date.strftime('%Y-%m-%dT%H:%M:%S') != \
new_updated_date.strftime('%Y-%m-%dT%H:%M:%S'):
existing_article.updated_date = \
new_updated_date.replace(tzinfo=None)
if existing_article.title != article['title']:
existing_article.title = article['title']
content = get_article_content(article)
if existing_article.content != content:
existing_article.content = content
existing_article.readed = False
art_contr.update({'entry_id': existing_article.entry_id},
existing_article.dump())
continue
article = construct_article(article, feed)
try:
new_articles.append(art_contr.create(**article))
logger.info("New article % (%r) added.", article['title'],
article['link'])
except Exception:
logger.exception("Error when inserting article in database:")
continue
return new_articles
async def insert_database(user, feed):
articles = await parse_feed(user, feed)
if None is articles:
return []
logger.debug('inserting articles for {}'.format(feed.title))
logger.info("Database insertion...")
new_articles = []
art_contr = ArticleController(user.id)
for article in articles:
existing_article_req = art_contr.read(feed_id=feed.id,
**extract_id(article))
exist = existing_article_req.count() != 0
if exist:
# if the article has been already retrieved, we only update
# the content or the title
logger.debug("Article %r (%r) already in the database.",
article['title'], article['link'])
existing_article = existing_article_req.first()
new_updated_date = None
try:
new_updated_date = dateutil.parser.parse(article['updated'])
except Exception as e:
print(e)#new_updated_date = existing_article.date
if None is existing_article.updated_date:
existing_article.updated_date = new_updated_date.replace(tzinfo=None)
if existing_article.updated_date.strftime('%Y-%m-%dT%H:%M:%S') != \
new_updated_date.strftime('%Y-%m-%dT%H:%M:%S'):
existing_article.updated_date = \
new_updated_date.replace(tzinfo=None)
if existing_article.title != article['title']:
existing_article.title = article['title']
content = get_article_content(article)
if existing_article.content != content:
existing_article.content = content
existing_article.readed = False
art_contr.update({'entry_id': existing_article.entry_id},
existing_article.dump())
continue
article = construct_article(article, feed)
try:
new_articles.append(art_contr.create(**article))
logger.info("New article % (%r) added.",
article['title'], article['link'])
except Exception:
logger.exception("Error when inserting article in database:")
continue
return new_articles