def get_articles(feed, fn): """ Parse the specified feed, gathering the latest new articles. If an article matches one that already exists, it is skipped. The minimum length of an entry is 500 characters. Anything under will be ignored. This will silently skip articles for which the full text can't be retrieved (i.e. if it returns 404). Some feeds, for whatever reason, do not include a `published` date in their entry data. In which case, it is left as an empty string. Args: | feed (Feed) -- the feed to fetch from. | fn (Callable) -- function to use an article """ # Fetch the feed data. data = feedparser.parse(feed.ext_url) # If the `bozo` value is anything # but 0, there was an error parsing (or connecting) to the feed. if data.bozo: # Some errors are ok. if not isinstance(data.bozo_exception, feedparser.CharacterEncodingOverride) and not isinstance(data.bozo_exception, feedparser.NonXMLContentType): raise data.bozo_exception for entry in data.entries: # URL for this entry. url = entry['links'][0]['href'] # Check for an existing Article. # If one exists, skip. if Article.query.filter_by(ext_url=url).count(): continue # Complete HTML content for this entry. try: entry_data, html = extractor.extract_entry_data(url) except (error.HTTPError, error.URLError, ConnectionResetError, BadStatusLine) as e: if type(e) == error.URLError or e.code == 404: # Can't reach, skip. logger.exception('Error extracting data for url {0}'.format(url)) continue else: # Just skip so things don't break! logger.exception('Error extracting data for url {0}'.format(url)) continue if entry_data is None: continue full_text = entry_data.cleaned_text # Skip over entries that are too short. if len(full_text) < 400: continue url = entry_data.canonical_link or url published = parse(entry.get('published')) if entry.get('published') else entry_data.publish_date updated = parse(entry.get('updated')) if entry.get('updated') else published title = entry.get('title', entry_data.title) # Secondary check for an existing Article, # by checking the title and source. existing = Article.query.filter_by(title=title).first() if existing and existing.source == feed.source: continue # Download and save the top article image. image_url = extractor.extract_image(entry_data, filename=hash(url)) fn(Article( ext_url=url, source=feed.source, feed=feed, html=html, text=fix_text_segment(full_text), authors=extractor.extract_authors(entry), tags=extractor.extract_tags(entry, known_tags=entry_data.tags), title=fix_text_segment(title), created_at=published, updated_at=updated, image=image_url, score=evaluator.score(url) ))
def get_articles(feed, fn): """ Parse the specified feed, gathering the latest new articles. If an article matches one that already exists, it is skipped. The minimum length of an entry is 500 characters. Anything under will be ignored. This will silently skip articles for which the full text can't be retrieved (i.e. if it returns 404). Some feeds, for whatever reason, do not include a `published` date in their entry data. In which case, it is left as an empty string. Args: | feed (Feed) -- the feed to fetch from. | fn (Callable) -- function to use an article """ # Fetch the feed data. data = feedparser.parse(feed.ext_url) # If the `bozo` value is anything # but 0, there was an error parsing (or connecting) to the feed. if data.bozo: # Some errors are ok. if not isinstance( data.bozo_exception, feedparser.CharacterEncodingOverride) and not isinstance( data.bozo_exception, feedparser.NonXMLContentType): raise data.bozo_exception for entry in data.entries: # URL for this entry. url = entry['links'][0]['href'] # Check for an existing Article. # If one exists, skip. if Article.query.filter_by(ext_url=url).count(): continue # Complete HTML content for this entry. try: entry_data, html = extractor.extract_entry_data(url) except (error.HTTPError, error.URLError, ConnectionResetError, BadStatusLine) as e: if type(e) == error.URLError or e.code == 404: # Can't reach, skip. logger.exception( 'Error extracting data for url {0}'.format(url)) continue else: # Just skip so things don't break! logger.exception( 'Error extracting data for url {0}'.format(url)) continue if entry_data is None: continue full_text = entry_data.cleaned_text # Skip over entries that are too short. if len(full_text) < 400: continue url = entry_data.canonical_link or url published = parse(entry.get('published')) if entry.get( 'published') else entry_data.publish_date updated = parse( entry.get('updated')) if entry.get('updated') else published title = entry.get('title', entry_data.title) # Secondary check for an existing Article, # by checking the title and source. existing = Article.query.filter_by(title=title).first() if existing and existing.source == feed.source: continue # Download and save the top article image. image_url = extractor.extract_image(entry_data, filename=hash(url)) fn( Article(ext_url=url, source=feed.source, feed=feed, html=html, text=fix_text_segment(full_text), authors=extractor.extract_authors(entry), tags=extractor.extract_tags(entry, known_tags=entry_data.tags), title=fix_text_segment(title), created_at=published, updated_at=updated, image=image_url, score=evaluator.score(url)))
def run(self, dumppath, use_patch): if use_patch: print('Patching out saving images to S3...') patcher = patch('argos.util.storage.save_from_url', autospec=True, return_value='https://i.imgur.com/Zf9mXlj.jpg') patcher.start() else: patcher = None print('Loading sources...') sources_map = {} with open(os.path.join(dumppath, 'sources.json'), 'r') as f: sources = json.load(f) for i, s in enumerate(sources): source = Source.query.filter(Source.name == s['name']).first() if not source: source = Source(name=s['name']) db.session.add(source) id = s['_id']['$oid'] sources_map[id] = source progress_bar(i/(len(sources) - 1) * 100) db.session.commit() print('\nLoading feeds...') feeds_map = {} with open(os.path.join(dumppath, 'feeds.json'), 'r') as f: feeds = json.load(f) for i, f in enumerate(feeds): feed = Feed.query.filter(Feed.ext_url == f['ext_url']).first() if not feed: feed = Feed(ext_url=f['ext_url']) db.session.add(feed) feed.source = sources_map[f['source']['$oid']] id = f['_id']['$oid'] feeds_map[id] = feed progress_bar(i/(len(feeds) - 1) * 100) db.session.commit() print('\nLoading articles...') with open(os.path.join(dumppath, 'articles.json'), 'r') as f: articles = json.load(f) for i, a in enumerate(articles): authors = [] for author in a['authors']: authors.append(Author.find_or_create(name=author)) existing = Article.query.filter(Article.ext_url == a['ext_url']).first() if not existing: feed = feeds_map[a['feed']['$oid']] article = Article( ext_url=a['ext_url'], source=feed.source, feed=feed, html=None, # not saved by argos.corpora text=fix_text_segment(a['text']), authors=authors, tags=[], title=fix_text_segment(a['title']), created_at=datetime.fromtimestamp(a['created_at']['$date']/1000), updated_at=datetime.fromtimestamp(a['updated_at']['$date']/1000), image=a['image'], score=evaluator.score(a['ext_url']) ) db.session.add(article) progress_bar(i/(len(articles) - 1) * 100) print('Loaded {0} sources, {1} feeds, and {2} articles.'.format(len(sources), len(feeds), len(articles))) print('Done!') if patcher is not None: patcher.stop()