def tag_kword_articles(credentials, items): jc = JsonConfig(".kwords") config = jc.read() kwords = config["keywords"] total_uids = [] for kword, kword_tag in kwords: uids = isolate_keyword_articles(items, kword.lower(), kword_tag) add_tags(credentials, uids, [kword_tag]) total_uids.extend(uids) return len(total_uids)
config_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), '.creds') config = json.load(open(config_path)) for key, entry in retrieve(config, verbose=True).iteritems(): if 'podcast' in entry.get('tags', {}): continue try: url = entry.get('resolved_url', entry.get('given_url', None)) if url is None: continue for pattern in PATTERNS: if not pattern.URL_PATT.search(url): continue page = urllib2.urlopen(url).read() m = pattern.M_PAT.search(page) if not m: continue fn = download_base_file(m.group(1)) m2 = pattern.T_PAT.search(page) if m2: title = m2.group(1) else: title = m.group(1) podcast.add_episode(title, fn, '') add_tags(config, [key], 'podcast') except: None podcast.write_to_file()
def tag_long_articles(credentials, items): uids = isolate_long_articles(items) add_tags(credentials, uids, ["long"]) return len(uids)