Q_i^dZddlZddlmZddlmZddlmZmZddlm Z m Z Gdd eZ y) z!Google Trends RSS feed collector.N)List)datetime) BaseCollector TrendItem)MAX_ITEMS_PER_SOURCE TRENDS_HOURScHeZdZdZfdZdedefdZdedeefdZ xZ S)GoogleTrendsCollectorz5Collects trending topics from Google Trends RSS feed.c$t|dy)N google_trends)super__init__)self __class__s B/home/sietch6/trending-topics-pipeline/collectors/google_trends.pyrzGoogleTrendsCollector.__init__s )regionreturncd|dtS)z$Build the RSS feed URL for a region.z+https://trends.google.com/trending/rss?geo=z&hours=)r )rrs r _get_feed_urlz#GoogleTrendsCollector._get_feed_urlsZZrc|j|}tj|}|jr$|jst d|j g}t|jdtdD]\}}t|dd}g}t|dr_|j} t| tr | D cgc]} | jdd}} n#t| tr| jddg}|j!t#|j$d ||t'j(j+|jd |||jd dd  |Scc} w)z Collect trending topics from Google Trends RSS. Args: region: Two-letter region code (US, CA) Returns: List of TrendItem objects zFailed to parse feed: Nr)startht_approx_traffic ht_news_itemht_news_item_titler link published)traffic news_itemsr) topic_namesourcerankr timestampurlmetadata)r feedparserparsebozoentries ValueErrorbozo_exception enumeratergetattrhasattrr isinstancelistgetdictappendrtitlerutcnow isoformat) rrr&feeditemsr$entryr r! news_datans rcollectzGoogleTrendsCollector.collectsS  ($ 99T\\5d6I6I5JKL L$T\\2G3G%HPQR KD%e%8"=GJun-!.. i.KT!Ua!%%( __classcell__)rs@rr r s8?*[C[C[.c.d9o.rr ) rBr(typingrrbaserrconfigrr r rrrIs#'*58M8r