|
177
|
0 |
0 |
if (my $sel = {www.5ch.com.tw => {content_text => 'div.text-edit', dateline => 'div.more-about div.date', headline => 'h3.m-ti', journalist => 'div.more-about div.reporter'}, www.cw.com.tw => {content_text => 'div.article__content', dateline => 'div.article__detail > time', headline => 'div.article__head h1', journalist => 'div.author--item > a'}, www.enewstw.com => {content_text => 'td.new_t p', dateline => 'td.blog_title tr:nth-child(2) > td.blog', headline => 'td.blog_title > strong', journalist => 'td.blog_title tr:nth-child(1) > td.blog'}, www.eventsinfocus.org => {content_text => 'div.content article div.clearfix.text-formatted', dateline => 'div.content time.datetime', headline => 'h1.title > span', journalist => 'div.content div.node__content div.clearfix.text-formatted > p:nth-child(1)'}, www.storm.mg => {content_text => 'div#article_inner_wrapper > article:nth-child(1)', dateline => 'span#info_time', headline => 'h1#article_title', journalist => '#article_info_wrapper #author_block a.link_author > span.info_author'}, www.taiwannews.com.tw => {content_text => 'article.article', dateline => 'div.article-date', headline => 'h1.article-title', journalist => 'div.article-author'}}->{$host}) { }
|
|
|
0 |
0 |
elsif (my $extractor_class = {ctee.com.tw => 'NewsExtractor::SiteSpecificExtractor::ctee_com_tw', estate.ltn.com.tw => 'NewsExtractor::SiteSpecificExtractor::estate_ltn_com_tw', focustaiwan.tw => 'NewsExtractor::SiteSpecificExtractor::focustaiwan_tw', health.ettoday.net => 'NewsExtractor::SiteSpecificExtractor::ETtoday', hk.crntt.com => 'NewsExtractor::SiteSpecificExtractor::hk_crntt_com', hk.on.cc => 'NewsExtractor::SiteSpecificExtractor::hk_on_cc', house.ettoday.net => 'NewsExtractor::SiteSpecificExtractor::ETtoday', house.udn.com => 'NewsExtractor::SiteSpecificExtractor::UDN', m.news.cctv.com => 'NewsExtractor::SiteSpecificExtractor::m_news_cctv_com', money.udn.com => 'NewsExtractor::SiteSpecificExtractor::UDN', new.ctv.com.tw => 'NewsExtractor::SiteSpecificExtractor::new_ctv_com_tw', newnet.tw => 'NewsExtractor::SiteSpecificExtractor::newnet_tw', news.cctv.com => 'NewsExtractor::SiteSpecificExtractor::news_cctv_com', news.cts.com.tw => 'NewsExtractor::SiteSpecificExtractor::news_cts_com_tw', news.ebc.net.tw => 'NewsExtractor::SiteSpecificExtractor::news_ebc_net_tw', news.pts.org.tw => 'NewsExtractor::SiteSpecificExtractor::news_pts_org_tw', news.tnn.tw => 'NewsExtractor::SiteSpecificExtractor::news_tnn_tw', news.ttv.com.tw => 'NewsExtractor::SiteSpecificExtractor::news_ttv_com_tw', news.tvbs.com.tw => 'NewsExtractor::SiteSpecificExtractor::news_tvbs_com_tw', newtalk.tw => 'NewsExtractor::SiteSpecificExtractor::newtalk_tw', opinion.udn.com => 'NewsExtractor::SiteSpecificExtractor::UDN', star.ettoday.net => 'NewsExtractor::SiteSpecificExtractor::ETtoday', stars.udn.com => 'NewsExtractor::SiteSpecificExtractor::UDN', talk.ltn.com.tw => 'NewsExtractor::SiteSpecificExtractor::talk_ltn_com_tw', turnnewsapp.com => 'NewsExtractor::SiteSpecificExtractor::turnnewsapp_com', udn.com => 'NewsExtractor::SiteSpecificExtractor::UDN', video.udn.com => 'NewsExtractor::JSONLDExtractor', www.aljazeera.com => 'NewsExtractor::SiteSpecificExtractor::www_aljazeera_com', www.allnews.tw => 'NewsExtractor::SiteSpecificExtractor::www_allnews_tw', www.bbc.com => 'NewsExtractor::SiteSpecificExtractor::www_bbc_com', www.bcc.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_bcc_com_tw', www.chinatimes.com => 'NewsExtractor::SiteSpecificExtractor::ChinaTimes', www.ctwant.com => 'NewsExtractor::JSONLDExtractor', www.digitimes.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_digitimes_com_tw', www.epochtimes.com => 'NewsExtractor::SiteSpecificExtractor::www_epochtimes_com', www.ettoday.net => 'NewsExtractor::SiteSpecificExtractor::ETtoday', www.fountmedia.io => 'NewsExtractor::SiteSpecificExtractor::www_fountmedia_io', www.hkcna.hk => 'NewsExtractor::SiteSpecificExtractor::www_hkcna_hk', www.hkcnews.com => 'NewsExtractor::SiteSpecificExtractor::www_hkcnews_com', www.idn.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_idn_com_tw', www.ksnews.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_ksnews_com_tw', www.mdnkids.com => 'NewsExtractor::SiteSpecificExtractor::www_mdnkids_com', www.nownews.com => 'NewsExtractor::SiteSpecificExtractor::www_nownews_com', www.ntdtv.com => 'NewsExtractor::SiteSpecificExtractor::www_ntdtv_com', www.penghutimes.com => 'NewsExtractor::SiteSpecificExtractor::www_penghutimes_com', www.peopo.org => 'NewsExtractor::SiteSpecificExtractor::www_peopo_org', www.rti.org.tw => 'NewsExtractor::SiteSpecificExtractor::www_rti_org_tw', www.rvn.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_rvn_com_tw', www.setn.com => 'NewsExtractor::SiteSpecificExtractor::www_setn_com', www.taipeitimes.com => 'NewsExtractor::SiteSpecificExtractor::www_taipeitimes_com', www.thestandnews.com => 'NewsExtractor::SiteSpecificExtractor::www_thestandnews_com', www.ttv.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_ttv_com_tw', www.twreporter.org => 'NewsExtractor::SiteSpecificExtractor::www_twreporter_org', www.upmedia.mg => 'NewsExtractor::SiteSpecificExtractor::www_upmedia_mg', www.ustv.com.tw => 'NewsExtractor::SiteSpecificExtractor::www_ustv_com_tw', www.xinhuanet.com => 'NewsExtractor::SiteSpecificExtractor::www_xinhuanet_com', yimedia.com.tw => 'NewsExtractor::SiteSpecificExtractor::yimedia_com_tw'}->{$host}) { }
|