From aa7b506bac610d5dfb2381d3bc0ecfa66c16c691 Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Tue, 26 Jul 2016 18:45:32 +0200 Subject: Update futurezone.at and heise.de (#188) * futurezone.at: Update site config. * heise.de: Fix extraction of author and single page link. --- futurezone.at.txt | 32 +++++++++++++++++++++----------- heise.de.txt | 5 +++-- 2 files changed, 24 insertions(+), 13 deletions(-) diff --git a/futurezone.at.txt b/futurezone.at.txt index 808c1f1..d975544 100644 --- a/futurezone.at.txt +++ b/futurezone.at.txt @@ -1,11 +1,21 @@ -date: //span[@class='date'] -strip: //div[@class='postsidebar'] -body: //div[@class='singlepost'] -title: //div[@class='singlepost']/h1 -move_into(//div[@class='singlepost']): //div[@class='info'] -strip: //div[@class='gallery'] -strip: //div[@class='biggallery'] -strip: //ul[@class='social'] -strip: //ul[@class='social_mail'] - -test_url: http://futurezone.at/future/5502-erste-galileo-satelliten-starten-ins-all.php \ No newline at end of file +single_page_link: //a[@class='print'] + +parser: html5php +tidy: no +prune: no + +body: //article +strip: //section[@class='heroheader'] +strip: //section[contains(concat(' ', normalize-space(@class), ' '), ' tags ')] +strip: //div[@class='gallery-title-container'] +strip: //figcaption/span +strip: //div[contains(concat(' ', normalize-space(@class), ' '), ' controls ')] +strip: //i[@class='icon zoom'] + +author: normalize-space(substring-after(//article/p[1], 'Autor: ')) +strip: //article/p[1] + +date: substring-after(//div[contains(concat(' ', normalize-space(@class), ' '), ' editedby ')]/span, 'Erstellt am ') +strip: //div[contains(concat(' ', normalize-space(@class), ' '), ' editedby ')] + +test_url: http://futurezone.at/produkte/sony-xperia-x-im-test-mittelklasse-zum-premium-preis/209.178.838 diff --git a/heise.de.txt b/heise.de.txt index 9a472ae..71faa87 100755 --- a/heise.de.txt +++ b/heise.de.txt @@ -5,7 +5,7 @@ prune: no title: //article/h1 | //h1 date: //p[@class='news_datum'] -author: //h4[@class='author'] +author: //span[@class='author'] body: //article | //div[@class='meldung_wrapper'] @@ -46,7 +46,7 @@ replace_string(): replace_string():
-single_page_link: //footer//a[contains(@href, '?view=print')] +single_page_link: //a[contains(@href, '?view=print')] next_page_link: //a[@class='next'] next_page_link: //a[@title='vor'] @@ -54,3 +54,4 @@ next_page_link: //a[@title='vor'] test_url: http://www.heise.de/open/artikel/Die-Neuerungen-von-Linux-3-15-2196231.html test_url: http://m.heise.de/open/artikel/Die-Neuerungen-von-Linux-3-15-2196231.html test_url: http://www.heise.de/newsticker/meldung/Ueberwachungstechnik-Die-globale-Handy-Standortueberwachung-2301494.html +test_url: http://www.heise.de/newsticker/meldung/Bodenradar-fuer-selbstfahrende-Autos-horcht-unter-die-Strasse-3273941.html -- cgit v1.3.1