From 2308312087fa37460ae4d4898d78321a7f6ae651 Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Wed, 18 May 2016 18:59:15 +0200 Subject: ORF.at: Update siteconfig. (#158) --- .orf.at.txt | 11 +++++++++-- orf.at.txt | 11 +++++++++-- science.orf.at.txt | 11 ----------- sport.orf.at.txt | 11 ----------- 4 files changed, 18 insertions(+), 26 deletions(-) delete mode 100644 science.orf.at.txt delete mode 100644 sport.orf.at.txt diff --git a/.orf.at.txt b/.orf.at.txt index ad3d7ce..dd0f20b 100644 --- a/.orf.at.txt +++ b/.orf.at.txt @@ -1,9 +1,10 @@ single_page_link: //div[@id='ss-storyText']//p[@class='readMore']/a +single_page_link: //div[@id='ss-storyText']//p[contains(., 'Mehr dazu')]/a title: substring-before(//title,' - ') body: //div[@id="ss-storyText"] -author: substring-before(//div[@id="ss-storyText"]//p[contains(text(), ', ORF.at')], ', ORF.at') -strip: //div[@id="ss-storyText"]//p[contains(text(), ', ORF.at')] +author: substring-before(//div[@id="ss-storyText"]//p[contains(text(), ', ORF')], ', ORF') +strip: //div[@id="ss-storyText"]//p[contains(text(), ', ORF')] date: substring-after(//div[@class='storyMeta socialshare']//p[@class='date'],'Publiziert am') strip: //p[@class='date'] @@ -14,6 +15,8 @@ strip: //div[@class='socialButtons'] strip: //div[@class='storyMeta socialshare'] strip: //div[@class='socialShareWrapper'] strip: //div[@id='socialshareprivacy'] +strip: //div[@class='storyMeta'] +strip: //div[@class='remote'] prune: no tidy: no @@ -23,3 +26,7 @@ test_url: http://orf.at/stories/2084731/ test_url: http://orf.at/stories/2317313/2317311/ test_url: http://wien.orf.at/news/stories/2746414/ test_url: http://ooe.orf.at/news/stories/2750613/ +test_url: http://science.orf.at/stories/2774991/ +test_url: http://orf.at/stories/2339962/ +test_url: http://orf.at/stories/2339958/ +test_url: http://help.orf.at/stories/1770242/ diff --git a/orf.at.txt b/orf.at.txt index ad3d7ce..dd0f20b 100644 --- a/orf.at.txt +++ b/orf.at.txt @@ -1,9 +1,10 @@ single_page_link: //div[@id='ss-storyText']//p[@class='readMore']/a +single_page_link: //div[@id='ss-storyText']//p[contains(., 'Mehr dazu')]/a title: substring-before(//title,' - ') body: //div[@id="ss-storyText"] -author: substring-before(//div[@id="ss-storyText"]//p[contains(text(), ', ORF.at')], ', ORF.at') -strip: //div[@id="ss-storyText"]//p[contains(text(), ', ORF.at')] +author: substring-before(//div[@id="ss-storyText"]//p[contains(text(), ', ORF')], ', ORF') +strip: //div[@id="ss-storyText"]//p[contains(text(), ', ORF')] date: substring-after(//div[@class='storyMeta socialshare']//p[@class='date'],'Publiziert am') strip: //p[@class='date'] @@ -14,6 +15,8 @@ strip: //div[@class='socialButtons'] strip: //div[@class='storyMeta socialshare'] strip: //div[@class='socialShareWrapper'] strip: //div[@id='socialshareprivacy'] +strip: //div[@class='storyMeta'] +strip: //div[@class='remote'] prune: no tidy: no @@ -23,3 +26,7 @@ test_url: http://orf.at/stories/2084731/ test_url: http://orf.at/stories/2317313/2317311/ test_url: http://wien.orf.at/news/stories/2746414/ test_url: http://ooe.orf.at/news/stories/2750613/ +test_url: http://science.orf.at/stories/2774991/ +test_url: http://orf.at/stories/2339962/ +test_url: http://orf.at/stories/2339958/ +test_url: http://help.orf.at/stories/1770242/ diff --git a/science.orf.at.txt b/science.orf.at.txt deleted file mode 100644 index c4b2183..0000000 --- a/science.orf.at.txt +++ /dev/null @@ -1,11 +0,0 @@ -body: //div[@class="storybox"] -title: //div[@class="storybox"]//h1 -strip: //p[@class='metaline'] -date: substring-after(//*[@class='time'],'Erstellt am') -strip: //div[@class='fact'] -strip: //p[@class='backlink'] -strip: //div[@class='mailto'] -strip: //div[@id='forumDisclaimer'] -strip: //div[@class='forum'] - -test_url: http://science.orf.at/stories/1700900/ \ No newline at end of file diff --git a/sport.orf.at.txt b/sport.orf.at.txt deleted file mode 100644 index f0be85c..0000000 --- a/sport.orf.at.txt +++ /dev/null @@ -1,11 +0,0 @@ -single_page_link: //div[@id='content']//p[@class='readMore']/a - -title: //div[@class='hidden offscreen']/h2 -body: //div[@id="storyText"] -move_into(//div[@id='storyText']): //div[@class='fact'] -strip: //small[@class='credit'] -strip: //small[@class='caption'] -date: substring-after(//div[@id='storyMeta']//p[@class='date'],'Publiziert am') -strip: //p[@class='toplink'] - -test_url: http://sport.orf.at/stories/2084851/ \ No newline at end of file -- cgit v1.3.1