From f6d8d5cc8b5a1156faa7859c3eb171c9a710a190 Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Thu, 31 Dec 2015 08:05:20 +0100 Subject: Update site config for {news.,}ORF.at --- .orf.at.txt | 25 +++++++++++++++++++++++++ news.orf.at.txt | 11 ----------- orf.at.txt | 32 +++++++++++++++++++++++--------- 3 files changed, 48 insertions(+), 20 deletions(-) create mode 100644 .orf.at.txt delete mode 100644 news.orf.at.txt diff --git a/.orf.at.txt b/.orf.at.txt new file mode 100644 index 0000000..ad3d7ce --- /dev/null +++ b/.orf.at.txt @@ -0,0 +1,25 @@ +single_page_link: //div[@id='ss-storyText']//p[@class='readMore']/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')] +date: substring-after(//div[@class='storyMeta socialshare']//p[@class='date'],'Publiziert am') +strip: //p[@class='date'] + +strip: //p[@class='credit'] +strip: //p[@class='toplink'] +strip: //div[@id="ss-storyText"]/h1 +strip: //div[@class='socialButtons'] +strip: //div[@class='storyMeta socialshare'] +strip: //div[@class='socialShareWrapper'] +strip: //div[@id='socialshareprivacy'] + +prune: no +tidy: no + +test_url: http://orf.at/stories/2317355/ +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/ diff --git a/news.orf.at.txt b/news.orf.at.txt deleted file mode 100644 index 3b1d3cc..0000000 --- a/news.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://news.orf.at/stories/2084731/ \ No newline at end of file diff --git a/orf.at.txt b/orf.at.txt index fb4f218..ad3d7ce 100644 --- a/orf.at.txt +++ b/orf.at.txt @@ -1,11 +1,25 @@ -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') +single_page_link: //div[@id='ss-storyText']//p[@class='readMore']/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')] +date: substring-after(//div[@class='storyMeta socialshare']//p[@class='date'],'Publiziert am') +strip: //p[@class='date'] + +strip: //p[@class='credit'] strip: //p[@class='toplink'] +strip: //div[@id="ss-storyText"]/h1 +strip: //div[@class='socialButtons'] +strip: //div[@class='storyMeta socialshare'] +strip: //div[@class='socialShareWrapper'] +strip: //div[@id='socialshareprivacy'] + +prune: no +tidy: no -test_url: http://orf.at/stories/2084731/ \ No newline at end of file +test_url: http://orf.at/stories/2317355/ +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/ -- cgit v1.3.1