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 From df66ecc064bf10658701c2fe4e8f302b43c066e2 Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Thu, 31 Dec 2015 09:32:55 +0100 Subject: fm4.orf.at: Fix date, author and side notes ("facts") --- fm4.orf.at.txt | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/fm4.orf.at.txt b/fm4.orf.at.txt index 5db3e58..0c8bf57 100644 --- a/fm4.orf.at.txt +++ b/fm4.orf.at.txt @@ -1,7 +1,13 @@ author: //div[@class='authorDescription']/h2 body: //div[@id='story'] -date: substring-before(substring-after(//p[@class='date'],'Erstellt am:'), '-') title: //h1[@class='detail'] -strip: //div[@class='fact'] -test_url: http://fm4.orf.at/stories/1689156/ \ No newline at end of file +date: translate(translate(substring-before(substring-after(//p[@class='date'],'Erstellt am:'), ' Uhr'), '. ', '.'), '-', ' ') +strip: //p[@class='date'] + +strip: //p[@class='credit'] + +tidy: no +prune: no + +test_url: http://fm4.orf.at/stories/1689156/ -- cgit v1.3.1 From 798dfbdacf3ca07b2a15229186c2da5b22e81edc Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Thu, 31 Dec 2015 14:30:25 +0100 Subject: derStandard.at: Update site config. Remove since it would require additional parsing and the original feed provides a proper date and time anyway. --- derstandard.at.txt | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/derstandard.at.txt b/derstandard.at.txt index 07db352..0bfeec0 100644 --- a/derstandard.at.txt +++ b/derstandard.at.txt @@ -1,13 +1,12 @@ -title: //div[@id='artikelHeader']/h1 +title: //div[@id='content-header']/h1 author: //span[@class='author'] -date: //span[@class='date'] body: //div[@class='copytext'] strip: //ul[@class='lookupLinksArtikel'] strip: //div[@id='pageTop'] strip: //div[@id='toolbar'] strip: //div[@id='articleTools'] -strip: //div[@id='weiterlesen'] +strip: //div[@id='weiterLesen'] strip: //div[@id='communityCanvas'] -test_url: http://derstandard.at/1318726018343/Breitband-LTE-Was-bringt-die-neue-Mobilfunk-Generation \ No newline at end of file +test_url: http://derstandard.at/1318726018343/Breitband-LTE-Was-bringt-die-neue-Mobilfunk-Generation -- cgit v1.3.1 From 572603492f12114a22c5070bcc08ab66420062b6 Mon Sep 17 00:00:00 2001 From: Lukas Anzinger Date: Sun, 3 Jan 2016 14:34:47 +0100 Subject: zeit.de: Update site config. --- zeit.de.txt | 102 ++++++++++++++++++++++++++++++++++-------------------------- 1 file changed, 58 insertions(+), 44 deletions(-) diff --git a/zeit.de.txt b/zeit.de.txt index 4e00894..0b6a837 100644 --- a/zeit.de.txt +++ b/zeit.de.txt @@ -1,48 +1,62 @@ -# 2015.07.08 [Marvin Dickhaus] fixed single_page_link -# 2013.10.30 [rezor92] fixed single_page_link -# 2012-12-23 [carlo@...] fixed half-assed headlines in articles, removed inline author profiles, adjusted picture captions -# 2012-03-17 [dkless@...] Cut metadata parts in the beginning and the ends of the content block; copyright entries for pictures removed; Author fixed, not sure if old entries still valid (I left them); Weird problems with some pages addressed (see last section for removing hidden section) -# 2011-12-09 [carlo@...] Removed "related articles" block -# 2011-08-23 [carlo@...] changed single page link to use print version: page works better, less ambiguity. Related cleanups and simplifications. -# 2011-08-20 [carlo@...] added author, fixed date +tidy: no +prune: no + +# Figures are wrapped in a noscript tag which is itself wrapped in a +# conditional comment. Feed readers will fail to parse this correctly +# so get rid of the noscript tag altogether. +replace_string(