From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Sun, 13 Jul 2014 10:15:40 +0200 Subject: updated specific configuration for parsing --- .../site_config/standard/post-gazette.com.txt | 50 +++++++++++----------- 1 file changed, 25 insertions(+), 25 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/post-gazette.com.txt (limited to 'inc/3rdparty/site_config/standard/post-gazette.com.txt') diff --git a/inc/3rdparty/site_config/standard/post-gazette.com.txt b/inc/3rdparty/site_config/standard/post-gazette.com.txt old mode 100644 new mode 100755 index 1ea945a0..baa9d69d --- a/inc/3rdparty/site_config/standard/post-gazette.com.txt +++ b/inc/3rdparty/site_config/standard/post-gazette.com.txt @@ -1,26 +1,26 @@ -title: //div[@class='story_headline'] -author: substring-before(substring-after(//div[@class='story_byline'],'By'),'/') -date: //div[@class='story_lastupdate'] -body: //div[@id='story'] -strip: //div[@class='story_byline'] -strip: //div[@class='story_lastupdate'] -strip: //div[@class='story_headline'] -strip: //div[@id='abuse'] -strip: //h2 -strip: //div[@class='pagenumbers_wrap'] -strip: //ul[@class='pagenumbers'] -strip: //div[starts-with(., 'To report inappropriate comments')] - -strip_id_or_class: story_share -strip_id_or_class: OUTBRAIN -strip_id_or_class: story_box_right -strip: //div[a[@href='http://www.post-gazette.com/pg/12062/1213990-42.stm']] -strip: //ul[@id='pikame']/li[position()>1] - -prune: no -tidy: no - -single_page_link: //a[contains(@href, '?p=0')] - -test_url: http://www.post-gazette.com/stories/sports/penguins/pens-crosby-expects-to-return-thursday-226648/ +title: //div[@class='story_headline'] +author: substring-before(substring-after(//div[@class='story_byline'],'By'),'/') +date: //div[@class='story_lastupdate'] +body: //div[@id='story'] +strip: //div[@class='story_byline'] +strip: //div[@class='story_lastupdate'] +strip: //div[@class='story_headline'] +strip: //div[@id='abuse'] +strip: //h2 +strip: //div[@class='pagenumbers_wrap'] +strip: //ul[@class='pagenumbers'] +strip: //div[starts-with(., 'To report inappropriate comments')] + +strip_id_or_class: story_share +strip_id_or_class: OUTBRAIN +strip_id_or_class: story_box_right +strip: //div[a[@href='http://www.post-gazette.com/pg/12062/1213990-42.stm']] +strip: //ul[@id='pikame']/li[position()>1] + +prune: no +tidy: no + +single_page_link: //a[contains(@href, '?p=0')] + +test_url: http://www.post-gazette.com/stories/sports/penguins/pens-crosby-expects-to-return-thursday-226648/ test_url: http://www.post-gazette.com/stories/sports/pirates/pirates-fork-over-changes-for-fans-at-pnc-park-629789 \ No newline at end of file -- cgit v1.2.3