diff options
| -rw-r--r-- | mobile.nytimes.com.txt | 4 | ||||
| -rw-r--r-- | nytimes.com.txt | 4 |
2 files changed, 7 insertions, 1 deletions
diff --git a/mobile.nytimes.com.txt b/mobile.nytimes.com.txt new file mode 100644 index 0000000..c60252e --- /dev/null +++ b/mobile.nytimes.com.txt @@ -0,0 +1,4 @@ +title: //h1[contains(@class, 'headline')] +body: //article[contains(@class, 'full-art')] +strip_id_or_class: image-credit +test_url: http://mobile.nytimes.com/2014/06/19/opinion/gail-collins-romney-and-the-2016-contenders-huddle.html
\ No newline at end of file diff --git a/nytimes.com.txt b/nytimes.com.txt index 5712b64..23c9ad1 100644 --- a/nytimes.com.txt +++ b/nytimes.com.txt @@ -33,6 +33,7 @@ replace_string: </div> date: substring-after(//*[contains(@class, 'dateline')], 'Published:') single_page_link: //link[contains(@href, 'pagewanted=all')] +single_page_link: //link[@rel='alternate' and contains(@href, 'mobile.nytimes.com')]/@href single_page_link: concat(substring-before(//div[@id='pageLinks']//a[contains(@href, 'pagewanted=')]/@href, 'pagewanted='), 'pagewanted=all') #single_page_link: //a[contains(@href, 'pagewanted=all') and not(contains(@href, 'login'))] @@ -44,4 +45,5 @@ test_url: http://www.nytimes.com/2011/07/24/books/review/an-academic-authors-uni test_url: http://www.nytimes.com/2012/06/10/arts/television/the-newsroom-aaron-sorkins-return-to-tv.html test_url: http://www.nytimes.com/2013/03/25/world/middleeast/israeli-military-responds-after-patrols-come-under-fire-from-syria.html test_url: http://www.nytimes.com/2013/08/15/nyregion/when-the-new-york-city-subway-ran-without-rails.html -test_url: http://www.nytimes.com/2004/02/29/weekinreview/correspondence-class-consciousness-china-s-wealthy-live-creed-hobbes-darwin-meet.html
\ No newline at end of file +test_url: http://www.nytimes.com/2004/02/29/weekinreview/correspondence-class-consciousness-china-s-wealthy-live-creed-hobbes-darwin-meet.html +test_url: http://www.nytimes.com/2014/06/19/opinion/gail-collins-romney-and-the-2016-contenders-huddle.html
\ No newline at end of file |