diff options
| -rw-r--r-- | .time.com.txt | 31 | ||||
| -rw-r--r-- | healthland.time.com.txt | 10 | ||||
| -rw-r--r-- | time.com.txt | 14 |
3 files changed, 23 insertions, 32 deletions
diff --git a/.time.com.txt b/.time.com.txt index f5d5c14..40ee26f 100644 --- a/.time.com.txt +++ b/.time.com.txt @@ -1,11 +1,26 @@ +date: //meta[@name='date']/@content +author: //meta[@name='byline']/@content date: //span[@class = 'date'] -body: //div[starts-with(@id, 'post-')]//div[@class = 'entry-content'] -strip://div[@class='more-ways'] -strip://div[@id = 'stayConnected'] -strip://p[child::a[@rel = 'bookmark']] -strip://p[starts-with(string(.),'(MORE:')] -strip://p[starts-with(string(.),'(PHOTOS:')] -move_into(//p[../@class = 'entry-content'][position() = last()])://div[@id = 'featbox'] +title: //div[@id='print']//h1 +title: //h1[@class="entry-title"] +body: //article//div[contains(@class,'entry-content')] +strip: //div[@class='more-ways'] +strip: //div[@id = 'stayConnected'] +strip: //p[child::a[@rel = 'bookmark']] +strip: //p[starts-with(string(.),'(MORE:')] +strip: //p[starts-with(string(.),'(PHOTOS:')] +strip: //aside +#move_into(//p[../@class = 'entry-content'][position() = last()])://div[@id = 'featbox'] prune: no -test_url: http://healthland.time.com/2011/07/24/amy-winehouse-and-the-pain-of-addiction/?preview=true&preview_id=39210&preview_nonce=0777d4e408
\ No newline at end of file +replace_string(<noscript>): <div> +replace_string(</noscript>): </div> + + +strip: //div[@id='print']//div[contains(@class, 'thumbnail')] + +single_page_link: //footer//a[contains(@href, '/printout/')] + +test_url: http://content.time.com/time/specials/packages/article/0,28804,2094921_2094923_2094924,00.html +test_url: http://healthland.time.com/2011/07/24/amy-winehouse-and-the-pain-of-addiction/ +test_url: http://nation.time.com/2013/09/17/navy-yard-shooter-had-been-treated-for-mental-health-problems/
\ No newline at end of file diff --git a/healthland.time.com.txt b/healthland.time.com.txt deleted file mode 100644 index 0542097..0000000 --- a/healthland.time.com.txt +++ /dev/null @@ -1,10 +0,0 @@ -date: //span[@class = 'date'] -body: //div[@class = 'entry-content'] -strip://div[@class='more-ways'] -strip://div[@id = 'stayConnected'] -strip://p[child::a[@rel = 'bookmark']] -strip://p[starts-with(string(.),'(MORE:')] -strip://p[starts-with(string(.),'(PHOTOS:')] -move_into(//p[../@class = 'entry-content'][position() = last()])://div[@id = 'featbox'] - -test_url: http://healthland.time.com/2011/07/24/amy-winehouse-and-the-pain-of-addiction/?preview=true&preview_id=39210&preview_nonce=0777d4e408
\ No newline at end of file diff --git a/time.com.txt b/time.com.txt deleted file mode 100644 index 0519d75..0000000 --- a/time.com.txt +++ /dev/null @@ -1,14 +0,0 @@ -# 2011-10-25 - carlo@... - Initial setup. - -single_page_link: //li[@class='print']/a/@href - -title: //h1 -author: //meta[@name="byline"]/@content -date: //meta[@name="date"]/@content - -strip: //span[@class="see"] -strip: //div[@class="byline"] -strip: //div[@id="date2"] -strip: //h1 - -test_url: http://www.time.com/time/specials/packages/article/0,28804,2094921_2094923_2094924,00.html
\ No newline at end of file |