diff options
| -rw-r--r-- | cbsnews.com.txt | 3 | ||||
| -rw-r--r-- | huffingtonpost.com.txt | 7 | ||||
| -rw-r--r-- | mashable.com.txt | 13 |
3 files changed, 18 insertions, 5 deletions
diff --git a/cbsnews.com.txt b/cbsnews.com.txt index 780bc15..04d2023 100644 --- a/cbsnews.com.txt +++ b/cbsnews.com.txt @@ -8,7 +8,8 @@ body: //div[@id='storyMediaBox'] | //div[contains(@class, 'storyText')] strip: //div[@class="scrollingArrows"] strip: //div[@class="timeLine"] strip: //dl[@class="storyBlogByline"] +strip: //span[@class='image-credit'] prune: no -test_url: http://www.cbsnews.com/8301-201_162-57366361/rescued-americans-dad-proud-of-the-u.s/
\ No newline at end of file +test_url: http://www.cbsnews.com/8301-201_162-57366361/rescued-americans-dad-proud-of-the-u.s/ diff --git a/huffingtonpost.com.txt b/huffingtonpost.com.txt index fb990a7..d4618c1 100644 --- a/huffingtonpost.com.txt +++ b/huffingtonpost.com.txt @@ -3,14 +3,19 @@ body: //div[img[starts-with(@id, 'img_caption')]] | //div[@class="big_photo"] | date: //meta[@name="publish_date"]/@content author: //a[@rel="author"] author: //meta[@name="author"]/@content + prune: no tidy: no + strip: //footer strip_id_or_class: ps-slideshow strip_id_or_class: fs-slideshow strip: //p[contains(., 'Related on HuffPost:')] +strip_id_or_class: contribute-story +strip_id_or_class: promo_holder + # end early replace_string(<div class="sbm-main): </body></html><div class="not-interested test_url: http://www.huffingtonpost.com/mitch-moxley/tracking-beijings-boom-th_b_1209828.html -test_url: http://www.huffingtonpost.com/2012/09/11/president-obama-iphone-throwdown_n_1873826.html
\ No newline at end of file +test_url: http://www.huffingtonpost.com/2012/09/11/president-obama-iphone-throwdown_n_1873826.html diff --git a/mashable.com.txt b/mashable.com.txt index 2c5a14a..b6efb6c 100644 --- a/mashable.com.txt +++ b/mashable.com.txt @@ -1,4 +1,11 @@ -title: //header[@class='entry-title']/h1 -body: //div[@class='description'] +title: //h1[@class='title'] +author: substring-after(//span[@class='author_name'], 'By ') +date: //time + +body: //article strip: //div[@class='ytm-gallery-box'] -test_url: http://mashable.com/2011/12/05/india-wants-google-and-facebook-to-censor-user-content/
\ No newline at end of file +strip: //div[contains(@class, 'adsense')] +strip: //aside[contains(@class, 'social')] +strip_id_or_class: article-topics + +test_url: http://mashable.com/2013/05/24/myspace-architects-rebuilding-a-brand/ |