diff options
| -rw-r--r-- | technologyreview.com.txt | 7 | ||||
| -rw-r--r-- | theintercept.com.txt | 4 | ||||
| -rw-r--r-- | theweek.com.txt | 10 |
3 files changed, 15 insertions, 6 deletions
diff --git a/technologyreview.com.txt b/technologyreview.com.txt index d405eb1..9871f0e 100644 --- a/technologyreview.com.txt +++ b/technologyreview.com.txt @@ -1,5 +1,4 @@ -title: //header[@class='article-meta']/h1 -title: substring-before(//title, '|') +title: //header[@class='article-topper__title'] body: //section[contains(@class, 'body')] @@ -12,5 +11,7 @@ date: substring-after(//ul[@class='byline']/li, 'on ') author: //div[@class='view-byline']/div[@class='meta']/h2[1] date: //div[@class='view-byline']/div[@class='meta']/h2[2] +strip_id_or_class: l-article-list + next_page_link: //section[@class='pagination']/a[contains(@class, 'continue')] -test_url: http://www.technologyreview.com/news/427567/facebooks-telescope-on-human-behavior/
\ No newline at end of file +test_url: http://www.technologyreview.com/news/427567/facebooks-telescope-on-human-behavior/ diff --git a/theintercept.com.txt b/theintercept.com.txt new file mode 100644 index 0000000..51f87d4 --- /dev/null +++ b/theintercept.com.txt @@ -0,0 +1,4 @@ +title: //h1[@class="Headline"] +body: //div[@class="PostContent"] + +test_url: https://theintercept.com/2014/10/30/inside-story-matt-taibbis-departure-first-look-media/ diff --git a/theweek.com.txt b/theweek.com.txt index f98749e..4bc9532 100644 --- a/theweek.com.txt +++ b/theweek.com.txt @@ -1,4 +1,8 @@ -body: //div[@class="briefingEntry"] -prune: no +body: //div[@class="article-body underline"] -test_url: http://theweek.com/article/index/215763/insider-trading-on-capitol-hill
\ No newline at end of file +strip_id_or_class: appendedAds +strip_id_or_class: right-ad-container +strip_id_or_class: ad_wrap + +test_url: http://theweek.com/article/index/215763/insider-trading-on-capitol-hill +test_url: https://theweek.com/articles/642979/why-poor-whites-flock-donald-trump |