Initial commit

1 files changed, 32 insertions, 0 deletions

diff --git a/npr.org.txt b/npr.org.txt
new file mode 100644
index 0000000..5dc7e74
--- /dev/null
+++ b/npr.org.txt
@@ -0,0 +1,32 @@
+title: //div[contains(@class, 'storytitle')]//h1
+author: //p[@class="byline"]/span
+body: //div[@id='storyspan02']//*[@class='duration' or @class='download' or contains(@class, 'photo')] | //div[@id='storytext'] | //div[@class='transcript']
+date: //meta[@name="date"]/@content
+
+strip: //div[@class='enlarge_measure']
+strip: //div[@class='enlarge_html']
+strip: //a[@class='enlargeicon']
+strip: //div[contains(@class, 'bookedition')]
+strip: //div[@class='textsize']
+strip: //ul[@class='genres']
+strip: //span[@class='bull']
+strip_id_or_class: secondary
+strip_id_or_class: con1col
+strip: //h3[@class='conheader']
+
+replace_string(<a name="more">&nbsp;</a>): <!-- no more -->
+replace_string(<div class="transcript">): <div class="transcript"><h2>Transcript</h2>
+
+prune: no
+strip://div[@class="ecommercepop"]
+strip://span[@class="bull"]
+strip://span[@class="purchaseLink"]
+strip://div[@class="enlarge_html"]
+strip://div[@class="enlarge_measure"]
+strip://div[@class="container con1col small"]
+strip://a[contains(@class, "enlargebtn")]
+strip://div[contains(@class, "bucketwrap internallink")]
+
+test_url: http://www.npr.org/blogs/thetwo-way/2011/07/12/137799301/sports-loses-its-escapist-gleam-in-a-summer-of-court-dates
+test_url: http://www.npr.org/2012/07/04/156190948/feeling-under-siege-catholic-leadership-shifts-right
+test_url: http://www.npr.org/2012/12/13/166480907/the-years-best-sci-fi-crosses-galaxies-and-genres
\ No newline at end of file