From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= <nicolas@loeuillet.org>
Date: Sun, 13 Jul 2014 10:15:40 +0200
Subject: updated specific configuration for parsing

---
 inc/3rdparty/site_config/standard/npr.org.txt | 66 ++++++++++++++-------------
 1 file changed, 34 insertions(+), 32 deletions(-)
 mode change 100644 => 100755 inc/3rdparty/site_config/standard/npr.org.txt

(limited to 'inc/3rdparty/site_config/standard/npr.org.txt')

diff --git a/inc/3rdparty/site_config/standard/npr.org.txt b/inc/3rdparty/site_config/standard/npr.org.txt
old mode 100644
new mode 100755
index afab0eb3..acd73e48
--- a/inc/3rdparty/site_config/standard/npr.org.txt
+++ b/inc/3rdparty/site_config/standard/npr.org.txt
@@ -1,32 +1,34 @@
-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
+title: //div[contains(@class, 'storytitle')]//h1
+author: //p[@class="byline"]/span
+body: //div[@id='primaryaudio']//*[@class='duration' or @class='download' or contains(@class, 'photo')] | //div[@id='storytext' or @id='supplementarycontent' or contains(@class, 'transcript')]
+date: //meta[@name="date"]/@content
+
+strip_id_or_class: enlarge_measure
+strip_id_or_class: enlarge_html
+strip: //a[contains(@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>
+replace_string(<div class="transcript storytext">): <div class="transcript storytext"><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
+test_url: http://www.npr.org/templates/story/story.php?storyId=229103221
\ No newline at end of file
-- 
cgit v1.2.3