diff options
| -rw-r--r-- | npr.org.txt | 12 | ||||
| -rw-r--r-- | theguardian.com.txt | 8 |
2 files changed, 15 insertions, 5 deletions
diff --git a/npr.org.txt b/npr.org.txt index 5dc7e74..acd73e4 100644 --- a/npr.org.txt +++ b/npr.org.txt @@ -1,11 +1,11 @@ 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'] +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: //div[@class='enlarge_measure'] -strip: //div[@class='enlarge_html'] -strip: //a[@class='enlargeicon'] +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'] @@ -16,6 +16,7 @@ strip: //h3[@class='conheader'] replace_string(<a name="more"> </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"] @@ -29,4 +30,5 @@ 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 +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 diff --git a/theguardian.com.txt b/theguardian.com.txt new file mode 100644 index 0000000..2b50949 --- /dev/null +++ b/theguardian.com.txt @@ -0,0 +1,8 @@ +title: //div[@id='main-article-info']//h1 +body: //div[@id='article-wrapper'] +date: //li[@class='publication']//time[@pubdate] | //li[@class='publication']//data[@pubdate] +strip: //div[contains(@class, 'email-subscription')] +author: //li[@class='byline'] +prune: no +tidy: no +test_url: http://www.theguardian.com/world/2013/oct/04/nsa-gchq-attack-tor-network-encryption
\ No newline at end of file |