diff options
author | Nicolas LÅ“uillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
---|---|---|
committer | Nicolas LÅ“uillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
commit | 4e067ceabd705201a16b4c92cf4b23f3b990326c (patch) | |
tree | 939f3a8e5ff3ab9ee414a57a895d3e78e1d46ce3 /inc/3rdparty/site_config/standard/newyorker.com.txt | |
parent | 58dbe103889148def78b0fc8744d3f94c56a1561 (diff) | |
download | wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.tar.gz wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.tar.zst wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.zip |
updated specific configuration for parsing
Diffstat (limited to 'inc/3rdparty/site_config/standard/newyorker.com.txt')
-rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/newyorker.com.txt | 21 |
1 files changed, 11 insertions, 10 deletions
diff --git a/inc/3rdparty/site_config/standard/newyorker.com.txt b/inc/3rdparty/site_config/standard/newyorker.com.txt index 5624aa8c..950324a3 100644..100755 --- a/inc/3rdparty/site_config/standard/newyorker.com.txt +++ b/inc/3rdparty/site_config/standard/newyorker.com.txt | |||
@@ -1,10 +1,11 @@ | |||
1 | title: //h1[@id='articlehed'] | //h2[@id="articleintro"] | 1 | title: //h1[@id='articlehed'] | //h2[@id="articleintro"] |
2 | body: //div[@id='articletext'] | 2 | body: //div[@id='articletext'] |
3 | 3 | ||
4 | strip: //ul[@id="bc"] | //div[@id="yrail"] | //div[@class="entry-keywords"] | //div[@class="entry-categories"] | //div[@class="socialUtils"] | //div[@id="footer"] | 4 | strip: //ul[@id="bc"] | //div[@id="yrail"] | //div[@class="entry-keywords"] | //div[@class="entry-categories"] | //div[@class="socialUtils"] | //div[@id="footer"] | //div[@class="cartoon"] |
5 | 5 | ||
6 | date: //h4[@id='articleauthor']/span[@class='dd dds'] | 6 | date: //h4[@id='articleauthor']/span[@class='dd dds'] |
7 | date: //div[@id="pagebody"]/div[@class='hentry entry']/div[@class='published'] | 7 | date: //div[@id="pagebody"]/div[@class='hentry entry']/div[@class='published'] |
8 | 8 | ||
9 | single_page_link: //div[@class='paginationViewSinglePage']/a | 9 | single_page_link: //div[@class='paginationViewSinglePage']/a |
10 | test_url: http://www.newyorker.com/online/blogs/culture/2012/06/mug-shot-web-sites.html \ No newline at end of file | 10 | test_url: http://www.newyorker.com/online/blogs/culture/2012/06/mug-shot-web-sites.html |
11 | test_url: http://www.newyorker.com/reporting/2013/04/22/130422fa_fact_bilger?currentPage=all&mobify=0 \ No newline at end of file | ||