diff options
author | Nicolas LÅ“uillet <nicolas.loeuillet@smile.fr> | 2014-10-10 13:33:54 +0200 |
---|---|---|
committer | Nicolas LÅ“uillet <nicolas.loeuillet@smile.fr> | 2014-10-10 13:33:54 +0200 |
commit | 44d35257e805856b4913c63fcbed3c0acb64bae8 (patch) | |
tree | 11e9d276c34b1b287706cb61182bdc71729661e2 /inc/3rdparty/site_config/standard/smithsonianmag.com.txt | |
parent | af8292c1de1886cd975d79f0f42df40e0bd1c5bd (diff) | |
parent | cf8a5e1eedbed484dbcb1ddc9f7a13fc19b7a27b (diff) | |
download | wallabag-1.8.0.tar.gz wallabag-1.8.0.tar.zst wallabag-1.8.0.zip |
Merge branch 'dev'1.8.0
Diffstat (limited to 'inc/3rdparty/site_config/standard/smithsonianmag.com.txt')
-rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/smithsonianmag.com.txt | 36 |
1 files changed, 18 insertions, 18 deletions
diff --git a/inc/3rdparty/site_config/standard/smithsonianmag.com.txt b/inc/3rdparty/site_config/standard/smithsonianmag.com.txt index 10a3f717..3e8fee95 100644..100755 --- a/inc/3rdparty/site_config/standard/smithsonianmag.com.txt +++ b/inc/3rdparty/site_config/standard/smithsonianmag.com.txt | |||
@@ -1,20 +1,20 @@ | |||
1 | # meta data | 1 | # meta data |
2 | title://h1[@id = 'articleTitle'] | 2 | title://h1[@id = 'articleTitle'] |
3 | author:substring-after(//ul[@id = 'byLine']/li[1],'By ') | 3 | author:substring-after(//ul[@id = 'byLine']/li[1],'By ') |
4 | date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',') | 4 | date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',') |
5 | body://div[@id = 'article-body'] | 5 | body://div[@id = 'article-body'] |
6 | 6 | ||
7 | # full content | 7 | # full content |
8 | single_page_link://td/li[@class = 'article-singlepage']/a | 8 | single_page_link://td/li[@class = 'article-singlepage']/a |
9 | 9 | ||
10 | # caption clean up | 10 | # caption clean up |
11 | wrap_in(i)://span[@class='articleImageCaptionwide'] | 11 | wrap_in(i)://span[@class='articleImageCaptionwide'] |
12 | move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p | 12 | move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p |
13 | 13 | ||
14 | 14 | ||
15 | # clean up | 15 | # clean up |
16 | strip://p[@id = 'articlePaginationWrapper'] | 16 | strip://p[@id = 'articlePaginationWrapper'] |
17 | strip://ul[contains(@class, 'cat-breadcrumb')] | 17 | strip://ul[contains(@class, 'cat-breadcrumb')] |
18 | strip://div [@class= 'viewMorePhotos'] | 18 | strip://div [@class= 'viewMorePhotos'] |
19 | 19 | ||
20 | test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html \ No newline at end of file | 20 | test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html \ No newline at end of file |