diff options
Diffstat (limited to 'inc/3rdparty/site_config/standard/theverge.com.txt')
-rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/theverge.com.txt | 79 |
1 files changed, 48 insertions, 31 deletions
diff --git a/inc/3rdparty/site_config/standard/theverge.com.txt b/inc/3rdparty/site_config/standard/theverge.com.txt index 11c5c153..1e1ce58f 100644..100755 --- a/inc/3rdparty/site_config/standard/theverge.com.txt +++ b/inc/3rdparty/site_config/standard/theverge.com.txt | |||
@@ -1,31 +1,48 @@ | |||
1 | title: //h1[contains(@class, "headline")] | 1 | author: //p[contains(@class, "byline")]/a[contains(@class, "author")] |
2 | 2 | ||
3 | author: //p[contains(@class, "byline")]/a[contains(@class, "author")] | 3 | date: //span[contains(@class, "publish-date")]/time[@pubdate]/@datetime |
4 | 4 | ||
5 | date: substring-after(normalize-space(//p[contains(@class, "byline")]/span[contains(@class, "publish-date")]), "on ") | 5 | body: //div[contains(@class, 'entry-content')] |
6 | 6 | # for vergecasts, e.g. http://www.theverge.com/2013/8/22/4648566/the-vergecast-090-august-22th-2013-video | |
7 | body: //article[contains(@class, 'feature-entry')] | 7 | body: //article |
8 | body: //article | 8 | prune: no |
9 | prune: no | 9 | #tidy: no |
10 | tidy: no | 10 | |
11 | 11 | strip: //article/header | |
12 | strip: //article/header | 12 | strip: //*[@id='sticky-menu'] |
13 | strip: //*[@id='sticky-menu'] | 13 | strip: //aside |
14 | strip: //aside | 14 | strip: //nav |
15 | strip: //nav | 15 | strip: //img[contains(@class, 'vox-lazy-load')] |
16 | 16 | # deal with bad parsing | |
17 | strip_id_or_class: gallery | 17 | strip: //div[contains(@class, 'story-image')]//div[contains(., 'function(')] |
18 | strip_id_or_class: article-meta | 18 | |
19 | strip_id_or_class: story-navigation | 19 | strip_id_or_class: gallery |
20 | strip_id_or_class: slegend | 20 | strip_id_or_class: article-meta |
21 | strip_id_or_class: related-product-meta | 21 | strip_id_or_class: story-navigation |
22 | strip_id_or_class: comments | 22 | strip_id_or_class: slegend |
23 | strip_id_or_class: ui-jump-list | 23 | strip_id_or_class: related-product-meta |
24 | strip_id_or_class: pullquote | 24 | strip_id_or_class: comments |
25 | 25 | strip_id_or_class: ui-jump-list | |
26 | strip: //q | 26 | strip_id_or_class: pullquote |
27 | 27 | strip_id_or_class: m-ad | |
28 | strip: //a[contains(@class, 'entry-section-title')] | 28 | strip_id_or_class: social-sharing |
29 | 29 | strip_id_or_class: m-video-entry__excerpt | |
30 | test_url: http://www.theverge.com/2012/2/29/2821763/lytro-review | 30 | strip_id_or_class: hidden |
31 | test_url: http://www.theverge.com/2011/11/3/2534861/nokia-lumia-800-review \ No newline at end of file | 31 | |
32 | replace_string(<noscript>): <div> | ||
33 | replace_string(</noscript>): </div> | ||
34 | |||
35 | find_string: <script | ||
36 | replace_string: <div style="display:none" | ||
37 | find_string: </script> | ||
38 | replace_string: </div> | ||
39 | |||
40 | strip: //q | ||
41 | |||
42 | strip: //a[contains(@class, 'entry-section-title')] | ||
43 | |||
44 | test_url: http://www.theverge.com/2012/2/29/2821763/lytro-review | ||
45 | test_url: http://www.theverge.com/2011/11/3/2534861/nokia-lumia-800-review | ||
46 | test_url: http://www.theverge.com/2013/2/24/4026114/barnes-noble-shifting-focus-away-from-nook-hardware | ||
47 | test_url: http://www.theverge.com/2014/6/19/5824072/top-shelf-living-the-dream | ||
48 | test_url: http://www.theverge.com/rss/frontpage \ No newline at end of file | ||