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/thestreet.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/thestreet.com.txt')
-rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/thestreet.com.txt | 48 |
1 files changed, 24 insertions, 24 deletions
diff --git a/inc/3rdparty/site_config/standard/thestreet.com.txt b/inc/3rdparty/site_config/standard/thestreet.com.txt index 5de75637..58eabf00 100644..100755 --- a/inc/3rdparty/site_config/standard/thestreet.com.txt +++ b/inc/3rdparty/site_config/standard/thestreet.com.txt | |||
@@ -1,25 +1,25 @@ | |||
1 | title: //div[@id='storyHdr']/h1 | 1 | title: //div[@id='storyHdr']/h1 |
2 | title: //div[@id='print']//h2 | 2 | title: //div[@id='print']//h2 |
3 | body: //div[@class="virtualpage"] | 3 | body: //div[@class="virtualpage"] |
4 | body: //div[@id='print']//div[@id='bd'] | 4 | body: //div[@id='print']//div[@id='bd'] |
5 | author: //meta[@name="AUTHOR"]/@content | 5 | author: //meta[@name="AUTHOR"]/@content |
6 | author: (//div[@id='print']//div[@id='bd']/h4)[1] | 6 | author: (//div[@id='print']//div[@id='bd']/h4)[1] |
7 | date: //meta[@name="DATE"]/@content | 7 | date: //meta[@name="DATE"]/@content |
8 | date: //div[@id='print']//div[@id='dte'] | 8 | date: //div[@id='print']//div[@id='dte'] |
9 | 9 | ||
10 | strip_id_or_class: articleFooter | 10 | strip_id_or_class: articleFooter |
11 | strip_id_or_class: sidebar | 11 | strip_id_or_class: sidebar |
12 | strip_id_or_class: ie6PrintSubhead | 12 | strip_id_or_class: ie6PrintSubhead |
13 | strip_id_or_class: subHdr | 13 | strip_id_or_class: subHdr |
14 | 14 | ||
15 | 15 | ||
16 | replace_string(<P/>): </p><p> | 16 | replace_string(<P/>): </p><p> |
17 | 17 | ||
18 | prune: no | 18 | prune: no |
19 | 19 | ||
20 | #TODO: redirects back - perhaps needs referer to work | 20 | #TODO: redirects back - perhaps needs referer to work |
21 | single_page_link: //div[@id='storyDetail']//a[contains(@href, '/print/')] | 21 | single_page_link: //div[@id='storyDetail']//a[contains(@href, '/print/')] |
22 | 22 | ||
23 | test_url: http://www.thestreet.com/story/11386556/1/which-of-these-10-dividend-stocks-is-worth-the-risk.html | 23 | test_url: http://www.thestreet.com/story/11386556/1/which-of-these-10-dividend-stocks-is-worth-the-risk.html |
24 | # multi page | 24 | # multi page |
25 | test_url: http://www.thestreet.com/story/11387090/1/7-ubs-stock-picks-for-2012.html \ No newline at end of file | 25 | test_url: http://www.thestreet.com/story/11387090/1/7-ubs-stock-picks-for-2012.html \ No newline at end of file |