diff options
| author | Nicolas Lœuillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
|---|---|---|
| committer | Nicolas Lœuillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
| commit | 4e067ceabd705201a16b4c92cf4b23f3b990326c (patch) | |
| tree | 939f3a8e5ff3ab9ee414a57a895d3e78e1d46ce3 /inc/3rdparty/site_config/standard/cnn.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/cnn.com.txt')
| -rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/cnn.com.txt | 42 |
1 files changed, 23 insertions, 19 deletions
diff --git a/inc/3rdparty/site_config/standard/cnn.com.txt b/inc/3rdparty/site_config/standard/cnn.com.txt index 995e2c79..6f69e4e8 100644..100755 --- a/inc/3rdparty/site_config/standard/cnn.com.txt +++ b/inc/3rdparty/site_config/standard/cnn.com.txt | |||
| @@ -1,19 +1,23 @@ | |||
| 1 | title: //div[@class="cnn_storyarea"]/h1 | 1 | body: //div[@id='cnnContentContainer']//div[contains(@class, 'cnn_strycntntlft')] |
| 2 | author: //div[@class="cnnByline"]/strong | 2 | title: //div[@class="cnn_storyarea"]/h1 |
| 3 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Sun') | 3 | author: //div[@class="cnnByline"]/strong |
| 4 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Mon') | 4 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Sun') |
| 5 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Tue') | 5 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Mon') |
| 6 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Wed') | 6 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Tue') |
| 7 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Thu') | 7 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Wed') |
| 8 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Fri') | 8 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Thu') |
| 9 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Sat') | 9 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Fri') |
| 10 | strip: //div[@class="cnn_storyarea"]/h1 | 10 | date: substring-after(//div[@class="cnn_strytmstmp"], 'Sat') |
| 11 | strip_id_or_class: cnnByline | 11 | strip: //div[@class="cnn_storyarea"]/h1 |
| 12 | strip_id_or_class: cnn_strytmstmp | 12 | strip_id_or_class: cnnByline |
| 13 | strip_id_or_class: cnn_strycaptiontxt | 13 | strip_id_or_class: cnn_strytmstmp |
| 14 | strip_id_or_class: cnn_strybtntoolsbttm | 14 | strip_id_or_class: cnn_strycaptiontxt |
| 15 | strip_id_or_class: cnn_strybtntools | 15 | strip_id_or_class: cnn_strybtntoolsbttm |
| 16 | strip_id_or_class: cnn_strybtmcntnt | 16 | strip_id_or_class: cnn_strybtntools |
| 17 | strip_id_or_class: cnn_containerwht | 17 | strip_id_or_class: cnn_strybtmcntnt |
| 18 | strip_id_or_class: cnn_stryathrtmp | 18 | strip_id_or_class: sharebar |
| 19 | test_url: http://www.cnn.com/2012/05/13/us/new-york-police-policy/index.html?eref=rss_topstories \ No newline at end of file | 19 | #strip_id_or_class: cnn_containerwht |
| 20 | strip_id_or_class: cnn_stryathrtmp | ||
| 21 | replace_string(<a name="em0"></a>): <!-- a name --> | ||
| 22 | test_url: http://www.cnn.com/2012/05/13/us/new-york-police-policy/index.html?eref=rss_topstories | ||
| 23 | test_url: http://rss.cnn.com/rss/edition.rss \ No newline at end of file | ||