diff options
Diffstat (limited to 'inc/3rdparty/site_config/standard/thenation.com.txt')
| -rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/thenation.com.txt | 22 |
1 files changed, 12 insertions, 10 deletions
diff --git a/inc/3rdparty/site_config/standard/thenation.com.txt b/inc/3rdparty/site_config/standard/thenation.com.txt index d88bcdd6..dab17f0b 100644..100755 --- a/inc/3rdparty/site_config/standard/thenation.com.txt +++ b/inc/3rdparty/site_config/standard/thenation.com.txt | |||
| @@ -1,11 +1,13 @@ | |||
| 1 | title: //h1[@class='print-title'] | 1 | title: //h2[@property='dc:title'] |
| 2 | body: //div[@class='print-content'] | 2 | #body: //div[@class='print-content'] |
| 3 | author: //a[contains(@href, '/authors')] | 3 | body: //div[@id='wysiwyg'] |
| 4 | author: substring-before(//div[@class='print-created'], '|') | 4 | author: //a[contains(@href, '/authors')] |
| 5 | date: //span[@class='article-date'] | 5 | author: substring-before(//div[@class='print-created'], '|') |
| 6 | date: substring-after(//div[@class='print-created'], '|') | 6 | date: //span[@class='article-date'] |
| 7 | prune: no | 7 | date: substring-after(//div[@class='print-created'], '|') |
| 8 | 8 | prune: no | |
| 9 | single_page_link: //ul[contains(@class, 'article-actions-bar')]//a[contains(@href, '/print/article/')] | 9 | |
| 10 | 10 | #single_page_link: //ul[contains(@class, 'article-actions-bar')]//a[contains(@href, '/print/article/')] | |
| 11 | single_page_link: //ul[contains(@class, 'article-actions-bar')]//a[contains(@href, '?page=full')] | ||
| 12 | |||
| 11 | test_url: http://www.thenation.com/article/162331/hard-against-time-roy-fisher \ No newline at end of file | 13 | test_url: http://www.thenation.com/article/162331/hard-against-time-roy-fisher \ No newline at end of file |