diff options
| -rw-r--r-- | bit-tech.net.txt | 19 | ||||
| -rw-r--r-- | eurogamer.net.txt | 13 | ||||
| -rw-r--r-- | lifehacker.co.uk.txt | 7 | ||||
| -rw-r--r-- | towerofthehand.com.txt | 10 |
4 files changed, 43 insertions, 6 deletions
diff --git a/bit-tech.net.txt b/bit-tech.net.txt new file mode 100644 index 0000000..c6f5b20 --- /dev/null +++ b/bit-tech.net.txt @@ -0,0 +1,19 @@ +body: //div[@id='column_1'] +next_page_link: //div[@class='next']/a[not(contains(@href, '/comments') or contains(@href, '/news/'))] +prune: no + +author: substring-after(//p[@class='byline'], 'by ') +date: substring-before(substring-after(//p[@class='byline'], 'on '), ' by') + +strip: //h1 +strip_id_or_class: socialLinks +strip_id_or_class: byline +strip_id_or_class: pageSelector +strip_id_or_class: articleTabs +strip_id_or_class: pageNav +strip_id_or_class: share +strip_id_or_class: commentsContainer +strip_id_or_class: below_article_related + +test_url: http://www.bit-tech.net/hardware/storage/2014/08/13/ocz-arc-100-240gb-review/1 +test_url: http://www.bit-tech.net/news/bits/2014/08/15/google-trojan/1 diff --git a/eurogamer.net.txt b/eurogamer.net.txt index 8a35166..8931bec 100644 --- a/eurogamer.net.txt +++ b/eurogamer.net.txt @@ -1,8 +1,9 @@ -body: //div[ @class='content' ] | //div[ @class='blog-entry' ] +body: //p[@class='strapline'] | //div[@class='cover-image'] | //article[@class='hd'] +strip: //div[@class='social top'] +strip: //p[@class='byline'] -strip: //h2/abbr | //div[ @class='lowleader' ] | //*[ @class='discussion' ] | //img[ @class='play-button' ] | //div[ @class='boxout' ] | //h2/a | //h2 | //h2/div | //p[ @class='timestamp' ] | //a[ @class='eurogamer-author' ] | //p[ @class='aPager' ] | //h1 | //div[ @id='lowleader' ] | //a[ @class='next' ] | //div[contains(concat(' ', normalize-space(@class), ' '), ' pullquote ')] +date: //span[@itemprop='datePublished'] +author: //a[@itemprop='author']/text() -date://p[ @class='timestamp' ] - -author://a[ @class='eurogamer-author' ] -test_url: http://www.eurogamer.net/articles/digitalfoundry-vs-unreal-engine-4
\ No newline at end of file +test_url: http://www.eurogamer.net/articles/2014-08-20-bungie-ordered-to-return-shares-to-composer-marty-odonnell +test_url: http://www.eurogamer.net/articles/2014-08-20-invisible-inc-does-espionage-justice diff --git a/lifehacker.co.uk.txt b/lifehacker.co.uk.txt new file mode 100644 index 0000000..c540f7f --- /dev/null +++ b/lifehacker.co.uk.txt @@ -0,0 +1,7 @@ +title: //div[@itemprop='headline'] +body: //noscript/img | //div[@itemprop='text'] +author: //div[@class='meta meta--post']//a[@class='is-author'] +date: //div[@class='meta meta--post']//time/@datetime + +test_url: http://www.lifehacker.co.uk/2014/08/22/dealhacker-10-google-chromecast-super-cheap-batteries-much +test_url: http://www.lifehacker.co.uk/2014/08/18/andrognito-hides-files-youd-like-keep-away-prying-eyes diff --git a/towerofthehand.com.txt b/towerofthehand.com.txt new file mode 100644 index 0000000..a4d87d1 --- /dev/null +++ b/towerofthehand.com.txt @@ -0,0 +1,10 @@ +title: //div[@id='headline'] +body: //div[@class='entry_text'] +author: //div[text() = 'Author:']/following-sibling::div/a +date: //div[text() = 'Published:']/following-sibling::div +single_page_link: //a[@href='noscript.html'] +prune: no + +test_url: http://towerofthehand.com/blog/2014/08/08-pitch-this-got-spinoff/index.html +test_url: http://towerofthehand.com/blog/2014/07/31-definitions-and-embodiments/index.html +test_url: http://towerofthehand.com/blog/2014/07/03-hero-with-thousand-faces/index.html |