diff options
| -rw-r--r-- | arstechnica.com.txt | 2 | ||||
| -rw-r--r-- | autocar.co.uk.txt | 8 | ||||
| -rw-r--r-- | bbc.com.txt | 2 | ||||
| -rw-r--r-- | buzzfeed.com.txt | 11 | ||||
| -rw-r--r-- | finance.yahoo.com.txt | 4 | ||||
| -rw-r--r-- | theatlantic.com.txt | 2 | ||||
| -rw-r--r-- | theglobeandmail.com.txt | 7 | ||||
| -rw-r--r-- | theguardian.com.txt | 6 |
8 files changed, 34 insertions, 8 deletions
diff --git a/arstechnica.com.txt b/arstechnica.com.txt index 767f680..eb92aa2 100644 --- a/arstechnica.com.txt +++ b/arstechnica.com.txt @@ -13,5 +13,7 @@ title: //div[@id='story']//h2[@class='title'] strip: //div[@class='pager'] next_page_link: //nav//a[span/@class='next']/@href +native_ad_clue: //meta[@property="og:url" and contains(@content, '/sponsored/')] + test_url: http://arstechnica.com/tech-policy/news/2012/02/gigabit-internet-for-80-the-unlikely-success-of-californias-sonicnet.ars test_url: http://arstechnica.com/apple/2005/04/macosx-10-4/ diff --git a/autocar.co.uk.txt b/autocar.co.uk.txt index 7851bfa..9f4fe18 100644 --- a/autocar.co.uk.txt +++ b/autocar.co.uk.txt @@ -1,11 +1,13 @@ title: //div[@class='col-center']/h1 author: //div[@class='personality']/a date: //div[@class='personality-date'] -body: //div[@class='content-top ']//div[@class='content'][1] | //div[contains(@class,'article-body')] +body: //div[@class='content-top ']//div[@class='content'][1] | //div[contains(@class,'article-body')] | //div[contains(@class,'main-article')] + +next_page_link: //div[@id='review-link']/a strip: //div[@class='author-block'] -# remove email subscription nag text strip: //p//iframe[contains(@src,'signup')]/preceding::p[1] -test_url: http://www.autocar.co.uk/car-review/volkswagen/e-golf +test_url: http://www.autocar.co.uk/car-review/volkswagen/golf test_url: http://www.autocar.co.uk/car-news/pebble-beach/saleen-unveils-performance-electric-vehicle-based-tesla-model-s +test_url: http://www.autocar.co.uk/car-review/rolls-royce/first-drives/rolls-royce-ghost-series-ii-first-drive-review diff --git a/bbc.com.txt b/bbc.com.txt index 93a3604..67a27c6 100644 --- a/bbc.com.txt +++ b/bbc.com.txt @@ -36,6 +36,8 @@ strip_id_or_class: comments_module replace_string(<noscript>): <div> replace_string(</noscript>): </div> +native_ad_clue: //meta[@property="og:url" and contains(@content, '/sponsored/')] + tidy: no prune: no diff --git a/buzzfeed.com.txt b/buzzfeed.com.txt index 97dddae..ea88ea4 100644 --- a/buzzfeed.com.txt +++ b/buzzfeed.com.txt @@ -10,6 +10,15 @@ date: //time[@data-print='date'] body: //div[@data-print='body'] body: //section[@data-print='body'] +find_string: rel:bf_image_src= +replace_string: src= +find_string: src="data: +replace_string: disabled_src="data: + +native_ad_clue: //meta[@property="article:section" and @content="Advertiser"] + # For various things... strip: *[@data-print="ignore"] -test_url: http://www.buzzfeed.com/hgrant/35-reasons-why-dogs-hate-the-holidays
\ No newline at end of file +test_url: http://www.buzzfeed.com/hgrant/35-reasons-why-dogs-hate-the-holidays +# Native ad +test_url: http://www.buzzfeed.com/bravo/ways-to-up-your-online-dating-game
\ No newline at end of file diff --git a/finance.yahoo.com.txt b/finance.yahoo.com.txt index 248522c..0c967db 100644 --- a/finance.yahoo.com.txt +++ b/finance.yahoo.com.txt @@ -5,8 +5,8 @@ strip: //div[contains(@class, 'related-companies')] strip: //div[@id='y-article-related'] strip: //div[@id='ypf-article-related'] prune: no +tidy: no single_page_link: //div[@class='ft']//a[contains(@href, 'page=all')] -test_url: http://sg.finance.yahoo.com/news/Motorola-takes-wraps-249-rsg-3508842732.html?x=0&.v=1 -test_url: http://finance.yahoo.com/news/super-young-retirement-savers.html
\ No newline at end of file +test_url: http://finance.yahoo.com/news/canadian-orebodies-gives-notice-exercise-130000032.html
\ No newline at end of file diff --git a/theatlantic.com.txt b/theatlantic.com.txt index aa41b15..3fc5611 100644 --- a/theatlantic.com.txt +++ b/theatlantic.com.txt @@ -15,6 +15,8 @@ strip: //div[@class='earthbox'] single_page_link: //article//a[contains(@class, 'print')] +native_ad_clue: //meta[@property="og:url" and contains(@content, '/sponsored/')] + test_url: http://www.theatlantic.com/technology/archive/2011/04/want-to-see-how-crazy-a-bot-run-market-can-be/237773/ test_url: http://www.theatlantic.com/magazine/archive/2007/11/the-autumn-of-the-multitaskers/6342/ test_url: http://www.theatlantic.com/entertainment/archive/2012/04/30-rock-live-a-funny-reminder-of-why-sitcoms-arent-shot-live-anymore/256447/
\ No newline at end of file diff --git a/theglobeandmail.com.txt b/theglobeandmail.com.txt index 750f847..2473cad 100644 --- a/theglobeandmail.com.txt +++ b/theglobeandmail.com.txt @@ -1,5 +1,10 @@ +body: //div[contains(@class, 'entry-content')]//div[contains(@class, 'column-2')] single_page_link: //div[contains(@class, 'pagination')]//a[contains(@title, 'ingle page')] +strip_id_or_class: entry-related +strip_id_or_class: entry-sidebar +strip_id_or_class: entry-pagination tidy: no prune: no -test_url: http://www.theglobeandmail.com/report-on-business/rob-magazine/how-a-novice-miner-survived-a-summer-in-the-klondike/article2345350/
\ No newline at end of file +test_url: http://www.theglobeandmail.com/report-on-business/rob-magazine/how-a-novice-miner-survived-a-summer-in-the-klondike/article2345350/ +test_url: http://www.theglobeandmail.com/report-on-business/industry-news/energy-and-resources/cliffs-natural-resources-looking-to-exit-ontarios-ring-of-fire/article20651617/
\ No newline at end of file diff --git a/theguardian.com.txt b/theguardian.com.txt index c803e4e..9ea44cb 100644 --- a/theguardian.com.txt +++ b/theguardian.com.txt @@ -6,8 +6,12 @@ strip: //div[contains(@class, 'kindleWidget')] #strip: //a[not(text())] strip_id_or_class: pocket-btn author: //li[@class='byline'] +native_ad_clue: //meta[@property="article:tag" and contains(@content, "Partner zone")] +native_ad_clue: //meta[@property="video:tag" and contains(@content, "Partner zone")] prune: no tidy: no test_url: http://www.theguardian.com/world/2013/oct/04/nsa-gchq-attack-tor-network-encryption test_url: http://www.theguardian.com/world/2013/oct/03/edward-snowden-files-john-lanchester -test_url: http://www.theguardian.com/commentisfree/2014/jun/15/britishness-search-identity-my-part-in-camerons-odyssey
\ No newline at end of file +test_url: http://www.theguardian.com/commentisfree/2014/jun/15/britishness-search-identity-my-part-in-camerons-odyssey +# Native ad +test_url: http://www.theguardian.com/sustainable-business/2014/jul/18/ben-jerry-turn-ice-cream-into-energy
\ No newline at end of file |