projects / github / wallabag / wallabag.git / blame
| Commit | Line | Data |
|---|---|---|
| 90a1a78b NL |
1 | body: //div[@class="story-body"] |
| 2 | # for video entries | |
| 3 | body: //div[contains(@class, "videoInStory") or @id="meta-information"] | |
| 4 | title: //h1[@class="story-header"] | |
| 5 | date: //span[@class="story-date"]/span[@class='date'] | |
| 6 | # for sport site | |
| 7 | date: //meta[@name='DCTERMS.created']/@content | |
| 8 | author: //div[@id='headline']//span[@class='byline-name'] | |
| 9 | ||
| 10 | # recipes, e.g. http://www.bbc.co.uk/food/recipes/mymincepies_71055 | |
| 11 | body: //div[contains(@class, 'hrecipe')]//div[@id='subcolumn-1'] | |
| 12 | ||
| 13 | #strip: //div[@class="story-feature narrow"] | |
| 14 | #strip: //div[@class="story-feature wide"] | |
| 15 | #strip: //div[@class="story-feature dslideshow-enclosure"] | |
| 16 | strip: //div[contains(@class, "story-feature") and not(contains(@class, 'full-width'))] | |
| 17 | strip: //span[@class="story-date"] | |
| 18 | #strip: //div[@class="caption body-narrow-width"] | |
| 19 | strip: //div[@class="warning"]//p | |
| 20 | strip: //div[@id='page-bookmark-links-head'] | |
| 21 | strip: //object | |
| 22 | strip: //div[contains(@class, "bbccom_advert_placeholder")] | |
| 23 | strip: //div[contains(@class, "embedded-hyper")] | |
| 24 | strip: //div[contains(@class, 'market-data')] | |
| 25 | strip: //a[contains(@class, 'hidden')] | |
| 26 | strip: //div[contains(@class, 'hypertabs')] | |
| 27 | strip: //div[contains(@class, 'related')] | |
| 28 | strip: //form[@id='comment-form'] | |
| 29 | strip: //div[contains(@class, 'comment-introduction')] | |
| 30 | strip: //div[contains(@class, 'share-tools')] | |
| 31 | strip: //div[@id='also-related-links'] | |
| 32 | ||
| 33 | strip_id_or_class: share-help | |
| 34 | strip_id_or_class: comments_module | |
| 35 | ||
| 36 | replace_string(<noscript>): <div> | |
| 37 | replace_string(</noscript>): </div> | |
| 38 | ||
| 39 | native_ad_clue: //meta[@property="og:url" and contains(@content, '/sponsored/')] | |
| 40 | ||
| 41 | tidy: no | |
| 42 | prune: no | |
| 43 | ||
| 44 | dissolve: //h2 | |
| 45 | ||
| 46 | test_url: http://www.bbc.com/sport/0/football/28918021 | |
| 47 | test_contains: Cameroonian footballer Albert Ebosse has died | |
| 48 | ||
| 49 | test_url: http://www.bbc.com/sport/0/football/23224017 | |
| 50 | ||
| 51 | test_url: http://www.bbc.com/news/business-15060862 | |
| 52 | test_contains: Europe's leaders are meeting again to try | |
| 53 | ||
| 54 | ||
| 55 | # news feed | |
| 56 | test_url: http://feeds.bbci.co.uk/news/rss.xml | |
| 57 | # sports feed | |
| 58 | test_url: http://feeds.bbci.co.uk/sport/0/football/rss.xml?edition=int | |
| 59 | # video entry | |
| 60 | test_url: http://www.bbc.com/news/world-asia-22056933 |