| | 1 | # A. Niepel, narya.de@...\r |
| | 2 | # - added single_page_link\r |
| | 3 | # - added author for default and single page view\r |
| | 4 | # - added date for single page view\r |
| | 5 | # fforst@...\r |
| | 6 | # - Fixed it\r |
| | 7 | # bode2104@...\r |
| | 8 | # - Fixed single_page_link\r |
| | 9 | # - Included intro text in single page view\r |
| | 10 | # - Added body in default view\r |
| | 11 | \r |
| | 12 | # set body\r |
| | 13 | tidy: no\r |
| | 14 | # body in single page view\r |
| | 15 | body: //div[@id="spArticleContent"]\r |
| | 16 | # body in default view\r |
| | 17 | body: //div[@id="spArticleSection"]\r |
| | 18 | # body in "Fotostrecke"\r |
| | 19 | body: //div[@id="spBigaContent"]\r |
| | 20 | \r |
| | 21 | # set date in single page view\r |
| | 22 | date: //div[@id="spArticleContent"]/h3\r |
| | 23 | # strip date\r |
| | 24 | strip: //div[@id="spArticleContent"]/h3\r |
| | 25 | # set date in "Fotostrecke"\r |
| | 26 | date: //div[@id="spBigaDatum"]\r |
| | 27 | \r |
| | 28 | #set title in single page view\r |
| | 29 | title: //div[@id='spArticleContent']/h2\r |
| | 30 | # strip title\r |
| | 31 | strip: //div[@id='spArticleContent']/h1\r |
| | 32 | strip: //div[@id='spArticleContent']/h2\r |
| | 33 | #set title in "Fotostrecke"\r |
| | 34 | title: //div[@class='spBigaHeadline']\r |
| | 35 | \r |
| | 36 | # set author\r |
| | 37 | author: //p[@class="spAuthor"]/a\r |
| | 38 | author: substring-after(//p[@class="spAuthor"], 'Von ')\r |
| | 39 | # strip author\r |
| | 40 | strip: //p[@class='spAuthor']\r |
| | 41 | \r |
| | 42 | # remove captions\r |
| | 43 | strip: //*/span[@class='spPicLayerText']\r |
| | 44 | strip: //*/div[@class='spPanoPlayerPaneControl']\r |
| | 45 | strip: //*/div[@class='spCredit']\r |
| | 46 | strip: //*/div[@class='spCredit']/following-sibling::p\r |
| | 47 | \r |
| | 48 | # remove ads\r |
| | 49 | strip: //div[@class='spMInline']\r |
| | 50 | \r |
| | 51 | # remove photogalleries and extras\r |
| | 52 | strip: //div[@class='spPhotoGallery']\r |
| | 53 | strip: //div[@class='spPhotoGallery']/following-sibling::br\r |
| | 54 | strip: //div[@class='spAssetAlignleft']\r |
| | 55 | strip: //div[contains(@class,'spAsset')]\r |
| | 56 | strip: //br[@clear='all']\r |
| | 57 | \r |
| | 58 | # remove community functions\r |
| | 59 | strip: //div[@id='spSocialBookmark']\r |
| | 60 | strip: //div[contains(@class, 'spCommunityBox')]\r |
| | 61 | strip: //div[contains(@class, 'spArticleNewsfeedBox')]\r |
| | 62 | strip: //div[@class='spArticleCredit']\r |
| | 63 | \r |
| | 64 | # remove clutter in "Fotostrecke"\r |
| | 65 | strip: //div[@id='spBreadcrumb']\r |
| | 66 | strip: //div[@id='spBigaLatestEntries']\r |
| | 67 | strip: //div[contains(@class, 'spBigaNavi')]\r |
| | 68 | strip: //div[@class='spDottedLine']\r |
| | 69 | \r |
| | 70 | # Use link to print article for single page view\r |
| | 71 | single_page_link: //a[contains(@href, '-druck')]\r |
| | 72 | \r |
| | 73 | # use next link in "Fotostrecke"\r |
| | 74 | next_page_link: //a[@class='spBigaControlForw']\r |
| | 75 | test_url: http://www.spiegel.de/politik/deutschland/0,1518,787602,00.html |
projects / github / wallabag / wallabag.git / blame_incremental