Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-10-31 10:37:00 -04:00
Code Issues Packages Projects Releases Wiki Activity

Fix #1170798 (Metro NL site changed, updated recipe)

Browse Source
This commit is contained in:
Kovid Goyal 2013-04-20 10:20:23 +05:30
parent 6be40de8f1
commit 00b8c40c63
1 changed files with 12 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
recipes/metro_news_nl.recipe
View File

@ -36,6 +36,9 @@ from BeautifulSoup import BeautifulSoup
Changed order of regex to speedup proces Changed order of regex to speedup proces
Version 1.9.3 23-05-2012 Version 1.9.3 23-05-2012
Updated Cover image Updated Cover image
Version 1.9.4 19-04-2013
Added regex filter for mailto
Updated for new layout of metro-site
''' '''
class AdvancedUserRecipe1306097511(BasicNewsRecipe): class AdvancedUserRecipe1306097511(BasicNewsRecipe):
@ -43,7 +46,7 @@ class AdvancedUserRecipe1306097511(BasicNewsRecipe):
oldest_article = 1.2 oldest_article = 1.2
max_articles_per_feed = 25 max_articles_per_feed = 25
__author__ = u'DrMerry' __author__ = u'DrMerry'
description = u'Metro Nederland' description = u'Metro Nederland v1.9.4 2013-04-19'
language = u'nl' language = u'nl'
simultaneous_downloads = 5 simultaneous_downloads = 5
masthead_url = 'http://blog.metronieuws.nl/wp-content/themes/metro/images/header.gif' masthead_url = 'http://blog.metronieuws.nl/wp-content/themes/metro/images/header.gif'
@ -68,13 +71,17 @@ class AdvancedUserRecipe1306097511(BasicNewsRecipe):
#(re.compile('(</?)h2', re.DOTALL|re.IGNORECASE),lambda match:'\1em') #(re.compile('(</?)h2', re.DOTALL|re.IGNORECASE),lambda match:'\1em')
] ]
remove_tags_before= dict(id='date') remove_tags_before= dict(id='subwrapper')
remove_tags_after = [dict(name='div', attrs={'class':['column-1-3','gallery-text']})]#id='share-and-byline')] remove_tags_after = dict(name='div', attrs={'class':['body-area','article-main-area']})
#name='div', attrs={'class':['subwrapper']})]
#'column-1-3','gallery-text']})]#id='share-and-byline')]
filter_regexps = [r'mailto:.*']
remove_tags = [ remove_tags = [
dict(name=['iframe','script','noscript','style']), dict(name=['iframe','script','noscript','style']),
dict(name='div', attrs={'class':['column-4-5','column-1-5','ad-msg','col-179 ','col-373 ','clear','ad','navigation',re.compile('share-tools(-top)?'),'tools','metroCommentFormWrap','article-tools-below-title','related-links','padding-top-15',re.compile('^promo.*?$'),'teaser-component',re.compile('fb(-comments|_iframe_widget)'),'promos','header-links','promo-2']}), dict(name='div', attrs={'class':['aside clearfix','aside clearfix middle-col-line','comments','share-tools','article-right-column','column-4-5','column-1-5','ad-msg','col-179 ','col-373 ','clear','ad','navigation',re.compile('share-tools(-top)?'),'tools','metroCommentFormWrap','article-tools-below-title','related-links','padding-top-15',re.compile('^promo.*?$'),'teaser-component',re.compile('fb(-comments|_iframe_widget)'),'promos','header-links','promo-2']}),
dict(id=['column-1-5-bottom','column-4-5',re.compile('^ad(\d+|adcomp.*?)?$'),'adadcomp-4','margin-5','sidebar',re.compile('^article-\d'),'comments','gallery-1']), dict(id=['article-2','googleads','column-1-5-bottom','column-4-5',re.compile('^ad(\d+|adcomp.*?)?$'),'adadcomp-4','margin-5','sidebar',re.compile('^article-\d'),'comments','gallery-1','sharez_container','ts-container','topshares','ts-title']),
dict(name='a', attrs={'name':'comments'}), dict(name='a', attrs={'name':'comments'}),
#dict(name='div', attrs={'data-href'}), #dict(name='div', attrs={'data-href'}),
dict(name='img', attrs={'class':'top-line','title':'volledig scherm'}), dict(name='img', attrs={'class':'top-line','title':'volledig scherm'}),