This commit is contained in:
Kovid Goyal 2021-01-04 22:15:31 +05:30
parent ea62b8b19d
commit a6840fb462
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C
3 changed files with 77 additions and 70 deletions

View File

@ -2,15 +2,17 @@ __license__ = 'GPL v3'
__copyright__ = '2013, Armin Geller' __copyright__ = '2013, Armin Geller'
## ##
## Written: May 2013 (new coding) # Written: May 2013 (new coding)
## Version: 4.4 # Version: 4.4
## Last update: 2020-12-29 # Last update: 2020-12-29
## ##
''' '''
Fetch WirtschaftsWoche Online Fetch WirtschaftsWoche Online
''' '''
import re
from calibre.web.feeds.news import BasicNewsRecipe from calibre.web.feeds.news import BasicNewsRecipe
class WirtschaftsWocheOnline(BasicNewsRecipe): class WirtschaftsWocheOnline(BasicNewsRecipe):
title = u'WirtschaftsWoche Online' title = u'WirtschaftsWoche Online'
__author__ = 'Armin Geller' # Update AGE 2013-01-05; 2018-03-01 __author__ = 'Armin Geller' # Update AGE 2013-01-05; 2018-03-01
@ -31,14 +33,11 @@ class WirtschaftsWocheOnline(BasicNewsRecipe):
cover_source = 'https://www.ikiosk.de/shop/epaper/wirtschaftswoche.html' cover_source = 'https://www.ikiosk.de/shop/epaper/wirtschaftswoche.html'
masthead_url = 'http://www.wiwo.de/images/wiwo_logo/5748610/1-formatOriginal.png' masthead_url = 'http://www.wiwo.de/images/wiwo_logo/5748610/1-formatOriginal.png'
def get_cover_url(self): def get_cover_url(self):
cover_source_soup = self.index_to_soup(self.cover_source) cover_source_soup = self.index_to_soup(self.cover_source)
preview_image_div = cover_source_soup.find(attrs={'class': 'gallery'}) preview_image_div = cover_source_soup.find(attrs={'class': 'gallery'})
return preview_image_div.a.img['src'] return preview_image_div.a.img['src']
extra_css = ''' extra_css = '''
h1, h2 {font-size: 1.6em; text-align: left} h1, h2 {font-size: 1.6em; text-align: left}
.c-leadtext {font-size: 1em; font-style: italic; font-weight: normal} .c-leadtext {font-size: 1em; font-style: italic; font-weight: normal}
@ -48,19 +47,30 @@ class WirtschaftsWocheOnline(BasicNewsRecipe):
''' '''
keep_only_tags = [ keep_only_tags = [
dict(name='div', attrs={'class':['o-article__element', dict(
'o-article__content c-richText', name='div',
attrs={
'class': [
'o-article__element', 'o-article__content c-richText',
'o-article__content-element o-article__content-element--richtext' 'o-article__content-element o-article__content-element--richtext'
]}) ]
}
)
] ]
remove_tags = [ remove_tags = [
dict(name='div', attrs={'class':['c-pagination u-flex', dict(
name='div',
attrs={
'class': [
'c-pagination u-flex',
'c-standard-article-teaser', 'c-standard-article-teaser',
'c-pagination u-flex ajaxify', 'c-pagination u-flex ajaxify',
'c-socialshare u-margin-xxl ', 'c-socialshare u-margin-xxl ',
'c-list', # AGe 2020-12-29 'c-list', # AGe 2020-12-29
]}) ]
}
)
] ]
feeds = [ feeds = [
@ -81,15 +91,12 @@ class WirtschaftsWocheOnline(BasicNewsRecipe):
# Wandel kostet Milliarden + ': ' + SUV und China sollen Audi wieder nach vorne bringen # Wandel kostet Milliarden + ': ' + SUV und China sollen Audi wieder nach vorne bringen
# https://www.wiwo.de/unternehmen/auto/wandel-kostet-milliarden-suv-und-china-sollen-audi-wieder-nach-vorne-bringen/21069566.html # https://www.wiwo.de/unternehmen/auto/wandel-kostet-milliarden-suv-und-china-sollen-audi-wieder-nach-vorne-bringen/21069566.html
# debug:
# input: <h2 class="c-headline c-headline--article u-margin-m"><span class="c-overline c-overline--alternate u-uppercase u-letter-spacing u-margin-m c-overline--article">Wandel kostet Milliarden</span> SUV und China sollen Audi wieder nach vorne bringen</h2>
# parsed: <h2 class="c-headline c-headline--article u-margin-m"><span class="c-overline c-overline--alternate u-uppercase u-letter-spacing u-margin-m c-overline--article">Wandel kostet Milliarden</span> SUV und China sollen Audi wieder nach vorne bringen</h2>
# processed: <h2 class="c-headline"><span class="c-overline">Wandel kostet Milliarden</span> SUV und China sollen Audi wieder nach vorne bringen</h2>
# structure: <h2 class="c-headline c-headline--article u-margin-m"><span class="c-overline c-overline--alternate u-uppercase u-letter-spacing u-margin-m c-overline--article">Wandel kostet Milliarden</span> SUV und China sollen Audi wieder nach vorne bringen</h2>
# epub: <h2 class="c-headline"><span class="c-overline">Wandel kostet Milliarden</span> SUV und China sollen Audi wieder nach vorne bringen</h2>
preprocess_regexps = [ preprocess_regexps = [
(re.compile(r'(c-overline--article">[^>]*)(</span>)', re.DOTALL|re.IGNORECASE), lambda match: match.group(1) + ': ' + match.group(2)), (
re.compile(
r'(c-overline--article">[^>]*)(</span>)', re.DOTALL | re.IGNORECASE
), lambda match: match.group(1) + ': ' + match.group(2)
),
] ]
# /For hegi # /For hegi