mirror of
https://github.com/kovidgoyal/calibre.git
synced 2026-02-25 04:30:11 -05:00
83 lines
3.7 KiB
Plaintext
83 lines
3.7 KiB
Plaintext
|
|
__license__ = 'GPL v3'
|
|
__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
|
|
'''
|
|
miamiherald.com
|
|
'''
|
|
|
|
from calibre.web.feeds.news import BasicNewsRecipe
|
|
|
|
class TheMiamiHerald(BasicNewsRecipe):
|
|
title = 'The Miami Herald'
|
|
__author__ = 'Darko Miletic and Sujata Raman'
|
|
description = "Miami-Dade and Broward's source for the latest breaking local news on sports, weather, business, jobs, real estate, shopping, health, travel, entertainment, & more."
|
|
oldest_article = 1
|
|
max_articles_per_feed = 100
|
|
publisher = u'The Miami Herald'
|
|
category = u'miami herald, weather, dolphins, news, miami news, local news, miamiherald, miami newspaper, miamiherald.com, miami, the miami herald, broward, miami-dade'
|
|
language = 'en'
|
|
|
|
no_stylesheets = True
|
|
use_embedded_content = False
|
|
encoding = 'cp1252'
|
|
remove_javascript = True
|
|
|
|
extra_css = '''
|
|
h1{font-family:Arial,Helvetica,sans-serif; font-size:large; color:#1A272F; }
|
|
.subheadline{font-family:Arial,Helvetica,sans-serif; font-size:30%; color: #666666;}
|
|
#storyBodyContent{font-family:Arial,Helvetica,sans-serif; font-size:xx-small; }
|
|
.byline{font-family:Arial,Helvetica,sans-serif; font-size:30%; color:#58595B; }
|
|
.credit_line{font-family:Arial,Helvetica,sans-serif; font-size:30%; color:#58595B; }
|
|
.storyPublishDate{font-family:Arial,Helvetica,sans-serif; font-size:30%; color:#666666; }
|
|
.shirttail{font-family:Arial,Helvetica,sans-serif; font-size:30%; color:#666666;font-style:italic }
|
|
.imageCaption{font-family:Arial,Helvetica,sans-serif; font-size:30%; color:#666666; }
|
|
'''
|
|
|
|
keep_only_tags = [dict(name='div', attrs={'id':['storyBody','storyPhotoContentArea']}),
|
|
]
|
|
|
|
remove_tags = [dict(name=['object','link','embed']),
|
|
dict(name='div', attrs={'class':["imageBuyButton","shareLinksArea","storyTools","spill_navigation pagination","circPromoArea","storyTools_footer","storyYahooContentMatch"]}) ,
|
|
dict(name='div', attrs={'id':["pluck","mlt","storyAssets"]}) ]
|
|
|
|
|
|
feeds = [
|
|
(u'Breaking News' , u'http://www.miamiherald.com/416/index.xml' )
|
|
,(u'Miami-Dade' , u'http://www.miamiherald.com/460/index.xml' )
|
|
,(u'Broward' , u'http://www.miamiherald.com/467/index.xml' )
|
|
,(u'Florida Keys' , u'http://www.miamiherald.com/505/index.xml' )
|
|
,(u'Florida' , u'http://www.miamiherald.com/569/index.xml' )
|
|
,(u'Nation' , u'http://www.miamiherald.com/509/index.xml' )
|
|
,(u'World' , u'http://www.miamiherald.com/578/index.xml' )
|
|
,(u'Americas' , u'http://www.miamiherald.com/579/index.xml' )
|
|
,(u'Cuba' , u'http://www.miamiherald.com/581/index.xml' )
|
|
,(u'Haiti' , u'http://www.miamiherald.com/582/index.xml' )
|
|
,(u'Politics' , u'http://www.miamiherald.com/515/index.xml' )
|
|
,(u'Education' , u'http://www.miamiherald.com/295/index.xml' )
|
|
,(u'Environment' , u'http://www.miamiherald.com/573/index.xml' )
|
|
]
|
|
|
|
|
|
|
|
|
|
|
|
def get_article_url(self, article):
|
|
ans = article.get('guid', None)
|
|
print ans
|
|
try:
|
|
self.log('Looking for full story link in', ans)
|
|
soup = self.index_to_soup(ans)
|
|
x = soup.find(text="Full Story")
|
|
|
|
if x is not None:
|
|
a = x.parent
|
|
if a and a.has_key('href'):
|
|
ans = 'http://www.miamiherald.com'+a['href']
|
|
self.log('Found full story link', ans)
|
|
except:
|
|
pass
|
|
return ans
|
|
|
|
|
|
|