calibre/recipes/blic.recipe

80 lines
4.4 KiB
Plaintext

__license__ = 'GPL v3'
__copyright__ = '2008-2011, Darko Miletic <darko.miletic at gmail.com>'
'''
blic.rs
'''
import re
from calibre.web.feeds.news import BasicNewsRecipe
class Blic(BasicNewsRecipe):
title = 'Blic'
__author__ = 'Darko Miletic'
description = 'Blic.rs online verzija najtiraznije novine u Srbiji donosi najnovije vesti iz Srbije i sveta, komentare, politicke analize, poslovne i ekonomske vesti, vesti iz regiona, intervjue, informacije iz kulture, reportaze, pokriva sve sportske dogadjaje, detaljan tv program, nagradne igre, zabavu, fenomenalni Blic strip, dnevni horoskop, arhivu svih dogadjaja'
publisher = 'RINGIER d.o.o.'
category = 'news, politics, Serbia'
oldest_article = 2
max_articles_per_feed = 100
no_stylesheets = True
use_embedded_content = False
masthead_url = 'http://www.blic.rs/resources/images/header/header_back.png'
language = 'sr'
publication_type = 'newspaper'
extra_css = """
@font-face {font-family: "serif1";src:url(res:///opt/sony/ebook/FONT/tt0011m_.ttf)}
@font-face {font-family: "sans1";src:url(res:///opt/sony/ebook/FONT/tt0003m_.ttf)}
body{font-family: Georgia, serif1, serif}
.articledescription,#nadnaslov,.article_info{font-family: Arial, sans1, sans-serif}
.img_full{float: none}
#nadnaslov{font-size: small}
#article_lead{font-size: 1.5em}
h1{color: red}
.potpis{font-size: x-small; color: gray}
.article_info{font-size: small}
img{margin-bottom: 0.8em; margin-top: 0.8em; display: block}
"""
conversion_options = {
'comment' : description
, 'tags' : category
, 'publisher': publisher
, 'language' : language
, 'linearize_tables' : True
}
preprocess_regexps = [(re.compile(u'\u0110'), lambda match: u'\u00D0')]
remove_tags_before = dict(name='div', attrs={'id':'article_info'})
remove_tags = [dict(name=['object','link','meta','base','object','embed'])]
remove_attributes = ['width','height','m_id','m_ext','mlg_id','poll_id','v_id']
feeds = [
(u'Politika' , u'http://www.blic.rs/rss/Vesti/Politika')
,(u'Tema Dana' , u'http://www.blic.rs/rss/Vesti/Tema-Dana')
,(u'Svet' , u'http://www.blic.rs/rss/Vesti/Svet')
,(u'Drustvo' , u'http://www.blic.rs/rss/Vesti/Drustvo')
,(u'Ekonomija' , u'http://www.blic.rs/rss/Vesti/Ekonomija')
,(u'Hronika' , u'http://www.blic.rs/rss/Vesti/Hronika')
,(u'Beograd' , u'http://www.blic.rs/rss/Vesti/Beograd')
,(u'Srbija' , u'http://www.blic.rs/rss/Vesti/Srbija')
,(u'Vojvodina' , u'http://www.blic.rs/rss/Vesti/Vojvodina')
,(u'Republika Srpska' , u'http://www.blic.rs/rss/Vesti/Republika-Srpska')
,(u'Reportaza' , u'http://www.blic.rs/rss/Vesti/Reportaza')
,(u'Dodatak' , u'http://www.blic.rs/rss/Vesti/Dodatak')
,(u'Zabava' , u'http://www.blic.rs/rss/Zabava')
,(u'Kultura' , u'http://www.blic.rs/rss/Kultura')
,(u'Slobodno Vreme' , u'http://www.blic.rs/rss/Slobodno-vreme')
,(u'IT' , u'http://www.blic.rs/rss/IT')
,(u'Komentar' , u'http://www.blic.rs/rss/Komentar')
,(u'Intervju' , u'http://www.blic.rs/rss/Intervju')
]
def print_version(self, url):
return url + '/print'
def preprocess_html(self, soup):
for item in soup.findAll(style=True):
del item['style']
return soup