Improved Zaman Gazetesi

This commit is contained in:
Kovid Goyal 2011-12-09 22:58:40 +05:30
parent c56391238b
commit c8bff7cf79

View File

@ -5,9 +5,10 @@ from calibre.web.feeds.news import BasicNewsRecipe
class Zaman (BasicNewsRecipe):
title = u'ZAMAN Gazetesi'
description = ' Zaman Gazetesi''nin internet sitesinden günlük haberler'
__author__ = u'thomass'
oldest_article = 2
max_articles_per_feed =100
max_articles_per_feed =50
# no_stylesheets = True
#delay = 1
#use_embedded_content = False
@ -16,19 +17,19 @@ class Zaman (BasicNewsRecipe):
category = 'news, haberler,TR,gazete'
language = 'tr'
publication_type = 'newspaper '
extra_css = ' body{ font-family: Verdana,Helvetica,Arial,sans-serif } .introduction{font-weight: bold} .story-feature{display: block; padding: 0; border: 1px solid; width: 40%; font-size: small} .story-feature h2{text-align: center; text-transform: uppercase} '
extra_css = '.buyukbaslik{font-weight: bold; font-size: 18px;color:#0000FF}'#body{ font-family: Verdana,Helvetica,Arial,sans-serif } .introduction{font-weight: bold} .story-feature{display: block; padding: 0; border: 1px solid; width: 40%; font-size: small} .story-feature h2{text-align: center; text-transform: uppercase} '
conversion_options = {
'tags' : category
,'language' : language
,'publisher' : publisher
,'linearize_tables': False
,'linearize_tables': True
}
cover_img_url = 'https://fbcdn-profile-a.akamaihd.net/hprofile-ak-snc4/188140_81722291869_2111820_n.jpg'
masthead_url = 'http://medya.zaman.com.tr/extentions/zaman.com.tr/img/section/logo-section.png'
keep_only_tags = [dict(name='div', attrs={'id':[ 'news-detail-content']}), dict(name='td', attrs={'class':['columnist-detail','columnist_head']}) ]
remove_tags = [ dict(name='div', attrs={'id':['news-detail-news-text-font-size','news-detail-gallery','news-detail-news-bottom-social']}),dict(name='div', attrs={'class':['radioEmbedBg','radyoProgramAdi']}),dict(name='a', attrs={'class':['webkit-html-attribute-value webkit-html-external-link']}),dict(name='table', attrs={'id':['yaziYorumTablosu']}),dict(name='img', attrs={'src':['http://medya.zaman.com.tr/pics/paylas.gif','http://medya.zaman.com.tr/extentions/zaman.com.tr/img/columnist/ma-16.png']})]
#keep_only_tags = [dict(name='div', attrs={'id':[ 'news-detail-content']}), dict(name='td', attrs={'class':['columnist-detail','columnist_head']}) ]
remove_tags = [ dict(name='img', attrs={'src':['http://medya.zaman.com.tr/zamantryeni/pics/zamanonline.gif']})]#,dict(name='div', attrs={'class':['radioEmbedBg','radyoProgramAdi']}),dict(name='a', attrs={'class':['webkit-html-attribute-value webkit-html-external-link']}),dict(name='table', attrs={'id':['yaziYorumTablosu']}),dict(name='img', attrs={'src':['http://medya.zaman.com.tr/pics/paylas.gif','http://medya.zaman.com.tr/extentions/zaman.com.tr/img/columnist/ma-16.png']})
#remove_attributes = ['width','height']
@ -37,7 +38,8 @@ class Zaman (BasicNewsRecipe):
feeds = [
( u'Anasayfa', u'http://www.zaman.com.tr/anasayfa.rss'),
( u'Son Dakika', u'http://www.zaman.com.tr/sondakika.rss'),
( u'En çok Okunanlar', u'http://www.zaman.com.tr/max_all.rss'),
#( u'En çok Okunanlar', u'http://www.zaman.com.tr/max_all.rss'),
#( u'Manşet', u'http://www.zaman.com.tr/manset.rss'),
( u'Gündem', u'http://www.zaman.com.tr/gundem.rss'),
( u'Yazarlar', u'http://www.zaman.com.tr/yazarlar.rss'),
( u'Politika', u'http://www.zaman.com.tr/politika.rss'),
@ -45,11 +47,20 @@ class Zaman (BasicNewsRecipe):
( u'Dış Haberler', u'http://www.zaman.com.tr/dishaberler.rss'),
( u'Yorumlar', u'http://www.zaman.com.tr/yorumlar.rss'),
( u'Röportaj', u'http://www.zaman.com.tr/roportaj.rss'),
( u'Dizi Yazı', u'http://www.zaman.com.tr/dizi.rss'),
( u'Bilişim', u'http://www.zaman.com.tr/bilisim.rss'),
( u'Otomotiv', u'http://www.zaman.com.tr/otomobil.rss'),
( u'Spor', u'http://www.zaman.com.tr/spor.rss'),
( u'Kürsü', u'http://www.zaman.com.tr/kursu.rss'),
( u'Eğitim', u'http://www.zaman.com.tr/egitim.rss'),
( u'Kültür Sanat', u'http://www.zaman.com.tr/kultursanat.rss'),
( u'Televizyon', u'http://www.zaman.com.tr/televizyon.rss'),
( u'Manşet', u'http://www.zaman.com.tr/manset.rss'),
( u'Aile', u'http://www.zaman.com.tr/aile.rss'),
( u'Cuma Eki', u'http://www.zaman.com.tr/cuma.rss'),
( u'Cumaertesi Eki', u'http://www.zaman.com.tr/cumaertesi.rss'),
( u'Pazar Eki', u'http://www.zaman.com.tr/pazar.rss'),
]
def print_version(self, url):
return url.replace('http://www.zaman.com.tr/haber.do?haberno=', 'http://www.zaman.com.tr/yazdir.do?haberno=')