Four Greek news recipes

SteliosGero · 10-31-2011, 07:54 AM

I am posting these in case you would like to add them.

Keep up the good work guys!

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe

class Capital(BasicNewsRecipe):
    title                  = 'Capital.gr'
    __author__             ='Stelios'
    description            = 'Financial News from Greece'
    #max_articles_per_feed  = 100
    oldest_article         = 3
    publisher              = 'Capital.gr'
    category               = 'news, GR'
    language               = 'el'
    encoding               = 'windows-1253'
    cover_url             = 'http://files.capital.gr/images/caplogo.gif'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    keep_only_tags = [
  dict(name='h1'),
  dict(name='p'),
  dict(name='span', attrs={'id' : ["textbody"]})
                        ]

#3 posts seemed to have utf8 encoding
    feeds          = [
                     (u'\u039F\u039B\u0395\u03A3 \u039F\u0399 \u0395\u0399\u0394\u0397\u03A3\u0395\u0399\u03A3', 'http://www.capital.gr/news/newsrss.asp?s=-1'),
	(u'\u0395\u03A0\u0399\u03A7\u0395\u0399\u03A1\u0397\u03A3\u0395\u0399\u03A3', 'http://www.capital.gr/news/newsrss.asp?s=-2'),
	(u'\u0391\u0393\u039F\u03A1\u0395\u03A3', 'http://www.capital.gr/news/newsrss.asp?s=-3'),
	(u'\u039F\u0399\u039A\u039F\u039D\u039F\u039C\u0399\u0391', 'http://www.capital.gr/news/newsrss.asp?s=-4'),
	(u'\u03A7\u03A1\u0397\u039C. \u0391\u039D\u0391\u039A\u039F\u0399\u039D\u03A9\u03A3\u0395\u0399\u03A3', 'http://www.capital.gr/news/newsrss.asp?s=-6'),
	(u'\u039C\u03CC\u03BD\u03B9\u03BC\u03B5\u03C2 \u03C3\u03C4\u03AE\u03BB\u03B5\u03C2: \u039C\u0395 \u0391\u03A0\u039F\u03A8\u0397', 'http://www.capital.gr/articles/articlesrss.asp?catid=4'),
	(u'\u039C\u03CC\u03BD\u03B9\u03BC\u03B5\u03C2 \u03C3\u03C4\u03AE\u03BB\u03B5\u03C2: \u03A3\u0399\u03A9\u03A0\u0397\u03A4\u0397\u03A1\u0399\u039F', 'http://www.capital.gr/articles/articlesrss.asp?catid=6'),
	(u'\u039C\u03CC\u03BD\u03B9\u03BC\u03B5\u03C2 \u03C3\u03C4\u03AE\u03BB\u03B5\u03C2: \u03A0\u0399\u03A3\u03A9 \u0391\u03A0\u039F \u03A4\u0399\u03A3 \u0393\u03A1\u0391\u039C\u039C\u0395\u03A3', 'http://www.capital.gr/articles/articlesrss.asp?catid=8'),
	#(u'\u039C\u03CC\u03BD\u03B9\u03BC\u03B5\u03C2 \u03C3\u03C4\u03AE\u03BB\u03B5\u03C2: \u03A4\u0395\u03A7\u039D\u039F\u039B\u039F\u0393\u0399\u0391', 'http://www.capital.gr/news/newsrss.asp?s=-8') not working for now
]

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe


class newsbeast(BasicNewsRecipe):
    title                  = 'Newsbeast'
    __author__             = 'Stelios'
    description            = 'News from Greece'
    oldest_article         = 2
    max_articles_per_feed  = 100
    publisher              = 'newsbeast'
    category               = 'news, GR'
    language               = 'el'
    encoding               = 'utf8'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    encoding              = 'utf8'
    keep_only_tags = [
  dict(name='div', attrs={'class' : ['article-title']}),
 # dict(name='img', attrs={'class' : ['article_photo']}),
 #If enabled feeds exceede 15MB
  dict(name='div', attrs={'class' : ['txt']})
]
    remove_tags = [
  dict(name='table', attrs={'id':['artFoot']}),
  dict(name='img'),
 #If removed feeds exceede 15MB
  dict(name='p', attrs={'class':['article-details']})
]

    feeds          = [
                 (u'\u0395\u03BB\u03BB\u03AC\u03B4\u03B1', 'http://www.newsbeast.gr/feeds/greece'),
	(u'\u039A\u03CC\u03C3\u03BC\u03BF\u03C2', 'http://www.newsbeast.gr/feeds/world'),
	(u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE', 'http://www.newsbeast.gr/feeds/politiki'),
	(u'\u039F\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1', 'http://www.newsbeast.gr/feeds/financial'),
	(u'\u0391\u03B8\u03BB\u03B7\u03C4\u03B9\u03BA\u03AC', 'http://www.newsbeast.gr/feeds/sports'),
	(u'\u039A\u03BF\u03B9\u03BD\u03C9\u03BD\u03AF\u03B1', 'http://www.newsbeast.gr/feeds/society'),
	(u'\u03A0\u03B5\u03C1\u03B9\u03B2\u03AC\u03BB\u03BB\u03BF\u03BD', 'http://www.newsbeast.gr/feeds/environment'),
	(u'Media', 'http://www.newsbeast.gr/feeds/media'),
	(u'\u0394\u03B9\u03B1\u03C3\u03BA\u03AD\u03B4\u03B1\u03C3\u03B7', 'http://www.newsbeast.gr/feeds/entertainment'),
	(u'Lifestyle', 'http://www.newsbeast.gr/feeds/lifestyle'),
	(u'\u03A4\u03B5\u03C7\u03BD\u03BF\u03BB\u03BF\u03B3\u03AF\u03B1', 'http://www.newsbeast.gr/feeds/technology'),
	(u'\u0391\u03C5\u03C4\u03BF\u03BA\u03AF\u03BD\u03B7\u03C4\u03BF', 'http://www.newsbeast.gr/feeds/car'),
	(u'\u0393\u03C5\u03BD\u03B1\u03AF\u03BA\u03B1', 'http://www.newsbeast.gr/feeds/woman'),
	(u'\u03A5\u03B3\u03B5\u03AF\u03B1', 'http://www.newsbeast.gr/feeds/health'),
	(u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://www.newsbeast.gr/feeds/culture'),
	(u'\u038C,\u03C4\u03B9 \u03BD\u03B1 \u03BD\u03B1\u03B9', 'http://www.newsbeast.gr/feeds/weird')
	]

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe


class SKAI(BasicNewsRecipe):
    title                  = 'SKAI'
    __author__             = 'Stelios'
    description            = 'News from Greece'
    oldest_article         = 2
    max_articles_per_feed  = 100
    publisher              = 'skai.gr'
    category               = 'news, GR'
    language               = 'el'
    encoding               = 'utf8'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    encoding              = 'utf8'
    keep_only_tags = [
 dict(name='h1'),
  dict(name='div', attrs={'class' : ['articleText']})
]


    feeds          = [
	(u'\u039A\u03C5\u03C1\u03B9\u03CC\u03C4\u03B5\u03C1\u03B5\u03C2 \u0395\u03B9\u03B4\u03AE\u03C3\u03B5\u03B9\u03C2', 'http://feeds.feedburner.com/skai/Uulu'),
	(u'\u0395\u03BB\u03BB\u03AC\u03B4\u03B1', 'http://feeds.feedburner.com/skai/PLwa'),
	(u'\u039A\u03CC\u03C3\u03BC\u03BF\u03C2', 'http://feeds.feedburner.com/skai/aqOL'),
	(u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE','http://feeds.feedburner.com/skai/yinm'),
	(u'\u039F\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1', 'http://feeds.feedburner.com/skai/oPUt'),
	(u'\u03A4\u03B5\u03C7\u03BD\u03BF\u03BB\u03BF\u03B3\u03AF\u03B1', 'http://feeds.feedburner.com/skai/fqsg'),
	(u'\u0391\u03B8\u03BB\u03B7\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://feeds.feedburner.com/skai/TfmK'),
	(u'\u03A5\u03B3\u03B5\u03AF\u03B1', 'http://feeds.feedburner.com/skai/TABn'),
	(u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://feeds.feedburner.com/skai/ppGl'),
	(u'\u0391\u03C5\u03C4\u03BF\u03BA\u03AF\u03BD\u03B7\u03C3\u03B7', 'http://feeds.feedburner.com/skai/HCCc'),
	(u'\u03A0\u03B5\u03C1\u03B9\u03B2\u03AC\u03BB\u03BB\u03BF\u03BD', 'http://feeds.feedburner.com/skai/jVWs'),
	(u'\u03A0\u03B1\u03C1\u03AC\u03BE\u03B5\u03BD\u03B1', 'http://feeds.feedburner.com/skai/bpAR')
]

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe

class ingr(BasicNewsRecipe):
    title                  = 'in.gr'
    __author__             = 'Stelios'
    description            = 'News from Greece'
#    max_articles_per_feed  = 100
    oldest_article         = 4
    publisher              = 'in.gr'
    category               = 'news, GR'
    language               = 'el'
    encoding               = 'utf8'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    encoding              = 'utf8'
    keep_only_tags = [
  dict(name='h1'),

  dict(name='div', attrs={'id' : ['in-news-article']})
                        ]
    remove_tags = [
dict(name='em', attrs={'class' : ['credits']}),
dict(name='div', attrs={'class' : ['article-tools-hor', 'promo-banners gAds', 'main', 'article-listen-player', 'article-tools-hor-bttm', 'tools-sec', 'article-tools', 'article-listen-player-ver']})
] 


    feeds          = [
                 (u'\u0395\u03BB\u03BB\u03AC\u03B4\u03B1', 'http://rss.in.gr/feed/news/greece'),
	(u'\u0395\u03B9\u03B4\u03AE\u03C3\u03B5\u03B9\u03C2', 'http://rss.in.gr/feed/news'),
	(u'\u039A\u03CC\u03C3\u03BC\u03BF\u03C2', 'http://rss.in.gr/feed/news/world'),
	(u'\u0395\u03C0\u03B9\u03C3\u03C4\u03AE\u03BC\u03B7', 'http://rss.in.gr/feed/news/science'),
	(u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://rss.in.gr/feed/news/culture')
	]

Also feel free to request more recipes I'd be glad to help.

SteliosGero · 10-31-2011, 11:50 AM

I will keep on dumping

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe

class Tovima(BasicNewsRecipe):
    title                  = 'To Vima'
    __author__             ='Stelios'
    description            = ' News from Greece'
    #max_articles_per_feed  = 100
    oldest_article         = 3
    publisher              = 'To Vima'
    category               = 'news, GR'
    language               = 'el'
    encoding               = 'utf8'
    cover_url             = 'http://www.tovima.gr/Themes/1/Default/Media/Home//small-n-short-logo.jpg'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    extra_css       = '''
                        .article_title{font-family :Arial,Helvetica,sans-serif; font-weight: bold; font-size:large;}
                        .article_text{font-family :Arial,Helvetica,sans-serif; font-size:x-small;}
                     '''
    keep_only_tags = [

  dict(name='div', attrs={'class' : ['article_title']}),
  dict(name='div', attrs={'class' : ['article_text']})
                        ]
    remove_tags = [
 dict(name='div', attrs={'class' : ['article_cat']})
	]
    feeds          = [
                     (u'\u03C0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE', 'http://www.tovima.gr/feed/politics/'),
	(u'\u03BF\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1', 'http://www.tovima.gr/feed/finance/'),
	(u'\u03B3\u03BD\u03CE\u03BC\u03B5\u03C2', 'http://www.tovima.gr/feed/opinions/'),
	(u'blogs', 'http://www.tovima.gr/feed/blogs/'),
	(u'\u03BA\u03CC\u03C3\u03BC\u03BF\u03C2','http://www.tovima.gr/feed/world/'),
	(u'science', 'http://www.tovima.gr/feed/science/'),
	(u'\u03BA\u03BF\u03B9\u03BD\u03C9\u03BD\u03AF\u03B1', 'http://www.tovima.gr/feed/society/'),
	(u'\u03C0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://www.tovima.gr/feed/culture/'),
	(u'\u03B1\u03B8\u03BB\u03B7\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2', 'http://www.tovima.gr/feed/sports/')
]

SteliosGero · 10-31-2011, 05:20 PM

Yet another two recipes

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe

class Men24(BasicNewsRecipe):
    title          = 'Men24.gr'
    __author__             = 'Stelios'
    description            = 'Greek Mens portal'
    oldest_article = 14
    max_articles_per_feed = 100
    language = 'el'
    cover_url = 'http://www.men24.gr/ast/img/men24Logo.jpg'
    category               = 'magazines, GR'
    language               = 'el'
    encoding               = 'windows-1253'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    extra_css       = '''
                        .artPrintTitle{font-family :Arial,Helvetica,sans-serif; font-weight: bold; font-size:large;}
                        .artPrintSubtitle{font-family :Arial,Helvetica,sans-serif; font-size:x-small;}
                     '''
    remove_tags = [
  dict(name='td', attrs={'class':['artPrintCategory']}),
  dict(name='table', attrs={'class':['footer']}),
  dict(name='img')
]
    feeds = [
        (u'\u038C\u03BB\u03B5\u03C2 \u03BF\u03B9 \u03B5\u03B9\u03B4\u03AE\u03C3\u03B5\u03B9\u03C2', 'http://www.men24.gr/svc/rss/lastNews/'),
        (u'\u03A3\u03C4\u03C5\u03BB', 'http://www.men24.gr/svc/rss/categoryNews/?category=style'),
        (u'Fitness', 'http://www.men24.gr/svc/rss/categoryNews/?category=fitness'),
        (u'Gadgets', 'http://www.men24.gr/svc/rss/categoryNews/?category=gadgets'),
        (u'\u0394\u03B9\u03B1\u03C3\u03BA\u03AD\u03B4\u03B1\u03C3\u03B7', 'http://www.men24.gr/svc/rss/categoryNews/?category=fun'),
        (u'\u03A7\u03C1\u03AE\u03BC\u03B1 \u03BA\u03B1\u03B9 \u039A\u03B1\u03C1\u03B9\u03AD\u03C1\u03B1', 'http://www.men24.gr/svc/rss/categoryNews/?category=money'),
        (u'Special Edition', 'http://www.men24.gr/svc/rss/categoryNews/?category=special'),
        (u'\u0388\u03C1\u03C9\u03C4\u03B1\u03C2 \u03BA\u03B1\u03B9 Sex', 'http://www.men24.gr/svc/rss/categoryNews/?category=love'),
        (u'\u0386\u03BD\u03C4\u03C1\u03B5\u03C2 \u03C4\u03BF\u03C5 24', 'http://www.men24.gr/svc/rss/categoryNews/?category=men'),
        (u'\u0393\u03C5\u03BD\u03B1\u03AF\u03BA\u03B5\u03C2', 'http://www.men24.gr/svc/rss/categoryNews/?category=women'),
        (u'\u039F\u03B4\u03B7\u03B3\u03BF\u03AF', 'http://www.men24.gr/svc/rss/categoryNews/?category=guides'),
        (u'\u03A4\u03B6\u03CC\u03B3\u03BF\u03C2', 'http://www.men24.gr/svc/rss/categoryNews/?category=gamble')

]

    def print_version(self, url):
        return url.replace('.asp', '.print.asp')

Code:

from calibre.web.feeds.recipes import BasicNewsRecipe


class protagon(BasicNewsRecipe):
    title                  = 'Protagon'
    __author__             = 'Stelios'
    description            = 'Opinion articles in Greek'
    oldest_article         = 7
    max_articles_per_feed  = 100
    publisher              = 'Various'
    category               = 'GR'
    language               = 'el'
    encoding               = 'utf8'
    no_stylesheets         = True
    use_embedded_content  = False
    remove_empty_feeds    = True
    
    keep_only_tags = [
     dict(name='h1', attrs={'id' : ['title']}),
     dict(name='div', attrs={'class' : ['freetext']})
]

    feeds          = [
                     (u'\u0398\u03AD\u03BC\u03B1\u03C4\u03B1', 'http://www.protagon.gr/rss?i=protagon.el.8emata')	
]

kovidgoyal · 10-31-2011, 11:31 PM

Your recipes will be in the next calibre release.

SteliosGero · 11-02-2011, 03:55 PM

Code:

class sigmalive(BasicNewsRecipe):
    title          = u'SigmaLive'
    oldest_article = 7
    max_articles_per_feed = 100
    auto_cleanup = True
    category               = 'news, CY'
    description            = 'Cypriot News'
    language               = 'el'
    encoding               = 'utf8'
    feeds          = [(u'sigmalive', u'http://sigmalive.com/rss/latest')]

SteliosGero · 11-02-2011, 04:05 PM

Code:

class AdvancedUserRecipe1320264153(BasicNewsRecipe):
    title          = u'zougla'
    oldest_article = 7
    max_articles_per_feed = 100
    auto_cleanup = True

    feeds          = [(u'zougla', u'http://www.zougla.gr/ArticleRss.xml')]

SteliosGero · 12-18-2011, 11:23 AM

1 more

Code:

class spn(BasicNewsRecipe):
    title          = u'Salonica Press News'
    oldest_article = 3
    max_articles_per_feed = 100
    auto_cleanup = True
    category               = 'news, GR'
    language               = 'el'
 

    feeds          = [(u'\u03a0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ae', u'http://www.spnews.gr/politiki?format=feed&amp;type=rss'), (u'\u039f\u03b9\u03ba\u03bf\u03bd\u03bf\u03bc\u03af\u03b1', u'http://www.spnews.gr/oikonomia?format=feed&amp;type=rss'), (u'\u0391\u03c5\u03c4\u03bf\u03b4\u03b9\u03bf\u03af\u03ba\u03b7\u03c3\u03b7', u'http://www.spnews.gr/aftodioikisi?format=feed&amp;type=rss'), (u'\u039a\u03bf\u03b9\u03bd\u03c9\u03bd\u03af\u03b1', u'http://www.spnews.gr/koinonia?format=feed&amp;type=rss'), (u'\u0391\u03b8\u03bb\u03b7\u03c4\u03b9\u03c3\u03bc\u03cc\u03c2', u'http://www.spnews.gr/sports?format=feed&amp;type=rss'), (u'\u0394\u03b9\u03b5\u03b8\u03bd\u03ae', u'http://www.spnews.gr/diethni?format=feed&amp;type=rss'), (u'\u03a0\u03bf\u03bb\u03b9\u03c4\u03b9\u03c3\u03bc\u03cc\u03c2', u'http://www.spnews.gr/politismos?format=feed&amp;type=rss'), (u'Media', u'http://www.spnews.gr/media-news?format=feed&amp;type=rss'), (u'\u0396\u03c9\u03ae', u'http://www.spnews.gr/zoi?format=feed&amp;type=rss'), (u'\u03a4\u03b5\u03c7\u03bd\u03bf\u03bb\u03bf\u03b3\u03af\u03b1', u'http://spnews.gr/texnologia?format=feed&amp;type=rss'), (u'\u03a0\u03b5\u03c1\u03b9\u03b2\u03ac\u03bb\u03bb\u03bf\u03bd', u'http://spnews.gr/periballon?format=feed&amp;type=rss'), (u'\u03a0\u03b1\u03c1\u03b1\u03c0\u03bf\u03bb\u03b9\u03c4\u03b9\u03ba\u03ac', u'http://spnews.gr/parapolitika?format=feed&amp;type=rss'), (u'\u03a0\u03b1\u03c1\u03b1\u03b4\u03b7\u03bc\u03bf\u03c4\u03b9\u03ba\u03ac', u'http://spnews.gr/paradimotika?format=feed&amp;type=rss'), (u'\u03a0\u03b1\u03c1\u03b1\u03b1\u03b8\u03bb\u03b7\u03c4\u03b9\u03ba\u03ac', u'http://spnews.gr/parathlitika?format=feed&amp;type=rss'), (u'\u0391\u03c0\u03cc\u03c8\u03b5\u03b9\u03c2', u'http://spnews.gr/apopseis?format=feed&amp;type=rss'), (u'\u03a3\u03c5\u03bd\u03b5\u03cd\u03be\u03b5\u03b9\u03c2', u'http://spnews.gr/synenteykseis?format=feed&amp;type=rss'), (u'Alert!', u'http://spnews.gr/alert?format=feed&amp;type=rss')]

    def print_version(self, url):
        return url+'?tmpl=component&print=1&layout=default&page='

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
More Greek news	drketikios	Recipes	3	03-25-2011 02:53 PM
Using News Recipes: Start Here	Starson17	Recipes	0	02-14-2011 04:47 PM
Arabic News Recipes	electromikey	Recipes	0	02-12-2011 08:50 AM
Indonesian News Recipes?!	movanet	Recipes	0	11-06-2010 07:11 AM
Baseball News Recipes?	geneaber	Calibre	1	11-12-2009 05:32 PM

10-31-2011, 11:31 PM	#4
kovidgoyal creator of calibre Posts: 45,305 Karma: 27111242 Join Date: Oct 2006 Location: Mumbai, India Device: Various	Your recipes will be in the next calibre release.

11-02-2011, 04:05 PM	#6
SteliosGero Junior Member Posts: 8 Karma: 10 Join Date: Oct 2011 Device: Kindle 3g	Code: class AdvancedUserRecipe1320264153(BasicNewsRecipe): title = u'zougla' oldest_article = 7 max_articles_per_feed = 100 auto_cleanup = True feeds = [(u'zougla', u'http://www.zougla.gr/ArticleRss.xml')]

Advert

Advert