From 51b819f6387286d921d56eecebd0c17b644b24c0 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Sat, 15 Oct 2011 06:15:28 +0530
Subject: [PATCH] Fix 874643

---
 recipes/icons/la_republica.png | Bin 0 -> 868 bytes
 recipes/la_republica.recipe    |  56 ++++++++++++++++++---------------
 2 files changed, 30 insertions(+), 26 deletions(-)
 create mode 100644 recipes/icons/la_republica.png

diff --git a/recipes/icons/la_republica.png b/recipes/icons/la_republica.png
new file mode 100644
index 0000000000000000000000000000000000000000..e24405b6d4511c7b54d9e4b61b7bd922926527b5
GIT binary patch
literal 868
zcmV-q1DpJbP)<h;3K|Lk000e1NJLTq002M$002M;0ssI2B@5<>0009kNkl<Zc%1E8
z?U5-#6df-BB?2}AW)d(Yphy4{!H@t&0*Jt@1bjq*2oS*|0wn?Ol=-9VmN~au_jt8?
ze(kC|xBJZf?w-u`dUgJ1WPS;xj2}SY?+u#EeB7X^<Xa>Y@O^)~-M(b8VzKDDZnavi
z*Xy}l?j<^t31l*vm#p-FX0sUtL8VgB)br;Rn^D1RHiIczt=9Q`*3pwpAe+tVEbttl
zQ79DF>$S3mWCHDWTUk4U%jE)7L{X&nIF&~6hQcsB91gr*#4ExS>Fablk&jHLQ-zl$
zr~{=^iR6IY@Av!oZ+JQUgn0_Sy4h@yrpW!o5Cn0z+a)FMQ{?C8XRTIao`NrdEx(@_
z0s_jFXfm0QlCW^U-xn#jkw(EBS)UXhkH;eAmg+#G(I6$^NTQ5niGZl$&gXMp*b)JR
zHBu8PEUIhvS_!^;gBS5??Ge!L+rEXMkioL3`HW<_Tt-QH_!0SWI2@8|@w-Z`8+`)s
z=Y{PN;Kw>q#e7Esq@%{LM_{+xk&+lVd9{`Z;PsQq5uKlCT((4juT6Z<8^aO-vd2XJ
zk(LV^1khlTa~XyQsdD245WD2-$4IF#au^|ikj1{^N!s%~CVbTfV#iP`Ap%BWc#bG_
z-XrB!qSxykk4NR*JLwWY&KV2_nunxm5<r}2o+~=(5<qZ|Mk7rnDaQaxgy!Xg5d!FE
z5S@&qz7dQNa9x*gX`(37i3C$Fgxo?7R+>gI7z6xamh#!UW0U~jSEAKhEEa037$ks5
z$mjE<FnS@iRm`SQzDHQCR%-dwEP)^hcwsdom?a?cBDGL9UWxfGUoDi46EN*X#tC3b
zB=5)^rBJrf%<&b?ksHAV0fYe`%5Qo{#)q=OJ2IWd9nC5K&4uWul_h|Vm)HBXLq>DT
zS7qc=-jR|CU@J(^zVGv2JwOuj4@n*KJNdYCvlz)C;amuFx955M!ViGkVBeiiC;pfJ
zZXyOkf4N*f|9T?90-C1=h{*d(Uy%`<n<c|Ad@f^ks26>6K*IK^gwzO$NFV_J<8Ro%
u0x|F~-NOi?TAMi5{eTZ&-yFXL(#AhWQgd%jDdNEZ0000<MNUMnLSTaL5|_FF

literal 0
HcmV?d00001

diff --git a/recipes/la_republica.recipe b/recipes/la_republica.recipe
index 14dc0d0aef..e55211c223 100644
--- a/recipes/la_republica.recipe
+++ b/recipes/la_republica.recipe
@@ -1,51 +1,55 @@
-#!/usr/bin/env  python
 __license__   = 'GPL v3'
 __author__    = 'Lorenzo Vigentini, based on Darko Miletic, Gabriele Marini'
-__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>, Lorenzo Vigentini <l.vigentini at gmail.com>'
+__copyright__ = '2009-2011, Darko Miletic <darko.miletic at gmail.com>, Lorenzo Vigentini <l.vigentini at gmail.com>'
 description   = 'Italian daily newspaper - v1.01 (04, January 2010); 16.05.2010 new version'
 
 '''
 http://www.repubblica.it/
 '''
 
+import re
 from calibre.web.feeds.news import BasicNewsRecipe
 
 class LaRepubblica(BasicNewsRecipe):
-    __author__        = 'Lorenzo Vigentini, Gabriele Marini'
-    description   = 'Italian daily newspaper'
-
-    cover_url      = 'http://www.repubblica.it/images/homepage/la_repubblica_logo.gif'
-    title          = u'La Repubblica'
-    publisher      = 'Gruppo editoriale L\'Espresso'
-    category       = 'News, politics, culture, economy, general interest'
-
-    language       = 'it'
-    timefmt        = '[%a, %d %b, %Y]'
-
-    oldest_article = 5
-    max_articles_per_feed = 100
-    use_embedded_content  = False
-    recursion             = 10
-
-    remove_javascript = True
-    no_stylesheets    = True
+    title                = 'La Repubblica'
+    __author__           = 'Lorenzo Vigentini, Gabriele Marini, Darko Miletic'
+    description          = 'il quotidiano online con tutte le notizie in tempo reale. News e ultime notizie. Tutti i settori: politica, cronaca, economia, sport, esteri, scienza, tecnologia, internet, spettacoli, musica, cultura, arte, mostre, libri, dvd, vhs, concerti, cinema, attori, attrici, recensioni, chat, cucina, mappe. Le citta di Repubblica: Roma, Milano, Bologna, Firenze, Palermo, Napoli, Bari, Torino.'
+    masthead_url         = 'http://www.repubblica.it/static/images/homepage/2010/la-repubblica-logo-home-payoff.png'
+    publisher            = 'Gruppo editoriale L\'Espresso'
+    category             = 'News, politics, culture, economy, general interest'
+    language             = 'it'
+    timefmt              = '[%a, %d %b, %Y]'
+    oldest_article       = 5
+    encoding             = 'utf8'
+    use_embedded_content = False
+    #recursion           = 10
+    no_stylesheets       = True
+    extra_css            = """
+                              img{display: block}
+                           """
 
+    preprocess_regexps = [
+        (re.compile(r'.*?<head>', re.DOTALL|re.IGNORECASE), lambda match: '<head>'),
+        (re.compile(r'<head>.*?<title>', re.DOTALL|re.IGNORECASE), lambda match: '<head><title>'),
+        (re.compile(r'</title>.*?</head>', re.DOTALL|re.IGNORECASE), lambda match: '</title></head>')
+    ]
+    
     def get_article_url(self, article):
         link = article.get('id', article.get('guid', None))
         if link is None:
             return article
         return link
-
-    keep_only_tags     = [dict(name='div', attrs={'class':'articolo'}),
-                          dict(name='div', attrs={'class':'body-text'}),
-#                          dict(name='div', attrs={'class':'page-content'}),
+    
+    keep_only_tags     = [
+                          dict(attrs={'class':'articolo'}),
+                          dict(attrs={'class':'body-text'}),
                           dict(name='p', attrs={'class':'disclaimer clearfix'}),
-                          dict(name='div', attrs={'id':'contA'})
+                          dict(attrs={'id':'contA'})
                          ]
 
 
     remove_tags        = [
-                            dict(name=['object','link']),
+                            dict(name=['object','link','meta']),
                             dict(name='span',attrs={'class':'linkindice'}),
                             dict(name='div', attrs={'class':'bottom-mobile'}),
                             dict(name='div', attrs={'id':['rssdiv','blocco']}),