New recipes by Darko Miletic for Cuban, Brasilian, Chilean and Montenegran news sources

src/calibre/web/feeds/recipes/__init__.py

@@ -24,7 +24,9 @@ recipe_modules = ['recipe_' + r for r in (
            'joelonsoftware', 'telepolis', 'common_dreams', 'nin', 'tomshardware_de',
            'pagina12', 'infobae', 'ambito', 'elargentino', 'sueddeutsche', 'the_age',
            'laprensa', 'amspec', 'freakonomics', 'criticadigital', 'elcronista',
-           'shacknews', 'teleread',
+           'shacknews', 'teleread', 'granma', 'juventudrebelde', 'juventudrebelde_english',
+           'la_tercera', 'el_mercurio_chile', 'la_cuarta', 'lanacion_chile', 'la_segunda',
+           'jb_online', 'estadao', 'o_globo', 'vijesti',
           )]
 
 import re, imp, inspect, time, os

src/calibre/web/feeds/recipes/recipe_el_mercurio_chile.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+emol.com
+'''
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class ElMercurio(BasicNewsRecipe):
+    title                 = 'El Mercurio online'
+    __author__            = 'Darko Miletic'
+    description           = 'El sitio de noticias online de Chile'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://www.emol.com/especiales/logo_emol/logo_emol.gif'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Chile'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [  
+                        dict(name='div', attrs={'class':'despliegue-txt_750px'})
+                       ,dict(name='div', attrs={'id':'div_cuerpo_participa'})
+                     ]
+                        
+
+    remove_tags = [  
+                     dict(name='div', attrs={'class':'contenedor_despliegue-col-left300'})
+                    ,dict(name='div', attrs={'id':['div_centro_dn_opc','div_cabezera','div_secciones','div_contenidos','div_pie','nav']})
+                    ]
+                        
+    feeds = [  
+               (u'Noticias de ultima hora', u'http://www.emol.com/rss20/rss.asp?canal=0')
+              ,(u'Nacional', u'http://www.emol.com/rss20/rss.asp?canal=1')
+              ,(u'Mundo', u'http://www.emol.com/rss20/rss.asp?canal=2')
+              ,(u'Deportes', u'http://www.emol.com/rss20/rss.asp?canal=4')
+              ,(u'Magazine', u'http://www.emol.com/rss20/rss.asp?canal=6')
+              ,(u'Tecnologia', u'http://www.emol.com/rss20/rss.asp?canal=5')
+              ,(u'La Musica', u'http://www.emol.com/rss20/rss.asp?canal=7')
+            ]
+        

src/calibre/web/feeds/recipes/recipe_estadao.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+estadao.com.br
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+
+class Estadao(BasicNewsRecipe):
+    title                 = 'O Estado de S. Paulo'
+    __author__            = 'Darko Miletic'
+    description           = 'News from Brasil'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'utf8'
+    cover_url             = 'http://www.estadao.com.br/img/logo_estadao.png'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Brasil'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'id':'c1'})]
+
+    remove_tags = [
+                     dict(name=['script','object','form','ul'])
+                    ,dict(name='div', attrs={'id':['votacao','estadaohoje']})
+                    ,dict(name='p', attrs={'id':'ctrl_texto'})
+                    ,dict(name='p', attrs={'class':'texto'})
+                  ]
+    
+    feeds = [
+               (u'Manchetes Estadao', u'http://www.estadao.com.br/rss/manchetes.xml')
+              ,(u'Ultimas noticias', u'http://www.estadao.com.br/rss/ultimas.xml')
+              ,(u'Nacional', u'http://www.estadao.com.br/rss/nacional.xml')
+              ,(u'Internacional', u'http://www.estadao.com.br/rss/internacional.xml')
+              ,(u'Cidades', u'http://www.estadao.com.br/rss/cidades.xml')
+              ,(u'Esportes', u'http://www.estadao.com.br/rss/esportes.xml')
+              ,(u'Arte & Lazer', u'http://www.estadao.com.br/rss/arteelazer.xml')
+              ,(u'Economia', u'http://www.estadao.com.br/rss/economia.xml')
+              ,(u'Vida &', u'http://www.estadao.com.br/rss/vidae.xml')
+            ]
+
+    def preprocess_html(self, soup):
+        ifr = soup.find('iframe')
+        if ifr:
+           ifr.extract()
+        return soup

src/calibre/web/feeds/recipes/recipe_granma.py

@@ -0,0 +1,42 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2008, Darko Miletic <darko.miletic at gmail.com>'
+'''
+granma.cubaweb.cu
+'''
+import urllib
+
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class Granma(BasicNewsRecipe):
+    title                 = 'Diario Granma'
+    __author__            = 'Darko Miletic'
+    description           = 'Organo oficial del Comite Central del Partido Comunista de Cuba'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://www.granma.cubaweb.cu/imagenes/granweb229d.jpg'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Cuba'
+                        , '--publisher'     , title
+                        , '--ignore-tables'
+                        ]
+
+    keep_only_tags = [dict(name='table', attrs={'height':'466'})]
+
+    feeds = [(u'Noticias', u'http://www.granma.cubaweb.cu/noticias.xml' )]
+
+    
+    def preprocess_html(self, soup):
+        del soup.body.table['style']
+        rtag = soup.find('td', attrs={'height':'458'})
+        if rtag:
+            del rtag['style']
+        return soup
+    

src/calibre/web/feeds/recipes/recipe_jb_online.py

@@ -0,0 +1,41 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+jbonline.terra.com.br
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class JBOnline(BasicNewsRecipe):
+    title                 = 'Jornal Brasileiro Online'
+    __author__            = 'Darko Miletic'
+    description           = 'News from Brasil'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://jbonline.terra.com.br/img/logo_01.gif'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Brasil'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'id':'corpoNoticia'})]
+
+    remove_tags = [dict(name=['script','object','form'])]
+    
+    feeds = [(u'Todos as editorias', u'http://jbonline.terra.com.br/extra/rsstrjb.xml')]
+
+    def preprocess_html(self, soup):
+        ifr = soup.find('iframe')
+        if ifr:
+           ifr.extract()
+        item = soup.find('div', attrs={'id':'corpoNoticia'})
+        if item:
+           del item['style']
+        return soup

src/calibre/web/feeds/recipes/recipe_juventudrebelde.py

@@ -0,0 +1,42 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2008, Darko Miletic <darko.miletic at gmail.com>'
+'''
+juventudrebelde.cu
+'''
+from calibre import strftime
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class Juventudrebelde(BasicNewsRecipe):
+    title                 = 'Juventud Rebelde'
+    __author__            = 'Darko Miletic'
+    description           = 'Diario de la Juventud Cubana'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = strftime('http://www.juventudrebelde.cu/UserFiles/File/impreso/iportada-%Y-%m-%d.jpg')
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Cuba'
+                        , '--publisher'     , title
+                        , '--ignore-tables'
+                        ]
+
+    keep_only_tags = [dict(name='div', attrs={'id':'noticia'})]
+
+    feeds = [
+               (u'Generales', u'http://www.juventudrebelde.cu/rss/generales.php' )
+              ,(u'Cuba', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=cuba' )
+              ,(u'Internacionales', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=internacionales' )
+              ,(u'Opinion', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=opinion' )
+              ,(u'Cultura', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=cultura' )
+              ,(u'Deportes', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=deportes' )
+              ,(u'Lectura', u'http://www.juventudrebelde.cu/rss/generales.php?seccion=lectura' )
+            ]
+
+            

src/calibre/web/feeds/recipes/recipe_juventudrebelde_english.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2008, Darko Miletic <darko.miletic at gmail.com>'
+'''
+juventudrebelde.co.cu
+'''
+from calibre import strftime
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class Juventudrebelde_english(BasicNewsRecipe):
+    title                 = 'Juventud Rebelde in english'
+    __author__            = 'Darko Miletic'
+    description           = 'The newspaper of Cuban Youth'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'iso-8859-1'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Cuba'
+                        , '--publisher'     , title
+                        , '--ignore-tables'
+                        ]
+
+    keep_only_tags = [dict(name='div', attrs={'class':'read'})]
+
+    feeds = [(u'All news', u'http://www.juventudrebelde.cip.cu/rss/all/' )]
+
+            

src/calibre/web/feeds/recipes/recipe_la_cuarta.py

@@ -0,0 +1,42 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+lacuarta.cl
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class LaCuarta(BasicNewsRecipe):
+    title                 = 'La Cuarta'
+    __author__            = 'Darko Miletic'
+    description           = 'El sitio de noticias online de Chile'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Chile'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'class':'articulo desplegado'}) ]
+
+    remove_tags = [  
+                     dict(name='script')
+                    ,dict(name='ul')
+                    ,dict(name='div', attrs={'id':['toolbox','articleImageDisplayer','enviarAmigo']})
+                    ,dict(name='div', attrs={'class':['par ad-1','par ad-2']})
+                    ,dict(name='input')
+                    ,dict(name='p', attrs={'id':['mensajeError','mensajeEnviandoNoticia','mensajeExito']})
+                    ,dict(name='strong', text='PUBLICIDAD')
+                  ]
+
+    
+    feeds = [(u'Noticias', u'http://lacuarta.cl/app/rss?sc=TEFDVUFSVEE=')]
+
+    

src/calibre/web/feeds/recipes/recipe_la_segunda.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+lasegunda.com
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class LaSegunda(BasicNewsRecipe):
+    title                 = 'La Segunda'
+    __author__            = 'Darko Miletic'
+    description           = 'El sitio de noticias online de Chile'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://www.lasegunda.com/imagenes/logotipo_lasegunda_Oli.gif'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Chile'
+                        , '--publisher'     , title
+                        , '--ignore-tables'
+                        ]
+                        
+    keep_only_tags = [dict(name='table')]
+                        
+    feeds = [ 
+               (u'Noticias de ultima hora', u'http://www.lasegunda.com/rss20/index.asp?canal=0')
+              ,(u'Politica', u'http://www.lasegunda.com/rss20/index.asp?canal=21')
+              ,(u'Cronica', u'http://www.lasegunda.com/rss20/index.asp?canal=20')
+              ,(u'Internacional', u'http://www.lasegunda.com/rss20/index.asp?canal=23')
+              ,(u'Deportes', u'http://www.lasegunda.com/rss20/index.asp?canal=24')
+              ,(u'Epectaculos/Cultura', u'http://www.lasegunda.com/rss20/index.asp?canal=25')
+              ,(u'Educacion', u'http://www.lasegunda.com/rss20/index.asp?canal=26')
+              ,(u'Ciencia y Tecnologia', u'http://www.lasegunda.com/rss20/index.asp?canal=27')
+              ,(u'Solidaridad', u'http://www.lasegunda.com/rss20/index.asp?canal=28')
+              ,(u'Buena Vida', u'http://www.lasegunda.com/rss20/index.asp?canal=32')
+            ]
+
+    def print_version(self, url):
+        rest, sep, article_id = url.partition('index.asp?idnoticia=')        
+        return u'http://www.lasegunda.com/edicionOnline/include/secciones/_detalle_impresion.asp?idnoticia=' + article_id
+    

src/calibre/web/feeds/recipes/recipe_la_tercera.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+latercera.com
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class LaTercera(BasicNewsRecipe):
+    title                 = 'La Tercera'
+    __author__            = 'Darko Miletic'
+    description           = 'El sitio de noticias online de Chile'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Chile'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'class':'span-16 articulo border'}) ]
+
+    remove_tags = [  
+                     dict(name='script')
+                    ,dict(name='ul')
+                    ,dict(name='div', attrs={'id':['boxComentarios','shim','enviarAmigo']})
+                    ,dict(name='div', attrs={'class':['ad640','span-10 imgSet A','infoRelCol']})
+                    ,dict(name='input')
+                    ,dict(name='p', attrs={'id':['mensajeError','mensajeEnviandoNoticia','mensajeExito']})
+                  ]
+
+    
+    feeds = [ 
+               (u'Noticias de ultima hora', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&ul=1')
+              ,(u'Pais', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=654')
+              ,(u'Mundo', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=678')
+              ,(u'Deportes', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=656')
+              ,(u'Negocios', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=655')
+              ,(u'Entretenimiento', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=661')
+              ,(u'Motores', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=665')
+              ,(u'Tendencias', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=659')
+              ,(u'Estilo', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=660')
+              ,(u'Educacion', u'http://www.latercera.com/app/rss?sc=TEFURVJDRVJB&category=657')
+            ]
+
+    

src/calibre/web/feeds/recipes/recipe_lanacion_chile.py

@@ -0,0 +1,44 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+lanacion.cl
+'''
+import urllib
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class LaNacionChile(BasicNewsRecipe):
+    title                 = 'La Nacion Chile'
+    __author__            = 'Darko Miletic'
+    description           = 'El sitio de noticias online de Chile'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://www.lanacion.cl/prontus_noticias_v2/imag/site/logo.gif'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Chile'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'class':'bloque'})]
+                        
+    feeds = [(u'Noticias', u'http://www.lanacion.cl/rss.xml')]
+
+    def print_version(self, url):
+        toprint = urllib.quote(url,':/')
+        return u'http://www.lanacion.cl/cgi-bx/imprimir.cgi?_URL=' + toprint
+
+    def preprocess_html(self, soup):
+        del soup.body['onload']
+        soup.head.base.extract()
+        item = soup.find('a', attrs={'href':'javascript:window.close()'})
+        if item:
+           item.extract()
+        return soup
+    

src/calibre/web/feeds/recipes/recipe_o_globo.py

@@ -0,0 +1,57 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
+'''
+oglobo.globo.com
+'''
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class OGlobo(BasicNewsRecipe):
+    title                 = 'O Globo'
+    __author__            = 'Darko Miletic'
+    description           = 'News from Brasil'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1252'
+    cover_url             = 'http://oglobo.globo.com/_img/o-globo.png'
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Brasil'
+                        , '--publisher'     , title
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'id':'ltintb'})]
+
+    remove_tags = [  
+                     dict(name='script')
+                    ,dict(name='object')
+                    ,dict(name='form')
+                    ,dict(name='div', attrs={'id':['linksPatGoogle','rdpm','cor','com','env','rcm_st']})
+                    ,dict(name='div', attrs={'class':'box-zap-anu2'})
+                    ,dict(name='a')
+                    ,dict(name='link')
+                  ]
+
+    
+    feeds = [
+               (u'Todos os canais', u'http://oglobo.globo.com/rss/plantao.xml')
+              ,(u'Ciencia', u'http://oglobo.globo.com/rss/plantaociencia.xml')
+              ,(u'Educacao', u'http://oglobo.globo.com/rss/plantaoeducacao.xml')
+              ,(u'Opiniao', u'http://oglobo.globo.com/rss/plantaoopiniao.xml')
+              ,(u'Sao Paulo', u'http://oglobo.globo.com/rss/plantaosaopaulo.xml')
+              ,(u'Viagem', u'http://oglobo.globo.com/rss/plantaoviagem.xml')
+              ,(u'Cultura', u'http://oglobo.globo.com/rss/plantaocultura.xml')
+              ,(u'Esportes', u'http://oglobo.globo.com/rss/plantaoesportes.xml')
+              ,(u'Mundo', u'http://oglobo.globo.com/rss/plantaomundo.xml')
+              ,(u'Pais', u'http://oglobo.globo.com/rss/plantaopais.xml')
+              ,(u'Rio', u'http://oglobo.globo.com/rss/plantaorio.xml')
+              ,(u'Saude', u'http://oglobo.globo.com/rss/plantaosaude.xml')
+              ,(u'Viver Melhor', u'http://oglobo.globo.com/rss/plantaovivermelhor.xml')
+              ,(u'Economia', u'http://oglobo.globo.com/rss/plantaoeconomia.xml')
+              ,(u'Tecnologia', u'http://oglobo.globo.com/rss/plantaotecnologia.xml')
+            ]

src/calibre/web/feeds/recipes/recipe_vijesti.py

@@ -0,0 +1,42 @@
+#!/usr/bin/env  python
+
+__license__   = 'GPL v3'
+__copyright__ = '2008, Darko Miletic <darko.miletic at gmail.com>'
+
+'''
+vijesti.cg.yu
+'''
+
+import string,re
+
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class Vijesti(BasicNewsRecipe):
+    title                 = 'Vijesti'
+    __author__            = 'Darko Miletic'
+    description           = 'News from Montenegro'    
+    oldest_article        = 2
+    max_articles_per_feed = 100
+    no_stylesheets        = True
+    use_embedded_content  = False
+    encoding              = 'cp1250'
+    cover_url             = 'http://www.vijesti.cg.yu/img/logo.gif'
+
+    preprocess_regexps = [(re.compile(u'\u0110'), lambda match: u'\u00D0')]
+
+    html2lrf_options = [
+                          '--comment'       , description
+                        , '--category'      , 'news, Montenegro'
+                        , '--publisher'     , 'Daily Press Vijesti'
+                        ]
+                        
+    keep_only_tags = [dict(name='div', attrs={'id':'mainnews'})]
+
+    feeds = [(u'Sve vijesti', u'http://www.vijesti.cg.yu/rss.php' )]
+
+    def preprocess_html(self, soup):
+        soup.html['xml:lang'] = 'sr-Latn-ME'
+        soup.html['lang']     = 'sr-Latn-ME'
+        mtag = '<meta http-equiv="Content-Language" content="sr-Latn-ME"/>'
+        soup.head.insert(0,mtag)
+        return soup