Update Dilema Veche

Merge branch 'patch-1' of https://github.com/mapopescu/calibre

recipes/dilemaveche.recipe

@@ -2,10 +2,11 @@
 #!/usr/bin/env  python
 
 __license__   = 'GPL v3'
-__copyright__ = u'2011, Silviu Cotoar\u0103'
+__copyright__ = u'2014-01-09, Silviu Cotoar\u0103, Marius Popescu'
 '''
 dilemaveche.ro
 '''
+
 from calibre.web.feeds.news import BasicNewsRecipe
 
 class DilemaVeche(BasicNewsRecipe):
@@ -15,78 +16,110 @@ class DilemaVeche(BasicNewsRecipe):
     description           = '"Sint vechi, domnule!" (I.L. Caragiale)'
     publisher             = 'Adevarul Holding'
     oldest_article        = 7
-    max_articles_per_feed = 200
-    encoding              = 'utf8'
     language              = 'ro'
+    max_articles_per_feed = 150
+    encoding              = 'utf-8'
+    simultaneous_downloads = 5
     masthead_url = 'http://www.dilemaveche.ro/sites/all/themes/dilema/theme/dilema_two/layouter/dilema_two_homepage/logo.png'
+    needs_subscription    = True
+    use_embedded_content  = False
     publication_type      = 'magazine'
+    remove_javascript     = True
+    no_stylesheets        = True
+    remove_empty_feeds    = True
     feeds = [    
+                #('Editoriale şi opinii', 'http://www.dilemaveche.ro/taxonomy/term/1/0/feed'),
                 ('Editoriale si opinii - Situatiunea', 'http://www.dilemaveche.ro/taxonomy/term/37/0/feed'),
                 ('Editoriale si opinii - Pe ce lume traim', 'http://www.dilemaveche.ro/taxonomy/term/38/0/feed'),
                 ('Editoriale si opinii - Bordeie si obiceie', 'http://www.dilemaveche.ro/taxonomy/term/44/0/feed'),
-                ('Editoriale si opinii - Talc Show', 'http://www.dilemaveche.ro/taxonomy/term/44/0/feed'),
+                ('Editoriale si opinii - Talc Show', 'http://www.dilemaveche.ro/taxonomy/term/39/0/feed'),
                 ('Tema saptamanii', 'http://www.dilemaveche.ro/taxonomy/term/19/0/feed'),
+                #('La zi în cultură', 'http://www.dilemaveche.ro/taxonomy/term/2/0/feed'),
                 ('La zi in cultura - Dilema va recomanda', 'http://www.dilemaveche.ro/taxonomy/term/58/0/feed'),
                 ('La zi in cultura - Carte', 'http://www.dilemaveche.ro/taxonomy/term/14/0/feed'),
                 ('La zi in cultura - Film', 'http://www.dilemaveche.ro/taxonomy/term/13/0/feed'),
                 ('La zi in cultura - Muzica', 'http://www.dilemaveche.ro/taxonomy/term/1341/0/feed'),
                 ('La zi in cultura - Arte performative', 'http://www.dilemaveche.ro/taxonomy/term/1342/0/feed'),
                 ('La zi in cultura - Arte vizuale', 'http://www.dilemaveche.ro/taxonomy/term/1512/0/feed'),
+                #('Societate', 'http://www.dilemaveche.ro/taxonomy/term/3/0/feed'),
                 ('Societate - Ieri cu vedere spre azi', 'http://www.dilemaveche.ro/taxonomy/term/15/0/feed'),
                 ('Societate - Din polul opus', 'http://www.dilemaveche.ro/taxonomy/term/41/0/feed'),
                 ('Societate - Mass comedia', 'http://www.dilemaveche.ro/taxonomy/term/43/0/feed'),
                 ('Societate - La singular si la plural', 'http://www.dilemaveche.ro/taxonomy/term/42/0/feed'),
+                #('Oameni şi idei', 'http://www.dilemaveche.ro/taxonomy/term/4/0/feed'),
                 ('Oameni si idei - Educatie', 'http://www.dilemaveche.ro/taxonomy/term/46/0/feed'),
                 ('Oameni si idei - Polemici si dezbateri', 'http://www.dilemaveche.ro/taxonomy/term/48/0/feed'),
-                ('Oameni si idei - Stiinta si tehnologie', 'http://www.dilemaveche.ro/taxonomy/term/46/0/feed'),
+                ('Oameni si idei - Stiinta si tehnologie', 'http://www.dilemaveche.ro/taxonomy/term/47/0/feed'),
-                ('Dileme on-line', 'http://www.dilemaveche.ro/taxonomy/term/005/0/feed')
+                ('Dileme on-line', 'http://www.dilemaveche.ro/taxonomy/term/5/0/feed'), #online only articles
+                ('Dilemateca', 'http://dilemaveche.ro/taxonomy/term/21/0/feed'), #once per month, 6-7 day of the month
+                ('Dilematix', 'http://dilemaveche.ro/taxonomy/term/20/0/feed'), #children, once-twice per year
+                ('Dilema Studiilor Postuniversitare', 'http://dilemaveche.ro/taxonomy/term/1635/0/feed') #once per year, July
                  ]
     remove_tags_before = dict(name='div',attrs={'class':'spacer_10'})
     remove_tags = [
-        dict(name='div', attrs={'class':'art_related_left'}),
+             dict(name='div', attrs={'id':['adshop_widget_428x60']}),
-        dict(name='div', attrs={'class':'controale'}),
+             dict(name='div', attrs={'id':['gallery']}),
+             dict(name='div', attrs={'class':['art_related_left']}),
+             #dict(name='div', attrs={'class':['box']}),
+             dict(name='a', attrs={'class':['prevPage']}),
+             dict(name='a', attrs={'class':['nextPage']}),
+             dict(name='div', attrs={'class':['article_details']}),
+             dict(name='div', attrs={'id':['comments']}),
+             dict(name='ul', attrs={'class':['social-buttons-list']}),
+             dict(name='a', attrs={'class':['editie']}),
              dict(name='div', attrs={'class':'simple_overlay'}),
+             dict(name='div', attrs={'class':'c_right_column'}),
+             dict(name='div', attrs={'id':'content_right'}),
+             dict(name='div', attrs={'class':'box_shadow_top'}),
+             dict(name='div', attrs={'class':'box_shadow_bottom'}),
+             dict(name='div', attrs={'id':['footer']}),
+             dict(name='div', attrs={'class':['clear spacer_20']}),
+             dict(name='div', attrs={'id':['adh-footer']}),
+             dict(name='div', attrs={'id':['skyright']}),
+             dict(name='div', attrs={'id':['closure']})
+             ]
+    remove_tags_after = [
+              dict(name='div', attrs={'id':['adshop_widget_428x60']})    
             ]
-    remove_tags_after = [dict(id='facebookLike')]
-    remove_javascript = True
-    no_stylesheets        = True
-    remove_empty_feeds = True
     extra_css             = """
         body{font-family: Georgia,Times,serif }
         img{margin-bottom: 0.4em; display:block}
                             """
-    needs_subscription = 'optional'
-    cover_margins = (10, 15, '#ffffff')
-
     def get_browser(self):
         br = BasicNewsRecipe.get_browser(self)
         if self.username is not None and self.password is not None:
-            br.open('http://dilemaveche.ro/user/login')
+            br.open('http://pay.dilemaveche.ro/autentificare/?redirect=http%3A%2F%2Fdilemaveche.ro%2F%2F&return=true')
-            br.select_form(nr=0)
+            br.select_form(nr = 0)
             br['username']   = self.username
             br['password'] = self.password
             br.submit()
         return br
 
+    def preprocess_html(self, soup):
+        return self.adeify_images(soup)
+
     def get_cover_url(self):
+        # small, from the current number article: http://dilemaveche.ro/sites/default/files/imagecache/articol_teaser/DV517web-1_copy.JPG
+        # medium, from the homepage PDF link: http://dilemaveche.ro/sites/default/files/imagecache/editie_small/DV517web-1_copy_0.JPG
+        # big, from the current number article, click on the samll image: http://dilemaveche.ro/sites/default/files/imagecache/image_gallery_large/DV517web-1_copy.JPG
         cover_url = None
         soup = self.index_to_soup('http://dilemaveche.ro')
         link_item = soup.find('div',attrs={'class':'box_dr_pdf_picture'})
         if link_item and link_item.a:
            cover_url = link_item.a['href']
-        br = BasicNewsRecipe.get_browser()
+        br = BasicNewsRecipe.get_browser(self)
         try:
             br.open(cover_url) 
-        except:  # daca nu gaseste pdf-ul
+        except: #daca nu gaseste pdf-ul
             self.log("\nPDF indisponibil")
             link_item = soup.find('div',attrs={'class':'box_dr_pdf_picture'})
             if link_item and link_item.img:
                 cover_url = link_item.img['src']
-            br = BasicNewsRecipe.get_browser()
+            br = BasicNewsRecipe.get_browser(self)
             try:
                  br.open(cover_url)
-            except:  # daca nu gaseste nici imaginea mica mica
+            except: #daca nu gaseste nici imaginea mica mica
-                print('Mama lor de nenorociti! nu este nici pdf nici imagine')
+                print('nu este nici pdf nici imagine')
                 cover_url ='http://www.dilemaveche.ro/sites/all/themes/dilema/theme/dilema_two/layouter/dilema_two_homepage/logo.png'
         return cover_url
-
+    cover_margins = (10, 15, '#ffffff')