Greensboro news and record by Walt Anthony

resources/recipes/greensboro_news_and_record.recipe

@@ -0,0 +1,54 @@
+__license__   = 'GPL v3'
+__copyright__ = '2010, Walt Anthony <workshop.northpole at gmail.com>'
+'''
+www.news-record.com
+'''
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class NewsandRecord(BasicNewsRecipe):
+    title          = u'Greensboro News & Record'
+    description    = "News from Greensboro, North Carolina"
+    __author__     = 'Walt Anthony'
+    publisher             = 'News & Record and Landmark Media Enterprises, LLC'
+    category              = 'news, USA'
+    oldest_article        = 3 #days
+    max_articles_per_feed = 25
+    summary_length        = 150
+    language              = 'en'
+    encoding              = 'utf-8'
+    remove_javascript     = True
+    no_stylesheets        = True
+
+
+    conversion_options = {
+                          'comment'   : description
+                        , 'tags'      : category
+                        , 'publisher' : publisher
+                        , 'language'  : language
+                        }
+
+
+
+    remove_tags_before = dict(name='h3', attrs={'class':'nrcTxt_headline'})
+    remove_tags_after  = dict(name='div', attrs={'id':'nrcBlk_ContentBody'})
+
+    remove_tags = [
+       dict(name='iframe'),
+       dict(name=['notags','embed','object','link','img']),
+
+    ]
+
+
+    feeds = [
+        ('News', 'http://www.news-record.com/news/archive/feed'),
+        ('Greensboro News', 'http://www.news-record.com/news/greensboro/feed'),
+        ('Education', 'http://www.news-record.com/news/education/feed'),
+        ('Government', 'http://www.news-record.com/news/government/feed'),
+        ('College Sports', 'http://www.news-record.com/sports/college/feed'),
+        ('Sports Extra', 'http://www.news-record.com/blog/sportsextra/feed'),
+        ('Life', 'http://www.news-record.com/life/top/feed'),
+        ('NASCAR', 'http://www.news-record.com/sports/nascar/top/feed'),
+        ('Editorials', 'http://www.news-record.com/opinion/editorials/feed'),
+        ('Letters to the Editor', 'http://www.news-record.com/opinion/letters/feed')
+    ]
+

src/calibre/ebooks/pdf/reflow.py

@@ -146,12 +146,14 @@ class Region(object):
         self.columns = []
         self.top = self.bottom = self.left = self.right = self.width = self.height = 0
 
-    def add_columns(self, columns):
+    def add(self, columns):
         if not self.columns:
             for x in sorted(columns, cmp=lambda x,y: cmp(x.left, y.left)):
                 self.columns.append(x)
         else:
-           pass
+            for i in range(len(columns)):
+                for elem in columns[i]:
+                    self.columns[i].add(elem)
 
     def contains(self, columns):
         if not self.columns:
@@ -168,6 +170,11 @@ class Region(object):
                 return False
         return True
 
+    @property
+    def is_empty(self):
+        return len(self.elements) == 0
+
+
 class Page(object):
 
     # Fraction of a character width that two strings have to be apart,
@@ -242,19 +249,25 @@ class Page(object):
                 self.texts.remove(match)
 
     def first_pass(self):
+        'Sort page into regions and columns'
         self.regions = []
         if not self.elements:
             return
         for i, x in enumerate(self.elements):
             x.idx = i
-        self.current_region = None
+        current_region = Region()
         processed = set([])
         for x in self.elements:
             if x in processed: continue
             elems = set(self.find_elements_in_row_of(x))
             columns = self.sort_into_columns(x, elems)
             processed.update(elems)
-            columns
+            if not current_region.contains(columns):
+                self.regions.append(self.current_region)
+                current_region = Region()
+            current_region.add(columns)
+        if not self.current_region.is_empty():
+            self.regions.append(current_region)
 
     def sort_into_columns(self, elem, neighbors):
         columns = [Column()]