merge from trunk

This commit is contained in:
ldolse 2011-01-19 14:36:11 +08:00
commit 9fcf7f84c3
3 changed files with 25 additions and 15 deletions

View File

@ -7,22 +7,29 @@ class DallasNews(BasicNewsRecipe):
max_articles_per_feed = 25 max_articles_per_feed = 25
no_stylesheets = True no_stylesheets = True
remove_tags_before = dict(name='h2', attrs={'class':'vitstoryheadline'}) use_embedded_content = False
remove_tags_after = dict(name='div', attrs={'style':'width: 100%; clear: right'}) remove_tags_before = dict(name='h1')
remove_tags_after = dict(name='div', attrs={'id':'article_tools_bottom'}) keep_only_tags = {'class':lambda x: x and 'article' in x}
remove_tags = [ remove_tags = [
dict(name='iframe'), {'class':['DMNSocialTools', 'article ', 'article first ', 'article premium']},
dict(name='div', attrs={'class':'biblockmore'}),
dict(name='div', attrs={'style':'width: 100%; clear: right'}),
dict(name='div', attrs={'id':'article_tools_bottom'}),
#dict(name='ul', attrs={'class':'articleTools'}),
] ]
feeds = [ feeds = [
('Latest News', 'http://www.dallasnews.com/newskiosk/rss/dallasnewslatestnews.xml'), ('Local News',
('Local News', 'http://www.dallasnews.com/newskiosk/rss/dallasnewslocalnews.xml'), 'http://www.dallasnews.com/news/politics/local-politics/?rss'),
('Nation and World', 'http://www.dallasnews.com/newskiosk/rss/dallasnewsnationworld.xml'), ('National Politics',
('Politics', 'http://www.dallasnews.com/newskiosk/rss/dallasnewsnationalpolitics.xml'), 'http://www.dallasnews.com/news/politics/national-politic/?rss'),
('Science', 'http://www.dallasnews.com/newskiosk/rss/dallasnewsscience.xml'), ('State Politics',
'http://www.dallasnews.com/news/politics/state-politics/?rss'),
('Religion',
'http://www.dallasnews.com/news/religion/?rss'),
('Crime',
'http://www.dallasnews.com/news/crime/headlines/?rss'),
('Celebrity News',
'http://www.dallasnews.com/entertainment/celebrity-news/?rss&listname=TopStories'),
('Nation',
'http://www.dallasnews.com/news/nation-world/nation/?rss'),
('World',
'http://www.dallasnews.com/news/nation-world/world/?rss'),
] ]

View File

@ -459,7 +459,7 @@ class HTMLPreProcessor(object):
try: try:
search_re = re.compile(search_pattern) search_re = re.compile(search_pattern)
replace_txt = getattr(self.extra_opts, replace, '') replace_txt = getattr(self.extra_opts, replace, '')
if replace_txt == None: if not replace_txt:
replace_txt = '' replace_txt = ''
rules.insert(0, (search_re, replace_txt)) rules.insert(0, (search_re, replace_txt))
except Exception as e: except Exception as e:

View File

@ -17,11 +17,14 @@
<item> <item>
<widget class="QLabel" name="label"> <widget class="QLabel" name="label">
<property name="text"> <property name="text">
<string>&lt;b&gt;Heuristic processing&lt;/b&gt; means that calibre will scan your book for common patterns and fix them. As the name implies, this involves guesswork, which means that it could end up worsening the result of a conversion, if calibre guesses wrong. Therefore, it is disabled by default. Often, if a conversion does not turn out as you expect, turning on heuristics can improve matters.</string> <string>&lt;b&gt;Heuristic processing&lt;/b&gt; means that calibre will scan your book for common patterns and fix them. As the name implies, this involves guesswork, which means that it could end up worsening the result of a conversion, if calibre guesses wrong. Therefore, it is disabled by default. Often, if a conversion does not turn out as you expect, turning on heuristics can improve matters. Read more about the various heuristic processing options in the &lt;a href=&quot;http://calibre-ebook.com/user_manual/conversion.html#heuristic-processing&quot;&gt;User Manual&lt;/a&gt;.</string>
</property> </property>
<property name="wordWrap"> <property name="wordWrap">
<bool>true</bool> <bool>true</bool>
</property> </property>
<property name="openExternalLinks">
<bool>true</bool>
</property>
</widget> </widget>
</item> </item>
<item> <item>