diff --git a/resources/recipes/cnn.recipe b/resources/recipes/cnn.recipe
index 868f3f3641..21c4139d6e 100644
--- a/resources/recipes/cnn.recipe
+++ b/resources/recipes/cnn.recipe
@@ -16,44 +16,60 @@ class CNN(BasicNewsRecipe):
no_stylesheets = True
use_embedded_content = False
oldest_article = 15
-
+ recursions = 1
+ match_regexps = [r'http://sportsillustrated.cnn.com/.*/[1-9].html']
+
extra_css = '''
- h1{font-family :Arial,Helvetica,sans-serif; font-size:large}
+ .cnn_strycntntlft{font-family :Arial,Helvetica,sans-serif;}
h2{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
.cnnTxtCmpnt{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
- .cnnTMcontent{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;color:#575757}
- .storytext{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
- .storybyline{font-family :Arial,Helvetica,sans-serif; font-size:xx-small; color:#575757}
+ .cnnTMcontent{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#575757}
+ .storytext{font-family :Arial,Helvetica,sans-serif; font-size:small}
+ .storybyline{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#575757}
.credit{font-family :Arial,Helvetica,sans-serif; font-size:xx-small; color:#575757}
- .storyBrandingBanner{font-family :Arial,Helvetica,sans-serif; font-size:xx-small; color:#575757}
- .storytimestamp{font-family :Arial,Helvetica,sans-serif; font-size:xx-small; color:#575757}
- .timestamp{font-family :Arial,Helvetica,sans-serif; font-size:xx-small; color:#575757}
- .subhead p{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;}
- .cnnStoryContent{font-family :Arial,Helvetica,sans-serif; font-size:xx-small}
- .cnnContentContainer{font-family :Arial,Helvetica,sans-serif; font-size:xx-small}
+ .storyBrandingBanner{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#575757}
+ .storytimestamp{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#575757}
+ .timestamp{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#575757}
+ .cnn_strytmstmp{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
+ .cnn_stryimg640caption{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
+ .cnn_strylccimg300cntr{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
+ .cnn_stryichgfcpt{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
+ .cnnByline{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
+ .cnn_bulletbin cnnStryHghLght{ font-size:xx-small;}
+ .subhead p{font-family :Arial,Helvetica,sans-serif; font-size:x-small;}
+ .cnnStoryContent{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
+ .cnnContentContainer{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
.col1{font-family :Arial,Helvetica,sans-serif; font-size:x-small; color:#666666;}
.col3{color:#333333; font-family :Arial,Helvetica,sans-serif; font-size:x-small;font-weight:bold;}
- .cnnInlineT1Caption{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;font-weight:bold;}
- .cnnInlineT1Credit{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;color:#333333;}
- .col10{color:#5A637E}
- .cnnTimeStamp{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;color:#333333;}
- .galleryhedDek{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;color:#575757;}
- .galleryWidgetHeader{font-family :Arial,Helvetica,sans-serif; font-size:xx-small;color:#004276;}
- .article-content{font-family :Arial,Helvetica,sans-serif; font-size:xx-small}
- .cnnRecapStory{font-family :Arial,Helvetica,sans-serif; font-size:xx-small}
+ .cnnInlineT1Caption{font-family :Arial,Helvetica,sans-serif; font-size:x-small;font-weight:bold;}
+ .cnnInlineT1Credit{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#333333;}
+ .col10{color:#5A637E;}
+ .cnnInlineRailBulletList{color:black;}
+ .cnnLine0{font-family :Arial,Helvetica,sans-serif; color:#666666;font-weight:bold;}
+ .cnnTimeStamp{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#333333;}
+ .galleryhedDek{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#575757;}
+ .galleryWidgetHeader{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#004276;}
+ .article-content{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
+ .cnnRecapStory{font-family :Arial,Helvetica,sans-serif; font-size:x-small}
+ h1{font-family :Arial,Helvetica,sans-serif; font-size:x-large}
+ .captionname{font-family :Arial,Helvetica,sans-serif; font-size:x-small;color:#575757;}
+ inStoryIE{{font-family :Arial,Helvetica,sans-serif; font-size:x-small;}
'''
keep_only_tags = [
- dict(name='div', attrs={'class':["cnnWCBoxContent","cnnContent","cnnMainBodySecs"]}),
+ dict(name='div', attrs={'class':["cnnWCBoxContent","cnnContent","cnnMainBodySecs","cnn_storyarea","cnn_strycntntlft","cnn_strytmstmp"]}),
dict(name='div', attrs={'id':["contentBody","content"]}),
- dict(name='td', attrs={'id':["cnnRecapStory"]}),]
+ dict(name='td', attrs={'id':["cnnRecapStory"]}),
+ dict(name='td', attrs={'class':["cnnBigSideLeft"]}),
+ ]
remove_tags = [
- dict(name='div', attrs={'class':["storyLink","article-tools clearfix","widget video related-video vList","cnnFooterBox","scrollArrows","boxHeading","cnnInlineMailbag","mainCol_lastBlock","cnn_bookmarks","cnnFooterBox","cnnEndOfStory","cnnInlineSL","cnnStoryHighlights","cnnFooterClick","cnnSnapShotHeader","cnnStoryToolsFooter","cnnWsnr","cnnUGCBox","cnnTopNewsModule","cnnStoryElementBox","cnnStoryPhotoBoxNavigation"]}),
+ dict(name='div', attrs={'class':["cnnInlineRailSelectList","cnn_strycntntrgt","cnn_strybtntools","cnn_strylctcntr cnn_strylctcqrelt","cnn_strylceclbtn","cnn_stryftsbttm","cnn_strybtmmorebx","storyLink","article-tools clearfix","widget video related-video vList","cnnFooterBox","scrollArrows","boxHeading","cnnInlineMailbag","mainCol_lastBlock","cnn_bookmarks","cnnFooterBox","cnnEndOfStory","cnnInlineSL","cnnStoryHighlights","cnnFooterClick","cnnSnapShotHeader","cnnStoryToolsFooter","cnnWsnr","cnnUGCBox","cnnTopNewsModule","cnnStoryElementBox","cnnStoryPhotoBoxNavigation"]}),
dict(name='span', attrs={'class':["cnnEmbeddedMosLnk"]}),
dict(name='div', attrs={'id':["cnnIncldHlder","articleCommentsContainer","featuredContent","superstarsWidget","shareMenuContainer","rssMenuContainer","storyBrandingBanner","cnnRightCol","siteFeatures","quigo628","rightColumn","clickIncludeBox","cnnHeaderRightCol","cnnSCFontLabel","cnnSnapShotBottomRight","cnnSCFontButtons","rightColumn"]}),
dict(name='p', attrs={'class':["cnnTopics"]}),
dict(name='td', attrs={'class':["cnnRightRail"]}),
dict(name='table', attrs={'class':["cnnTMbox"]}),
dict(name='ul', attrs={'id':["cnnTopNav","cnnBotNav","cnnSBNav"]}),
+ dict(name='div', attrs={'id':["cnn_ftrcntnt"]})
]
# def print_version(self, url):
@@ -75,3 +91,10 @@ class CNN(BasicNewsRecipe):
('Offbeat', 'http://rss.cnn.com/rss/cnn_offbeat.rss'),
('Most Popular', 'http://rss.cnn.com/rss/cnn_mostpopular.rss')
]
+
+ def preprocess_html(self, soup):
+
+ for tag in soup.findAll(name=['ul','li']):
+ tag.name = 'div'
+
+ return soup
diff --git a/src/calibre/gui2/wizard/__init__.py b/src/calibre/gui2/wizard/__init__.py
index adde8c7fde..d3552f3a99 100644
--- a/src/calibre/gui2/wizard/__init__.py
+++ b/src/calibre/gui2/wizard/__init__.py
@@ -468,6 +468,11 @@ class LibraryPage(QWizardPage, LibraryUI):
self.init_languages()
self.connect(self.language, SIGNAL('currentIndexChanged(int)'),
self.change_language)
+ self.connect(self.location, SIGNAL('textChanged(QString)'),
+ self.location_text_changed)
+
+ def location_text_changed(self, newtext):
+ self.emit(SIGNAL('completeChanged()'))
def init_languages(self):
self.language.blockSignals(True)
@@ -525,9 +530,13 @@ class LibraryPage(QWizardPage, LibraryUI):
self.location.setText(lp)
def isComplete(self):
- lp = unicode(self.location.text())
- return lp and os.path.exists(lp) and os.path.isdir(lp) and os.access(lp,
- os.W_OK)
+ try:
+ lp = unicode(self.location.text())
+ ans = bool(lp) and os.path.exists(lp) and os.path.isdir(lp) and os.access(lp,
+ os.W_OK)
+ except:
+ ans = False
+ return ans
def commit(self, completed):
oldloc = prefs['library_path']
diff --git a/src/calibre/library/cli.py b/src/calibre/library/cli.py
index f3fddcd637..e49da584ca 100644
--- a/src/calibre/library/cli.py
+++ b/src/calibre/library/cli.py
@@ -20,7 +20,7 @@ from calibre.utils.genshi.template import MarkupTemplate
FIELDS = set(['title', 'authors', 'author_sort', 'publisher', 'rating',
'timestamp', 'size', 'tags', 'comments', 'series', 'series_index',
- 'formats', 'isbn', 'uuid', 'cover'])
+ 'formats', 'isbn', 'uuid', 'pubdate', 'cover'])
XML_TEMPLATE = '''\
@@ -38,6 +38,7 @@ XML_TEMPLATE = '''\
${record['publisher']}
${record['rating']}
${record['timestamp']}
+ ${record['pubdate']}
${record['size']}
diff --git a/src/calibre/library/database2.py b/src/calibre/library/database2.py
index e30b11cb20..2a693fee42 100644
--- a/src/calibre/library/database2.py
+++ b/src/calibre/library/database2.py
@@ -1584,7 +1584,7 @@ class LibraryDatabase2(LibraryDatabase):
prefix = self.library_path
FIELDS = set(['title', 'authors', 'author_sort', 'publisher', 'rating',
'timestamp', 'size', 'tags', 'comments', 'series', 'series_index',
- 'isbn', 'uuid'])
+ 'isbn', 'uuid', 'pubdate'])
data = []
for record in self.data:
if record is None: continue