Update Something Awful

This commit is contained in:
Kovid Goyal 2013-07-27 07:42:51 +05:30
parent 0ca5ffc49f
commit 2a0f6bbeae

View File

@ -1,4 +1,3 @@
import re
from calibre.web.feeds.news import BasicNewsRecipe from calibre.web.feeds.news import BasicNewsRecipe
class SomethingAwfulRecipe(BasicNewsRecipe): class SomethingAwfulRecipe(BasicNewsRecipe):
@ -6,8 +5,7 @@ class SomethingAwfulRecipe(BasicNewsRecipe):
__author__ = 'atordo' __author__ = 'atordo'
description = 'The Internet Makes You Stupid' description = 'The Internet Makes You Stupid'
cover_url = 'http://i.somethingawful.com/core/head-logo-bluegren.png' cover_url = 'http://i.somethingawful.com/core/head-logo-bluegren.png'
masthead_url = 'http://i.somethingawful.com/core/head-logo-bluegren.png' oldest_article = 15
oldest_article = 7
max_articles_per_feed = 50 max_articles_per_feed = 50
auto_cleanup = False auto_cleanup = False
no_stylesheets = True no_stylesheets = True
@ -16,27 +14,21 @@ class SomethingAwfulRecipe(BasicNewsRecipe):
use_embedded_content = False use_embedded_content = False
remove_empty_feeds = True remove_empty_feeds = True
publication_type = 'magazine' publication_type = 'magazine'
reverse_article_order = True
recursions = 1 recursions = 1
match_regexps = [r'\?page=\d+$'] match_regexps = [r'^http://www.somethingawful.com/.+/.+/\d{1,2}/$']
preprocess_regexps = [
(re.compile(r'<!-- content end-->.*</body>', re.DOTALL), lambda match: '</body>')
]
remove_attributes = ['align', 'alt', 'valign'] remove_attributes = ['align', 'alt', 'valign']
keep_only_tags = [ keep_only_tags = [
dict(name='div', attrs={'class':'content_area'}) dict(name='div', attrs={'class':'article_head'})
# ,dict(name='p', attrs={'class':'pagebar'}) ,dict(name='div', attrs={'class':'organ article'})
] ,dict(name='ul', attrs={'class':'pager'})
remove_tags = [
dict(name='div', attrs={'class':['column_box','featurenav','social']})
,dict(name='div', attrs={'id':'sidebar'})
,dict(name='a', attrs={'class':'curpage'})
] ]
extra_css = ''' extra_css = '''
.author{font-size:small} .date{font-size:small}
.byline{font-size:small} .font_big{font-size:large} .byline{font-size:small} .font_big{font-size:large}
.compat5{font-weight:bold} .accentbox{background-color:#E3E3E3; border:solid black} .compat5{font-weight:bold} .accentbox{background-color:#E3E3E3; border:solid black}
img{margin-bottom:0.4em; display:block; margin-left: auto; margin-right:auto} img{margin-bottom:0.4em; display:block; margin-left: auto; margin-right:auto}
@ -53,7 +45,7 @@ class SomethingAwfulRecipe(BasicNewsRecipe):
,('The Great Goon Database', 'http://www.somethingawful.com/rss/great-goon-database.rss.xml') ,('The Great Goon Database', 'http://www.somethingawful.com/rss/great-goon-database.rss.xml')
,('Livejournal Theater', 'http://www.somethingawful.com/rss/livejournal-theater.rss.xml') ,('Livejournal Theater', 'http://www.somethingawful.com/rss/livejournal-theater.rss.xml')
,('Joystick Token Healthpack', 'http://www.somethingawful.com/rss/token-healthpack.rss.xml') ,('Joystick Token Healthpack', 'http://www.somethingawful.com/rss/token-healthpack.rss.xml')
#,('Webcam Ward', 'http://www.somethingawful.com/rss/webcam-ward.rss.xml') ,('Webcam Ward', 'http://www.somethingawful.com/rss/webcam-ward.rss.xml')
,('Features / Articles', 'http://www.somethingawful.com/rss/feature-articles.rss.xml') ,('Features / Articles', 'http://www.somethingawful.com/rss/feature-articles.rss.xml')
,('Guides', 'http://www.somethingawful.com/rss/guides.rss.xml') ,('Guides', 'http://www.somethingawful.com/rss/guides.rss.xml')
,('Legal Threats', 'http://www.somethingawful.com/rss/legal-threats.rss.xml') ,('Legal Threats', 'http://www.somethingawful.com/rss/legal-threats.rss.xml')
@ -77,6 +69,7 @@ class SomethingAwfulRecipe(BasicNewsRecipe):
,('Johnston Checks In', 'http://www.somethingawful.com/rss/levi-johnston.rss.xml') ,('Johnston Checks In', 'http://www.somethingawful.com/rss/levi-johnston.rss.xml')
,('Twitter Tuesday', 'http://www.somethingawful.com/rss/twitter-tuesday.rss.xml') ,('Twitter Tuesday', 'http://www.somethingawful.com/rss/twitter-tuesday.rss.xml')
,('Music Article', 'http://www.somethingawful.com/rss/music-article.rss.xml') ,('Music Article', 'http://www.somethingawful.com/rss/music-article.rss.xml')
,('The Everdraed Showcase', 'http://www.somethingawful.com/rss/everdraed-showcase.xml')
,('Reviews [Games]', 'http://www.somethingawful.com/rss/game-reviews.rss.xml') ,('Reviews [Games]', 'http://www.somethingawful.com/rss/game-reviews.rss.xml')
,('Reviews [Movies]', 'http://www.somethingawful.com/rss/movie-reviews.rss.xml') ,('Reviews [Movies]', 'http://www.somethingawful.com/rss/movie-reviews.rss.xml')
,('Rom Pit', 'http://www.somethingawful.com/rss/rom-pit.rss.xml') ,('Rom Pit', 'http://www.somethingawful.com/rss/rom-pit.rss.xml')
@ -92,4 +85,6 @@ class SomethingAwfulRecipe(BasicNewsRecipe):
,('Garbage Day', 'http://www.somethingawful.com/rss/garbage-day.rss.xml') ,('Garbage Day', 'http://www.somethingawful.com/rss/garbage-day.rss.xml')
,('WTF, D&D!?', 'http://www.somethingawful.com/rss/dungeons-and-dragons.rss.xml') ,('WTF, D&D!?', 'http://www.somethingawful.com/rss/dungeons-and-dragons.rss.xml')
,('Current Releases', 'http://www.somethingawful.com/rss/current-movie-reviews.rss.xml') ,('Current Releases', 'http://www.somethingawful.com/rss/current-movie-reviews.rss.xml')
,('Awful Things for Sale', 'http://www.somethingawful.com/rss/awful-things-sale.xml')
] ]