This commit is contained in:
Kovid Goyal 2011-10-27 14:44:35 +05:30
parent c212da613d
commit c8b2795416

View File

@ -1,14 +1,13 @@
from calibre.web.feeds.news import BasicNewsRecipe from calibre.web.feeds.news import BasicNewsRecipe
import re import re
from calibre.ebooks.BeautifulSoup import BeautifulSoup from calibre.utils.magick import Image
from calibre.utils.magick import Image, PixelWand, create_canvas
''' Version 1.2, updated cover image to match the changed website. ''' Version 1.2, updated cover image to match the changed website.
added info date on title added info date on title
version 1.4 Updated tags, delay and added autoclean 22-09-2011 version 1.4 Updated tags, delay and added autoclean 22-09-2011
version 1.5 Changes due to changes in site version 1.5 Changes due to changes in site
version 1.6 Added css, removed auto cleanup, added buitenland section, added use_embedded_content, added remove_attributes version 1.6 Added css, removed auto cleanup, added buitenland section, added use_embedded_content, added remove_attributes
Added som processing on pictures Added som processing on pictures
Removed links in html Removed links in html
Removed extre white characters Removed extre white characters
@ -48,12 +47,12 @@ class AdvancedUserRecipe1306097511(BasicNewsRecipe):
div.article-image-caption {width: 246px;margin-bottom: 5px;margin-left: 10px;}\ div.article-image-caption {width: 246px;margin-bottom: 5px;margin-left: 10px;}\
div.article-image-caption-2column {margin-bottom: 10px;width: 373px;} div.article-image-caption-3column {}\ div.article-image-caption-2column {margin-bottom: 10px;width: 373px;} div.article-image-caption-3column {}\
img {border:0px;} .img-mask {position:absolute;top:0px;left:0px;}' img {border:0px;} .img-mask {position:absolute;top:0px;left:0px;}'
keep_only_tags = [dict(name='div', attrs={'class':[ 'article-image-caption-2column', 'article-image-caption-3column', 'article-body', 'article-box-fact']}), keep_only_tags = [dict(name='div', attrs={'class':[ 'article-image-caption-2column', 'article-image-caption-3column', 'article-body', 'article-box-fact']}),
dict(name='div', attrs={'id':['date']}), dict(name='div', attrs={'id':['date']}),
dict(name='h1', attrs={'class':['title']}), dict(name='h1', attrs={'class':['title']}),
dict(name='h2', attrs={'class':['subtitle']})] dict(name='h2', attrs={'class':['subtitle']})]
remove_tags = [dict(name='div', attrs={'class':[ 'metroCommentFormWrap', remove_tags = [dict(name='div', attrs={'class':[ 'metroCommentFormWrap',
'commentForm', 'metroCommentInnerWrap', 'article-slideshow-counter-container', 'article-slideshow-control', 'ad', 'header-links', 'commentForm', 'metroCommentInnerWrap', 'article-slideshow-counter-container', 'article-slideshow-control', 'ad', 'header-links',
'art-rgt','pluck-app pluck-comm', 'share-and-byline', 'article-tools-below-title', 'col-179 ', 'related-links', 'clear padding-top-15', 'share-tools', 'article-page-auto-pushes', 'footer-edit']}), 'art-rgt','pluck-app pluck-comm', 'share-and-byline', 'article-tools-below-title', 'col-179 ', 'related-links', 'clear padding-top-15', 'share-tools', 'article-page-auto-pushes', 'footer-edit']}),
@ -72,28 +71,27 @@ class AdvancedUserRecipe1306097511(BasicNewsRecipe):
img = Image() img = Image()
img.open(iurl) img.open(iurl)
#width, height = img.size #width, height = img.size
#print '***img is: ', iurl, '\n****width is: ', width, 'height is: ', height #print '***img is: ', iurl, '\n****width is: ', width, 'height is: ', height
pw = PixelWand()
img.trim(0) img.trim(0)
img.save(iurl) img.save(iurl)
''' '''
#width, height = img.size #width, height = img.size
#print '***TRIMMED img width is: ', width, 'height is: ', height #print '***TRIMMED img width is: ', width, 'height is: ', height
left=0 left=0
top=0 top=0
border_color='#ffffff' border_color='#ffffff'
width, height = img.size width, height = img.size
#print '***retrieved img width is: ', width, 'height is: ', height #print '***retrieved img width is: ', width, 'height is: ', height
height_correction = 1.17 height_correction = 1.17
canvas = create_canvas(width, height*height_correction,border_color) canvas = create_canvas(width, height*height_correction,border_color)
canvas.compose(img, left, top) canvas.compose(img, left, top)
#img = canvas #img = canvas
canvas.save(iurl) canvas.save(iurl)
#width, height = canvas.size #width, height = canvas.size
#print '***NEW img width is: ', width, 'height is: ', height #print '***NEW img width is: ', width, 'height is: ', height
''' '''
return soup return soup
feeds = [ feeds = [
(u'Binnenland', u'http://www.metronieuws.nl/rss.xml?c=1277377288-3'), (u'Binnenland', u'http://www.metronieuws.nl/rss.xml?c=1277377288-3'),
(u'Economie', u'http://www.metronieuws.nl/rss.xml?c=1278070988-0'), (u'Economie', u'http://www.metronieuws.nl/rss.xml?c=1278070988-0'),