mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-09-29 15:31:08 -04:00
51 lines
2.4 KiB
Plaintext
51 lines
2.4 KiB
Plaintext
__license__ = 'GPL v3'
|
|
__copyright__ = '2011, Pat Stapleton <pat.stapleton at gmail.com>'
|
|
'''
|
|
abc.net.au/news
|
|
'''
|
|
import re
|
|
from calibre.web.feeds.recipes import BasicNewsRecipe
|
|
|
|
class TheDailyNewsEG(BasicNewsRecipe):
|
|
title = u'al-masry al-youm'
|
|
__author__ = 'Omm Mishmishah'
|
|
description = 'Independent News from Egypt'
|
|
masthead_url = 'http://www.almasryalyoum.com/sites/default/files/img/english_logo.png'
|
|
cover_url = 'http://www.almasryalyoum.com/sites/default/files/img/english_logo.png'
|
|
|
|
auto_cleanup = True
|
|
oldest_article = 7
|
|
max_articles_per_feed = 100
|
|
no_stylesheets = False
|
|
#delay = 1
|
|
use_embedded_content = False
|
|
encoding = 'utf8'
|
|
publisher = 'Independent News Egypt'
|
|
category = 'News, Egypt, World'
|
|
language = 'en_EG'
|
|
publication_type = 'newsportal'
|
|
# preprocess_regexps = [(re.compile(r'<!--.*?-->', re.DOTALL), lambda m: '')]
|
|
#Remove annoying map links (inline-caption class is also used for some image captions! hence regex to match maps.google)
|
|
preprocess_regexps = [(re.compile(r'<a class="inline-caption" href="http://maps\.google\.com.*?/a>', re.DOTALL), lambda m: '')]
|
|
conversion_options = {
|
|
'comments' : description
|
|
,'tags' : category
|
|
,'language' : language
|
|
,'publisher' : publisher
|
|
,'linearize_tables': False
|
|
}
|
|
|
|
keep_only_tags = [dict(attrs={'class':['article section']})]
|
|
|
|
remove_tags = [dict(attrs={'class':['related', 'tags', 'tools', 'attached-content ready',
|
|
'inline-content story left', 'inline-content map left contracted', 'published',
|
|
'story-map', 'statepromo', 'topics', ]})]
|
|
|
|
remove_attributes = ['width','height']
|
|
|
|
feeds = [(u'English News', u'http://www.almasryalyoum.com/en/rss_feed_term/113/rss.xml'),
|
|
(u'News Features', u'http://www.almasryalyoum.com/en/rss_feed_term/115/rss.xml'),
|
|
(u'Culture', u'http://www.almasryalyoum.com/en/rss_feed_term/133/rss.xml'),
|
|
(u'Cinema', u'http://www.almasryalyoum.com/en/rss_feed_term/134/rss.xml')
|
|
]
|