[github/wallabag/wallabag.git] / inc / 3rdparty / site_config / standard / smithsonianmag.com.txt

# meta data
title://h1[@id = 'articleTitle']
author:substring-after(//ul[@id = 'byLine']/li[1],'By ')
date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',')
body://div[@id = 'article-body']

# full content
single_page_link://td/li[@class = 'article-singlepage']/a

# caption clean up
wrap_in(i)://span[@class='articleImageCaptionwide']
move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p


# clean up
strip://p[@id = 'articlePaginationWrapper']
strip://ul[contains(@class, 'cat-breadcrumb')]
strip://div [@class= 'viewMorePhotos']

test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html
Commit	Line	Data
4e067cea NL	1	# meta data
	2	title://h1[@id = 'articleTitle']
	3	author:substring-after(//ul[@id = 'byLine']/li[1],'By ')
	4	date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',')
	5	body://div[@id = 'article-body']
	6
	7	# full content
	8	single_page_link://td/li[@class = 'article-singlepage']/a
	9
	10	# caption clean up
	11	wrap_in(i)://span[@class='articleImageCaptionwide']
	12	move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p
	13
	14
	15	# clean up
	16	strip://p[@id = 'articlePaginationWrapper']
	17	strip://ul[contains(@class, 'cat-breadcrumb')]
	18	strip://div [@class= 'viewMorePhotos']
ac4d1142 NL	19
ac4d1142 NL	20	test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html