Sha256: 50fe13a81f4bc997a45b989bd620e76c81db9add4021f502799656d3eb7af493
Contents?: true
Size: 1.63 KB
Versions: 18
Compression:
Stored size: 1.63 KB
Contents
require 'nokogiri'
module Awestruct
module ContextHelper
def html_to_text(str)
str.gsub( /<[^>]+>/, '' ).gsub( / /, ' ' )
end
def clean_html(str)
str.gsub( / /, ' ' )
end
def without_images(str)
str.gsub(/<img[^>]+>/,'').gsub(/<a[^>]+>([^<]*)<\/a>/, '\1')
end
def close_tags(s)
stack = []
s.scan(/<\/?[^>]+>/).each do |tag|
if tag[1] != '/'
tag = tag[1..-1].scan(/\w+/).first
stack = [ tag ] + stack
else
tag = tag[2..-1].scan(/\w+/).first
if stack[0] == tag
stack = stack.drop(1)
else
raise "Malformed HTML expected #{tag[0]} but got #{tag} '#{s}'"
end
end
end
stack.inject(s) { |memo,tag| memo += "</#{tag}>" }
end
def summarize(text, numwords=20, ellipsis='...')
close_tags(text.split(/ /)[0, numwords].join(' ') + ellipsis)
end
def fully_qualify_urls(base_url, text)
doc = Nokogiri::HTML.fragment( text )
doc.css( "a" ).each do |a|
a['href'] = fix_url( base_url, a['href'] ) if a['href']
end
doc.css( "link" ).each do |link|
link['href'] = fix_url( base_url, link['href'] )
end
doc.css( "img" ).each do |img|
img['src'] = fix_url( base_url, img['src'] )
end
if RUBY_VERSION.start_with? '1.8'
doc.to_s
else
doc.to_s.tap do |d|
d.force_encoding(text.encoding) if d.encoding != text.encoding
end
end
end
def fix_url(base_url, url)
return url unless ( url =~ /^\// )
"#{base_url}#{url}"
end
end
end
Version data entries
18 entries across 18 versions & 1 rubygems