Sha256: baf3207d0dc5918e432e5948b421d3859274e2c53e1c69f756b1f0bd0b34101c
Contents?: true
Size: 1 KB
Versions: 12
Compression:
Stored size: 1 KB
Contents
require 'nokogiri'
require 'stringio'
module TextRank
module CharFilter
##
# Character filter to remove HTML tags and convert HTML entities to text.
#
# = Example
#
# StripHtml.new.filter!(""Optimism", said Cacambo, "What is that?"")
# => "\"Optimism\", said Cacambo, \"What is that?\""
#
# StringHtml.new.filter!("<b>Alas! It is the <u>obstinacy</u> of maintaining that everything is best when it is worst.</b>")
# => "Alas! It is the obstinacy of maintaining that everything is best when it is worst."
##
class StripHtml < Nokogiri::XML::SAX::Document
def initialize
@text = StringIO.new
end
# Perform the filter
# @param text [String]
# @return [String]
def filter!(text)
@text.rewind
Nokogiri::HTML::SAX::Parser.new(self).parse(text)
@text.string
end
protected
def characters(string)
@text << ' '
@text << string
end
end
end
end
Version data entries
12 entries across 12 versions & 1 rubygems