Method: WWMD::Page#set_data

Defined in:
lib/wwmd/page/page.rb

#set_data ⇒ Object

set reporting data for the page

Scan for comments, anchors, links and javascript includes and set page flags. The heavy lifting for parsing is done in the scrape class.

returns: array [ code, page_status, body_data.size ]



96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
 
# File 'lib/wwmd/page/page.rb', line 96

def set_data
  # reset scrape and inputs object
  # transparently gunzip
  begin
    io = StringIO.new(self.body_data)
    gz = Zlib::GzipReader.new(io)
    self.body_data.replace(gz.read)
  rescue => e
  end
  @scrape.reset(self.body_data)
  @inputs.set

  # remove comments that are css selectors for IE silliness
  @comments = @scrape.for_comments.reject do |c|
    c =~ /\[if IE\]/ ||
    c =~ /\[if IE \d/ ||
    c =~ /\[if lt IE \d/
  end
  @links = @scrape.for_links.map do |url|
    l = @urlparse.parse(self.last_effective_url,url).to_s
  end
  @jlinks = @scrape.for_javascript_links
  @forms = @scrape.for_forms
  @spider.add(self.last_effective_url,@links)
  return [self.code,self.body_data.size]
end