Module: Kudzu::Model::Page

Included in:
Adapter::Memory::Page, Agent::Response
Defined in:
lib/kudzu/model/page.rb

Instance Method Summary collapse

Instance Method Details

#body ⇒ Object 



63
64
65
 
# File 'lib/kudzu/model/page.rb', line 63

def body
  @body
end

#body=(body) ⇒ Object 



67
68
69
 
# File 'lib/kudzu/model/page.rb', line 67

def body=(body)
  @body = body
end

#css? ⇒ Boolean

Returns:

  • (Boolean) 


23
24
25
 
# File 'lib/kudzu/model/page.rb', line 23

def css?
  !mime_type.to_s.match(%r{text/css}).nil?
end

#decoded_body ⇒ Object 



79
80
81
 
# File 'lib/kudzu/model/page.rb', line 79

def decoded_body
  @decoded_body ||= decode_body(body)
end

#etag ⇒ Object 



11
12
13
 
# File 'lib/kudzu/model/page.rb', line 11

def etag
  response_header['etag']
end

#filtered ⇒ Object 



71
72
73
 
# File 'lib/kudzu/model/page.rb', line 71

def filtered
  @filtered
end

#filtered=(filtered) ⇒ Object 



75
76
77
 
# File 'lib/kudzu/model/page.rb', line 75

def filtered=(filtered)
  @filtered = filtered
end

#html? ⇒ Boolean

Returns:

  • (Boolean) 


15
16
17
 
# File 'lib/kudzu/model/page.rb', line 15

def html?
  !mime_type.to_s.match(%r{text/html|application/xhtml\+xml}).nil?
end

#js? ⇒ Boolean

Returns:

  • (Boolean) 


27
28
29
 
# File 'lib/kudzu/model/page.rb', line 27

def js?
  !mime_type.to_s.match(%r{text/javascript|application/javascript|application/x-javascript}).nil?
end

#last_modified ⇒ Object 



4
5
6
7
8
9
 
# File 'lib/kudzu/model/page.rb', line 4

def last_modified
  last_modified = response_header['last-modified']
  Time.parse(last_modified).localtime if last_modified
rescue
  nil
end

#parsed_doc ⇒ Object 



83
84
85
86
87
88
89
 
# File 'lib/kudzu/model/page.rb', line 83

def parsed_doc
  @parsed_doc ||= if html?
                    Nokogiri::HTML(decoded_body)
                  elsif xml?
                    Nokogiri::XML(decoded_body)
                  end
end

#status_client_error? ⇒ Boolean

Returns:

  • (Boolean) 


43
44
45
 
# File 'lib/kudzu/model/page.rb', line 43

def status_client_error?
  400 <= status && status <= 499
end

#status_gone? ⇒ Boolean

Returns:

  • (Boolean) 


59
60
61
 
# File 'lib/kudzu/model/page.rb', line 59

def status_gone?
  status == 410
end

#status_not_found? ⇒ Boolean

Returns:

  • (Boolean) 


55
56
57
 
# File 'lib/kudzu/model/page.rb', line 55

def status_not_found?
  status == 404
end

#status_not_modified? ⇒ Boolean

Returns:

  • (Boolean) 


51
52
53
 
# File 'lib/kudzu/model/page.rb', line 51

def status_not_modified?
  status == 304
end

#status_redirection? ⇒ Boolean

Returns:

  • (Boolean) 


39
40
41
 
# File 'lib/kudzu/model/page.rb', line 39

def status_redirection?
  300 <= status && status <= 399
end

#status_server_error? ⇒ Boolean

Returns:

  • (Boolean) 


47
48
49
 
# File 'lib/kudzu/model/page.rb', line 47

def status_server_error?
  500 <= status && status <= 599
end

#status_success? ⇒ Boolean

Returns:

  • (Boolean) 


35
36
37
 
# File 'lib/kudzu/model/page.rb', line 35

def status_success?
  200 <= status && status <= 299
end

#text? ⇒ Boolean

Returns:

  • (Boolean) 


31
32
33
 
# File 'lib/kudzu/model/page.rb', line 31

def text?
  html? || xml? || !mime_type.to_s.match(%r{text/}).nil?
end

#xml? ⇒ Boolean

Returns:

  • (Boolean) 


19
20
21
 
# File 'lib/kudzu/model/page.rb', line 19

def xml?
  !mime_type.to_s.match(%r{text/xml|application/xml|application/rss\+xml|application/atom\+xml}).nil?
end