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 



65
66
67
 
# File 'lib/kudzu/model/page.rb', line 65

def body
  @body
end

#body=(body) ⇒ Object 



69
70
71
 
# File 'lib/kudzu/model/page.rb', line 69

def body=(body)
  @body = body
end

#css? ⇒ Boolean

Returns:

  • (Boolean) 


25
26
27
 
# File 'lib/kudzu/model/page.rb', line 25

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

#decoded_body ⇒ Object 



81
82
83
 
# File 'lib/kudzu/model/page.rb', line 81

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

#etag ⇒ Object 



13
14
15
 
# File 'lib/kudzu/model/page.rb', line 13

def etag
  response_header['etag']
end

#filtered ⇒ Object 



73
74
75
 
# File 'lib/kudzu/model/page.rb', line 73

def filtered
  @filtered
end

#filtered=(filtered) ⇒ Object 



77
78
79
 
# File 'lib/kudzu/model/page.rb', line 77

def filtered=(filtered)
  @filtered = filtered
end

#html? ⇒ Boolean

Returns:

  • (Boolean) 


17
18
19
 
# File 'lib/kudzu/model/page.rb', line 17

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

#js? ⇒ Boolean

Returns:

  • (Boolean) 


29
30
31
 
# File 'lib/kudzu/model/page.rb', line 29

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

#last_modified ⇒ Object 



6
7
8
9
10
11
 
# File 'lib/kudzu/model/page.rb', line 6

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

#parsed_doc ⇒ Object 



85
86
87
88
89
90
91
 
# File 'lib/kudzu/model/page.rb', line 85

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) 


45
46
47
 
# File 'lib/kudzu/model/page.rb', line 45

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

#status_gone? ⇒ Boolean

Returns:

  • (Boolean) 


61
62
63
 
# File 'lib/kudzu/model/page.rb', line 61

def status_gone?
  status == 410
end

#status_not_found? ⇒ Boolean

Returns:

  • (Boolean) 


57
58
59
 
# File 'lib/kudzu/model/page.rb', line 57

def status_not_found?
  status == 404
end

#status_not_modified? ⇒ Boolean

Returns:

  • (Boolean) 


53
54
55
 
# File 'lib/kudzu/model/page.rb', line 53

def status_not_modified?
  status == 304
end

#status_redirection? ⇒ Boolean

Returns:

  • (Boolean) 


41
42
43
 
# File 'lib/kudzu/model/page.rb', line 41

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

#status_server_error? ⇒ Boolean

Returns:

  • (Boolean) 


49
50
51
 
# File 'lib/kudzu/model/page.rb', line 49

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

#status_success? ⇒ Boolean

Returns:

  • (Boolean) 


37
38
39
 
# File 'lib/kudzu/model/page.rb', line 37

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

#text? ⇒ Boolean

Returns:

  • (Boolean) 


33
34
35
 
# File 'lib/kudzu/model/page.rb', line 33

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

#xml? ⇒ Boolean

Returns:

  • (Boolean) 


21
22
23
 
# File 'lib/kudzu/model/page.rb', line 21

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