Class: HTTPage

Inherits:
Object
  • Object
show all
Includes:
BufferAffects
Defined in:
lib/httpage/httpage.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(url, redir_retry = 5, conn_retry = 8, timeout = 40) ⇒ HTTPage

Returns a new instance of HTTPage.



14
15
16
17
18
19
20
21
22
 
# File 'lib/httpage/httpage.rb', line 14

def initialize(url,redir_retry=5,conn_retry=8,timeout=40)
  @encoding = nil
  @response = nil
  @http_req = nil
  @redir_retry = redir_retry
  @conn_retry = conn_retry
  @timeout = timeout
  self.url = url
end

Instance Attribute Details

#conn_retry ⇒ Object

Returns the value of attribute conn_retry.



11
12
13
 
# File 'lib/httpage/httpage.rb', line 11

def conn_retry
  @conn_retry
end

#encoding ⇒ Object

Returns page encoding.



41
42
43
 
# File 'lib/httpage/httpage.rb', line 41

def encoding
  @encoding ||= get_page_encoding
end

#redir_retry ⇒ Object

Returns the value of attribute redir_retry.



11
12
13
 
# File 'lib/httpage/httpage.rb', line 11

def redir_retry
  @redir_retry
end

#timeout ⇒ Object

Returns the value of attribute timeout.



11
12
13
 
# File 'lib/httpage/httpage.rb', line 11

def timeout
  @timeout
end

#url ⇒ Object

Returns the value of attribute url.



11
12
13
 
# File 'lib/httpage/httpage.rb', line 11

def url
  @url
end

Instance Method Details

#body ⇒ Object

Returns document body.



149
150
151
152
 
# File 'lib/httpage/httpage.rb', line 149

def body
  r = self.response
  return r.respond_to?(:body) ? r.body : nil
end

#clean ⇒ Object 



193
 
# File 'lib/httpage/httpage.rb', line 193

def clean; clean_text end

#clean_text(text = nil, enc = nil) ⇒ Object

Transliterates text to ASCII and removes unknown characters.



171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
 
# File 'lib/httpage/httpage.rb', line 171

def clean_text(text=nil, enc=nil)
  text ||= self.body
  enc ||= self.encoding
  page = Iconv.iconv('UTF-8//IGNORE', enc, text).join
  page = strip_html(page)
  page.gsub!(/['`]/m, '_amp__')
  page = Iconv.iconv('ASCII//TRANSLIT//IGNORE', 'UTF-8', page).join.downcase
  page.tr!(".!?", ' ')
  page.gsub!(/[^\x00-\x7F]+/, '')
  page.gsub!(/[^a-z0-9\-_\[\]\(\)\*\=\@\#\$\%\^\&\{\}\:\;\,\<\>\+\s\n\.\!\?]+/im, '')
  page.gsub!('_amp__',"'")
  page.squeeze!(" \n")
  page.gsub!(/^\s?\n\s?$/m, '')
  page.gsub!(/\n\s/,"\n")
  page.gsub!(/\s\n/,"\n")
  page.gsub!(/^\s+/,'')
  page.gsub!(/(^|\s)\'+(.*?)\'+(\s|$)/m,'\1\2\3')
  page.gsub!(/(^|\s)\'+(\s|$)/, '')
  page.squeeze!("\n ")
  return page
end

#clean_words(text = nil, enc = nil) ⇒ Object

Transliterates text to ASCII and removes unknown characters leaving just words.



197
198
199
200
201
 
# File 'lib/httpage/httpage.rb', line 197

def clean_words(text=nil, enc=nil)
  clean_text(text, enc).
  gsub(%r{[.*?]}mi, ' ').
  gsub(/[^a-z0-9]+/im, ' ')
end

#reset_buffers ⇒ Object

Resets encoding and response buffers.



26
27
28
29
 
# File 'lib/httpage/httpage.rb', line 26

def reset_buffers
  @encoding = nil
  @response = nil
end

#response ⇒ Object

Fetches document using HTTP and returns response object. It also sets encoding.



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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
 
# File 'lib/httpage/httpage.rb', line 97

def response
  return @response unless @response.nil?
  found = false
  response = nil
  url = @url
  http_req = @http_req
  redir_retry = @redir_retry
  conn_retry = @conn_retry
  
  until found do
    begin
      status = Timeout::timeout(@timeout) do
        case url.scheme.downcase.to_sym
        when :http
          response = Net::HTTP.start(url.host, url.port) { |http|  http.request(http_req) }
        when :https
          https             = Net::HTTP.new(url.host, url.port)
          https.use_ssl     = true
          https.verify_mode = OpenSSL::SSL::VERIFY_NONE
          response = https.start { |http| http.request(http_req) }
        else
          return nil
        end
      end
      response.value
    rescue Net::HTTPRetriableError
      conn_retry -= 1
      if response.respond_to?(:header) && !response.header['location'].nil? && !response.header['location'].empty?
        url = URI.parse(response.header['location'])
        http_req = Net::HTTP::Get.new(url.path)
        redir_retry -= 1
      end
    rescue
      return nil
    end
    if response.kind_of?(Net::HTTPOK)
      found = true
      break
    end
    break if (redir_retry < 0 || conn_retry < 0)
  end
  if found
    @response = response
    @encoding = get_page_encoding
    return response
  else
    return nil
  end
end

#strip_html(text = nil) ⇒ Object

Strips HTML tags from document.



156
157
158
159
160
161
162
163
164
165
166
167
 
# File 'lib/httpage/httpage.rb', line 156

def strip_html(text=nil)
  text ||= self.body
  coder=HTMLEntities.new
  coder.decode(text.tr("\t", ' ').
               tr("\r", '').
               sub(%r{<body.*?>(.*?)</body>}mi, '\1').
               gsub(%r{<script.*?>(.*?)</script>}mi, ' ').
               gsub(%r{<style.*?>(.*?)</style>}mi, ' ').
               gsub(%r{<!--.*?-->}mi, ' ').
               gsub(/<br\s*\/?>|<p>/mi, "\n").
               gsub(/<.*?>/m, ''))
end