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.



15
16
17
18
19
20
21
22
23
24
25
 
# File 'lib/httpage/httpage.rb', line 15

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



44
45
46
47
 
# File 'lib/httpage/httpage.rb', line 44

def encoding
  @encoding, @content_type = get_page_info if @encoding.nil?
  return @encoding
end

#real_url ⇒ Object (readonly)

Returns the value of attribute real_url.



12
13
14
 
# File 'lib/httpage/httpage.rb', line 12

def real_url
  @real_url
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.



176
177
178
179
 
# File 'lib/httpage/httpage.rb', line 176

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

#clean ⇒ Object 



220
 
# File 'lib/httpage/httpage.rb', line 220

def clean; clean_text end

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

Transliterates text to ASCII and removes unknown characters.



198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
 
# File 'lib/httpage/httpage.rb', line 198

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.



224
225
226
227
228
 
# File 'lib/httpage/httpage.rb', line 224

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

#content_type ⇒ Object

Returns page content-type.



51
52
53
54
 
# File 'lib/httpage/httpage.rb', line 51

def content_type
  @encoding, @content_type = get_page_info if @content_type.nil?
  return @content_type
end

#reset_buffers ⇒ Object

Resets encoding and response buffers.



29
30
31
32
 
# File 'lib/httpage/httpage.rb', line 29

def reset_buffers
  @encoding = nil
  @response = nil
end

#response ⇒ Object

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



122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
 
# File 'lib/httpage/httpage.rb', line 122

def response
  return @response unless @response.nil?
  url         = @url
  found       = false
  response    = nil
  @real_url   = nil
  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
    @real_url = url
    @response = response
    @encoding, @content_type = get_page_info
    return response
  else
    return nil
  end
end

#strip_html(text = nil) ⇒ Object

Strips HTML tags from document.



183
184
185
186
187
188
189
190
191
192
193
194
 
# File 'lib/httpage/httpage.rb', line 183

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