siefca-httpage 0.2.0 → 0.2.2

data/lib/httpage/httpage.rb

@@ -9,16 +9,19 @@ class HTTPage
   attr_affects_buffers  :url, :encoding
 
   attr_accessor :redir_retry, :conn_retry, :timeout, :url
-  attr_writer :encoding
+  attr_reader   :real_url
+  attr_writer   :encoding
   
   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
+    @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
   
   # Resets encoding and response buffers.
@@ -39,31 +42,42 @@ class HTTPage
   # Returns page encoding.
 
   def encoding
-    @encoding ||= get_page_encoding
+    @encoding, @content_type = get_page_info if @encoding.nil?
+    return @encoding
+  end
+
+  # Returns page content-type.
+
+  def content_type
+    @encoding, @content_type = get_page_info if @content_type.nil?
+    return @content_type
   end
 
   # Obtains encoding from document body or server response header.
 
-  def get_page_encoding(default_encoding='ascii')
-    return default_encoding if self.response.nil?
+  def get_page_info(default_encoding='ascii', default_content_type='text/html')
+    return [default_content_type, default_encoding] if self.response.nil?
     
     # try meta-tag header
     header  = self.response.body.scan(/<meta http-equiv\s*=\s*['"]*content-type['"]*\s*content\s*=\s*['"]*\s*(.*?)\s*['"]*\s*\/?>/i)
     header  = header.flatten.first
     enc     = extract_encoding(header)
+    ctype   = extract_content_type(header)
 
     # try server header
-    if enc.nil?  
+    if ctype.nil?
       header = response.header['content-type']
+      ctype = extract_content_type(header)
       enc = extract_encoding(header)
     end
-
+    
     # try default
-    enc = default_encoding if enc.nil?
+    enc   = default_encoding      if enc.nil?
+    ctype = default_content_type  if ctype.nil?
 
-    return enc
+    return [ctype, enc]
   end
-  private :get_page_encoding
+  private :get_page_info
 
   # Extracts enconding from content-type string.
 
@@ -92,16 +106,28 @@ class HTTPage
   end
   private :extract_encoding
 
+  # Extracts content-type from content-type string.
+
+  def extract_content_type(ctype_string)
+    return nil if ctype_string.nil? || ctype_string.empty?
+    ct = ctype_string.chomp.squeeze(' ').split(';').first
+    ct = ct.strip.downcase.to_sym unless ct.nil?
+    return ct
+  end
+  private :extract_content_type
+
+
   # Fetches document using HTTP and returns response object. It also sets encoding.
 
   def response
     return @response unless @response.nil?
-    found = false
-    response = nil
-    url = @url
-    http_req = @http_req
+    url         = @url
+    found       = false
+    response    = nil
+    @real_url   = nil
+    http_req    = @http_req
     redir_retry = @redir_retry
-    conn_retry = @conn_retry
+    conn_retry  = @conn_retry
     
     until found do
       begin
@@ -136,8 +162,9 @@ class HTTPage
       break if (redir_retry < 0 || conn_retry < 0)
     end
     if found
+      @real_url = url
       @response = response
-      @encoding = get_page_encoding
+      @encoding, @content_type = get_page_info
       return response
     else
       return nil

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: siefca-httpage
 version: !ruby/object:Gem::Version 
-  version: 0.2.0
+  version: 0.2.2
 platform: ruby
 authors: 
 - "Pawe\xC5\x82 Wilk"