+ def decompress_body(str)
+ method = self['content-encoding']
+ case method
+ when nil
+ return str
+ when /gzip/ # Matches gzip, x-gzip, and the non-rfc-compliant gzip;q=\d sent by some servers
+ debug "gunzipping body"
+ begin
+ return Zlib::GzipReader.new(StringIO.new(str)).read
+ rescue Zlib::Error => e
+ # If we can't unpack the whole stream (e.g. because we're doing a
+ # partial read
+ debug "full gunzipping failed (#{e}), trying to recover as much as possible"
+ ret = ''
+ ret.force_encoding(Encoding::ASCII_8BIT)
+ begin
+ Zlib::GzipReader.new(StringIO.new(str)).each_byte { |byte|
+ ret << byte
+ }
+ rescue
+ end
+ return ret
+ end
+ when 'deflate'
+ debug "inflating body"
+ # From http://www.koders.com/ruby/fid927B4382397E5115AC0ABE21181AB5C1CBDD5C17.aspx?s=thread:
+ # -MAX_WBITS stops zlib from looking for a zlib header
+ inflater = Zlib::Inflate.new(-Zlib::MAX_WBITS)
+ begin
+ return inflater.inflate(str)
+ rescue Zlib::Error => e
+ raise e
+ # TODO
+ # debug "full inflation failed (#{e}), trying to recover as much as possible"
+ end
+ when /^(?:iso-8859-\d+|windows-\d+|utf-8|utf8)$/i
+ # B0rked servers (Freshmeat being one of them) sometimes return the charset
+ # in the content-encoding; in this case we assume that the document has
+ # a standard content-encoding
+ old_hsh = self.to_hash
+ self['content-type']= self['content-type']+"; charset="+method.downcase
+ warning "Charset vs content-encoding confusion, trying to recover: from\n#{old_hsh.pretty_inspect}to\n#{self.to_hash.pretty_inspect}"
+ return str
+ else
+ debug self.to_hash
+ raise "Unhandled content encoding #{method}"
+ end