parser.rb in combine_pdf-0.2.6

- old
+ new

@@ -63,14 +63,17 @@
 			return @parsed unless @parsed.empty?
 			@scanner = StringScanner.new @string_to_parse
 			@scanner.pos = 0
 			if @scanner.scan /\%PDF\-[\d\-\.]+/
 				@version = @scanner.matched.scan(/[\d\.]+/)[0].to_f
+				@scanner.skip_until /[\n\r]+/
+				# @scanner.skip /[^\d]*/
 			end
-
 			@parsed = _parse_
 
+			raise "Unknown PDF parsing error - maleformed PDF file?" unless (@parsed.select {|i| !i.is_a?(Hash)}).empty?
+
 			if @root_object == {}
 				xref_streams = @parsed.select {|obj| obj.is_a?(Hash) && obj[:Type] == :XRef}
 				xref_streams.each do |xref_dictionary|
 					@root_object.merge! xref_dictionary
 				end
@@ -173,12 +176,12 @@
 					str = @scanner.scan_until(/endstream/)
 					# raise error if the stream doesn't end.
 					raise "Parsing Error: PDF file error - a stream object wasn't properly colsed using 'endstream'!" unless str
 					# need to remove end of stream
 					if out.last.is_a? Hash
-						out.last[:raw_stream_content] = str[0...-10] #cuts only one EON char (\n or \r)
-						# out.last[:raw_stream_content] = str.gsub(/[\n\r]?[\n\r]?endstream/, "")
+						# out.last[:raw_stream_content] = str[0...-10] #cuts only one EON char (\n or \r)
+						out.last[:raw_stream_content] = str.gsub(/[\n\r]?[\n\r]endstream\z/, "")
 					else
 						warn "Stream not attached to dictionary!"
 						out << str[0...-10].force_encoding(Encoding::ASCII_8BIT)
 					end
 				##########################################
@@ -321,10 +324,10 @@
 					@scanner.scan_until(/(trailer)|(\%EOF)/)
 
 					if @scanner.matched[-1] == 'r'
 						if @scanner.skip_until(/<</)
 							data = _parse_
-							@root_object = {}
+							@root_object ||= {}
 							@root_object[data.shift] = data.shift while data[0]						
 						end
 						##########
 						## skip untill end of segment, maked by %%EOF
 						@scanner.skip_until(/\%\%EOF/)
\ No newline at end of file