make sure we have a newline in a chunk length before parsing it, tighten up the regex. fixes 2063 (I hope)

git-svn-id: file:///home/svn/framework3/trunk@9448 4d416f70-5f16-0410-b530-b9f4589650da
2010-06-07 23:43:32 +00:00 · 2010-06-07 23:43:32 +00:00 · 7044c65f22
parent a0e252ad75
commit 7044c65f22
1 changed files with 11 additions and 1 deletions
--- a/lib/rex/proto/http/packet.rb
+++ b/lib/rex/proto/http/packet.rb
@ -92,6 +92,7 @@ class Packet
 				end
 			end
 		rescue
+			# This rescue might be a problem because it will swallow TimeoutError
 			self.error = $!
 			return ParseCode::Error
 		end
@ -331,8 +332,17 @@ protected
 			# Remove any leading newlines or spaces
 			self.bufq.lstrip!

+			# If we didn't get a newline, then this might not be the full
+			# length, go back and get more.
+			# e.g. 
+			#  first packet: "200"
+			#  second packet: "0\r\n\r\n<html>..."
+			if not bufq.index("\n")
+				return
+			end
+
 			# Extract the actual hexadecimal length value
-			clen = self.bufq.slice!(/^[a-zA-Z0-9]*\r?\n/)
+			clen = self.bufq.slice!(/^[a-fA-F0-9]+\r?\n/)

 			clen.rstrip! if (clen)