Check UTF-8-ness of shortstrs in non-content-bearing methods. Note that we use the previously developed "fast" checker to avoid a dependency on xmerl, not so much because it is fast.

1 files changed, 6 insertions, 2 deletions

diff --git a/codegen.py b/codegen.py
index 842549cf..91fa1154 100644
--- a/codegen.py
+++ b/codegen.py
@@ -187,7 +187,7 @@ def genErl(spec):
         elif type == 'table':
             return p+'Len:32/unsigned, '+p+'Tab:'+p+'Len/binary'
 
-    def genFieldPostprocessing(packed):
+    def genFieldPostprocessing(packed, hasContent):
         for f in packed:
             type = erlType(f.domain)
             if type == 'bit':
@@ -199,6 +199,10 @@ def genErl(spec):
             elif type == 'table':
                 print "  F%d = rabbit_binary_parser:parse_table(F%dTab)," % \
                       (f.index, f.index)
+            # We skip the check on content-bearing methods for
+            # speed. This is a sanity check, not a security thing.
+            elif type == 'shortstr' and not hasContent:
+                print "  rabbit_binary_parser:assert_utf8(F%d)," % (f.index)
             else:
                 pass
 
@@ -214,7 +218,7 @@ def genErl(spec):
             restSeparator = ''
         recordConstructorExpr = '#%s{%s}' % (m.erlangName(), fieldMapList(m.arguments))
         print "decode_method_fields(%s, <<%s>>) ->" % (m.erlangName(), binaryPattern)
-        genFieldPostprocessing(packedFields)
+        genFieldPostprocessing(packedFields, m.hasContent)
         print "  %s;" % (recordConstructorExpr,)
 
     def genDecodeProperties(c):