sm correctly read strings as utf8 even if JS_EncodeCharacters doesn't work as documented

2 files changed, 22 insertions, 2 deletions

diff --git a/dbtests/jstests.cpp b/dbtests/jstests.cpp
index 49cac4acfaa..b6e2b3a3e8d 100644
--- a/dbtests/jstests.cpp
+++ b/dbtests/jstests.cpp
@@ -467,6 +467,22 @@ namespace JSTests {
         }        
         static const char *ns() { return "unittest.jstests.utf8check"; }
     };
+
+    class LongUtf8String {
+    public:
+        LongUtf8String() { reset(); }
+        ~LongUtf8String() { reset(); }
+        void run() {
+            if( !globalScriptEngine->utf8Ok() )
+                return;
+            client.eval( "unittest", "db.jstests.longutf8string.save( {_id:'\\uffff\uffff\uffff\uffff'} )" );
+        }
+    private:
+        void reset() {
+            client.dropCollection( ns() );
+        }        
+        static const char *ns() { return "unittest.jstests.longutf8string"; }
+    };
     
     class All : public Suite {
     public:
@@ -484,6 +500,7 @@ namespace JSTests {
             add< TypeConservation >();
             add< WeirdObjects >();
             add< Utf8Check >();
+            add< LongUtf8String >();
         }
     };
     
diff --git a/scripting/engine_spidermonkey.cpp b/scripting/engine_spidermonkey.cpp
index e947d05ea08..0f95a6817e2 100644
--- a/scripting/engine_spidermonkey.cpp
+++ b/scripting/engine_spidermonkey.cpp
@@ -86,10 +86,13 @@ namespace mongo {
             if( srclen == 0 )
                 return "";
             
-            size_t len = srclen * 4;
+            size_t len = srclen * 6; // we only need *3, but see note on len below
             char * dst = (char*)malloc( len );
             
-            len /= 2; // weird JS_EncodeCharacters api expects len in 16bit units but modifies it to represent size in 8bit units.
+            len /= 2;
+            // doc re weird JS_EncodeCharacters api claims len expected in 16bit
+            // units, but experiments suggest 8bit units expected.  We allocate
+            // enough memory that either will work.
 
             assert( JS_EncodeCharacters( _context , s , srclen , dst , &len) );