commit a6509f7b378ed6591d550134fdda18e4a436fe77
author Phongphan Phuttha <phongphan@acm.org> Sat Oct 31 01:09:47 2015 +0700
committer Jens Geyer <jensg@apache.org> Fri Oct 30 21:45:36 2015 +0200
tree 4f71a22310692b111a58d2a8009607e908d3118a
parent d6ca81b2cc7bed049a4f06242695f29549b49309

THRIFT-3404 Fixed JSON String reader doesn't recognize UTF-16 surrogate pairs.
Client: Delphi
Patch: Phongphan Phuttha <phongphan@acm.org>

This closes #671

lib/delphi/src/Thrift.Protocol.JSON.pas

diff --git a/lib/delphi/src/Thrift.Protocol.JSON.pas b/lib/delphi/src/Thrift.Protocol.JSON.pas
index f491f53..36c3d72 100644
--- a/lib/delphi/src/Thrift.Protocol.JSON.pas
+++ b/lib/delphi/src/Thrift.Protocol.JSON.pas
@@ -24,6 +24,7 @@
 interface
 
 uses
+  Character,
   Classes,
   SysUtils,
   Math,
@@ -821,9 +822,12 @@
 var buffer : TMemoryStream;
     ch  : Byte;
     wch : Word;
+    highSurogate: Char;
+    surrogatePairs: Array[0..1] of Char;
     off : Integer;
     tmp : TBytes;
 begin
+  highSurogate := #0;
   buffer := TMemoryStream.Create;
   try
     if not skipContext
@@ -862,11 +866,31 @@
            + (HexVal(tmp[1]) shl 8)
            + (HexVal(tmp[2]) shl 4)
            +  HexVal(tmp[3]);
+
       // we need to make UTF8 bytes from it, to be decoded later
-      tmp := SysUtils.TEncoding.UTF8.GetBytes(Char(wch));
-      buffer.Write( tmp[0], length(tmp));
+      if Character.IsHighSurrogate(char(wch)) then begin
+        if highSurogate <> #0
+        then raise TProtocolException.Create( TProtocolException.INVALID_DATA, 'Expected low surrogate char');
+        highSurogate := char(wch);
+      end
+      else if Character.IsLowSurrogate(char(wch)) then begin
+        if highSurogate = #0
+        then TProtocolException.Create( TProtocolException.INVALID_DATA, 'Expected high surrogate char');
+        surrogatePairs[0] := highSurogate;
+        surrogatePairs[1] := char(wch);
+        tmp := TEncoding.UTF8.GetBytes(surrogatePairs);
+        buffer.Write( tmp[0], Length(tmp));
+        highSurogate := #0;
+      end
+      else begin
+        tmp := SysUtils.TEncoding.UTF8.GetBytes(Char(wch));
+        buffer.Write( tmp[0], Length(tmp));
+      end;
     end;
 
+    if highSurogate <> #0
+    then raise TProtocolException.Create( TProtocolException.INVALID_DATA, 'Expected low surrogate char');
+
     SetLength( result, buffer.Size);
     if buffer.Size > 0 then Move( buffer.Memory^, result[0], Length(result));