webservices: Add support for decoding supplementary characters' references.

dlls/webservices/reader.c

@@ -1190,6 +1190,7 @@ static int codepoint_to_utf8( int cp, unsigned char *dst )
         dst[0] = 0xe0 | cp;
         return 3;
     }
+    if (cp >= 0x110000) return -1;
     dst[3] = 0x80 | (cp & 0x3f);
     cp >>= 6;
     dst[2] = 0x80 | (cp & 0x3f);
@@ -1259,7 +1260,7 @@ static HRESULT decode_text( const unsigned char *str, ULONG len, unsigned char *
                     if (!len) return WS_E_INVALID_FORMAT;
 
                     p -= nb_digits = start - len;
-                    if (!nb_digits || nb_digits > 5 || p[nb_digits] != ';') return WS_E_INVALID_FORMAT;
+                    if (!nb_digits || nb_digits > 6 || p[nb_digits] != ';') return WS_E_INVALID_FORMAT;
                     for (i = 0; i < nb_digits; i++)
                     {
                         cp *= 16;

dlls/webservices/tests/reader.c

@@ -3614,11 +3614,16 @@ static void test_entities(void)
     static const char str29[] = "<t>&#0;</t>";
     static const char str30[] = "<t>&#65;</t>";
     static const char str31[] = "<t>&#65393;</t>";
+    static const char str32[] = "<t>&#x10ffff;</t>";
+    static const char str33[] = "<t>&#x110000;</t>";
+    static const char str34[] = "<t>&#1114111;</t>";
+    static const char str35[] = "<t>&#1114112;</t>";
     static const char res4[] = {0xea, 0xaa, 0xaa, 0x00};
     static const char res5[] = {0xf2, 0xaa, 0xaa, 0xaa, 0x00};
     static const char res21[] = {0xed, 0x9f, 0xbf, 0x00};
     static const char res24[] = {0xee, 0x80, 0x80, 0x00};
     static const char res31[] = {0xef, 0xbd, 0xb1, 0x00};
+    static const char res32[] = {0xf4, 0x8f, 0xbf, 0xbf, 0x00};
     static const struct
     {
         const char *str;
@@ -3658,6 +3663,10 @@ static void test_entities(void)
         { str29, WS_E_INVALID_FORMAT },
         { str30, S_OK, "A" },
         { str31, S_OK, res31 },
+        { str32, S_OK, res32 },
+        { str33, WS_E_INVALID_FORMAT },
+        { str34, S_OK, res32 },
+        { str35, WS_E_INVALID_FORMAT },
     };
     HRESULT hr;
     WS_XML_READER *reader;