Fix minLength & maxLength with code point instead of code unit

include/rapidjson/encodings.h

@@ -616,6 +616,22 @@ struct Transcoder<Encoding, Encoding> {
     }
 };
 
+//! Returns number of code points in a encoded string.
+template<typename Encoding>
+bool CountStringCodePoint(const typename Encoding::Ch* s, SizeType length, SizeType* outCount) {
+    GenericStringStream<Encoding> is(s);
+    const typename Encoding::Ch* end = s + length;
+    SizeType count = 0;
+    while (is.src_ < end) {
+        unsigned codepoint;
+        if (!Encoding::Decode(is, &codepoint))
+            return false;
+        count++;
+    }
+    *outCount = count;
+    return true;
+}
+
 RAPIDJSON_NAMESPACE_END
 
 #if defined(__GNUC__) || defined(_MSV_VER)

include/rapidjson/schema.h

@@ -428,8 +428,13 @@ public:
         if ((type_ & (1 << kStringSchemaType)) == 0)
             return false;
 
-        if (length < minLength_ || length > maxLength_)
-            return false;
+        //if (length < minLength_ || length > maxLength_)
+        //    return false;
+        if (minLength_ != 0 || maxLength_ != SizeType(~0)) {
+            SizeType count;
+            if (CountStringCodePoint<Encoding>(str, length, &count) && (count < minLength_ || count > maxLength_))
+                return false;
+        }
 
 #if RAPIDJSON_SCHEMA_HAS_REGEX
         if (pattern_ && !IsPatternMatch(*pattern_, str, length))

test/unittest/encodingstest.cpp

@@ -424,3 +424,14 @@ TEST(EncodingsTest, UTF32) {
         }
     }
 }
+
+TEST(EncodingsTest, CountStringCodePoint) {
+    SizeType count;
+    EXPECT_TRUE(CountStringCodePoint<UTF8<> >("", 0, &count));
+    EXPECT_EQ(0u, count);
+    EXPECT_TRUE(CountStringCodePoint<UTF8<> >("Hello", 5, &count));
+    EXPECT_EQ(5u, count);
+    EXPECT_TRUE(CountStringCodePoint<UTF8<> >("\xC2\xA2\xE2\x82\xAC\xF0\x9D\x84\x9E", 9, &count)); // cents euro G-clef
+    EXPECT_EQ(3u, count);
+    EXPECT_FALSE(CountStringCodePoint<UTF8<> >("\xC2\xA2\xE2\x82\xAC\xF0\x9D\x84\x9E\x80", 10, &count));
+}