Add automata to recognize utf8 in strings

src/fuzz.cpp

@@ -1,5 +1,6 @@
 #include "callbacks.h"
 #include "json_value.h"
+#include "parser3.h"
 #include "weaseljson.h"
 
 #include <simdjson.h>
@@ -132,5 +133,19 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
   testStreaming(s);
   compareWithSimdjson(s);
   testStringRoundTrip(s);
+  bool json_utf8 = true;
+  for (int i = 0; i < int(size); ++i) {
+    uint8_t c = data[i];
+    json_utf8 = json_utf8 && c >= 0x20 && c != '"' && c != '\\';
+  }
+  if (json_utf8) {
+    parser3::Utf8Dfa dfa;
+    auto result = dfa.scan((const char *)data, (const char *)data + size);
+    bool ok = result == (const char *)data + size && dfa.accept();
+    bool valid = simdjson::validate_utf8(s.data(), s.size());
+    if (ok != valid) {
+      abort();
+    }
+  }
   return 0;
 }