Validate utf8

src/fuzz.cpp

@@ -1,6 +1,9 @@
+#include "callbacks.h"
 #include "minify.h"
 #include "parser3.h"
 
+#include <simdjson.h>
+
 std::pair<std::string, parser3::Status> runStreaming(std::string copy) {
   MinifyState state;
   auto c = minifyCallbacks();
@@ -41,7 +44,52 @@ void testStreaming(std::string const &json) {
   }
 }
 
+void compareWithSimdjson(std::string const &json) {
+  parser3::Status ours;
+  {
+    auto copy = json;
+    auto c = noopCallbacks();
+    parser3::Parser3 parser3(&c, nullptr);
+    ours = parser3.parse(copy.data(), copy.size());
+    if (ours == parser3::S_AGAIN) {
+      ours = parser3.parse(nullptr, 0);
+    }
+  }
+
+  using namespace simdjson;
+  simdjson::padded_string my_padded_data(json.data(), json.size());
+  simdjson::dom::parser parser;
+  auto doc = parser.parse(my_padded_data);
+  auto theirs = doc.error();
+  if (ours == parser3::S_OVERFLOW || theirs == simdjson::DEPTH_ERROR) {
+    return;
+  }
+  if ((ours == parser3::S_OK) != (theirs == simdjson::SUCCESS)) {
+    if (json.starts_with("\xef\xbb\xbf")) {
+      // What to do with byte order mark?
+      return;
+    }
+    if (theirs == simdjson::NUMBER_ERROR || theirs == simdjson::BIGINT_ERROR) {
+      // This gets returned for precision errors sometimes?
+      return;
+    }
+    if (theirs == simdjson::STRING_ERROR) {
+      // why god why god do I gotta suffer
+      return;
+    }
+    if (theirs == simdjson::NUMBER_OUT_OF_RANGE) {
+      // We don't validate the precision of numbers
+      return;
+    }
+    printf("ours: %d\n", ours);
+    printf("theirs: %d\n", theirs);
+    abort();
+  }
+}
+
 extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
-  testStreaming(std::string((const char *)data, size));
+  auto s = std::string((const char *)data, size);
+  testStreaming(s);
+  compareWithSimdjson(s);
   return 0;
 }