Validate that names are valid UTF-8 (#6682)

Add an `isUTF8` utility and use it in both the text and binary parsers.
Add missing checks for overlong encodings and overlarge code points in
our WTF8 reader, which the new utility uses. Re-enable the spec tests
that test UTF-8 validation.

2 files changed, 25 insertions, 0 deletions

diff --git a/src/support/string.cpp b/src/support/string.cpp
index 31d0e9170..01fe4e522 100644
--- a/src/support/string.cpp
+++ b/src/support/string.cpp
@@ -195,9 +195,21 @@ std::optional<uint32_t> takeWTF8CodePoint(std::string_view& str) {
   }
 
   str = str.substr(1 + trailingBytes);
+
   if (!valid) {
     return std::nullopt;
   }
+
+  size_t expectedTrailing = u < 0x80       ? 0
+                            : u < 0x800    ? 1
+                            : u < 0x10000  ? 2
+                            : u < 0x110000 ? 3
+                                           : -1;
+  if (trailingBytes != expectedTrailing) {
+    // Overlong encoding or overlarge code point.
+    return std::nullopt;
+  }
+
   return u;
 }
 
@@ -404,4 +416,14 @@ std::ostream& printEscapedJSON(std::ostream& os, std::string_view str) {
   return os << '"';
 }
 
+bool isUTF8(std::string_view str) {
+  while (str.size()) {
+    auto u = takeWTF8CodePoint(str);
+    if (!u || (0xD800 <= *u && *u < 0xE000)) {
+      return false;
+    }
+  }
+  return true;
+}
+
 } // namespace wasm::String
diff --git a/src/support/string.h b/src/support/string.h
index af120ab4e..24eb570c2 100644
--- a/src/support/string.h
+++ b/src/support/string.h
@@ -99,6 +99,9 @@ bool convertWTF16ToWTF8(std::ostream& os, std::string_view str);
 // unit. Returns `true` if the input was valid UTF-16.
 bool convertUTF16ToUTF8(std::ostream& os, std::string_view str);
 
+// Whether the string is valid UTF-8.
+bool isUTF8(std::string_view str);
+
 } // namespace wasm::String
 
 #endif // wasm_support_string_h