Fix #122

peglib.h

@@ -180,11 +180,11 @@ inline size_t codepoint_length(const char *s8, size_t l) {
     auto b = static_cast<uint8_t>(s8[0]);
     if ((b & 0x80) == 0) {
       return 1;
-    } else if ((b & 0xE0) == 0xC0) {
+    } else if ((b & 0xE0) == 0xC0 && l >= 2) {
       return 2;
-    } else if ((b & 0xF0) == 0xE0) {
+    } else if ((b & 0xF0) == 0xE0 && l >= 3) {
       return 3;
-    } else if ((b & 0xF8) == 0xF0) {
+    } else if ((b & 0xF8) == 0xF0 && l >= 4) {
       return 4;
     }
   }

test/test1.cc

@@ -45,6 +45,17 @@ TEST_CASE("Start rule with ignore operator test", "[general]")
     REQUIRE(ret == false);
 }
 
+TEST_CASE("Invalid UTF-8 text test", "[general]")
+{
+    std::string s = "a <- '";
+    s += static_cast<char>(0xe8); // Make invalid utf8 text...
+
+    parser parser(s.c_str());
+
+    bool ret = parser;
+    REQUIRE(ret == false);
+}
+
 TEST_CASE("Backslash escape sequence test", "[general]")
 {
     parser parser(R"(