Script streaming: fix split UTF-8 character handling.

test/cctest/test-api.cc

Index: test/cctest/test-api.cc
diff --git a/test/cctest/test-api.cc b/test/cctest/test-api.cc
index b4cf5cf4a9891e7442ac307a71b34bc761ac5927..66dc5a0c4ab0c8bb95f51eac826f971a448b3c66 100644
--- a/test/cctest/test-api.cc
+++ b/test/cctest/test-api.cc
@@ -23381,3 +23381,23 @@ TEST(StreamingProducesParserCache) {
   CHECK(cached_data->data != NULL);
   CHECK_GT(cached_data->length, 0);
 }
+
+
+TEST(StreamingScriptWithInvalidUtf8) {
+  // Regression test for a crash: test that invalid UTF-8 bytes in the end of a
+  // chunk don't produce a crash.
+  const char* reference = "\xeb\x91\x80\x80\x80";
+  char chunk1[] =
+      "function foo() {\n"
+      "  // This function will contain an UTF-8 character which is not in\n"
+      "  // ASCII.\n"
+      "  var foobXXXXX";  // Too many bytes which look like incomplete chars!
+  char chunk2[] =
+      "r = 13;\n"
+      "  return foob\xeb\x91\x80\x80\x80r;\n"
+      "}\n";
+  for (int i = 0; i < 5; ++i) chunk1[strlen(chunk1) - 5 + i] = reference[i];
+
+  const char* chunks[] = {chunk1, chunk2, "foo();", NULL};
+  RunStreamingTest(chunks, v8::ScriptCompiler::StreamedSource::UTF8, false);
+}