Streaming API: detect UTF-8 BOM.

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 5fa2a2fb3a0af7e0102c0b5fc980281db03ab6d6..25c5049e45140155574f027fe8a29bf8d2f061d8 100644
--- a/test/cctest/test-api.cc
+++ b/test/cctest/test-api.cc
@@ -23848,15 +23848,31 @@ class TestSourceStream : public v8::ScriptCompiler::ExternalSourceStream {
   // too).
   static char* FullSourceString(const char** chunks) {
     size_t total_len = 0;
+    bool has_bom = false;
     for (size_t i = 0; chunks[i] != NULL; ++i) {
       total_len += strlen(chunks[i]);
+
+      // Remove BOM when constructing the full source string; this simluates
+      // what the embedder does.
+      if (i == 0 && strlen(chunks[i]) >= 3 &&
+          chunks[i][0] == static_cast<char>(0xef) &&
+          chunks[i][1] == static_cast<char>(0xbb) &&
+          chunks[i][2] == static_cast<char>(0xbf)) {
+        total_len -= 3;
+        has_bom = true;
+      }
     }
     char* full_string = new char[total_len + 1];
     size_t offset = 0;
     for (size_t i = 0; chunks[i] != NULL; ++i) {
       size_t len = strlen(chunks[i]);
-      memcpy(full_string + offset, chunks[i], len);
-      offset += len;
+      if (has_bom) {
+        memcpy(full_string + offset, chunks[i] + 3, len - 3);
+        offset += len - 3;
+      } else {
+        memcpy(full_string + offset, chunks[i], len);
+        offset += len;
+      }
     }
     full_string[total_len] = 0;
     return full_string;
@@ -23900,6 +23916,7 @@ void RunStreamingTest(const char** chunks,
     CHECK(!script.IsEmpty());
     v8::Handle<Value> result(script->Run());
     // All scripts are supposed to return the fixed value 13 when ran.
+    CHECK_EQ(false, try_catch.HasCaught());
     CHECK_EQ(13, result->Int32Value());
   } else {
     CHECK(script.IsEmpty());
@@ -24191,3 +24208,24 @@ TEST(StreamingUtf8ScriptWithMultipleMultibyteCharactersSomeSplit2) {
   const char* chunks[] = {chunk1, chunk2, "foo();", NULL};
   RunStreamingTest(chunks, v8::ScriptCompiler::StreamedSource::UTF8);
 }
+
+
+TEST(StreamingUtf8ScriptWithByteOrderMark) {
+  // Some Windows editors add the "UTF-8 BOM". If we don't handle it properly,
+  // some scripts might be parsed successfully but but the positions of lazy
+  // functions will be off.
+  i::FLAG_min_preparse_length = 0;
+  i::FLAG_lazy = true;
+  // Note that in this case the byte order mark doesn't actually cause a parse
+  // error, since it's a UTF-8 character followed by a comment.
+  char chunk1[] =
+      "XXX// That's the BOM.\n"
+      "function this_is_lazy() {\n"
+      "  return 13;\n"
+      "}\n";
+  chunk1[0] = static_cast<char>(0xef);
+  chunk1[1] = static_cast<char>(0xbb);
+  chunk1[2] = static_cast<char>(0xbf);
+  const char* chunks[] = {chunk1, "this_is_lazy();", NULL};
+  RunStreamingTest(chunks, v8::ScriptCompiler::StreamedSource::UTF8);
+}