add function to test whether string contents are definitely one byte

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/16530003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14976 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

include/v8.h

@@ -1524,10 +1524,18 @@ class V8EXPORT String : public Primitive {
   V8_DEPRECATED(V8_INLINE(bool MayContainNonAscii()) const) { return true; }
 
   /**
-   * Returns whether this string contains only one byte data.
+   * Returns whether this string is known to contain only one byte data.
+   * Does not read the string.
+   * False negatives are possible.
    */
   bool IsOneByte() const;
 
+  /**
+   * Returns whether this string contain only one byte data.
+   * Will read the entire string in some cases.
+   */
+  bool ContainsOnlyOneByte() const;
+
   /**
    * Write the contents of the string to an external buffer.
    * If no arguments are given, expects the buffer to be large

src/api.cc

@@ -4304,6 +4304,85 @@ bool String::IsOneByte() const {
 }
 
 
+class ContainsOnlyOneByteHelper {
+ public:
+  ContainsOnlyOneByteHelper() : is_one_byte_(true) {}
+  bool Check(i::String* string) {
+    i::ConsString* cons_string = i::String::VisitFlat(this, string, 0);
+    if (cons_string == NULL) return is_one_byte_;
+    return CheckCons(cons_string);
+  }
+  void VisitOneByteString(const uint8_t* chars, int length) {
+    // Nothing to do.
+  }
+  // TODO(dcarney): do word aligned read.
+  void VisitTwoByteString(const uint16_t* chars, int length) {
+    // Check whole string without breaking.
+    uint16_t total = 0;
+    for (int i = 0; i < length; i++) {
+      total |= chars[i] >> 8;
+    }
+    if (total != 0) is_one_byte_ = false;
+  }
+
+ private:
+  bool CheckCons(i::ConsString* cons_string) {
+    while (true) {
+      // Check left side if flat.
+      i::String* left = cons_string->first();
+      i::ConsString* left_as_cons =
+          i::String::VisitFlat(this, left, 0);
+      if (!is_one_byte_) return false;
+      // Check right side if flat.
+      i::String* right = cons_string->second();
+      i::ConsString* right_as_cons =
+          i::String::VisitFlat(this, right, 0);
+      if (!is_one_byte_) return false;
+      // Standard recurse/iterate trick.
+      if (left_as_cons != NULL && right_as_cons != NULL) {
+        if (left->length() < right->length()) {
+          CheckCons(left_as_cons);
+          cons_string = right_as_cons;
+        } else {
+          CheckCons(right_as_cons);
+          cons_string = left_as_cons;
+        }
+        // Check fast return.
+        if (!is_one_byte_) return false;
+        continue;
+      }
+      // Descend left in place.
+      if (left_as_cons != NULL) {
+        cons_string = left_as_cons;
+        continue;
+      }
+      // Descend right in place.
+      if (right_as_cons != NULL) {
+        cons_string = right_as_cons;
+        continue;
+      }
+      // Terminate.
+      break;
+    }
+    return is_one_byte_;
+  }
+  bool is_one_byte_;
+  DISALLOW_COPY_AND_ASSIGN(ContainsOnlyOneByteHelper);
+};
+
+
+bool String::ContainsOnlyOneByte() const {
+  i::Handle<i::String> str = Utils::OpenHandle(this);
+  if (IsDeadCheck(str->GetIsolate(),
+                  "v8::String::ContainsOnlyOneByte()")) {
+    return false;
+  }
+  if (str->HasOnlyOneByteChars()) return true;
+  ContainsOnlyOneByteHelper helper;
+  return helper.Check(*str);
+}
+
+
 class Utf8LengthHelper : public i::AllStatic {
  public:
   enum State {

src/objects.cc

@@ -1205,6 +1205,11 @@ bool String::MakeExternal(v8::String::ExternalAsciiStringResource* resource) {
   if (FLAG_enable_slow_asserts) {
     // Assert that the resource and the string are equivalent.
     ASSERT(static_cast<size_t>(this->length()) == resource->length());
+    if (this->IsTwoByteRepresentation()) {
+      ScopedVector<uint16_t> smart_chars(this->length());
+      String::WriteToFlat(this, smart_chars.start(), 0, this->length());
+      ASSERT(String::IsOneByte(smart_chars.start(), this->length()));
+    }
     ScopedVector<char> smart_chars(this->length());
     String::WriteToFlat(this, smart_chars.start(), 0, this->length());
     ASSERT(memcmp(smart_chars.start(),

test/cctest/test-api.cc

@@ -16881,6 +16881,51 @@ THREADED_TEST(TwoByteStringInAsciiCons) {
 }
 
 
+TEST(ContainsOnlyOneByte) {
+  v8::V8::Initialize();
+  v8::Isolate* isolate = v8::Isolate::GetCurrent();
+  v8::HandleScope scope(isolate);
+  // Make a buffer long enough that it won't automatically be converted.
+  const int length = 200;
+  i::SmartArrayPointer<uint16_t> string_contents(new uint16_t[length]);
+  // Set to contain only one byte.
+  for (int i = 0; i < length-1; i++) {
+    string_contents[i] = 0x41;
+  }
+  string_contents[length-1] = 0;
+  // Simple case.
+  Handle<String> string;
+  string = String::NewExternal(new TestResource(*string_contents));
+  CHECK(!string->IsOneByte() && string->ContainsOnlyOneByte());
+  // Counter example.
+  string = String::NewFromTwoByte(isolate, *string_contents);
+  CHECK(string->IsOneByte() && string->ContainsOnlyOneByte());
+  // Test left right and balanced cons strings.
+  Handle<String> base = String::NewFromUtf8(isolate, "a");
+  Handle<String> left = base;
+  Handle<String> right = base;
+  for (int i = 0; i < 1000; i++) {
+    left = String::Concat(base, left);
+    right = String::Concat(right, base);
+  }
+  Handle<String> balanced = String::Concat(left, base);
+  balanced = String::Concat(balanced, right);
+  Handle<String> cons_strings[] = {left, balanced, right};
+  Handle<String> two_byte =
+      String::NewExternal(new TestResource(*string_contents));
+  for (size_t i = 0; i < ARRAY_SIZE(cons_strings); i++) {
+    // Base assumptions.
+    string = cons_strings[i];
+    CHECK(string->IsOneByte() && string->ContainsOnlyOneByte());
+    // Test left and right concatentation.
+    string = String::Concat(two_byte, cons_strings[i]);
+    CHECK(!string->IsOneByte() && string->ContainsOnlyOneByte());
+    string = String::Concat(cons_strings[i], two_byte);
+    CHECK(!string->IsOneByte() && string->ContainsOnlyOneByte());
+  }
+}
+
+
 // Failed access check callback that performs a GC on each invocation.
 void FailedAccessCheckCallbackGC(Local<v8::Object> target,
                                  v8::AccessType type,