Clean up is-ASCII checks.

Review URL: http://codereview.chromium.org/5963003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6109 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/heap-inl.h

@@ -43,16 +43,14 @@ int Heap::MaxObjectSizeInPagedSpace() {
 MaybeObject* Heap::AllocateStringFromUtf8(Vector<const char> str,
                                           PretenureFlag pretenure) {
   // Check for ASCII first since this is the common case.
-  for (int i = 0; i < str.length(); ++i) {
+  if (String::IsAscii(str.start(), str.length())) {
-    if (static_cast<uint8_t>(str[i]) > String::kMaxAsciiCharCodeU) {
-      // Non-ASCII and we need to decode.
-      return AllocateStringFromUtf8Slow(str, pretenure);
-    }
-  }
     // If the string is ASCII, we do not need to convert the characters
     // since UTF8 is backwards compatible with ASCII.
     return AllocateStringFromAscii(str, pretenure);
   }
+  // Non-ASCII and we need to decode.
+  return AllocateStringFromUtf8Slow(str, pretenure);
+}
 
 
 MaybeObject* Heap::AllocateSymbol(Vector<const char> str,

src/heap.cc

@@ -2549,20 +2549,10 @@ MaybeObject* Heap::AllocateExternalStringFromTwoByte(
   }
 
   // For small strings we check whether the resource contains only
-  // ascii characters.  If yes, we use a different string map.
+  // ASCII characters.  If yes, we use a different string map.
-  bool is_ascii = true;
+  static const size_t kAsciiCheckLengthLimit = 32;
-  if (length >= static_cast<size_t>(String::kMinNonFlatLength)) {
+  bool is_ascii = length <= kAsciiCheckLengthLimit &&
-    is_ascii = false;
+      String::IsAscii(resource->data(), length);
-  } else {
-    const uc16* data = resource->data();
-    for (size_t i = 0; i < length; i++) {
-      if (data[i] > String::kMaxAsciiCharCode) {
-        is_ascii = false;
-        break;
-      }
-    }
-  }
-
   Map* map = is_ascii ?
       Heap::external_string_with_ascii_data_map() : Heap::external_string_map();
   Object* result;
@@ -3342,11 +3332,8 @@ MaybeObject* Heap::AllocateStringFromUtf8Slow(Vector<const char> string,
 MaybeObject* Heap::AllocateStringFromTwoByte(Vector<const uc16> string,
                                              PretenureFlag pretenure) {
   // Check if the string is an ASCII string.
-  int i = 0;
-  while (i < string.length() && string[i] <= String::kMaxAsciiCharCode) i++;
-
   MaybeObject* maybe_result;
-  if (i == string.length()) {  // It's an ASCII string.
+  if (String::IsAscii(string.start(), string.length())) {
     maybe_result = AllocateRawAsciiString(string.length(), pretenure);
   } else {  // It's not an ASCII string.
     maybe_result = AllocateRawTwoByteString(string.length(), pretenure);

src/ia32/regexp-macro-assembler-ia32.cc

@@ -211,9 +211,7 @@ void RegExpMacroAssemblerIA32::CheckCharacters(Vector<const uc16> str,
   // If input is ASCII, don't even bother calling here if the string to
   // match contains a non-ascii character.
   if (mode_ == ASCII) {
-    for (int i = 0; i < str.length(); i++) {
+    ASSERT(String::IsAscii(str.start(), str.length()));
-      ASSERT(str[i] <= String::kMaxAsciiCharCodeU);
-    }
   }
 #endif
   int byte_length = str.length() * char_size();

src/objects.h

@@ -5245,6 +5245,34 @@ class String: public HeapObject {
                           int from,
                           int to);
 
+  static inline bool IsAscii(const char* chars, int length) {
+    const char* limit = chars + length;
+#ifdef V8_HOST_CAN_READ_UNALIGNED
+    ASSERT(kMaxAsciiCharCode == 0x7F);
+    const uintptr_t non_ascii_mask = kUintptrAllBitsSet / 0xFF * 0x80;
+    while (chars <= limit - sizeof(uintptr_t)) {
+      if (*reinterpret_cast<const uintptr_t*>(chars) & non_ascii_mask) {
+        return false;
+      }
+      chars += sizeof(uintptr_t);
+    }
+#endif
+    while (chars < limit) {
+      if (static_cast<uint8_t>(*chars) > kMaxAsciiCharCodeU) return false;
+      ++chars;
+    }
+    return true;
+  }
+
+  static inline bool IsAscii(const uc16* chars, int length) {
+    const uc16* limit = chars + length;
+    while (chars < limit) {
+      if (*chars > kMaxAsciiCharCodeU) return false;
+      ++chars;
+    }
+    return true;
+  }
+
  protected:
   class ReadBlockBuffer {
    public:

src/string-search.h

@@ -66,12 +66,7 @@ class StringSearchBase {
   }
 
   static inline bool IsAsciiString(Vector<const uc16> string) {
-    for (int i = 0, n = string.length(); i < n; i++) {
+    return String::IsAscii(string.start(), string.length());
-      if (static_cast<unsigned>(string[i]) > String::kMaxAsciiCharCodeU) {
-        return false;
-      }
-    }
-    return true;
   }
 
   // The following tables are shared by all searches.

src/x64/regexp-macro-assembler-x64.cc

@@ -223,9 +223,7 @@ void RegExpMacroAssemblerX64::CheckCharacters(Vector<const uc16> str,
   // If input is ASCII, don't even bother calling here if the string to
   // match contains a non-ascii character.
   if (mode_ == ASCII) {
-    for (int i = 0; i < str.length(); i++) {
+    ASSERT(String::IsAscii(str.start(), str.length()));
-      ASSERT(str[i] <= String::kMaxAsciiCharCodeU);
-    }
   }
 #endif
   int byte_length = str.length() * char_size();