Tentative implementation of string slices (hidden under the flag --string-slices).

src/objects.cc

Index: src/objects.cc
diff --git a/src/objects.cc b/src/objects.cc
index a423ae4c25df2caaff6ade6de2f2684f94286522..3ffd6898103c91e863896c324f04d217b8be9c1b 100644
--- a/src/objects.cc
+++ b/src/objects.cc
@@ -1169,6 +1169,9 @@ void HeapObject::IterateBody(InstanceType type, int object_size,
       case kConsStringTag:
         ConsString::BodyDescriptor::IterateBody(this, v);
         break;
+      case kSlicedStringTag:
+        SlicedString::BodyDescriptor::IterateBody(this, v);
+        break;
       case kExternalStringTag:
         if ((type & kStringEncodingMask) == kAsciiStringTag) {
           reinterpret_cast<ExternalAsciiString*>(this)->
@@ -5246,14 +5249,21 @@ Vector<const char> String::ToAsciiVector() {
     string = cons->first();
     string_tag = StringShape(string).representation_tag();

[antonm, 2011/07/27 14:04:49]

not related to your change, but I am curious what will happen if have the
following structure:

cons string
  first -> const string
  second -> empty

[Yang, 2011/07/28 15:43:06]

Ditto. Said structure is impossible to get. I'll add an assert here.

On 2011/07/27 14:04:49, antonm wrote:

   }
-  if (string_tag == kSeqStringTag) {
-    SeqAsciiString* seq = SeqAsciiString::cast(string);
-    char* start = seq->GetChars();
+  if (string_tag == kExternalStringTag) {
+    // Note that a sliced string cannot contain an external string.

[antonm, 2011/07/27 14:04:49]

maybe be more elaborate why it's needed.

+    ExternalAsciiString* ext = ExternalAsciiString::cast(string);
+    const char* start = ext->resource()->data();
     return Vector<const char>(start + offset, length);
   }
-  ASSERT(string_tag == kExternalStringTag);
-  ExternalAsciiString* ext = ExternalAsciiString::cast(string);
-  const char* start = ext->resource()->data();
+  if (string_tag == kSlicedStringTag) {
+    SlicedString* slice = SlicedString::cast(string);
+    offset = slice->offset();
+    string = slice->parent();
+    string_tag = StringShape(string).representation_tag();
+  }
+  ASSERT(string_tag == kSeqStringTag);
+  SeqAsciiString* seq = SeqAsciiString::cast(string);
+  char* start = seq->GetChars();
   return Vector<const char>(start + offset, length);
 }
 
@@ -5272,15 +5282,22 @@ Vector<const uc16> String::ToUC16Vector() {
     string = cons->first();
     string_tag = StringShape(string).representation_tag();
   }
-  if (string_tag == kSeqStringTag) {
-    SeqTwoByteString* seq = SeqTwoByteString::cast(string);
-    return Vector<const uc16>(seq->GetChars() + offset, length);
+  if (string_tag == kExternalStringTag) {
+    // Note that a sliced string cannot contain an external string.
+    ExternalTwoByteString* ext = ExternalTwoByteString::cast(string);
+    const uc16* start =
+        reinterpret_cast<const uc16*>(ext->resource()->data());
+    return Vector<const uc16>(start + offset, length);
+  }
+  if (string_tag == kSlicedStringTag) {
+    SlicedString* slice = SlicedString::cast(string);
+    offset = slice->offset();
+    string = slice->parent();
+    string_tag = StringShape(string).representation_tag();
   }
-  ASSERT(string_tag == kExternalStringTag);
-  ExternalTwoByteString* ext = ExternalTwoByteString::cast(string);
-  const uc16* start =
-      reinterpret_cast<const uc16*>(ext->resource()->data());
-  return Vector<const uc16>(start + offset, length);
+  ASSERT(string_tag == kSeqStringTag);
+  SeqTwoByteString* seq = SeqTwoByteString::cast(string);
+  return Vector<const uc16>(seq->GetChars() + offset, length);
 }
 
 
@@ -5358,6 +5375,10 @@ const uc16* String::GetTwoByteData(unsigned start) {
     case kExternalStringTag:
       return ExternalTwoByteString::cast(this)->
         ExternalTwoByteStringGetData(start);
+    case kSlicedStringTag: {
+      SlicedString* slice = SlicedString::cast(this);
+      return slice->parent()->GetTwoByteData(start + slice->offset());
+    }
     case kConsStringTag:
       UNREACHABLE();
       return NULL;
@@ -5635,6 +5656,10 @@ const unibrow::byte* String::ReadBlock(String* input,
       return ConsString::cast(input)->ConsStringReadBlock(rbb,
                                                           offset_ptr,
                                                           max_chars);
+    case kSlicedStringTag:
+      return SlicedString::cast(input)->SlicedStringReadBlock(rbb,
+                                                              offset_ptr,
+                                                              max_chars);
     case kExternalStringTag:
       if (input->IsAsciiRepresentation()) {
         return ExternalAsciiString::cast(input)->ExternalAsciiStringReadBlock(
@@ -5778,6 +5803,11 @@ void String::ReadBlockIntoBuffer(String* input,
                                                              offset_ptr,
                                                              max_chars);
       return;
+    case kSlicedStringTag:
+      SlicedString::cast(input)->SlicedStringReadBlockIntoBuffer(rbb,
+                                                                 offset_ptr,
+                                                                 max_chars);
+      return;
     case kExternalStringTag:
       if (input->IsAsciiRepresentation()) {
         ExternalAsciiString::cast(input)->
@@ -5923,6 +5953,31 @@ uint16_t ConsString::ConsStringGet(int index) {
 }
 
 
+uint16_t SlicedString::SlicedStringGet(int index) {
+  return parent()->Get(offset() + index);
+}
+
+
+const unibrow::byte* SlicedString::SlicedStringReadBlock(
+    ReadBlockBuffer* buffer, unsigned* offset_ptr, unsigned chars) {
+  unsigned offset = this->offset();
+  *offset_ptr += offset;

[antonm, 2011/07/27 14:04:49]

looks sketchy, but I am not sure I am aware of better solution.

+  const unibrow::byte* answer = String::ReadBlock(String::cast(parent()),
+                                                  buffer, offset_ptr, chars);
+  *offset_ptr -= offset;
+  return answer;
+}
+
+
+void SlicedString::SlicedStringReadBlockIntoBuffer(
+    ReadBlockBuffer* buffer, unsigned* offset_ptr, unsigned chars) {
+  unsigned offset = this->offset();
+  *offset_ptr += offset;
+  String::ReadBlockIntoBuffer(String::cast(parent()),
+                              buffer, offset_ptr, chars);
+  *offset_ptr -= offset;
+}
+
 template <typename sinkchar>
 void String::WriteToFlat(String* src,
                          sinkchar* sink,
@@ -5990,6 +6045,13 @@ void String::WriteToFlat(String* src,
         }
         break;
       }
+      case kAsciiStringTag | kSlicedStringTag:
+      case kTwoByteStringTag | kSlicedStringTag: {
+        SlicedString* slice = SlicedString::cast(source);
+        unsigned offset = slice->offset();
+        WriteToFlat(slice->parent(), sink, from + offset, to + offset);
+        return;
+      }
     }
   }
 }