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..745305f8302a4b8c4f3d7baaa8c6542becf0de9b 100644
--- a/src/objects.cc
+++ b/src/objects.cc
@@ -733,62 +733,56 @@ MaybeObject* String::SlowTryFlatten(PretenureFlag pretenure) {
 #ifdef DEBUG
   // Do not attempt to flatten in debug mode when allocation is not
   // allowed.  This is to avoid an assertion failure when allocating.
-  // Flattening strings is the only case where we always allow
-  // allocation because no GC is performed if the allocation fails.
+  // Flattening and truncating strings are the only cases where we always
+  // allow allocation because no GC is performed if the allocation fails.
   if (!HEAP->IsAllocationAllowed()) return this;
 #endif
 
+  ASSERT(StringShape(this).representation_tag() == kConsStringTag);
+
   Heap* heap = GetHeap();
-  switch (StringShape(this).representation_tag()) {
-    case kConsStringTag: {
-      ConsString* cs = ConsString::cast(this);
-      if (cs->second()->length() == 0) {
-        return cs->first();
-      }
-      // There's little point in putting the flat string in new space if the
-      // cons string is in old space.  It can never get GCed until there is
-      // an old space GC.
-      PretenureFlag tenure = heap->InNewSpace(this) ? pretenure : TENURED;
-      int len = length();
-      Object* object;
-      String* result;
-      if (IsAsciiRepresentation()) {
-        { MaybeObject* maybe_object = heap->AllocateRawAsciiString(len, tenure);
-          if (!maybe_object->ToObject(&object)) return maybe_object;
-        }
-        result = String::cast(object);
-        String* first = cs->first();
-        int first_length = first->length();
-        char* dest = SeqAsciiString::cast(result)->GetChars();
-        WriteToFlat(first, dest, 0, first_length);
-        String* second = cs->second();
-        WriteToFlat(second,
-                    dest + first_length,
-                    0,
-                    len - first_length);
-      } else {
-        { MaybeObject* maybe_object =
-              heap->AllocateRawTwoByteString(len, tenure);
-          if (!maybe_object->ToObject(&object)) return maybe_object;
-        }
-        result = String::cast(object);
-        uc16* dest = SeqTwoByteString::cast(result)->GetChars();
-        String* first = cs->first();
-        int first_length = first->length();
-        WriteToFlat(first, dest, 0, first_length);
-        String* second = cs->second();
-        WriteToFlat(second,
-                    dest + first_length,
-                    0,
-                    len - first_length);
-      }
-      cs->set_first(result);
-      cs->set_second(heap->empty_string());
-      return result;
-    }
-    default:
-      return this;
-  }
+  ConsString* cs = ConsString::cast(this);

[Vitaly Repeshko, 2011/08/17 19:20:23]

If there are no intended semantic changes here, please revert and keep the old
version. 

+  // There's little point in putting the flat string in new space if the
+  // cons string is in old space.  It can never get GCed until there is
+  // an old space GC.
+  PretenureFlag tenure = heap->InNewSpace(this) ? pretenure : TENURED;
+  int len = length();
+  Object* object;
+  String* result;
+  if (IsAsciiRepresentation()) {
+    { MaybeObject* maybe_object = heap->AllocateRawAsciiString(len, tenure);
+      if (!maybe_object->ToObject(&object)) return maybe_object;
+    }
+    result = String::cast(object);
+    String* first = cs->first();
+    int first_length = first->length();
+    char* dest = SeqAsciiString::cast(result)->GetChars();
+    WriteToFlat(first, dest, 0, first_length);
+    String* second = cs->second();
+    WriteToFlat(second,
+                dest + first_length,
+                0,
+                len - first_length);
+  } else {
+    { MaybeObject* maybe_object =
+          heap->AllocateRawTwoByteString(len, tenure);
+      if (!maybe_object->ToObject(&object)) return maybe_object;
+    }
+    result = String::cast(object);
+    uc16* dest = SeqTwoByteString::cast(result)->GetChars();
+    String* first = cs->first();
+    int first_length = first->length();
+    WriteToFlat(first, dest, 0, first_length);
+    String* second = cs->second();
+    WriteToFlat(second,
+                dest + first_length,
+                0,
+                len - first_length);
+  }
+  cs->set_first(result);
+  cs->set_second(heap->empty_string());
+  ASSERT(this->IsFlat());
+  return result;
 }
 
 
@@ -1169,6 +1163,9 @@ 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)->
@@ -5245,15 +5242,22 @@ Vector<const char> String::ToAsciiVector() {
     ASSERT(cons->second()->length() == 0);
     string = cons->first();
     string_tag = StringShape(string).representation_tag();
+    ASSERT(string_tag != kConsStringTag);
   }
-  if (string_tag == kSeqStringTag) {
-    SeqAsciiString* seq = SeqAsciiString::cast(string);
-    char* start = seq->GetChars();
+  if (string_tag == kSlicedStringTag) {
+    // Note that the parent of a slice cannot be a cons.

[Vitaly Repeshko, 2011/08/17 19:20:23]

... or a slice.

+    SlicedString* slice = SlicedString::cast(string);
+    offset = slice->offset();
+    string = slice->parent();

[Vitaly Repeshko, 2011/08/17 19:20:23]

string_tag also should be updated. Either add a TODO or fix now.

+  }
+  if (string_tag == kExternalStringTag) {
+    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();
+  ASSERT(StringShape(string).representation_tag() == kSeqStringTag);
+  SeqAsciiString* seq = SeqAsciiString::cast(string);
+  char* start = seq->GetChars();
   return Vector<const char>(start + offset, length);
 }
 
@@ -5271,16 +5275,23 @@ Vector<const uc16> String::ToUC16Vector() {
     ASSERT(cons->second()->length() == 0);
     string = cons->first();
     string_tag = StringShape(string).representation_tag();
+    ASSERT(string_tag != kConsStringTag);
+  }
+  if (string_tag == kSlicedStringTag) {
+    // Note that the parent of a slice cannot be a cons.

[Vitaly Repeshko, 2011/08/17 19:20:23]

See above.

+    SlicedString* slice = SlicedString::cast(string);
+    offset = slice->offset();
+    string = slice->parent();
   }
-  if (string_tag == kSeqStringTag) {
-    SeqTwoByteString* seq = SeqTwoByteString::cast(string);
-    return Vector<const uc16>(seq->GetChars() + offset, length);
+  if (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 == kExternalStringTag);
-  ExternalTwoByteString* ext = ExternalTwoByteString::cast(string);
-  const uc16* start =
-      reinterpret_cast<const uc16*>(ext->resource()->data());
-  return Vector<const uc16>(start + offset, length);
+  ASSERT(StringShape(string).representation_tag() == kSeqStringTag);
+  SeqTwoByteString* seq = SeqTwoByteString::cast(string);
+  return Vector<const uc16>(seq->GetChars() + offset, length);
 }
 
 
@@ -5358,6 +5369,10 @@ 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;
@@ -5648,6 +5663,10 @@ const unibrow::byte* String::ReadBlock(String* input,
                                                      max_chars);
         return rbb->util_buffer;
       }
+    case kSlicedStringTag:
+      return SlicedString::cast(input)->SlicedStringReadBlock(rbb,
+                                                              offset_ptr,
+                                                              max_chars);
     default:
       break;
   }
@@ -5789,6 +5808,11 @@ void String::ReadBlockIntoBuffer(String* input,
                                                      max_chars);
        }
        return;
+    case kSlicedStringTag:
+      SlicedString::cast(input)->SlicedStringReadBlockIntoBuffer(rbb,
+                                                                 offset_ptr,
+                                                                 max_chars);
+      return;
     default:
       break;
   }
@@ -5923,6 +5947,31 @@ 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;
+  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 +6039,13 @@ 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;
+      }
     }
   }
 }