[builtins] Move builtins into own files

src/builtins/builtins-string.cc

+// Copyright 2016 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "src/builtins/builtins.h"
+#include "src/builtins/builtins-utils.h"
+
+#include "src/code-factory.h"
+
+namespace v8 {
+namespace internal {
+
+// -----------------------------------------------------------------------------
+// ES6 section 21.1 String Objects
+
+// ES6 section 21.1.2.1 String.fromCharCode ( ...codeUnits )
+void Builtins::Generate_StringFromCharCode(CodeStubAssembler* assembler) {
+  typedef CodeStubAssembler::Label Label;
+  typedef compiler::Node Node;
+  typedef CodeStubAssembler::Variable Variable;
+
+  Node* code = assembler->Parameter(1);
+  Node* context = assembler->Parameter(4);
+
+  // Check if we have exactly one argument (plus the implicit receiver), i.e.
+  // if the parent frame is not an arguments adaptor frame.
+  Label if_oneargument(assembler), if_notoneargument(assembler);
+  Node* parent_frame_pointer = assembler->LoadParentFramePointer();
+  Node* parent_frame_type =
+      assembler->Load(MachineType::Pointer(), parent_frame_pointer,
+                      assembler->IntPtrConstant(
+                          CommonFrameConstants::kContextOrFrameTypeOffset));
+  assembler->Branch(
+      assembler->WordEqual(
+          parent_frame_type,
+          assembler->SmiConstant(Smi::FromInt(StackFrame::ARGUMENTS_ADAPTOR))),
+      &if_notoneargument, &if_oneargument);
+
+  assembler->Bind(&if_oneargument);
+  {
+    // Single argument case, perform fast single character string cache lookup
+    // for one-byte code units, or fall back to creating a single character
+    // string on the fly otherwise.
+    Node* code32 = assembler->TruncateTaggedToWord32(context, code);
+    Node* code16 = assembler->Word32And(
+        code32, assembler->Int32Constant(String::kMaxUtf16CodeUnit));
+    Node* result = assembler->StringFromCharCode(code16);
+    assembler->Return(result);
+  }
+
+  assembler->Bind(&if_notoneargument);
+  {
+    // Determine the resulting string length.
+    Node* parent_frame_length =
+        assembler->Load(MachineType::Pointer(), parent_frame_pointer,
+                        assembler->IntPtrConstant(
+                            ArgumentsAdaptorFrameConstants::kLengthOffset));
+    Node* length = assembler->SmiToWord(parent_frame_length);
+
+    // Assume that the resulting string contains only one-byte characters.
+    Node* result = assembler->AllocateSeqOneByteString(context, length);
+
+    // Truncate all input parameters and append them to the resulting string.
+    Variable var_offset(assembler, MachineType::PointerRepresentation());
+    Label loop(assembler, &var_offset), done_loop(assembler);
+    var_offset.Bind(assembler->IntPtrConstant(0));
+    assembler->Goto(&loop);
+    assembler->Bind(&loop);
+    {
+      // Load the current {offset}.
+      Node* offset = var_offset.value();
+
+      // Check if we're done with the string.
+      assembler->GotoIf(assembler->WordEqual(offset, length), &done_loop);
+
+      // Load the next code point and truncate it to a 16-bit value.
+      Node* code = assembler->Load(
+          MachineType::AnyTagged(), parent_frame_pointer,
+          assembler->IntPtrAdd(
+              assembler->WordShl(assembler->IntPtrSub(length, offset),
+                                 assembler->IntPtrConstant(kPointerSizeLog2)),
+              assembler->IntPtrConstant(
+                  CommonFrameConstants::kFixedFrameSizeAboveFp -
+                  kPointerSize)));
+      Node* code32 = assembler->TruncateTaggedToWord32(context, code);
+      Node* code16 = assembler->Word32And(
+          code32, assembler->Int32Constant(String::kMaxUtf16CodeUnit));
+
+      // Check if {code16} fits into a one-byte string.
+      Label if_codeisonebyte(assembler), if_codeistwobyte(assembler);
+      assembler->Branch(
+          assembler->Int32LessThanOrEqual(
+              code16, assembler->Int32Constant(String::kMaxOneByteCharCode)),
+          &if_codeisonebyte, &if_codeistwobyte);
+
+      assembler->Bind(&if_codeisonebyte);
+      {
+        // The {code16} fits into the SeqOneByteString {result}.
+        assembler->StoreNoWriteBarrier(
+            MachineRepresentation::kWord8, result,
+            assembler->IntPtrAdd(
+                assembler->IntPtrConstant(SeqOneByteString::kHeaderSize -
+                                          kHeapObjectTag),
+                offset),
+            code16);
+        var_offset.Bind(
+            assembler->IntPtrAdd(offset, assembler->IntPtrConstant(1)));
+        assembler->Goto(&loop);
+      }
+
+      assembler->Bind(&if_codeistwobyte);
+      {
+        // Allocate a SeqTwoByteString to hold the resulting string.
+        Node* cresult = assembler->AllocateSeqTwoByteString(context, length);
+
+        // Copy all characters that were previously written to the
+        // SeqOneByteString in {result} over to the new {cresult}.
+        Variable var_coffset(assembler, MachineType::PointerRepresentation());
+        Label cloop(assembler, &var_coffset), done_cloop(assembler);
+        var_coffset.Bind(assembler->IntPtrConstant(0));
+        assembler->Goto(&cloop);
+        assembler->Bind(&cloop);
+        {
+          Node* coffset = var_coffset.value();
+          assembler->GotoIf(assembler->WordEqual(coffset, offset), &done_cloop);
+          Node* ccode = assembler->Load(
+              MachineType::Uint8(), result,
+              assembler->IntPtrAdd(
+                  assembler->IntPtrConstant(SeqOneByteString::kHeaderSize -
+                                            kHeapObjectTag),
+                  coffset));
+          assembler->StoreNoWriteBarrier(
+              MachineRepresentation::kWord16, cresult,
+              assembler->IntPtrAdd(
+                  assembler->IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                            kHeapObjectTag),
+                  assembler->WordShl(coffset, 1)),
+              ccode);
+          var_coffset.Bind(
+              assembler->IntPtrAdd(coffset, assembler->IntPtrConstant(1)));
+          assembler->Goto(&cloop);
+        }
+
+        // Write the pending {code16} to {offset}.
+        assembler->Bind(&done_cloop);
+        assembler->StoreNoWriteBarrier(
+            MachineRepresentation::kWord16, cresult,
+            assembler->IntPtrAdd(
+                assembler->IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                          kHeapObjectTag),
+                assembler->WordShl(offset, 1)),
+            code16);
+
+        // Copy the remaining parameters to the SeqTwoByteString {cresult}.
+        Label floop(assembler, &var_offset), done_floop(assembler);
+        assembler->Goto(&floop);
+        assembler->Bind(&floop);
+        {
+          // Compute the next {offset}.
+          Node* offset = assembler->IntPtrAdd(var_offset.value(),
+                                              assembler->IntPtrConstant(1));
+
+          // Check if we're done with the string.
+          assembler->GotoIf(assembler->WordEqual(offset, length), &done_floop);
+
+          // Load the next code point and truncate it to a 16-bit value.
+          Node* code = assembler->Load(
+              MachineType::AnyTagged(), parent_frame_pointer,
+              assembler->IntPtrAdd(
+                  assembler->WordShl(
+                      assembler->IntPtrSub(length, offset),
+                      assembler->IntPtrConstant(kPointerSizeLog2)),
+                  assembler->IntPtrConstant(
+                      CommonFrameConstants::kFixedFrameSizeAboveFp -
+                      kPointerSize)));
+          Node* code32 = assembler->TruncateTaggedToWord32(context, code);
+          Node* code16 = assembler->Word32And(
+              code32, assembler->Int32Constant(String::kMaxUtf16CodeUnit));
+
+          // Store the truncated {code} point at the next offset.
+          assembler->StoreNoWriteBarrier(
+              MachineRepresentation::kWord16, cresult,
+              assembler->IntPtrAdd(
+                  assembler->IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                            kHeapObjectTag),
+                  assembler->WordShl(offset, 1)),
+              code16);
+          var_offset.Bind(offset);
+          assembler->Goto(&floop);
+        }
+
+        // Return the SeqTwoByteString.
+        assembler->Bind(&done_floop);
+        assembler->Return(cresult);
+      }
+    }
+
+    assembler->Bind(&done_loop);
+    assembler->Return(result);
+  }
+}
+
+namespace {  // for String.fromCodePoint
+
+bool IsValidCodePoint(Isolate* isolate, Handle<Object> value) {
+  if (!value->IsNumber() && !Object::ToNumber(value).ToHandle(&value)) {
+    return false;
+  }
+
+  if (Object::ToInteger(isolate, value).ToHandleChecked()->Number() !=
+      value->Number()) {
+    return false;
+  }
+
+  if (value->Number() < 0 || value->Number() > 0x10FFFF) {
+    return false;
+  }
+
+  return true;
+}
+
+uc32 NextCodePoint(Isolate* isolate, BuiltinArguments args, int index) {
+  Handle<Object> value = args.at<Object>(1 + index);
+  ASSIGN_RETURN_ON_EXCEPTION_VALUE(isolate, value, Object::ToNumber(value), -1);
+  if (!IsValidCodePoint(isolate, value)) {
+    isolate->Throw(*isolate->factory()->NewRangeError(
+        MessageTemplate::kInvalidCodePoint, value));
+    return -1;
+  }
+  return DoubleToUint32(value->Number());
+}
+
+}  // namespace
+
+// ES6 section 21.1.2.2 String.fromCodePoint ( ...codePoints )
+BUILTIN(StringFromCodePoint) {
+  HandleScope scope(isolate);
+  int const length = args.length() - 1;
+  if (length == 0) return isolate->heap()->empty_string();
+  DCHECK_LT(0, length);
+
+  // Optimistically assume that the resulting String contains only one byte
+  // characters.
+  List<uint8_t> one_byte_buffer(length);
+  uc32 code = 0;
+  int index;
+  for (index = 0; index < length; index++) {
+    code = NextCodePoint(isolate, args, index);
+    if (code < 0) {
+      return isolate->heap()->exception();
+    }
+    if (code > String::kMaxOneByteCharCode) {
+      break;
+    }
+    one_byte_buffer.Add(code);
+  }
+
+  if (index == length) {
+    RETURN_RESULT_OR_FAILURE(isolate, isolate->factory()->NewStringFromOneByte(
+                                          one_byte_buffer.ToConstVector()));
+  }
+
+  List<uc16> two_byte_buffer(length - index);
+
+  while (true) {
+    if (code <= unibrow::Utf16::kMaxNonSurrogateCharCode) {
+      two_byte_buffer.Add(code);
+    } else {
+      two_byte_buffer.Add(unibrow::Utf16::LeadSurrogate(code));
+      two_byte_buffer.Add(unibrow::Utf16::TrailSurrogate(code));
+    }
+
+    if (++index == length) {
+      break;
+    }
+    code = NextCodePoint(isolate, args, index);
+    if (code < 0) {
+      return isolate->heap()->exception();
+    }
+  }
+
+  Handle<SeqTwoByteString> result;
+  ASSIGN_RETURN_FAILURE_ON_EXCEPTION(
+      isolate, result,
+      isolate->factory()->NewRawTwoByteString(one_byte_buffer.length() +
+                                              two_byte_buffer.length()));
+
+  CopyChars(result->GetChars(), one_byte_buffer.ToConstVector().start(),
+            one_byte_buffer.length());
+  CopyChars(result->GetChars() + one_byte_buffer.length(),
+            two_byte_buffer.ToConstVector().start(), two_byte_buffer.length());
+
+  return *result;
+}
+
+// ES6 section 21.1.3.1 String.prototype.charAt ( pos )
+void Builtins::Generate_StringPrototypeCharAt(CodeStubAssembler* assembler) {
+  typedef CodeStubAssembler::Label Label;
+  typedef compiler::Node Node;
+  typedef CodeStubAssembler::Variable Variable;
+
+  Node* receiver = assembler->Parameter(0);
+  Node* position = assembler->Parameter(1);
+  Node* context = assembler->Parameter(4);
+
+  // Check that {receiver} is coercible to Object and convert it to a String.
+  receiver =
+      assembler->ToThisString(context, receiver, "String.prototype.charAt");
+
+  // Convert the {position} to a Smi and check that it's in bounds of the
+  // {receiver}.
+  // TODO(bmeurer): Find an abstraction for this!
+  {
+    // Check if the {position} is already a Smi.
+    Variable var_position(assembler, MachineRepresentation::kTagged);
+    var_position.Bind(position);
+    Label if_positionissmi(assembler),
+        if_positionisnotsmi(assembler, Label::kDeferred);
+    assembler->Branch(assembler->WordIsSmi(position), &if_positionissmi,
+                      &if_positionisnotsmi);
+    assembler->Bind(&if_positionisnotsmi);
+    {
+      // Convert the {position} to an Integer via the ToIntegerStub.
+      Callable callable = CodeFactory::ToInteger(assembler->isolate());
+      Node* index = assembler->CallStub(callable, context, position);
+
+      // Check if the resulting {index} is now a Smi.
+      Label if_indexissmi(assembler, Label::kDeferred),
+          if_indexisnotsmi(assembler, Label::kDeferred);
+      assembler->Branch(assembler->WordIsSmi(index), &if_indexissmi,
+                        &if_indexisnotsmi);
+
+      assembler->Bind(&if_indexissmi);
+      {
+        var_position.Bind(index);
+        assembler->Goto(&if_positionissmi);
+      }
+
+      assembler->Bind(&if_indexisnotsmi);
+      {
+        // The ToIntegerStub canonicalizes everything in Smi range to Smi
+        // representation, so any HeapNumber returned is not in Smi range.
+        // The only exception here is -0.0, which we treat as 0.
+        Node* index_value = assembler->LoadHeapNumberValue(index);
+        Label if_indexiszero(assembler, Label::kDeferred),
+            if_indexisnotzero(assembler, Label::kDeferred);
+        assembler->Branch(assembler->Float64Equal(
+                              index_value, assembler->Float64Constant(0.0)),
+                          &if_indexiszero, &if_indexisnotzero);
+
+        assembler->Bind(&if_indexiszero);
+        {
+          var_position.Bind(assembler->SmiConstant(Smi::FromInt(0)));
+          assembler->Goto(&if_positionissmi);
+        }
+
+        assembler->Bind(&if_indexisnotzero);
+        {
+          // The {index} is some other integral Number, that is definitely
+          // neither -0.0 nor in Smi range.
+          assembler->Return(assembler->EmptyStringConstant());
+        }
+      }
+    }
+    assembler->Bind(&if_positionissmi);
+    position = var_position.value();
+
+    // Determine the actual length of the {receiver} String.
+    Node* receiver_length =
+        assembler->LoadObjectField(receiver, String::kLengthOffset);
+
+    // Return "" if the Smi {position} is outside the bounds of the {receiver}.
+    Label if_positioninbounds(assembler),
+        if_positionnotinbounds(assembler, Label::kDeferred);
+    assembler->Branch(assembler->SmiAboveOrEqual(position, receiver_length),
+                      &if_positionnotinbounds, &if_positioninbounds);
+    assembler->Bind(&if_positionnotinbounds);
+    assembler->Return(assembler->EmptyStringConstant());
+    assembler->Bind(&if_positioninbounds);
+  }
+
+  // Load the character code at the {position} from the {receiver}.
+  Node* code = assembler->StringCharCodeAt(receiver, position);
+
+  // And return the single character string with only that {code}.
+  Node* result = assembler->StringFromCharCode(code);
+  assembler->Return(result);
+}
+
+// ES6 section 21.1.3.2 String.prototype.charCodeAt ( pos )
+void Builtins::Generate_StringPrototypeCharCodeAt(
+    CodeStubAssembler* assembler) {
+  typedef CodeStubAssembler::Label Label;
+  typedef compiler::Node Node;
+  typedef CodeStubAssembler::Variable Variable;
+
+  Node* receiver = assembler->Parameter(0);
+  Node* position = assembler->Parameter(1);
+  Node* context = assembler->Parameter(4);
+
+  // Check that {receiver} is coercible to Object and convert it to a String.
+  receiver =
+      assembler->ToThisString(context, receiver, "String.prototype.charCodeAt");
+
+  // Convert the {position} to a Smi and check that it's in bounds of the
+  // {receiver}.
+  // TODO(bmeurer): Find an abstraction for this!
+  {
+    // Check if the {position} is already a Smi.
+    Variable var_position(assembler, MachineRepresentation::kTagged);
+    var_position.Bind(position);
+    Label if_positionissmi(assembler),
+        if_positionisnotsmi(assembler, Label::kDeferred);
+    assembler->Branch(assembler->WordIsSmi(position), &if_positionissmi,
+                      &if_positionisnotsmi);
+    assembler->Bind(&if_positionisnotsmi);
+    {
+      // Convert the {position} to an Integer via the ToIntegerStub.
+      Callable callable = CodeFactory::ToInteger(assembler->isolate());
+      Node* index = assembler->CallStub(callable, context, position);
+
+      // Check if the resulting {index} is now a Smi.
+      Label if_indexissmi(assembler, Label::kDeferred),
+          if_indexisnotsmi(assembler, Label::kDeferred);
+      assembler->Branch(assembler->WordIsSmi(index), &if_indexissmi,
+                        &if_indexisnotsmi);
+
+      assembler->Bind(&if_indexissmi);
+      {
+        var_position.Bind(index);
+        assembler->Goto(&if_positionissmi);
+      }
+
+      assembler->Bind(&if_indexisnotsmi);
+      {
+        // The ToIntegerStub canonicalizes everything in Smi range to Smi
+        // representation, so any HeapNumber returned is not in Smi range.
+        // The only exception here is -0.0, which we treat as 0.
+        Node* index_value = assembler->LoadHeapNumberValue(index);
+        Label if_indexiszero(assembler, Label::kDeferred),
+            if_indexisnotzero(assembler, Label::kDeferred);
+        assembler->Branch(assembler->Float64Equal(
+                              index_value, assembler->Float64Constant(0.0)),
+                          &if_indexiszero, &if_indexisnotzero);
+
+        assembler->Bind(&if_indexiszero);
+        {
+          var_position.Bind(assembler->SmiConstant(Smi::FromInt(0)));
+          assembler->Goto(&if_positionissmi);
+        }
+
+        assembler->Bind(&if_indexisnotzero);
+        {
+          // The {index} is some other integral Number, that is definitely
+          // neither -0.0 nor in Smi range.
+          assembler->Return(assembler->NaNConstant());
+        }
+      }
+    }
+    assembler->Bind(&if_positionissmi);
+    position = var_position.value();
+
+    // Determine the actual length of the {receiver} String.
+    Node* receiver_length =
+        assembler->LoadObjectField(receiver, String::kLengthOffset);
+
+    // Return NaN if the Smi {position} is outside the bounds of the {receiver}.
+    Label if_positioninbounds(assembler),
+        if_positionnotinbounds(assembler, Label::kDeferred);
+    assembler->Branch(assembler->SmiAboveOrEqual(position, receiver_length),
+                      &if_positionnotinbounds, &if_positioninbounds);
+    assembler->Bind(&if_positionnotinbounds);
+    assembler->Return(assembler->NaNConstant());
+    assembler->Bind(&if_positioninbounds);
+  }
+
+  // Load the character at the {position} from the {receiver}.
+  Node* value = assembler->StringCharCodeAt(receiver, position);
+  Node* result = assembler->SmiFromWord32(value);
+  assembler->Return(result);
+}
+
+// ES6 section 21.1.3.25 String.prototype.toString ()
+void Builtins::Generate_StringPrototypeToString(CodeStubAssembler* assembler) {
+  typedef compiler::Node Node;
+
+  Node* receiver = assembler->Parameter(0);
+  Node* context = assembler->Parameter(3);
+
+  Node* result = assembler->ToThisValue(
+      context, receiver, PrimitiveType::kString, "String.prototype.toString");
+  assembler->Return(result);
+}
+
+// ES6 section 21.1.3.27 String.prototype.trim ()
+BUILTIN(StringPrototypeTrim) {
+  HandleScope scope(isolate);
+  TO_THIS_STRING(string, "String.prototype.trim");
+  return *String::Trim(string, String::kTrim);
+}
+
+// Non-standard WebKit extension
+BUILTIN(StringPrototypeTrimLeft) {
+  HandleScope scope(isolate);
+  TO_THIS_STRING(string, "String.prototype.trimLeft");
+  return *String::Trim(string, String::kTrimLeft);
+}
+
+// Non-standard WebKit extension
+BUILTIN(StringPrototypeTrimRight) {
+  HandleScope scope(isolate);
+  TO_THIS_STRING(string, "String.prototype.trimRight");
+  return *String::Trim(string, String::kTrimRight);
+}
+
+// ES6 section 21.1.3.28 String.prototype.valueOf ( )
+void Builtins::Generate_StringPrototypeValueOf(CodeStubAssembler* assembler) {
+  typedef compiler::Node Node;
+
+  Node* receiver = assembler->Parameter(0);
+  Node* context = assembler->Parameter(3);
+
+  Node* result = assembler->ToThisValue(
+      context, receiver, PrimitiveType::kString, "String.prototype.valueOf");
+  assembler->Return(result);
+}
+
+}  // namespace internal
+}  // namespace v8