[regexp] Create property on result for each named capture

src/builtins/builtins-regexp.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-constructor.h"
 #include "src/builtins/builtins-utils.h"
 #include "src/builtins/builtins.h"
 #include "src/code-factory.h"
 #include "src/code-stub-assembler.h"
 #include "src/regexp/jsregexp.h"
@@ skipping 15 matching lines @@
  protected:
   Node* FastLoadLastIndex(Node* regexp);
   Node* SlowLoadLastIndex(Node* context, Node* regexp);
   Node* LoadLastIndex(Node* context, Node* regexp, bool is_fastpath);
 
   void FastStoreLastIndex(Node* regexp, Node* value);
   void SlowStoreLastIndex(Node* context, Node* regexp, Node* value);
   void StoreLastIndex(Node* context, Node* regexp, Node* value,
                       bool is_fastpath);
 
-  Node* ConstructNewResultFromMatchInfo(Node* context, Node* match_info,
-                                        Node* string);
+  Node* ConstructNewResultFromMatchInfo(Node* const context, Node* const regexp,
+                                        Node* const match_info,
+                                        Node* const string);
 
   Node* RegExpPrototypeExecBodyWithoutResult(Node* const context,
                                              Node* const regexp,
                                              Node* const string,
                                              Label* if_didnotmatch,
                                              const bool is_fastpath);
   Node* RegExpPrototypeExecBody(Node* const context, Node* const regexp,
                                 Node* const string, const bool is_fastpath);
 
   Node* ThrowIfNotJSReceiver(Node* context, Node* maybe_receiver,
@@ skipping 86 matching lines @@
 
 void RegExpBuiltinsAssembler::StoreLastIndex(Node* context, Node* regexp,
                                              Node* value, bool is_fastpath) {
   if (is_fastpath) {
     FastStoreLastIndex(regexp, value);
   } else {
     SlowStoreLastIndex(context, regexp, value);
   }
 }
 
-Node* RegExpBuiltinsAssembler::ConstructNewResultFromMatchInfo(Node* context,
-                                                               Node* match_info,
-                                                               Node* string) {
-  Label out(this);
+Node* RegExpBuiltinsAssembler::ConstructNewResultFromMatchInfo(
+    Node* const context, Node* const regexp, Node* const match_info,
+    Node* const string) {
+  Label named_captures(this), out(this);
 
   Node* const num_indices = SmiUntag(LoadFixedArrayElement(
       match_info, RegExpMatchInfo::kNumberOfCapturesIndex));
   Node* const num_results = SmiTag(WordShr(num_indices, 1));
   Node* const start =
       LoadFixedArrayElement(match_info, RegExpMatchInfo::kFirstCaptureIndex);
   Node* const end = LoadFixedArrayElement(
       match_info, RegExpMatchInfo::kFirstCaptureIndex + 1);
 
   // Calculate the substring of the first match before creating the result array
   // to avoid an unnecessary write barrier storing the first result.
   Node* const first = SubString(context, string, start, end);
 
   Node* const result =
       AllocateRegExpResult(context, num_results, start, string);
   Node* const result_elements = LoadElements(result);
 
   StoreFixedArrayElement(result_elements, 0, first, SKIP_WRITE_BARRIER);
 
-  GotoIf(SmiEqual(num_results, SmiConstant(Smi::FromInt(1))), &out);
+  // If no captures exist we can skip named capture handling as well.
+  GotoIf(SmiEqual(num_results, SmiConstant(1)), &out);
 
   // Store all remaining captures.
   Node* const limit = IntPtrAdd(
       IntPtrConstant(RegExpMatchInfo::kFirstCaptureIndex), num_indices);
 
   Variable var_from_cursor(this, MachineType::PointerRepresentation());
   Variable var_to_cursor(this, MachineType::PointerRepresentation());
 
   var_from_cursor.Bind(IntPtrConstant(RegExpMatchInfo::kFirstCaptureIndex + 2));
   var_to_cursor.Bind(IntPtrConstant(1));
 
   Variable* vars[] = {&var_from_cursor, &var_to_cursor};
   Label loop(this, 2, vars);
 
   Goto(&loop);
   Bind(&loop);
   {
     Node* const from_cursor = var_from_cursor.value();
     Node* const to_cursor = var_to_cursor.value();
     Node* const start = LoadFixedArrayElement(match_info, from_cursor);
 
     Label next_iter(this);
-    GotoIf(SmiEqual(start, SmiConstant(Smi::FromInt(-1))), &next_iter);
+    GotoIf(SmiEqual(start, SmiConstant(-1)), &next_iter);
 
     Node* const from_cursor_plus1 = IntPtrAdd(from_cursor, IntPtrConstant(1));
     Node* const end = LoadFixedArrayElement(match_info, from_cursor_plus1);
 
     Node* const capture = SubString(context, string, start, end);
     StoreFixedArrayElement(result_elements, to_cursor, capture);
     Goto(&next_iter);
 
     Bind(&next_iter);
     var_from_cursor.Bind(IntPtrAdd(from_cursor, IntPtrConstant(2)));
     var_to_cursor.Bind(IntPtrAdd(to_cursor, IntPtrConstant(1)));
-    Branch(UintPtrLessThan(var_from_cursor.value(), limit), &loop, &out);
+    Branch(UintPtrLessThan(var_from_cursor.value(), limit), &loop,
+           &named_captures);
+  }
+
+  Bind(&named_captures);
+  {
+    // We reach this point only if captures exist, implying that this is an
+    // IRREGEXP JSRegExp.
+
+    CSA_ASSERT(this, HasInstanceType(regexp, JS_REGEXP_TYPE));
+    CSA_ASSERT(this, SmiGreaterThan(num_results, SmiConstant(1)));
+
+    // Preparations for named capture properties. Exit early if the result does
+    // not have any named captures to minimize performance impact.
+
+    Node* const data = LoadObjectField(regexp, JSRegExp::kDataOffset);
+    CSA_ASSERT(this, SmiEqual(LoadFixedArrayElement(data, JSRegExp::kTagIndex),
+                              SmiConstant(JSRegExp::IRREGEXP)));
+
+    // The names fixed array associates names at even indices with a capture
+    // index at odd indices.
+    Node* const names =
+        LoadFixedArrayElement(data, JSRegExp::kIrregexpCaptureNameMapIndex);
+    GotoIf(SmiEqual(names, SmiConstant(0)), &out);
+
+    // Allocate a new object to store the named capture properties.
+    // TODO(jgruber): Could be optimized by adding the object map to the heap
+    // root list.
+
+    Node* const native_context = LoadNativeContext(context);
+    Node* const object_function =
+        LoadContextElement(native_context, Context::OBJECT_FUNCTION_INDEX);
+    Node* const object_function_map = LoadObjectField(
+        object_function, JSFunction::kPrototypeOrInitialMapOffset);
+
+    Node* const group_object = AllocateJSObjectFromMap(object_function_map);
+
+    // Store it on the result as a 'group' property.
+
+    {
+      Node* const name = HeapConstant(isolate()->factory()->group_string());
+      Node* const language_mode = SmiConstant(Smi::FromInt(STRICT));
+      CallRuntime(Runtime::kSetProperty, context, result, name, group_object,
+                  language_mode);
+    }
+
+    // One or more named captures exist, add a property for each one.
+
+    CSA_ASSERT(this, HasInstanceType(names, FIXED_ARRAY_TYPE));
+    Node* const names_length = LoadAndUntagFixedArrayBaseLength(names);
+    CSA_ASSERT(this, IntPtrGreaterThan(names_length, IntPtrConstant(0)));
+
+    Variable var_i(this, MachineType::PointerRepresentation());
+    var_i.Bind(IntPtrConstant(0));
+
+    Variable* vars[] = {&var_i};
+    const int vars_count = sizeof(vars) / sizeof(vars[0]);
+    Label loop(this, vars_count, vars);
+
+    Goto(&loop);
+    Bind(&loop);
+    {
+      Node* const i = var_i.value();
+      Node* const i_plus_1 = IntPtrAdd(i, IntPtrConstant(1));
+      Node* const i_plus_2 = IntPtrAdd(i_plus_1, IntPtrConstant(1));
+
+      Node* const name = LoadFixedArrayElement(names, i);
+      Node* const index = LoadFixedArrayElement(names, i_plus_1);
+      Node* const capture =
+          LoadFixedArrayElement(result_elements, SmiUntag(index));
+
+      Node* const language_mode = SmiConstant(Smi::FromInt(STRICT));
+      CallRuntime(Runtime::kSetProperty, context, group_object, name, capture,
+                  language_mode);

[Dan Ehrenberg, 2017/01/20 18:23:46]

We should really be calling DefineOwnProperty here, not SetProperty. SetProperty
will invoke setters that a user might define on Object.prototype, which is not
the semantics that I'd expect. Generally, the RegExp part of JS calls
CreateDataProperty, which is based on Define, not Set.

+
+      var_i.Bind(i_plus_2);
+      Branch(IntPtrGreaterThanOrEqual(var_i.value(), names_length), &out,
+             &loop);
+    }
   }
 
   Bind(&out);
   return result;
 }
 
 // ES#sec-regexp.prototype.exec
 // RegExp.prototype.exec ( string )
 // Implements the core of RegExp.prototype.exec but without actually
 // constructing the JSRegExpResult. Returns either null (if the RegExp did not
@@ skipping 132 matching lines @@
   Variable var_result(this, MachineRepresentation::kTagged);
 
   Label if_didnotmatch(this), out(this);
   Node* const indices_or_null = RegExpPrototypeExecBodyWithoutResult(
       context, regexp, string, &if_didnotmatch, is_fastpath);
 
   // Successful match.
   {
     Node* const match_indices = indices_or_null;
     Node* const result =
-        ConstructNewResultFromMatchInfo(context, match_indices, string);
+        ConstructNewResultFromMatchInfo(context, regexp, match_indices, string);
     var_result.Bind(result);
     Goto(&out);
   }
 
   Bind(&if_didnotmatch);
   {
     var_result.Bind(null);
     Goto(&out);
   }
 
@@ skipping 2149 matching lines @@
 
   Label if_matched(this), if_didnotmatch(this);
   Branch(WordEqual(match_indices, null), &if_didnotmatch, &if_matched);
 
   Bind(&if_didnotmatch);
   Return(null);
 
   Bind(&if_matched);
   {
     Node* result =
-        ConstructNewResultFromMatchInfo(context, match_indices, string);
+        ConstructNewResultFromMatchInfo(context, regexp, match_indices, string);
     Return(result);
   }
 }
 
 }  // namespace internal
 }  // namespace v8