[regexp] Create property on result for each named capture

src/builtins/builtins-regexp.cc

Index: src/builtins/builtins-regexp.cc
diff --git a/src/builtins/builtins-regexp.cc b/src/builtins/builtins-regexp.cc
index 2191268441c92687715ff86cbfecd9774b770ebc..229cd5d3d84931f0507f531fa900831db9a14558 100644
--- a/src/builtins/builtins-regexp.cc
+++ b/src/builtins/builtins-regexp.cc
@@ -33,8 +33,9 @@ class RegExpBuiltinsAssembler : public CodeStubAssembler {
   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,
@@ -141,10 +142,10 @@ void RegExpBuiltinsAssembler::StoreLastIndex(Node* context, Node* regexp,
   }
 }
 
-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));
@@ -164,7 +165,7 @@ Node* RegExpBuiltinsAssembler::ConstructNewResultFromMatchInfo(Node* context,
 
   StoreFixedArrayElement(result_elements, 0, first, SKIP_WRITE_BARRIER);
 
-  GotoIf(SmiEqual(num_results, SmiConstant(Smi::FromInt(1))), &out);
+  GotoIf(SmiEqual(num_results, SmiConstant(1)), &named_captures);
 
   // Store all remaining captures.
   Node* const limit = IntPtrAdd(
@@ -187,7 +188,7 @@ Node* RegExpBuiltinsAssembler::ConstructNewResultFromMatchInfo(Node* context,
     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);
@@ -199,7 +200,60 @@ Node* RegExpBuiltinsAssembler::ConstructNewResultFromMatchInfo(Node* context,
     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);
+  {
+    // Preparations for named capture properties. Exit early if the result is
+    // either not an IRREGEXP or does not have any named captures to minimize
+    // performance impact.
+
+    CSA_ASSERT(this, HasInstanceType(regexp, JS_REGEXP_TYPE));
+    Node* const data = LoadObjectField(regexp, JSRegExp::kDataOffset);

[jgruber, 2017/01/17 09:26:51]

This unconditionally adds two loads and two branches to each exec path. 

I'd like to see the impact before optimizing; packing a new internal flag into
JSRegExp::Flags isn't completely trivial since e.g. flags are sometimes
reconstructed from the flags string.

[Yang, 2017/01/18 13:25:36]

I think we can already make this a bit simpler: we only need to care about named
captures if we have any captures at all. If we have captures, then it cannot be
an atom regexp, so the tag check is redundant too.

[jgruber, 2017/01/19 15:59:12]

Very good point, done. 

+    Node* const tag = LoadFixedArrayElement(data, JSRegExp::kTagIndex);
+    GotoUnless(SmiEqual(tag, SmiConstant(JSRegExp::IRREGEXP)), &out);
+
+    // 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);
+
+    // 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, result, name, capture,
+                  language_mode);
+
+      var_i.Bind(i_plus_2);
+      Branch(IntPtrGreaterThanOrEqual(var_i.value(), names_length), &out,
+             &loop);
+    }
   }
 
   Bind(&out);
@@ -352,7 +406,7 @@ Node* RegExpBuiltinsAssembler::RegExpPrototypeExecBody(Node* const context,
   {
     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);
   }
@@ -2522,7 +2576,7 @@ TF_BUILTIN(RegExpInternalMatch, RegExpBuiltinsAssembler) {
   Bind(&if_matched);
   {
     Node* result =
-        ConstructNewResultFromMatchInfo(context, match_indices, string);
+        ConstructNewResultFromMatchInfo(context, regexp, match_indices, string);
     Return(result);
   }
 }