Add ES2015 RegExp full subclassing semantics behind a flag

src/js/regexp.js

 // Copyright 2012 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.
 
 (function(global, utils) {
 
 'use strict';
 
 %CheckIsBootstrapping();
 
 // -------------------------------------------------------------------
 // Imports
 
+var AddIndexedProperty;
 var ExpandReplacement;
+var GlobalArray = global.Array;
 var GlobalObject = global.Object;
 var GlobalRegExp = global.RegExp;
 var GlobalRegExpPrototype;
 var InternalArray = utils.InternalArray;
 var InternalPackedArray = utils.InternalPackedArray;
 var MakeTypeError;
+var MathMax = global.Math.max;
+var MathMin = global.Math.min;
 var matchSymbol = utils.ImportNow("match_symbol");
 var replaceSymbol = utils.ImportNow("replace_symbol");
 var searchSymbol = utils.ImportNow("search_symbol");
 var splitSymbol = utils.ImportNow("split_symbol");
+var SpeciesConstructor;
 
 utils.Import(function(from) {
+  AddIndexedProperty = from.AddIndexedProperty;
   ExpandReplacement = from.ExpandReplacement;
   MakeTypeError = from.MakeTypeError;
+  SpeciesConstructor = from.SpeciesConstructor;
 });
 
 // -------------------------------------------------------------------
 
 // Property of the builtins object for recording the result of the last
 // regexp match.  The property RegExpLastMatchInfo includes the matchIndices
 // array of the last successful regexp match (an array of start/end index
 // pairs for the match and all the captured substrings), the invariant is
 // that there are at least two capture indeces.  The array also contains
 // the subject string for the last successful match.
 var RegExpLastMatchInfo = new InternalPackedArray(
  2,                 // REGEXP_NUMBER_OF_CAPTURES
  "",                // Last subject.
  UNDEFINED,         // Last input - settable with RegExpSetInput.
  0,                 // REGEXP_FIRST_CAPTURE + 0
  0                  // REGEXP_FIRST_CAPTURE + 1
 );
 
 // -------------------------------------------------------------------
 
+// ES6 7.2.8
 function IsRegExp(o) {
   if (!IS_RECEIVER(o)) return false;
   var is_regexp = o[matchSymbol];
   if (!IS_UNDEFINED(is_regexp)) return TO_BOOLEAN(is_regexp);
   return IS_REGEXP(o);
 }
 
 
 // ES6 section 21.2.3.2.2
 function RegExpInitialize(object, pattern, flags) {
@@ skipping 99 matching lines @@
   if (matchInfo !== null) {
     // ES6 21.2.5.2.2 step 18.
     if (REGEXP_STICKY(regexp)) regexp.lastIndex = matchInfo[CAPTURE1];
     RETURN_NEW_RESULT_FROM_MATCH_INFO(matchInfo, string);
   }
   regexp.lastIndex = 0;
   return null;
 }
 
 
+function RegExpSubclassExecJS(string) {
+  if (!IS_REGEXP(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        'RegExp.prototype.exec', this);
+  }
+
+  string = TO_STRING(string);
+  var lastIndex = this.lastIndex;
+
+  // Conversion is required by the ES2015 specification (RegExpBuiltinExec
+  // algorithm, step 4) even if the value is discarded for non-global RegExps.
+  var i = TO_LENGTH(lastIndex);
+
+  var global = TO_BOOLEAN(this.global);
+  var sticky = TO_BOOLEAN(this.sticky);
+  var updateLastIndex = global || sticky;
+  if (updateLastIndex) {
+    if (i < 0 || i > string.length) {
+      this.lastIndex = 0;
+      return null;
+    }
+  } else {
+    i = 0;
+  }
+
+  // matchIndices is either null or the RegExpLastMatchInfo array.
+  // TODO(littledan): Whether a RegExp is sticky is compiled into the RegExp
+  // itself, but ES2015 allows monkey-patching this property to differ from
+  // the internal flags. If it differs, recompile a different RegExp?
+  var matchIndices = %_RegExpExec(this, string, i, RegExpLastMatchInfo);
+
+  if (IS_NULL(matchIndices)) {
+    this.lastIndex = 0;
+    return null;
+  }
+
+  // Successful match.
+  if (updateLastIndex) {
+    this.lastIndex = RegExpLastMatchInfo[CAPTURE1];
+  }
+  RETURN_NEW_RESULT_FROM_MATCH_INFO(matchIndices, string);
+}
+%FunctionRemovePrototype(RegExpSubclassExecJS);

[Yang, 2016/03/22 12:49:38]

Like previously noted, let's always do this in OverrideFunction.

[Dan Ehrenberg, 2016/03/22 18:26:31]

See v8:4567 for the bug that this is working around.

+
+
 function RegExpExecJS(string) {
   if (!IS_REGEXP(this)) {
     throw MakeTypeError(kIncompatibleMethodReceiver,
                         'RegExp.prototype.exec', this);
   }
 
   string = TO_STRING(string);
   var lastIndex = this.lastIndex;
 
   // Conversion is required by the ES2015 specification (RegExpBuiltinExec
@@ skipping 19 matching lines @@
   }
 
   // Successful match.
   if (updateLastIndex) {
     this.lastIndex = RegExpLastMatchInfo[CAPTURE1];
   }
   RETURN_NEW_RESULT_FROM_MATCH_INFO(matchIndices, string);
 }
 
 
+function RegExpSubclassExec(regexp, string) {
+  var exec = regexp.exec;
+  if (IS_CALLABLE(exec)) {
+    var result = %_Call(exec, regexp, string);
+    if (!IS_OBJECT(result) && !IS_NULL(result)) {
+      throw MakeTypeError(kInvalidRegExpExecResult);
+    }
+    return result;
+  }
+  return %_Call(RegExpExecJS, regexp, string);
+}
+
+
 // One-element cache for the simplified test regexp.
 var regexp_key;
 var regexp_val;
 
 // Section 15.10.6.3 doesn't actually make sense, but the intention seems to be
 // that test is defined in terms of String.prototype.exec. However, it probably
 // means the original value of String.prototype.exec, which is what everybody
 // else implements.
 function RegExpTest(string) {
   if (!IS_REGEXP(this)) {
@@ skipping 37 matching lines @@
     // matchIndices is either null or the RegExpLastMatchInfo array.
     var matchIndices = %_RegExpExec(regexp, string, 0, RegExpLastMatchInfo);
     if (IS_NULL(matchIndices)) {
       this.lastIndex = 0;
       return false;
     }
     return true;
   }
 }
 
+function RegExpSubclassTest(string) {
+  if (!IS_OBJECT(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        'RegExp.prototype.test', this);
+  }
+  string = TO_STRING(string);
+  var match = RegExpSubclassExec(this, string);
+  return !IS_NULL(match);
+}
+%FunctionRemovePrototype(RegExpSubclassTest);
+
 function TrimRegExp(regexp) {
   if (regexp_key !== regexp) {
     regexp_key = regexp;
     regexp_val =
       new GlobalRegExp(
           %_SubString(REGEXP_SOURCE(regexp), 2, REGEXP_SOURCE(regexp).length),
           (REGEXP_IGNORE_CASE(regexp) ? REGEXP_MULTILINE(regexp) ? "im" : "i"
                                       : REGEXP_MULTILINE(regexp) ? "m" : ""));
   }
   return regexp_val;
@@ skipping 101 matching lines @@
 
     startIndex = currentIndex = endIndex;
   }
 
   var array_result = [];
   %MoveArrayContents(result, array_result);
   return array_result;
 }
 
 
+function RegExpSubclassSplit(string, limit) {
+  if (!IS_RECEIVER(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        "RegExp.prototype.@@split", this);
+  }
+  string = TO_STRING(string);
+  var constructor = SpeciesConstructor(this, GlobalRegExp);
+  var flags = TO_STRING(this.flags);
+  var unicode = %StringIndexOf(flags, 'u', 0) >= 0;
+  var sticky = %StringIndexOf(flags, 'y', 0) >= 0;
+  var new_flags = sticky ? flags : flags + "y";
+  var splitter = new constructor(this, new_flags);
+  var array = new GlobalArray();
+  var array_index = 0;
+  var lim = (IS_UNDEFINED(limit)) ? kMaxUint32 : TO_UINT32(limit);
+  var size = string.length;
+  var prev_string_index = 0;
+  if (lim === 0) return array;
+  var result;
+  if (size === 0) {
+    result = RegExpSubclassExec(splitter, string);
+    if (IS_NULL(result)) AddIndexedProperty(array, 0, string);
+    return array;
+  }
+  var string_index = prev_string_index;
+  while (string_index < size) {
+    splitter.lastIndex = string_index;
+    result = RegExpSubclassExec(splitter, string);
+    if (IS_NULL(result)) {
+      string_index += GetUnicodeAdvancedIncrement(string, string_index,
+                                                  unicode);
+    } else {
+      var end = MathMin(splitter.lastIndex, size);
+      if (end === prev_string_index) {
+        string_index += GetUnicodeAdvancedIncrement(string, string_index,
+                                                    unicode);
+      } else {
+        AddIndexedProperty(
+            array, array_index,
+            %_SubString(string, prev_string_index, string_index));
+        array_index++;
+        if (array_index === lim) return array;
+        prev_string_index = end;
+        var number_of_captures = MathMax(TO_LENGTH(result.length), 0);
+        for (var i = 1; i < number_of_captures; i++) {
+          AddIndexedProperty(array, array_index, result[i]);
+          array_index++;
+          if (array_index === lim) return array;
+        }
+        string_index = prev_string_index;
+      }
+    }
+  }
+  AddIndexedProperty(array, array_index,
+                     %_SubString(string, prev_string_index, size));
+  return array;
+}
+%FunctionRemovePrototype(RegExpSubclassSplit);
+
+
 // ES6 21.2.5.6.
 function RegExpMatch(string) {
-  // TODO(yangguo): allow non-regexp receivers.
   if (!IS_REGEXP(this)) {
     throw MakeTypeError(kIncompatibleMethodReceiver,
                         "RegExp.prototype.@@match", this);
   }
   var subject = TO_STRING(string);
 
   if (!REGEXP_GLOBAL(this)) return RegExpExecNoTests(this, subject, 0);
   this.lastIndex = 0;
   var result = %StringMatch(subject, this, RegExpLastMatchInfo);
   return result;
 }
 
 
+function RegExpSubclassMatch(string) {
+  if (!IS_OBJECT(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        "RegExp.prototype.@@match", this);
+  }
+  string = TO_STRING(string);
+  var global = this.global;
+  if (!global) return RegExpSubclassExec(this, string);
+  var unicode = this.unicode;
+  this.lastIndex = 0;
+  var array = [];
+  var n = 0;
+  var result;
+  while (true) {
+    result = RegExpSubclassExec(this, string);
+    if (IS_NULL(result)) {
+      if (n === 0) return null;
+      return array;
+    }
+    var matchStr = TO_STRING(result[0]);
+    %AddElement(array, n, matchStr);
+    if (matchStr === "") AdvanceStringIndex(this, string, unicode);
+    n++;
+  }
+}
+%FunctionRemovePrototype(RegExpSubclassMatch);
+
+
 // ES6 21.2.5.8.
 
 // TODO(lrn): This array will survive indefinitely if replace is never
 // called again. However, it will be empty, since the contents are cleared
 // in the finally block.
 var reusableReplaceArray = new InternalArray(4);
 
 // Helper function for replacing regular expressions with the result of a
 // function application in String.prototype.replace.
 function StringReplaceGlobalRegExpWithFunction(subject, regexp, replace) {
@@ skipping 107 matching lines @@
   }
 
   result += replacement;  // The add method converts to string if necessary.
   // Can't use matchInfo any more from here, since the function could
   // overwrite it.
   return result + %_SubString(subject, endOfMatch, subject.length);
 }
 
 
 function RegExpReplace(string, replace) {
-  // TODO(littledan): allow non-regexp receivers.
   if (!IS_REGEXP(this)) {
     throw MakeTypeError(kIncompatibleMethodReceiver,
                         "RegExp.prototype.@@replace", this);
   }
   var subject = TO_STRING(string);
   var search = this;
 
   if (!IS_CALLABLE(replace)) {
     replace = TO_STRING(replace);
 
@@ skipping 21 matching lines @@
 
   if (REGEXP_GLOBAL(search)) {
     // Global regexp search, function replace.
     return StringReplaceGlobalRegExpWithFunction(subject, search, replace);
   }
   // Non-global regexp search, function replace.
   return StringReplaceNonGlobalRegExpWithFunction(subject, search, replace);
 }
 
 
+// Expand the $-expressions in the string and return a new string with
+// the result.
+function GetCaptures(matched, string, position, captures, replacement) {
+  var match_length = matched.length;
+  var string_length = string.length;
+  var captures_length = captures.length;
+  var tail_pos = position + match_length;
+  var result = "";
+  var pos, expansion, peek, next, scaled_index, advance, new_scaled_index;
+
+  var next = %StringIndexOf(replacement, '$', 0);
+  if (next < 0) {
+    result += replacement;
+    return result;
+  }
+
+  if (next > 0) result += %_SubString(replacement, 0, next);
+
+  while (true) {
+    expansion = '$';
+    pos = next + 1;
+    if (pos < replacement.length) {
+      peek = %_StringCharCodeAt(replacement, pos);
+      if (peek == 36) {         // $$
+        ++pos;
+        result += '$';
+      } else if (peek == 38) {  // $& - match
+        ++pos;
+        result += matched;
+      } else if (peek == 96) {  // $` - prefix
+        ++pos;
+        result += %_SubString(string, 0, position);
+      } else if (peek == 39) {  // $' - suffix
+        ++pos;
+        result += %_SubString(string, tail_pos, string_length);
+      } else if (peek >= 48 && peek <= 57) {
+        // Valid indices are $1 .. $9, $01 .. $09 and $10 .. $99
+        scaled_index = (peek - 48);
+        advance = 1;
+        if (pos + 1 < replacement.length) {
+          next = %_StringCharCodeAt(replacement, pos + 1);
+          if (next >= 48 && next <= 57) {
+            new_scaled_index = scaled_index * 10 + ((next - 48));
+            if (new_scaled_index < captures_length) {
+              scaled_index = new_scaled_index;
+              advance = 2;
+            }
+          }
+        }
+        if (scaled_index != 0 && scaled_index < captures_length) {
+          var capture = captures[scaled_index];
+          if (!IS_UNDEFINED(capture)) result += capture;
+          pos += advance;
+        } else {
+          result += '$';
+        }
+      } else {
+        result += '$';
+      }
+    } else {
+      result += '$';
+    }
+
+    // Go the the next $ in the replacement.
+    next = %StringIndexOf(replacement, '$', pos);
+
+    // Return if there are no more $ characters in the replacement. If we
+    // haven't reached the end, we need to append the suffix.
+    if (next < 0) {
+      if (pos < replacement.length) {
+        result += %_SubString(replacement, pos, replacement.length);
+      }
+      return result;
+    }
+
+    // Append substring between the previous and the next $ character.
+    if (next > pos) {
+      result += %_SubString(replacement, pos, next);
+    }
+  }
+  return result;
+}
+
+
+function GetUnicodeAdvancedIncrement(string, index, unicode) {
+  var increment = 1;
+  if (unicode) {
+    var first = %_StringCharCodeAt(string, index);
+    if (first >= 0xD800 && first <= 0xDBFF && string.length > index + 1) {
+      var second = %_StringCharCodeAt(string, index + 1);
+      if (second >= 0xDC00 && second <= 0xDFFF) {
+        increment = 2;
+      }
+    }
+  }
+  return increment;
+}
+
+
+function AdvanceStringIndex(regexp, string, unicode) {
+  var last_index = regexp.lastIndex;
+  regexp.lastIndex = last_index +
+      GetUnicodeAdvancedIncrement(string, last_index, unicode);
+}
+
+
+function RegExpSubclassReplace(string, replace) {
+  if (!IS_OBJECT(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        "RegExp.prototype.@@replace", this);
+  }
+  string = TO_STRING(string);
+  var length = string.length;
+  var functional_replace = IS_CALLABLE(replace);
+  if (!functional_replace) replace = TO_STRING(replace);
+  var global = this.global;
+  if (global) {
+    var unicode = this.unicode;
+    this.lastIndex = 0;
+  }
+  var results = new InternalArray();
+  var result, replacement;
+  while (true) {
+    result = RegExpSubclassExec(this, string);
+    if (IS_NULL(result)) {
+      break;
+    } else {
+      results.push(result);
+      if (!global) break;
+      var match_str = TO_STRING(result[0]);
+      if (match_str === "") AdvanceStringIndex(this, string, unicode);
+    }
+  }
+  var accumulated_result = "";
+  var next_source_position = 0;
+  for (var i = 0; i < results.length; i++) {
+    result = results[i];
+    var captures_length = MathMax(TO_LENGTH(result.length), 0);
+    var matched = TO_STRING(result[0]);
+    var matched_length = matched.length;
+    var position = MathMax(MathMin(TO_INTEGER(result.index), length), 0);
+    var captures = new InternalArray();
+    for (var n = 0; n < captures_length; n++) {
+      var capture = result[n];
+      if (!IS_UNDEFINED(capture)) capture = TO_STRING(capture);
+      captures[n] = capture;
+    }
+    if (functional_replace) {
+      var parameters = new InternalArray(captures_length + 2);
+      for (var j = 0; j < captures_length; j++) {
+        parameters[j] = captures[j];
+      }
+      parameters[j] = position;
+      parameters[j + 1] = string;
+      replacement = %reflect_apply(replace, UNDEFINED, parameters, 0,
+                                   parameters.length);
+    } else {
+      replacement = GetCaptures(matched, string, position, captures, replace);
+    }
+    if (position >= next_source_position) {
+      accumulated_result +=
+        %_SubString(string, next_source_position, position) + replacement;
+      next_source_position = position + matched_length;
+    }
+  }
+  if (next_source_position >= length) return accumulated_result;
+  return accumulated_result + %_SubString(string, next_source_position, length);
+}
+%FunctionRemovePrototype(RegExpSubclassReplace);
+
+
 // ES6 21.2.5.9.
 function RegExpSearch(string) {
-  // TODO(yangguo): allow non-regexp receivers.
   if (!IS_REGEXP(this)) {
     throw MakeTypeError(kIncompatibleMethodReceiver,
                         "RegExp.prototype.@@search", this);
   }
   var match = DoRegExpExec(this, TO_STRING(string), 0);
   if (match) return match[CAPTURE0];
   return -1;
 }
 
 
+function RegExpSubclassSearch(string) {
+  if (!IS_OBJECT(this)) {
+    throw MakeTypeError(kIncompatibleMethodReceiver,
+                        "RegExp.prototype.@@search", this);
+  }
+  string = TO_STRING(string);
+  var previousLastIndex = this.lastIndex;
+  this.lastIndex = 0;
+  var result = RegExpSubclassExec(this, string);
+  this.lastIndex = previousLastIndex;
+  if (IS_NULL(result)) return -1;
+  return result.index;
+}
+%FunctionRemovePrototype(RegExpSubclassSearch);
+
+
 // Getters for the static properties lastMatch, lastParen, leftContext, and
 // rightContext of the RegExp constructor.  The properties are computed based
 // on the captures array of the last successful match and the subject string
 // of the last successful match.
 function RegExpGetLastMatch() {
   var regExpSubject = LAST_SUBJECT(RegExpLastMatchInfo);
   return %_SubString(regExpSubject,
                      RegExpLastMatchInfo[CAPTURE0],
                      RegExpLastMatchInfo[CAPTURE1]);
 }
@@ skipping 104 matching lines @@
       %IncrementUseCounter(kRegExpPrototypeSourceGetter);
       return UNDEFINED;
     }
     throw MakeTypeError(kRegExpNonRegExp, "RegExp.prototype.source");
   }
   return REGEXP_SOURCE(this);
 }
 %FunctionSetName(RegExpGetSource, "RegExp.prototype.source");
 %SetNativeFlag(RegExpGetSource);
 
+

[Yang, 2016/03/22 12:49:38]

stray edit?

[Dan Ehrenberg, 2016/03/22 18:26:31]

Reverted

 // -------------------------------------------------------------------
 
 %FunctionSetInstanceClassName(GlobalRegExp, 'RegExp');
 GlobalRegExpPrototype = new GlobalObject();
 %FunctionSetPrototype(GlobalRegExp, GlobalRegExpPrototype);
 %AddNamedProperty(
     GlobalRegExp.prototype, 'constructor', GlobalRegExp, DONT_ENUM);
 %SetCode(GlobalRegExp, RegExpConstructor);
 
 utils.InstallFunctions(GlobalRegExp.prototype, DONT_ENUM, [
@@ skipping 57 matching lines @@
 }
 %ToFastProperties(GlobalRegExp);
 
 // -------------------------------------------------------------------
 // Exports
 
 utils.Export(function(to) {
   to.RegExpExec = DoRegExpExec;
   to.RegExpExecNoTests = RegExpExecNoTests;
   to.RegExpLastMatchInfo = RegExpLastMatchInfo;
+  to.RegExpSubclassExecJS = RegExpSubclassExecJS;
+  to.RegExpSubclassMatch = RegExpSubclassMatch;
+  to.RegExpSubclassReplace = RegExpSubclassReplace;
+  to.RegExpSubclassSearch = RegExpSubclassSearch;
+  to.RegExpSubclassSplit = RegExpSubclassSplit;
+  to.RegExpSubclassTest = RegExpSubclassTest;
   to.RegExpTest = RegExpTest;
   to.IsRegExp = IsRegExp;
 });
 
 })