[regexp] experimental implementation for \p property class.

Currently we only support general categories for property classes.

R=erik.corry@gmail.com, erikcorry@chromium.org, littledan@chromium.org
BUG=v8:4743
LOG=N

Review URL: https://codereview.chromium.org/1685593002

Cr-Commit-Position: refs/heads/master@{#33872}

src/bootstrapper.cc

@@ -2317,6 +2317,7 @@ EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_regexps)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_unicode_regexps)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_do_expressions)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_regexp_lookbehind)
+EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_regexp_property)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_function_name)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(harmony_function_sent)
 EMPTY_INITIALIZE_GLOBAL_FOR_FEATURE(promise_extra)
@@ -2961,6 +2962,7 @@ bool Genesis::InstallExperimentalNatives() {
   static const char* harmony_do_expressions_natives[] = {nullptr};
   static const char* harmony_regexp_subclass_natives[] = {nullptr};
   static const char* harmony_regexp_lookbehind_natives[] = {nullptr};
+  static const char* harmony_regexp_property_natives[] = {nullptr};
   static const char* harmony_function_name_natives[] = {nullptr};
   static const char* harmony_function_sent_natives[] = {nullptr};
   static const char* promise_extra_natives[] = {"native promise-extra.js",

src/flag-definitions.h

@@ -213,7 +213,8 @@ DEFINE_IMPLICATION(es_staging, move_object_start)
   V(harmony_tailcalls, "harmony tail calls")                                 \
   V(harmony_object_values_entries, "harmony Object.values / Object.entries") \
   V(harmony_object_own_property_descriptors,                                 \
-    "harmony Object.getOwnPropertyDescriptors()")
+    "harmony Object.getOwnPropertyDescriptors()")                            \
+  V(harmony_regexp_property, "harmony unicode regexp property classes")
 
 // Features that are complete (but still behind --harmony/es-staging flag).
 #define HARMONY_STAGED(V)                                   \

src/regexp/regexp-parser.cc

@@ -352,6 +352,23 @@ RegExpTree* RegExpParser::ParseDisjunction() {
             builder->AddCharacterClass(cc);
             break;
           }
+          case 'p':
+          case 'P': {
+            uc32 p = Next();
+            Advance(2);
+            if (unicode()) {
+              ZoneList<CharacterRange>* ranges = ParsePropertyClass();
+              if (ranges == nullptr) {
+                return ReportError(CStrVector("Invalid property name"));
+              }
+              RegExpCharacterClass* cc =
+                  new (zone()) RegExpCharacterClass(ranges, p == 'P');
+              builder->AddCharacterClass(cc);
+            } else {
+              builder->AddCharacter(p);
+            }
+            break;
+          }
           case '1':
           case '2':
           case '3':
@@ -801,6 +818,55 @@ bool RegExpParser::ParseUnicodeEscape(uc32* value) {
   return ParseHexEscape(4, value);
 }
 
+ZoneList<CharacterRange>* RegExpParser::ParsePropertyClass() {
+#ifdef V8_I18N_SUPPORT
+  char property_name[3];
+  memset(property_name, 0, sizeof(property_name));
+  if (current() == '{') {
+    Advance();
+    if (current() < 'A' || current() > 'Z') return nullptr;
+    property_name[0] = static_cast<char>(current());
+    Advance();
+    if (current() >= 'a' && current() <= 'z') {
+      property_name[1] = static_cast<char>(current());
+      Advance();
+    }
+    if (current() != '}') return nullptr;
+  } else if (current() >= 'A' && current() <= 'Z') {
+    property_name[0] = static_cast<char>(current());
+  } else {
+    return nullptr;
+  }
+  Advance();
+
+  int32_t category =
+      u_getPropertyValueEnum(UCHAR_GENERAL_CATEGORY_MASK, property_name);
+  if (category == UCHAR_INVALID_CODE) return nullptr;
+
+  USet* set = uset_openEmpty();
+  UErrorCode ec = U_ZERO_ERROR;
+  uset_applyIntPropertyValue(set, UCHAR_GENERAL_CATEGORY_MASK, category, &ec);
+  ZoneList<CharacterRange>* ranges = nullptr;
+  if (ec == U_ZERO_ERROR && !uset_isEmpty(set)) {
+    uset_removeAllStrings(set);
+    int item_count = uset_getItemCount(set);
+    ranges = new (zone()) ZoneList<CharacterRange>(item_count, zone());
+    int item_result = 0;
+    for (int i = 0; i < item_count; i++) {
+      uc32 start = 0;
+      uc32 end = 0;
+      item_result += uset_getItem(set, i, &start, &end, nullptr, 0, &ec);
+      ranges->Add(CharacterRange::Range(start, end), zone());
+    }
+    DCHECK_EQ(U_ZERO_ERROR, ec);
+    DCHECK_EQ(0, item_result);
+  }
+  uset_close(set);
+  return ranges;
+#else   // V8_I18N_SUPPORT
+  return nullptr;
+#endif  // V8_I18N_SUPPORT
+}
 
 bool RegExpParser::ParseUnlimitedLengthHexNumber(int max_value, uc32* value) {
   uc32 x = 0;

src/regexp/regexp-parser.h

@@ -173,6 +173,7 @@ class RegExpParser BASE_EMBEDDED {
   bool ParseHexEscape(int length, uc32* value);
   bool ParseUnicodeEscape(uc32* value);
   bool ParseUnlimitedLengthHexNumber(int max_value, uc32* value);
+  ZoneList<CharacterRange>* ParsePropertyClass();
 
   uc32 ParseOctalLiteral();
 

test/mjsunit/harmony/unicode-regexp-property-class.js

+// 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.
+
+// Flags: --harmony-regexp-property --harmony-unicode-regexps
+
+assertThrows("/\\p/u");
+assertThrows("/\\p{garbage}/u");
+assertThrows("/\\p{}/u");
+assertThrows("/\\p{/u");
+assertThrows("/\\p}/u");
+assertThrows("/\p{Math}/u");
+assertThrows("/\p{Bidi_M}/u");
+assertThrows("/\p{Hex}/u");
+
+assertTrue(/\p{Ll}/u.test("a"));
+assertFalse(/\P{Ll}/u.test("a"));
+assertTrue(/\P{Ll}/u.test("A"));
+assertFalse(/\p{Ll}/u.test("A"));
+assertTrue(/\p{Ll}/u.test("\u{1D7BE}"));
+assertFalse(/\P{Ll}/u.test("\u{1D7BE}"));
+assertFalse(/\p{Ll}/u.test("\u{1D5E3}"));
+assertTrue(/\P{Ll}/u.test("\u{1D5E3}"));
+
+assertTrue(/\p{Ll}/iu.test("a"));
+assertTrue(/\p{Ll}/iu.test("\u{118D4}"));
+assertTrue(/\p{Ll}/iu.test("A"));
+assertTrue(/\p{Ll}/iu.test("\u{118B4}"));
+assertFalse(/\P{Ll}/iu.test("a"));
+assertFalse(/\P{Ll}/iu.test("\u{118D4}"));
+assertFalse(/\P{Ll}/iu.test("A"));
+assertFalse(/\P{Ll}/iu.test("\u{118B4}"));
+
+assertTrue(/\p{Lu}/u.test("A"));
+assertFalse(/\P{Lu}/u.test("A"));
+assertTrue(/\P{Lu}/u.test("a"));
+assertFalse(/\p{Lu}/u.test("a"));
+assertTrue(/\p{Lu}/u.test("\u{1D5E3}"));
+assertFalse(/\P{Lu}/u.test("\u{1D5E3}"));
+assertFalse(/\p{Lu}/u.test("\u{1D7BE}"));
+assertTrue(/\P{Lu}/u.test("\u{1D7BE}"));
+
+assertTrue(/\p{Lu}/iu.test("a"));
+assertTrue(/\p{Lu}/iu.test("\u{118D4}"));
+assertTrue(/\p{Lu}/iu.test("A"));
+assertTrue(/\p{Lu}/iu.test("\u{118B4}"));
+assertFalse(/\P{Lu}/iu.test("a"));
+assertFalse(/\P{Lu}/iu.test("\u{118D4}"));
+assertFalse(/\P{Lu}/iu.test("A"));
+assertFalse(/\P{Lu}/iu.test("\u{118B4}"));
+
+assertTrue(/\p{Sm}/u.test("+"));
+assertFalse(/\P{Sm}/u.test("+"));
+assertTrue(/\p{Sm}/u.test("\u{1D6C1}"));
+assertFalse(/\P{Sm}/u.test("\u{1D6C1}"));
+
+assertTrue(/\pL/u.test("a"));
+assertFalse(/\PL/u.test("a"));
+assertFalse(/\pL/u.test("1"));
+assertTrue(/\PL/u.test("1"));
+assertTrue(/\pL/u.test("\u1FAB"));
+assertFalse(/\PL/u.test("\u1FAB"));
+assertFalse(/\p{L}/u.test("\uA6EE"));
+assertTrue(/\P{L}/u.test("\uA6EE"));

test/mjsunit/mjsunit.status

@@ -290,9 +290,11 @@
   # TODO(titzer): SSE 4.1 required for asm-wasm test (floor).
   'wasm/asm-wasm': [SKIP],
 
-    # case-insensitive unicode regexp relies on case mapping provided by ICU.
+  # case-insensitive unicode regexp relies on case mapping provided by ICU.
   'harmony/unicode-regexp-ignore-case': [PASS, ['no_i18n == True', FAIL]],
   'harmony/unicode-regexp-ignore-case-noi18n': [FAIL, ['no_i18n == True', PASS]],
+  # desugaring regexp property class relies on ICU.
+  'harmony/unicode-regexp-property-class': [PASS, ['no_i18n == True', FAIL]],
 }],  # ALWAYS
 
 ['novfp3 == True', {