Issue #1302 - Add self-hosted implementation for string regex .matchAll

This resolves #1302.

13 files changed, 287 insertions, 58 deletions

diff --git a/js/public/Class.h b/js/public/Class.h
index f7533654b..67c0cbca8 100644
--- a/js/public/Class.h
+++ b/js/public/Class.h
@@ -779,7 +779,7 @@ struct JSClass {
 // application.
 #define JSCLASS_GLOBAL_APPLICATION_SLOTS 5
 #define JSCLASS_GLOBAL_SLOT_COUNT                                             \
-    (JSCLASS_GLOBAL_APPLICATION_SLOTS + JSProto_LIMIT * 2 + 39)
+    (JSCLASS_GLOBAL_APPLICATION_SLOTS + JSProto_LIMIT * 2 + 40)
 #define JSCLASS_GLOBAL_FLAGS_WITH_SLOTS(n)                                    \
     (JSCLASS_IS_GLOBAL | JSCLASS_HAS_RESERVED_SLOTS(JSCLASS_GLOBAL_SLOT_COUNT + (n)))
 #define JSCLASS_GLOBAL_FLAGS                                                  \
diff --git a/js/src/builtin/RegExp.cpp b/js/src/builtin/RegExp.cpp
index b7853d533..55e0c8578 100644
--- a/js/src/builtin/RegExp.cpp
+++ b/js/src/builtin/RegExp.cpp
@@ -795,6 +795,7 @@ const JSFunctionSpec js::regexp_methods[] = {
     JS_SELF_HOSTED_FN("exec", "RegExp_prototype_Exec", 1,0),
     JS_SELF_HOSTED_FN("test", "RegExpTest" ,    1,0),
     JS_SELF_HOSTED_SYM_FN(match, "RegExpMatch", 1,0),
+    JS_SELF_HOSTED_SYM_FN(matchAll, "RegExpMatchAll", 1, 0),
     JS_SELF_HOSTED_SYM_FN(replace, "RegExpReplace", 2,0),
     JS_SELF_HOSTED_SYM_FN(search, "RegExpSearch", 1,0),
     JS_SELF_HOSTED_SYM_FN(split, "RegExpSplit", 2,0),
diff --git a/js/src/builtin/RegExp.js b/js/src/builtin/RegExp.js
index 1a2276594..25827743e 100644
--- a/js/src/builtin/RegExp.js
+++ b/js/src/builtin/RegExp.js
@@ -1031,3 +1031,135 @@ function RegExpSpecies() {
     return this;
 }
 _SetCanonicalName(RegExpSpecies, "get [Symbol.species]");
+
+// String.prototype.matchAll proposal.
+//
+// RegExp.prototype [ @@matchAll ] ( string )
+function RegExpMatchAll(string) {
+    // Step 1.
+    var rx = this;
+
+    // Step 2.
+    if (!IsObject(rx))
+        ThrowTypeError(JSMSG_NOT_NONNULL_OBJECT, rx === null ? "null" : typeof rx);
+
+    // Step 3.
+    var str = ToString(string);
+
+    // Step 4.
+    var C = SpeciesConstructor(rx, GetBuiltinConstructor("RegExp"));
+
+    // Step 5.
+    var flags = ToString(rx.flags);
+    
+    // Step 2.b.iii; located here because it needs |flags|.
+    if (!callFunction(std_String_includes, flags, "g")) {
+        ThrowTypeError(JSMSG_BAD_REGEXP_FLAG, "- matchAll requires g");
+    }
+    
+    // Step 6.
+    var matcher = new C(rx, flags);
+
+    // Steps 7-8.
+    matcher.lastIndex = ToLength(rx.lastIndex);
+
+    // Steps 9-12.
+    // Note, always global because non-global throws as per 
+    // https://github.com/tc39/ecma262/pull/1716
+    var flags = REGEXP_GLOBAL_FLAG |
+                (callFunction(std_String_includes, flags, "u") ? REGEXP_UNICODE_FLAG : 0);
+
+    // Step 13.
+    return CreateRegExpStringIterator(matcher, str, flags);
+}
+
+// String.prototype.matchAll proposal.
+//
+// CreateRegExpStringIterator ( R, S, global, fullUnicode )
+function CreateRegExpStringIterator(regexp, string, flags) {
+    // Step 1.
+    assert(typeof string === "string", "|string| is a string value");
+
+    // Steps 2-3.
+    assert(typeof flags === "number", "|flags| is a number value");
+
+    // Steps 4-9.
+    var iterator = NewRegExpStringIterator();
+    UnsafeSetReservedSlot(iterator, REGEXP_STRING_ITERATOR_REGEXP_SLOT, regexp);
+    UnsafeSetReservedSlot(iterator, REGEXP_STRING_ITERATOR_STRING_SLOT, string);
+    UnsafeSetReservedSlot(iterator, REGEXP_STRING_ITERATOR_FLAGS_SLOT, flags | 0);
+    UnsafeSetReservedSlot(iterator, REGEXP_STRING_ITERATOR_DONE_SLOT, false);
+
+    // Step 10.
+    return iterator;
+}
+
+// String.prototype.matchAll proposal.
+//
+// %RegExpStringIteratorPrototype%.next ( )
+function RegExpStringIteratorNext() {
+    // Steps 1-3.
+    var obj;
+    if (!IsObject(this) || (obj = GuardToRegExpStringIterator(this)) === null) {
+        return callFunction(CallRegExpStringIteratorMethodIfWrapped, this,
+                            "RegExpStringIteratorNext");
+    }
+
+    var result = { value: undefined, done: false };
+
+    // Step 4.
+    var done = UnsafeGetReservedSlot(obj, REGEXP_STRING_ITERATOR_DONE_SLOT);
+    if (done) {
+        result.done = true;
+        return result;
+    }
+
+    // Step 5.
+    var regexp = UnsafeGetObjectFromReservedSlot(obj, REGEXP_STRING_ITERATOR_REGEXP_SLOT);
+
+    // Step 6.
+    var string = UnsafeGetStringFromReservedSlot(obj, REGEXP_STRING_ITERATOR_STRING_SLOT);
+
+    // Steps 7-8.
+    var flags = UnsafeGetInt32FromReservedSlot(obj, REGEXP_STRING_ITERATOR_FLAGS_SLOT);
+    var global = !!(flags & REGEXP_GLOBAL_FLAG);
+    var fullUnicode = !!(flags & REGEXP_UNICODE_FLAG);
+
+    // Step 9.
+    var match = RegExpExec(regexp, string, false);
+
+    // Step 10.
+    if (match === null) {
+        // Step 10.a.
+        UnsafeSetReservedSlot(obj, REGEXP_STRING_ITERATOR_DONE_SLOT, true);
+
+        // Step 10.b.
+        result.done = true;
+        return result;
+    }
+
+    // Step 11.a.
+    if (global) {
+        // Step 11.a.i.
+        var matchStr = ToString(match[0]);
+
+        // Step 11.a.ii.
+        if (matchStr.length === 0) {
+            // Step 11.a.ii.1.
+            var thisIndex = ToLength(regexp.lastIndex);
+
+            // Step 11.a.ii.2.
+            var nextIndex = fullUnicode ? AdvanceStringIndex(string, thisIndex) : thisIndex + 1;
+
+            // Step 11.a.ii.3.
+            regexp.lastIndex = nextIndex;
+        }
+    } else {
+        // Step 11.b.i.
+        UnsafeSetReservedSlot(obj, REGEXP_STRING_ITERATOR_DONE_SLOT, true);
+    }
+
+    // Steps 11.a.iii and 11.b.ii.
+    result.value = match;
+    return result;
+}
diff --git a/js/src/builtin/SelfHostingDefines.h b/js/src/builtin/SelfHostingDefines.h
index 6512810ca..117ac7ffd 100644
--- a/js/src/builtin/SelfHostingDefines.h
+++ b/js/src/builtin/SelfHostingDefines.h
@@ -92,6 +92,11 @@
 #define REGEXP_UNICODE_FLAG     0x10
 #define REGEXP_DOTALL_FLAG      0x20
 
+#define REGEXP_STRING_ITERATOR_REGEXP_SLOT 0
+#define REGEXP_STRING_ITERATOR_STRING_SLOT 1
+#define REGEXP_STRING_ITERATOR_FLAGS_SLOT 2
+#define REGEXP_STRING_ITERATOR_DONE_SLOT 3
+
 #define MODULE_OBJECT_ENVIRONMENT_SLOT 2
 
 #define MODULE_STATE_FAILED       0
diff --git a/js/src/builtin/String.js b/js/src/builtin/String.js
index f830b1aa2..d07ec6127 100644
--- a/js/src/builtin/String.js
+++ b/js/src/builtin/String.js
@@ -63,6 +63,33 @@ function String_generic_match(thisValue, regexp) {
     return callFunction(String_match, thisValue, regexp);
 }
 
+// String.prototype.matchAll proposal.
+//
+// String.prototype.matchAll ( regexp )
+function String_matchAll(regexp) {
+    // Step 1.
+    RequireObjectCoercible(this);
+
+    // Step 2.
+    if (regexp !== undefined && regexp !== null) {
+        // Step 2.a.
+        var matcher = GetMethod(regexp, std_matchAll);
+
+        // Step 2.b.
+        if (matcher !== undefined)
+            return callContentFunction(matcher, regexp, this);
+    }
+
+    // Step 3.
+    var string = ToString(this);
+
+    // Step 4.
+    var rx = RegExpCreate(regexp, "g");
+
+    // Step 5.
+    return callContentFunction(GetMethod(rx, std_matchAll), rx, string);
+}
+
 /**
  * A helper function implementing the logic for both String.prototype.padStart
  * and String.prototype.padEnd as described in ES7 Draft March 29, 2016
diff --git a/js/src/jsapi.h b/js/src/jsapi.h
index 1a69b1513..4dad68e6e 100644
--- a/js/src/jsapi.h
+++ b/js/src/jsapi.h
@@ -5087,7 +5087,8 @@ GetSymbolDescription(HandleSymbol symbol);
     macro(split) \
     macro(toPrimitive) \
     macro(toStringTag) \
-    macro(unscopables)
+    macro(unscopables) \
+    macro(matchAll)
 
 enum class SymbolCode : uint32_t {
     // There is one SymbolCode for each well-known symbol.
diff --git a/js/src/jsiter.cpp b/js/src/jsiter.cpp
index 3e222ca6f..c4da86bdb 100644
--- a/js/src/jsiter.cpp
+++ b/js/src/jsiter.cpp
@@ -24,6 +24,7 @@
 #include "jstypes.h"
 #include "jsutil.h"
 
+#include "builtin/SelfHostingDefines.h"
 #include "ds/Sort.h"
 #include "gc/Marking.h"
 #include "js/Proxy.h"
@@ -1135,6 +1136,38 @@ static const JSFunctionSpec string_iterator_methods[] = {
     JS_FS_END
 };
 
+static const Class RegExpStringIteratorPrototypeClass = {
+    "RegExp String Iterator",
+    0
+};
+
+enum {
+    RegExpStringIteratorSlotRegExp,
+    RegExpStringIteratorSlotString,
+    RegExpStringIteratorSlotFlags,
+    RegExpStringIteratorSlotDone,
+    RegExpStringIteratorSlotCount
+};
+
+static_assert(RegExpStringIteratorSlotRegExp == REGEXP_STRING_ITERATOR_REGEXP_SLOT,
+              "RegExpStringIteratorSlotRegExp must match self-hosting define for regexp slot.");
+static_assert(RegExpStringIteratorSlotString == REGEXP_STRING_ITERATOR_STRING_SLOT,
+              "RegExpStringIteratorSlotString must match self-hosting define for string slot.");
+static_assert(RegExpStringIteratorSlotFlags == REGEXP_STRING_ITERATOR_FLAGS_SLOT,
+              "RegExpStringIteratorSlotFlags must match self-hosting define for flags slot.");
+static_assert(RegExpStringIteratorSlotDone == REGEXP_STRING_ITERATOR_DONE_SLOT,
+              "RegExpStringIteratorSlotDone must match self-hosting define for done slot.");
+
+const Class RegExpStringIteratorObject::class_ = {
+    "RegExp String Iterator",
+    JSCLASS_HAS_RESERVED_SLOTS(RegExpStringIteratorSlotCount)
+};
+
+static const JSFunctionSpec regexp_string_iterator_methods[] = {
+    JS_SELF_HOSTED_FN("next", "RegExpStringIteratorNext", 0, 0),
+    JS_FS_END
+};
+
 JSObject*
 js::ValueToIterator(JSContext* cx, unsigned flags, HandleValue vp)
 {
@@ -1541,6 +1574,30 @@ GlobalObject::initStringIteratorProto(JSContext* cx, Handle<GlobalObject*> globa
     return true;
 }
 
+/* static */ bool
+GlobalObject::initRegExpStringIteratorProto(JSContext* cx, Handle<GlobalObject*> global)
+{
+    if (global->getReservedSlot(REGEXP_STRING_ITERATOR_PROTO).isObject())
+        return true;
+
+    RootedObject iteratorProto(cx, GlobalObject::getOrCreateIteratorPrototype(cx, global));
+    if (!iteratorProto)
+        return false;
+
+    const Class* cls = &RegExpStringIteratorPrototypeClass;
+    RootedObject proto(cx, GlobalObject::createBlankPrototypeInheriting(cx, global, cls,
+                                                                        iteratorProto));
+    if (!proto ||
+        !DefinePropertiesAndFunctions(cx, proto, nullptr, regexp_string_iterator_methods) ||
+        !DefineToStringTag(cx, proto, cx->names().RegExpStringIterator))
+    {
+        return false;
+    }
+
+    global->setReservedSlot(REGEXP_STRING_ITERATOR_PROTO, ObjectValue(*proto));
+    return true;
+}
+
 JSObject*
 js::InitLegacyIteratorClass(JSContext* cx, HandleObject obj)
 {
diff --git a/js/src/jsiter.h b/js/src/jsiter.h
index a3035ddd0..52eb045c5 100644
--- a/js/src/jsiter.h
+++ b/js/src/jsiter.h
@@ -151,6 +151,12 @@ class StringIteratorObject : public JSObject
     static const Class class_;
 };
 
+class RegExpStringIteratorObject : public JSObject
+{
+  public:
+    static const Class class_;
+};
+
 bool
 GetIterator(JSContext* cx, HandleObject obj, unsigned flags, MutableHandleObject objp);
 
diff --git a/js/src/jsstr.cpp b/js/src/jsstr.cpp
index 74f61b87d..3964ab84e 100644
--- a/js/src/jsstr.cpp
+++ b/js/src/jsstr.cpp
@@ -2584,6 +2584,7 @@ static const JSFunctionSpec string_methods[] = {
 
     /* Perl-ish methods (search is actually Python-esque). */
     JS_SELF_HOSTED_FN("match", "String_match",        1,0),
+    JS_SELF_HOSTED_FN("matchAll", "String_matchAll",  1,0),
     JS_SELF_HOSTED_FN("search", "String_search",      1,0),
     JS_SELF_HOSTED_FN("replace", "String_replace",    2,0),
     JS_SELF_HOSTED_FN("split",  "String_split",       2,0),
diff --git a/js/src/vm/CommonPropertyNames.h b/js/src/vm/CommonPropertyNames.h
index 4ae49d577..99cb02e58 100644
--- a/js/src/vm/CommonPropertyNames.h
+++ b/js/src/vm/CommonPropertyNames.h
@@ -282,6 +282,7 @@
     macro(RegExpFlagsGetter, RegExpFlagsGetter, "RegExpFlagsGetter") \
     macro(RegExpMatcher, RegExpMatcher, "RegExpMatcher") \
     macro(RegExpSearcher, RegExpSearcher, "RegExpSearcher") \
+    macro(RegExpStringIterator, RegExpStringIterator, "RegExp String Iterator") \
     macro(RegExpTester, RegExpTester, "RegExpTester") \
     macro(RegExp_prototype_Exec, RegExp_prototype_Exec, "RegExp_prototype_Exec") \
     macro(Reify, Reify, "Reify") \
diff --git a/js/src/vm/GlobalObject.cpp b/js/src/vm/GlobalObject.cpp
index 85707e1c6..013208f66 100644
--- a/js/src/vm/GlobalObject.cpp
+++ b/js/src/vm/GlobalObject.cpp
@@ -468,62 +468,29 @@ GlobalObject::initSelfHostingBuiltins(JSContext* cx, Handle<GlobalObject*> globa
         return false;
     }
 
-    RootedValue std_isConcatSpreadable(cx);
-    std_isConcatSpreadable.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::isConcatSpreadable));
-    if (!JS_DefineProperty(cx, global, "std_isConcatSpreadable", std_isConcatSpreadable,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    // Define a top-level property 'std_iterator' with the name of the method
-    // used by for-of loops to create an iterator.
-    RootedValue std_iterator(cx);
-    std_iterator.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::iterator));
-    if (!JS_DefineProperty(cx, global, "std_iterator", std_iterator,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    RootedValue std_match(cx);
-    std_match.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::match));
-    if (!JS_DefineProperty(cx, global, "std_match", std_match,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    RootedValue std_replace(cx);
-    std_replace.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::replace));
-    if (!JS_DefineProperty(cx, global, "std_replace", std_replace,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    RootedValue std_search(cx);
-    std_search.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::search));
-    if (!JS_DefineProperty(cx, global, "std_search", std_search,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    RootedValue std_species(cx);
-    std_species.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::species));
-    if (!JS_DefineProperty(cx, global, "std_species", std_species,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
-    }
-
-    RootedValue std_split(cx);
-    std_split.setSymbol(cx->wellKnownSymbols().get(JS::SymbolCode::split));
-    if (!JS_DefineProperty(cx, global, "std_split", std_split,
-                           JSPROP_PERMANENT | JSPROP_READONLY))
-    {
-        return false;
+    struct SymbolAndName {
+        JS::SymbolCode code;
+        const char* name;
+    };
+
+    SymbolAndName wellKnownSymbols[] = {
+        {JS::SymbolCode::isConcatSpreadable, "std_isConcatSpreadable"},
+        {JS::SymbolCode::iterator, "std_iterator"},
+        {JS::SymbolCode::match, "std_match"},
+        {JS::SymbolCode::matchAll, "std_matchAll"},
+        {JS::SymbolCode::replace, "std_replace"},
+        {JS::SymbolCode::search, "std_search"},
+        {JS::SymbolCode::species, "std_species"},
+        {JS::SymbolCode::split, "std_split"},
+    };
+
+    RootedValue symVal(cx);
+    for (const auto& sym : wellKnownSymbols) {
+        symVal.setSymbol(cx->wellKnownSymbols().get(sym.code));
+        if (!JS_DefineProperty(cx, global, sym.name, symVal,
+                               JSPROP_PERMANENT | JSPROP_READONLY)) {
+            return false;
+        }
     }
 
     return InitBareBuiltinCtor(cx, global, JSProto_Array) &&
diff --git a/js/src/vm/GlobalObject.h b/js/src/vm/GlobalObject.h
index 5aacfc5dc..9179abbb7 100644
--- a/js/src/vm/GlobalObject.h
+++ b/js/src/vm/GlobalObject.h
@@ -93,6 +93,7 @@ class GlobalObject : public NativeObject
         ITERATOR_PROTO,
         ARRAY_ITERATOR_PROTO,
         STRING_ITERATOR_PROTO,
+        REGEXP_STRING_ITERATOR_PROTO,
         LEGACY_GENERATOR_OBJECT_PROTO,
         STAR_GENERATOR_OBJECT_PROTO,
         STAR_GENERATOR_FUNCTION_PROTO,
@@ -583,6 +584,12 @@ class GlobalObject : public NativeObject
     }
 
     static NativeObject*
+    getOrCreateRegExpStringIteratorPrototype(JSContext* cx, Handle<GlobalObject*> global) {
+        return MaybeNativeObject(getOrCreateObject(cx, global, REGEXP_STRING_ITERATOR_PROTO,
+                                                   initRegExpStringIteratorProto));
+    }
+
+    static NativeObject*
     getOrCreateLegacyGeneratorObjectPrototype(JSContext* cx, Handle<GlobalObject*> global) {
         return MaybeNativeObject(getOrCreateObject(cx, global, LEGACY_GENERATOR_OBJECT_PROTO,
                                                    initLegacyGeneratorProto));
@@ -767,6 +774,7 @@ class GlobalObject : public NativeObject
     static bool initIteratorProto(JSContext* cx, Handle<GlobalObject*> global);
     static bool initArrayIteratorProto(JSContext* cx, Handle<GlobalObject*> global);
     static bool initStringIteratorProto(JSContext* cx, Handle<GlobalObject*> global);
+    static bool initRegExpStringIteratorProto(JSContext* cx, Handle<GlobalObject*> global);
 
     // Implemented in vm/GeneratorObject.cpp.
     static bool initLegacyGeneratorProto(JSContext* cx, Handle<GlobalObject*> global);
diff --git a/js/src/vm/SelfHosting.cpp b/js/src/vm/SelfHosting.cpp
index 833410465..ffd707b14 100644
--- a/js/src/vm/SelfHosting.cpp
+++ b/js/src/vm/SelfHosting.cpp
@@ -857,6 +857,24 @@ intrinsic_NewStringIterator(JSContext* cx, unsigned argc, Value* vp)
 }
 
 static bool
+intrinsic_NewRegExpStringIterator(JSContext* cx, unsigned argc, Value* vp)
+{
+    CallArgs args = CallArgsFromVp(argc, vp);
+    MOZ_ASSERT(args.length() == 0);
+
+    RootedObject proto(cx, GlobalObject::getOrCreateRegExpStringIteratorPrototype(cx, cx->global()));
+    if (!proto)
+        return false;
+
+    JSObject* obj = NewObjectWithGivenProto(cx, &RegExpStringIteratorObject::class_, proto);
+    if (!obj)
+        return false;
+
+    args.rval().setObject(*obj);
+    return true;
+}
+
+static bool
 intrinsic_SetCanonicalName(JSContext* cx, unsigned argc, Value* vp)
 {
     CallArgs args = CallArgsFromVp(argc, vp);
@@ -2288,6 +2306,8 @@ static const JSFunctionSpec intrinsic_functions[] = {
     JS_INLINABLE_FN("GuardToStringIterator",
                     intrinsic_GuardToBuiltin<StringIteratorObject>, 1,0,
                     IntrinsicGuardToStringIterator),
+    JS_FN("GuardToRegExpStringIterator",
+          intrinsic_GuardToBuiltin<RegExpStringIteratorObject>, 1,0),
 
     JS_FN("_CreateMapIterationResultPair", intrinsic_CreateMapIterationResultPair, 0, 0),
     JS_INLINABLE_FN("_GetNextMapEntryForIterator", intrinsic_GetNextMapEntryForIterator, 2,0,
@@ -2305,6 +2325,9 @@ static const JSFunctionSpec intrinsic_functions[] = {
     JS_FN("NewStringIterator",       intrinsic_NewStringIterator,       0,0),
     JS_FN("CallStringIteratorMethodIfWrapped",
           CallNonGenericSelfhostedMethod<Is<StringIteratorObject>>,     2,0),
+    JS_FN("NewRegExpStringIterator", intrinsic_NewRegExpStringIterator, 0,0),
+    JS_FN("CallRegExpStringIteratorMethodIfWrapped",
+          CallNonGenericSelfhostedMethod<Is<RegExpStringIteratorObject>>, 2,0),
 
     JS_FN("IsStarGeneratorObject",
           intrinsic_IsInstanceOfBuiltin<StarGeneratorObject>,           1,0),