Enable unicode regular expressions

Add support for the 'u' flag for regular expressions.

Change-Id: I409054eaa9c50183619752d14f2638f5a38c0ea7
Reviewed-by: Simon Hausmann <simon.hausmann@qt.io>

5 files changed, 19 insertions, 7 deletions

diff --git a/src/qml/jsruntime/qv4engine.cpp b/src/qml/jsruntime/qv4engine.cpp
index 69b23484a8..6d5a43dd1f 100644
--- a/src/qml/jsruntime/qv4engine.cpp
+++ b/src/qml/jsruntime/qv4engine.cpp
@@ -377,6 +377,8 @@ ExecutionEngine::ExecutionEngine(QJSEngine *jsEngine)
     Q_ASSERT(index == RegExpObject::Index_IgnoreCase);
     ic = ic->addMember((str = newIdentifier(QStringLiteral("multiline")))->propertyKey(), Attr_ReadOnly, &index);
     Q_ASSERT(index == RegExpObject::Index_Multiline);
+    ic = ic->addMember((str = newIdentifier(QStringLiteral("unicode")))->propertyKey(), Attr_ReadOnly, &index);
+    Q_ASSERT(index == RegExpObject::Index_Unicode);
     jsObjects[RegExpProto] = memoryManager->allocObject<RegExpPrototype>(ic->d());
     classes[Class_RegExpObject] = ic->changePrototype(regExpPrototype()->d());
 
@@ -787,9 +789,10 @@ Heap::RegExpObject *ExecutionEngine::newRegExpObject(const QString &pattern, int
     bool global = (flags & QV4::CompiledData::RegExp::RegExp_Global);
     bool ignoreCase = (flags & QV4::CompiledData::RegExp::RegExp_IgnoreCase);
     bool multiline = (flags & QV4::CompiledData::RegExp::RegExp_Multiline);
+    bool unicode = (flags & QV4::CompiledData::RegExp::RegExp_Unicode);
 
     Scope scope(this);
-    Scoped<RegExp> re(scope, RegExp::create(this, pattern, ignoreCase, multiline, global));
+    Scoped<RegExp> re(scope, RegExp::create(this, pattern, ignoreCase, multiline, global, unicode));
     return newRegExpObject(re);
 }
 
diff --git a/src/qml/jsruntime/qv4regexp.cpp b/src/qml/jsruntime/qv4regexp.cpp
index 89fd9fc233..e562482395 100644
--- a/src/qml/jsruntime/qv4regexp.cpp
+++ b/src/qml/jsruntime/qv4regexp.cpp
@@ -70,7 +70,7 @@ uint RegExp::match(const QString &string, int start, uint *matchOffsets)
     return JSC::Yarr::interpret(byteCode(), s.characters16(), string.length(), start, matchOffsets);
 }
 
-Heap::RegExp *RegExp::create(ExecutionEngine* engine, const QString& pattern, bool ignoreCase, bool multiline, bool global)
+Heap::RegExp *RegExp::create(ExecutionEngine* engine, const QString& pattern, bool ignoreCase, bool multiline, bool global, bool unicode)
 {
     RegExpCacheKey key(pattern, ignoreCase, multiline, global);
 
@@ -83,7 +83,7 @@ Heap::RegExp *RegExp::create(ExecutionEngine* engine, const QString& pattern, bo
         return result->d();
 
     Scope scope(engine);
-    Scoped<RegExp> result(scope, engine->memoryManager->alloc<RegExp>(engine, pattern, ignoreCase, multiline, global));
+    Scoped<RegExp> result(scope, engine->memoryManager->alloc<RegExp>(engine, pattern, ignoreCase, multiline, global, unicode));
 
     result->d()->cache = cache;
     cachedValue.set(engine, result);
@@ -91,12 +91,13 @@ Heap::RegExp *RegExp::create(ExecutionEngine* engine, const QString& pattern, bo
     return result->d();
 }
 
-void Heap::RegExp::init(ExecutionEngine *engine, const QString &pattern, bool ignoreCase, bool multiline, bool global)
+void Heap::RegExp::init(ExecutionEngine *engine, const QString &pattern, bool ignoreCase, bool multiline, bool global, bool unicode)
 {
     Base::init();
     this->pattern = new QString(pattern);
     this->ignoreCase = ignoreCase;
     this->multiLine = multiline;
+    this->unicode = unicode;
     this->global = global;
 
     valid = false;
@@ -109,6 +110,8 @@ void Heap::RegExp::init(ExecutionEngine *engine, const QString &pattern, bool ig
         flags = static_cast<JSC::RegExpFlags>(flags | JSC::FlagMultiline);
     if (global)
         flags = static_cast<JSC::RegExpFlags>(flags | JSC::FlagGlobal);
+    if (unicode)
+        flags = static_cast<JSC::RegExpFlags>(flags | JSC::FlagUnicode);
 
     JSC::Yarr::YarrPattern yarrPattern(WTF::String(pattern), flags, error);
     if (error != JSC::Yarr::ErrorCode::NoError)
diff --git a/src/qml/jsruntime/qv4regexp_p.h b/src/qml/jsruntime/qv4regexp_p.h
index 597e42538a..9090aaa7d5 100644
--- a/src/qml/jsruntime/qv4regexp_p.h
+++ b/src/qml/jsruntime/qv4regexp_p.h
@@ -76,7 +76,7 @@ struct RegExpCacheKey;
 namespace Heap {
 
 struct RegExp : Base {
-    void init(ExecutionEngine *engine, const QString& pattern, bool ignoreCase, bool multiline, bool global);
+    void init(ExecutionEngine *engine, const QString& pattern, bool ignoreCase, bool multiline, bool global, bool unicode);
     void destroy();
 
     QString *pattern;
@@ -96,6 +96,7 @@ struct RegExp : Base {
     bool ignoreCase;
     bool multiLine;
     bool global;
+    bool unicode;
     bool valid;
 
     int captureCount() const { return subPatternCount + 1; }
@@ -122,7 +123,7 @@ struct RegExp : public Managed
     bool multiLine() const { return d()->multiLine; }
     bool global() const { return d()->global; }
 
-    static Heap::RegExp *create(ExecutionEngine* engine, const QString& pattern, bool ignoreCase = false, bool multiline = false, bool global = false);
+    static Heap::RegExp *create(ExecutionEngine* engine, const QString& pattern, bool ignoreCase = false, bool multiline = false, bool global = false, bool unicode = false);
 
     bool isValid() const { return d()->valid; }
 
diff --git a/src/qml/jsruntime/qv4regexpobject.cpp b/src/qml/jsruntime/qv4regexpobject.cpp
index 8429b96baa..f8caf404e9 100644
--- a/src/qml/jsruntime/qv4regexpobject.cpp
+++ b/src/qml/jsruntime/qv4regexpobject.cpp
@@ -151,6 +151,7 @@ void RegExpObject::initProperties()
     setProperty(Index_Global, Primitive::fromBoolean(global()));
     setProperty(Index_IgnoreCase, Primitive::fromBoolean(value()->ignoreCase));
     setProperty(Index_Multiline, Primitive::fromBoolean(value()->multiLine));
+    setProperty(Index_Unicode, Primitive::fromBoolean(value()->unicode));
 }
 
 // Converts a JS RegExp to a QRegExp.
@@ -191,6 +192,8 @@ uint RegExpObject::flags() const
         f |= QV4::RegExpObject::RegExp_IgnoreCase;
     if (value()->multiLine)
         f |= QV4::RegExpObject::RegExp_Multiline;
+    if (value()->unicode)
+        f |= QV4::RegExpObject::RegExp_Unicode;
     return f;
 }
 
diff --git a/src/qml/jsruntime/qv4regexpobject_p.h b/src/qml/jsruntime/qv4regexpobject_p.h
index 0d4fe760eb..e52220c257 100644
--- a/src/qml/jsruntime/qv4regexpobject_p.h
+++ b/src/qml/jsruntime/qv4regexpobject_p.h
@@ -108,7 +108,8 @@ struct RegExpObject: Object {
     enum Flags {
         RegExp_Global     = 0x01,
         RegExp_IgnoreCase = 0x02,
-        RegExp_Multiline  = 0x04
+        RegExp_Multiline  = 0x04,
+        RegExp_Unicode    = 0x08
     };
 
     enum {
@@ -117,6 +118,7 @@ struct RegExpObject: Object {
         Index_Global = 2,
         Index_IgnoreCase = 3,
         Index_Multiline = 4,
+        Index_Unicode = 5,
         Index_ArrayIndex = Heap::ArrayObject::LengthPropertyIndex + 1,
         Index_ArrayInput = Index_ArrayIndex + 1
     };