Try best to convert a regexp to a ECMAScript expression

2 files changed, 60 insertions, 13 deletions

diff --git a/src/corelib/tools/qregexp.cpp b/src/corelib/tools/qregexp.cpp
index 3108f31..b0f2054 100644
--- a/src/corelib/tools/qregexp.cpp
+++ b/src/corelib/tools/qregexp.cpp
@@ -1252,25 +1252,33 @@ struct QRegExpLookahead
 };
 #endif
 
-QRegExpEngine::QRegExpEngine(const QRegExpEngineKey &key)
-    : cs(key.cs), greedyQuantifiers(key.patternSyntax == QRegExp::RegExp2)
-{
-    setup();
-
-    QString rx;
+/*! \internal
+    convert the pattern string to the RegExp syntax.
 
-    switch (key.patternSyntax) {
-    case QRegExp::Wildcard:
+    This is also used by QScriptEngine::newRegExp to convert to a pattern that JavaScriptCore can understan
+ */
+Q_CORE_EXPORT QString qt_regexp_toCanonical(const QString &pattern, QRegExp::PatternSyntax patternSyntax)
+{
+    switch (patternSyntax) {
 #ifndef QT_NO_REGEXP_WILDCARD
-        rx = wc2rx(key.pattern);
-#endif
+    case QRegExp::Wildcard:
+        return wc2rx(pattern);
         break;
+#endif
     case QRegExp::FixedString:
-        rx = QRegExp::escape(key.pattern);
+        return QRegExp::escape(pattern);
         break;
     default:
-        rx = key.pattern;
+        return pattern;
     }
+}
+
+QRegExpEngine::QRegExpEngine(const QRegExpEngineKey &key)
+    : cs(key.cs), greedyQuantifiers(key.patternSyntax == QRegExp::RegExp2)
+{
+    setup();
+
+    QString rx = qt_regexp_toCanonical(key.pattern, key.patternSyntax);
 
     valid = (parse(rx.unicode(), rx.length()) == rx.length());
     if (!valid) {
diff --git a/src/script/api/qscriptengine.cpp b/src/script/api/qscriptengine.cpp
index 0a3fbf1..43f14fa 100644
--- a/src/script/api/qscriptengine.cpp
+++ b/src/script/api/qscriptengine.cpp
@@ -1435,7 +1435,46 @@ QScriptValue QScriptEngine::newRegExp(const QRegExp &regexp)
     JSC::ExecState* exec = d->currentFrame;
     JSC::JSValue buf[2];
     JSC::ArgList args(buf, sizeof(buf));
-    JSC::UString jscPattern = QScript::qtStringToJSCUString(regexp.pattern());
+
+    //convert the pattern to a ECMAScript pattern
+    extern QString qt_regexp_toCanonical(const QString &, QRegExp::PatternSyntax);
+    QString pattern = qt_regexp_toCanonical(regexp.pattern(), regexp.patternSyntax());
+    if (regexp.isMinimal()) {
+        QString ecmaPattern;
+        int len = pattern.length();
+        ecmaPattern.reserve(len);
+        int i = 0;
+        const QChar *wc = pattern.unicode();
+        bool inBracket = false;
+        while (i < len) {
+            QChar c = wc[i++];
+            ecmaPattern += c;
+            switch (c.unicode()) {
+            case '?':
+            case '+':
+            case '*':
+            case '}':
+                if (!inBracket)
+                    ecmaPattern += QLatin1Char('?');
+                break;
+            case '\\':
+                if (i < len)
+                    ecmaPattern += wc[i++];
+                break;
+            case '[':
+                inBracket = true;
+                break;
+            case ']':
+                inBracket = false;
+                break;
+            default:
+                break;
+            }
+        }
+        pattern = ecmaPattern;
+    }
+
+    JSC::UString jscPattern = QScript::qtStringToJSCUString(pattern);
     QString flags;
     if (regexp.caseSensitivity() == Qt::CaseInsensitive)
         flags.append(QLatin1Char('i'));