Improve conversion between JS RegExp and QRegularExpression

Add support for InvertedGreedinessOption and MultilineOption.

Change-Id: I19dce6e356a7ec406640bb8858885cd576b4aa2f
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>

2 files changed, 110 insertions, 44 deletions

diff --git a/src/qml/jsruntime/qv4regexpobject.cpp b/src/qml/jsruntime/qv4regexpobject.cpp
index a8a37cda37..f763237ea3 100644
--- a/src/qml/jsruntime/qv4regexpobject.cpp
+++ b/src/qml/jsruntime/qv4regexpobject.cpp
@@ -84,6 +84,42 @@ void Heap::RegExpObject::init(QV4::RegExp *value)
     o->initProperties();
 }
 
+static QString minimalPattern(const QString &pattern)
+{
+    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;
+        }
+    }
+    return ecmaPattern;
+}
+
 // Converts a QRegExp to a JS RegExp.
 // The conversion is not 100% exact since ECMA regexp and QRegExp
 // have different semantics/flags, but we try to do our best.
@@ -93,40 +129,8 @@ void Heap::RegExpObject::init(const QRegExp &re)
 
     // Convert the pattern to a ECMAScript pattern.
     QString pattern = QT_PREPEND_NAMESPACE(qt_regexp_toCanonical)(re.pattern(), re.patternSyntax());
-    if (re.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;
-    }
+    if (re.isMinimal())
+        pattern = minimalPattern(pattern);
 
     Scope scope(internalClass->engine);
     Scoped<QV4::RegExpObject> o(scope, this);
@@ -148,10 +152,16 @@ void Heap::RegExpObject::init(const QRegularExpression &re)
     Scope scope(internalClass->engine);
     Scoped<QV4::RegExpObject> o(scope, this);
 
-    const uint flags = (re.patternOptions() & QRegularExpression::CaseInsensitiveOption)
-            ? CompiledData::RegExp::RegExp_IgnoreCase
-            : CompiledData::RegExp::RegExp_NoFlags;
-    o->d()->value.set(scope.engine, QV4::RegExp::create(scope.engine, re.pattern(), flags));
+    QRegularExpression::PatternOptions options = re.patternOptions();
+    uint flags = (options & QRegularExpression::CaseInsensitiveOption)
+                ? CompiledData::RegExp::RegExp_IgnoreCase
+                : CompiledData::RegExp::RegExp_NoFlags;
+    if (options & QRegularExpression::MultilineOption)
+        flags |= CompiledData::RegExp::RegExp_Multiline;
+    QString pattern = re.pattern();
+    if (options & QRegularExpression::InvertedGreedinessOption)
+        pattern = minimalPattern(pattern);
+    o->d()->value.set(scope.engine, QV4::RegExp::create(scope.engine, pattern, flags));
     o->initProperties();
 }
 #endif
@@ -178,11 +188,12 @@ QRegExp RegExpObject::toQRegExp() const
 // have different semantics/flags, but we try to do our best.
 QRegularExpression RegExpObject::toQRegularExpression() const
 {
-    QRegularExpression::PatternOptions caseSensitivity
-            = (value()->flags & CompiledData::RegExp::RegExp_IgnoreCase)
-            ? QRegularExpression::CaseInsensitiveOption
-            : QRegularExpression::NoPatternOption;
-    return QRegularExpression(*value()->pattern, caseSensitivity);
+    QRegularExpression::PatternOptions options = QRegularExpression::NoPatternOption;
+    if (value()->flags & CompiledData::RegExp::RegExp_IgnoreCase)
+        options |= QRegularExpression::CaseInsensitiveOption;
+    if (value()->flags & CompiledData::RegExp::RegExp_Multiline)
+        options |= QRegularExpression::MultilineOption;
+    return QRegularExpression(*value()->pattern, options);
 }
 #endif
 
diff --git a/tests/auto/qml/qjsengine/tst_qjsengine.cpp b/tests/auto/qml/qjsengine/tst_qjsengine.cpp
index 8dd0f7b188..1d859021ad 100644
--- a/tests/auto/qml/qjsengine/tst_qjsengine.cpp
+++ b/tests/auto/qml/qjsengine/tst_qjsengine.cpp
@@ -142,6 +142,8 @@ private slots:
     void qRegExpInport();
     void qRegularExpressionImport_data();
     void qRegularExpressionImport();
+    void qRegularExpressionExport_data();
+    void qRegularExpressionExport();
     void dateRoundtripJSQtJS();
     void dateRoundtripQtJSQt();
     void dateConversionJSQt();
@@ -3287,7 +3289,6 @@ void tst_QJSEngine::qRegularExpressionImport_data()
 {
     QTest::addColumn<QRegularExpression>("rx");
     QTest::addColumn<QString>("string");
-    QTest::addColumn<QString>("matched");
 
     QTest::newRow("normal")            << QRegularExpression("(test|foo)") << "test _ foo _ test _ Foo";
     QTest::newRow("normal2")           << QRegularExpression("(Test|Foo)") << "test _ foo _ test _ Foo";
@@ -3311,6 +3312,14 @@ void tst_QJSEngine::qRegularExpressionImport_data()
     QTest::newRow(".+ minimal")        << QRegularExpression("^.+$") << ".+";
     QTest::newRow("[.?] minimal")      << QRegularExpression("^[.?]$") << ".?";
     QTest::newRow("[.+] minimal")      << QRegularExpression("^[.+]$") << ".+";
+    QTest::newRow("aaa inverted greedyness")  << QRegularExpression("a{2,5}", QRegularExpression::InvertedGreedinessOption) << "aAaAaaaaaAa";
+    QTest::newRow("inverted greedyness")  << QRegularExpression(".*\\} [*8]", QRegularExpression::InvertedGreedinessOption) << "}?} ?} *";
+    QTest::newRow(".? inverted greedyness")  << QRegularExpression(".?", QRegularExpression::InvertedGreedinessOption) << ".?";
+    QTest::newRow(".+ inverted greedyness")  << QRegularExpression(".+", QRegularExpression::InvertedGreedinessOption) << ".+";
+    QTest::newRow("[.?] inverted greedyness")  << QRegularExpression("[.?]", QRegularExpression::InvertedGreedinessOption) << ".?";
+    QTest::newRow("[.+] inverted greedyness")  << QRegularExpression("[.+]", QRegularExpression::InvertedGreedinessOption) << ".+";
+    QTest::newRow("two lines")  << QRegularExpression("^.*$") << "abc\ndef";
+    QTest::newRow("multiline")  << QRegularExpression("^.*$", QRegularExpression::MultilineOption) << "abc\ndef";
 }
 
 void tst_QJSEngine::qRegularExpressionImport()
@@ -3333,6 +3342,52 @@ void tst_QJSEngine::qRegularExpressionImport()
         QCOMPARE(result.property(i).toString(), match.captured(i));
 }
 
+void tst_QJSEngine::qRegularExpressionExport_data()
+{
+    QTest::addColumn<QString>("js");
+    QTest::addColumn<QRegularExpression>("regularexpression");
+
+    QTest::newRow("normal")            << "/(test|foo)/" << QRegularExpression("(test|foo)");
+    QTest::newRow("normal2")           << "/(Test|Foo)/" << QRegularExpression("(Test|Foo)");
+    QTest::newRow("case insensitive")  << "/(test|foo)/i" << QRegularExpression("(test|foo)", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("case insensitive2") << "/(Test|Foo)/i" << QRegularExpression("(Test|Foo)", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("b(a*)(b*)")         << "/b(a*)(b*)/i" << QRegularExpression("b(a*)(b*)", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("greedy")            << "/a*(a*)/i" << QRegularExpression("a*(a*)", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("wildcard")          << "/.*\\.txt/" << QRegularExpression(".*\\.txt");
+    QTest::newRow("wildcard 2")        << "/a.b\\.txt/" << QRegularExpression("a.b\\.txt");
+    QTest::newRow("slash")             << "/g\\/.*\\/s/i" << QRegularExpression("g\\/.*\\/s", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("slash2")            << "/g \\/ .* \\/ s/i" << QRegularExpression("g \\/ .* \\/ s", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("fixed")             << "/a\\*aa\\.a\\(ba\\)\\*a\\\\ba/i" << QRegularExpression("a\\*aa\\.a\\(ba\\)\\*a\\\\ba", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("fixed insensitive") << "/A\\*A/i" << QRegularExpression("A\\*A", QRegularExpression::CaseInsensitiveOption);
+    QTest::newRow("fixed sensitive")   << "/A\\*A/" << QRegularExpression("A\\*A");
+    QTest::newRow("html")              << "/<b>(.*)<\\/b>/" << QRegularExpression("<b>(.*)<\\/b>");
+    QTest::newRow("html minimal")      << "/^<b>(.*)<\\/b>$/" << QRegularExpression("^<b>(.*)<\\/b>$");
+    QTest::newRow("aaa")               << "/a{2,5}/" << QRegularExpression("a{2,5}");
+    QTest::newRow("aaa minimal")       << "/^a{2,5}$/" << QRegularExpression("^a{2,5}$");
+    QTest::newRow("minimal")           << "/^.*\\} [*8]$/" << QRegularExpression("^.*\\} [*8]$");
+    QTest::newRow(".? minimal")        << "/^.?$/" << QRegularExpression("^.?$");
+    QTest::newRow(".+ minimal")        << "/^.+$/" << QRegularExpression("^.+$");
+    QTest::newRow("[.?] minimal")      << "/^[.?]$/" << QRegularExpression("^[.?]$");
+    QTest::newRow("[.+] minimal")      << "/^[.+]$/" << QRegularExpression("^[.+]$");
+    QTest::newRow("multiline")  << "/^.*$/m" << QRegularExpression("^.*$", QRegularExpression::MultilineOption);
+}
+
+void tst_QJSEngine::qRegularExpressionExport()
+{
+    QFETCH(QString, js);
+    QFETCH(QRegularExpression, regularexpression);
+
+    QJSEngine eng;
+    QJSValue rexp;
+    rexp = eng.evaluate(js);
+
+    QCOMPARE(rexp.isRegExp(), true);
+    QCOMPARE(rexp.isCallable(), false);
+
+    QRegularExpression rx = qjsvalue_cast<QRegularExpression>(rexp);
+    QCOMPARE(rx, regularexpression);
+}
+
 // QScriptValue::toDateTime() returns a local time, whereas JS dates
 // are always stored as UTC. Qt Script must respect the current time
 // zone, and correctly adjust for daylight saving time that may be in