Optimize mime type matching

Don't count "*", don't search for "[" and for "?"
inside m_pattern on every call to matchFileName().
Do it once, when constructing the MimeGlobPattern.

Fix matching the pattern for names without any
wildcard: index of question mark should be -1, not
just different from 0.

This shortens loading a Qt6 project in Creator
by about 500 ms.

Pick-to: 6.2 5.15
Change-Id: Ifa40c2cec4aba07a0312ef36877e571a8c8fb151
Reviewed-by: Qt CI Bot <qt_ci_bot@qt-project.org>
Reviewed-by: David Faure <david.faure@kdab.com>

3 files changed, 92 insertions, 36 deletions

diff --git a/src/corelib/mimetypes/qmimeglobpattern.cpp b/src/corelib/mimetypes/qmimeglobpattern.cpp
index c3ee37c526..a61a13e723 100644
--- a/src/corelib/mimetypes/qmimeglobpattern.cpp
+++ b/src/corelib/mimetypes/qmimeglobpattern.cpp
@@ -91,6 +91,40 @@ void QMimeGlobMatchResult::addMatch(const QString &mimeType, int weight, const Q
     }
 }
 
+QMimeGlobPattern::PatternType QMimeGlobPattern::detectPatternType(const QString &pattern) const
+{
+    const int patternLength = pattern.length();
+    if (!patternLength)
+        return OtherPattern;
+
+    const bool starCount = pattern.count(QLatin1Char('*')) == 1;
+    const bool hasSquareBracket = pattern.indexOf(QLatin1Char('[')) != -1;
+    const bool hasQuestionMark = pattern.indexOf(QLatin1Char('?')) != -1;
+
+    if (!hasSquareBracket && !hasQuestionMark) {
+        if (starCount == 1) {
+            // Patterns like "*~", "*.extension"
+            if (pattern.at(0) == QLatin1Char('*'))
+                return SuffixPattern;
+            // Patterns like "README*" (well this is currently the only one like that...)
+            if (pattern.at(patternLength - 1) == QLatin1Char('*'))
+                return PrefixPattern;
+        }
+        // Names without any wildcards like "README"
+        if (starCount == 0)
+            return LiteralPattern;
+    }
+
+    if (pattern == QLatin1String("[0-9][0-9][0-9].vdr"))
+        return VdrPattern;
+
+    if (pattern == QLatin1String("*.anim[1-9j]"))
+        return AnimPattern;
+
+    return OtherPattern;
+}
+
+
 /*!
     \internal
     \class QMimeGlobPattern
@@ -100,58 +134,66 @@ void QMimeGlobMatchResult::addMatch(const QString &mimeType, int weight, const Q
     \sa QMimeType, QMimeDatabase, QMimeMagicRuleMatcher, QMimeMagicRule
 */
 
-bool QMimeGlobPattern::matchFileName(const QString &inputFilename) const
+bool QMimeGlobPattern::matchFileName(const QString &inputFileName) const
 {
     // "Applications MUST match globs case-insensitively, except when the case-sensitive
     // attribute is set to true."
     // The constructor takes care of putting case-insensitive patterns in lowercase.
-    const QString filename = m_caseSensitivity == Qt::CaseInsensitive ? inputFilename.toLower() : inputFilename;
+    const QString fileName = m_caseSensitivity == Qt::CaseInsensitive
+            ? inputFileName.toLower() : inputFileName;
 
-    const int pattern_len = m_pattern.length();
-    if (!pattern_len)
+    const int patternLength = m_pattern.length();
+    if (!patternLength)
         return false;
-    const int len = filename.length();
-
-    const int starCount = m_pattern.count(QLatin1Char('*'));
+    const int fileNameLength = fileName.length();
 
-    // Patterns like "*~", "*.extension"
-    if (m_pattern[0] == QLatin1Char('*') && m_pattern.indexOf(QLatin1Char('[')) == -1 && starCount == 1)
-    {
-        if (len + 1 < pattern_len) return false;
+    switch (m_patternType) {
+    case SuffixPattern: {
+        if (fileNameLength + 1 < patternLength)
+            return false;
 
-        const QChar *c1 = m_pattern.unicode() + pattern_len - 1;
-        const QChar *c2 = filename.unicode() + len - 1;
+        const QChar *c1 = m_pattern.unicode() + patternLength - 1;
+        const QChar *c2 = fileName.unicode() + fileNameLength - 1;
         int cnt = 1;
-        while (cnt < pattern_len && *c1-- == *c2--)
+        while (cnt < patternLength && *c1-- == *c2--)
             ++cnt;
-        return cnt == pattern_len;
+        return cnt == patternLength;
     }
-
-    // Patterns like "README*" (well this is currently the only one like that...)
-    if (starCount == 1 && m_pattern.at(pattern_len - 1) == QLatin1Char('*')) {
-        if (len + 1 < pattern_len) return false;
-        if (m_pattern.at(0) == QLatin1Char('*'))
-            return filename.indexOf(QStringView{m_pattern}.mid(1, pattern_len - 2)) != -1;
+    case PrefixPattern: {
+        if (fileNameLength + 1 < patternLength)
+            return false;
 
         const QChar *c1 = m_pattern.unicode();
-        const QChar *c2 = filename.unicode();
+        const QChar *c2 = fileName.unicode();
         int cnt = 1;
-        while (cnt < pattern_len && *c1++ == *c2++)
+        while (cnt < patternLength && *c1++ == *c2++)
            ++cnt;
-        return cnt == pattern_len;
+        return cnt == patternLength;
     }
-
-    // Names without any wildcards like "README"
-    if (m_pattern.indexOf(QLatin1Char('[')) == -1 && starCount == 0 && m_pattern.indexOf(QLatin1Char('?')))
-        return (m_pattern == filename);
-
-    // Other (quite rare) patterns, like "*.anim[1-9j]": use slow but correct method
+    case LiteralPattern:
+        return (m_pattern == fileName);
+    case VdrPattern: // "[0-9][0-9][0-9].vdr" case
+        return fileNameLength == 7
+                && fileName.at(0).isDigit() && fileName.at(1).isDigit() && fileName.at(2).isDigit()
+                && QStringView{fileName}.mid(3, 4) == QLatin1String(".vdr");
+    case AnimPattern: { // "*.anim[1-9j]" case
+        if (fileNameLength < 6)
+            return false;
+        const QChar lastChar = fileName.at(fileNameLength - 1);
+        const bool lastCharOK = (lastChar.isDigit() && lastChar != QLatin1Char('0'))
+                              || lastChar == QLatin1Char('j');
+        return lastCharOK && QStringView{fileName}.mid(fileNameLength - 6, 5) == QLatin1String(".anim");
+    }
+    case OtherPattern:
+        // Other fallback patterns: slow but correct method
 #if QT_CONFIG(regularexpression)
-    auto rx = QRegularExpression::fromWildcard(m_pattern);
-    return rx.match(filename).hasMatch();
+        auto rx = QRegularExpression::fromWildcard(m_pattern);
+        return rx.match(fileName).hasMatch();
 #else
-    return false;
+        return false;
 #endif
+    }
+    return false;
 }
 
 static bool isSimplePattern(const QString &pattern)
diff --git a/src/corelib/mimetypes/qmimeglobpattern_p.h b/src/corelib/mimetypes/qmimeglobpattern_p.h
index 88284521d1..584e3329a6 100644
--- a/src/corelib/mimetypes/qmimeglobpattern_p.h
+++ b/src/corelib/mimetypes/qmimeglobpattern_p.h
@@ -80,7 +80,10 @@ public:
 
     explicit QMimeGlobPattern(const QString &thePattern, const QString &theMimeType, unsigned theWeight = DefaultWeight, Qt::CaseSensitivity s = Qt::CaseInsensitive) :
         m_pattern(s == Qt::CaseInsensitive ? thePattern.toLower() : thePattern),
-        m_mimeType(theMimeType), m_weight(theWeight), m_caseSensitivity(s)
+        m_mimeType(theMimeType),
+        m_weight(theWeight),
+        m_caseSensitivity(s),
+        m_patternType(detectPatternType(m_pattern))
     {
     }
 
@@ -90,9 +93,10 @@ public:
         qSwap(m_mimeType,        other.m_mimeType);
         qSwap(m_weight,          other.m_weight);
         qSwap(m_caseSensitivity, other.m_caseSensitivity);
+        qSwap(m_patternType,     other.m_patternType);
     }
 
-    bool matchFileName(const QString &filename) const;
+    bool matchFileName(const QString &inputFileName) const;
 
     inline const QString &pattern() const { return m_pattern; }
     inline unsigned weight() const { return m_weight; }
@@ -100,10 +104,21 @@ public:
     inline bool isCaseSensitive() const { return m_caseSensitivity == Qt::CaseSensitive; }
 
 private:
+    enum PatternType {
+        SuffixPattern,
+        PrefixPattern,
+        LiteralPattern,
+        VdrPattern,        // special handling for "[0-9][0-9][0-9].vdr" pattern
+        AnimPattern,       // special handling for "*.anim[1-9j]" pattern
+        OtherPattern
+    };
+    PatternType detectPatternType(const QString &pattern) const;
+
     QString m_pattern;
     QString m_mimeType;
     int m_weight;
     Qt::CaseSensitivity m_caseSensitivity;
+    PatternType m_patternType;
 };
 Q_DECLARE_SHARED(QMimeGlobPattern)
 
diff --git a/src/corelib/mimetypes/qmimeprovider.cpp b/src/corelib/mimetypes/qmimeprovider.cpp
index 92be90658e..f152094dcd 100644
--- a/src/corelib/mimetypes/qmimeprovider.cpp
+++ b/src/corelib/mimetypes/qmimeprovider.cpp
@@ -278,7 +278,6 @@ void QMimeBinaryProvider::matchGlobList(QMimeGlobMatchResult &result, CacheFile
         //qDebug() << pattern << mimeType << weight << caseSensitive;
         QMimeGlobPattern glob(pattern, QString() /*unused*/, weight, qtCaseSensitive);
 
-        // TODO: this could be done faster for literals where a simple == would do.
         if (glob.matchFileName(fileName))
             result.addMatch(QLatin1String(mimeType), weight, pattern);
     }