kdecore: optimize KMimeType::findByContent()

read once, match until match is found or otherwise. that means less disk
I/O

Signed-off-by: Ivailo Monev <xakepa10@gmail.com>

kdecore/services/kmimemagicrule.cpp

@@ -41,12 +41,12 @@
  *
  */
 
-static bool testMatches(QIODevice* device, qint64 deviceSize, QByteArray& availableData, const QList<KMimeMagicMatch>& matches, const QString& mimeType)
+static bool testMatches(const QByteArray &availableData, const QList<KMimeMagicMatch>& matches, const QString& mimeType)
 {
     for ( QList<KMimeMagicMatch>::const_iterator it = matches.begin(), end = matches.end() ;
           it != end ; ++it ) {
         const KMimeMagicMatch& match = *it;
-        if (match.match(device, deviceSize, availableData, mimeType)) {
+        if (match.match(availableData, mimeType)) {
             // One of the hierarchies matched -> mimetype recognized.
             return true;
         }
@@ -54,53 +54,29 @@ static bool testMatches(QIODevice* device, qint64 deviceSize, QByteArray& availa
     return false;
 }
 
-bool KMimeMagicRule::match(QIODevice* device, qint64 deviceSize, QByteArray& availableData) const
+bool KMimeMagicRule::match(const QByteArray &availableData) const
 {
-    return testMatches(device, deviceSize, availableData, m_matches, m_mimetype);
+    return testMatches(availableData, m_matches, m_mimetype);
 }
 
-bool KMimeMagicMatch::match(QIODevice* device, qint64 deviceSize, QByteArray& availableData, const QString& mimeType) const
+bool KMimeMagicMatch::match(const QByteArray &availableData, const QString& mimeType) const
 {
     // First, check that "this" matches, then we'll dive into subMatches if any.
 
     const qint64 mDataSize = m_data.size();
-    if (m_rangeStart + mDataSize > deviceSize)
+    if (m_rangeStart + mDataSize > availableData.size())
         return false; // file is too small
 
     // Read in one block all the data we'll need
     // Example: m_data="ABC", m_rangeLength=3 -> we need 3+3-1=5 bytes (ABCxx,xABCx,xxABC would match)
-    const int dataNeeded = qMin(mDataSize + m_rangeLength - 1, deviceSize - m_rangeStart);
-    QByteArray readData;
-
-    /*kDebug() << "need " << dataNeeded << " bytes of data starting at " << m_rangeStart
-             << "  - availableData has " << availableData.size() << " bytes,"
-             << " device has " << deviceSize << " bytes.";*/
-
-    if (m_rangeStart + dataNeeded > availableData.size() && availableData.size() < deviceSize) {
-        // Need to read from device
-        if (!device->seek(m_rangeStart))
-            return false;
-        readData.resize(dataNeeded);
-        const int nread = device->read(readData.data(), dataNeeded);
-        //kDebug() << "readData (from device): reading" << dataNeeded << "bytes.";
-        if (nread < mDataSize)
-            return false; // error (or not enough data but we checked for that already)
-        if (m_rangeStart == 0 && readData.size() > availableData.size()) {
-            availableData = readData; // update cache
-        }
-        if (nread < readData.size()) {
-            // File big enough to contain m_data, but not big enough for the full rangeLength.
-            // Pad with zeros.
-            memset(readData.data() + nread, 0, dataNeeded - nread);
-        }
-        //kDebug() << "readData (from device) at pos " << m_rangeStart << ":" << readData;
-    } else {
-        readData = QByteArray::fromRawData(availableData.constData() + m_rangeStart,
-                                           dataNeeded);
+    const int dataNeeded = qMin(mDataSize + m_rangeLength - 1, availableData.size() - m_rangeStart);
+    QByteArray readData = QByteArray::fromRawData(
+        availableData.constData() + m_rangeStart,
+        dataNeeded
+    );
     // Warning, readData isn't null-terminated so this kDebug
     // gives valgrind warnings (when printing as char* data).
     //kDebug() << "readData (from availableData) at pos " << m_rangeStart << ":" << readData;
-    }
 
     // All we need to do now, is to look for m_data in readData (whose size is dataNeeded).
     // Either as a simple indexOf search, or applying the mask.
@@ -141,5 +117,5 @@ bool KMimeMagicMatch::match(QIODevice* device, qint64 deviceSize, QByteArray& av
         return true;
 
     // Check that one of the submatches matches too
-    return testMatches(device, deviceSize, availableData, m_subMatches, mimeType);
+    return testMatches(availableData, m_subMatches, mimeType);
 }

kdecore/services/kmimemagicrule_p.h

@@ -31,7 +31,7 @@
  */
 struct KMimeMagicMatch
 {
-    bool match(QIODevice* device, qint64 deviceSize, QByteArray& availableData, const QString& mimeType) const;
+    bool match(const QByteArray &availableData, const QString& mimeType) const;
 
     qint64 m_rangeStart;
     qint64 m_rangeLength;
@@ -57,7 +57,7 @@ public:
     KMimeMagicRule(const QString &mimetype, int priority, const QList<KMimeMagicMatch> &matches)
         : m_mimetype(mimetype), m_priority(priority), m_matches(matches) {}
 
-    bool match(QIODevice* device, qint64 deviceSize, QByteArray& availableData) const;
+    bool match(const QByteArray &availableData) const;
 
     QString mimetype() const { return m_mimetype; }
     int priority() const { return m_priority; }

kdecore/services/kmimetype.cpp

@@ -223,7 +223,8 @@ KMimeType::Ptr KMimeType::findByUrl(const KUrl &url, mode_t mode,
         QFile file(localfile);
         if (file.open(QIODevice::ReadOnly)) {
             int magicAccuracy = 0;
-            KMimeType::Ptr mime = KMimeTypeRepository::self()->findFromContent(&file, &magicAccuracy);
+            // provide enough data for most rules (there are exceptions which require twice as much tho)
+            KMimeType::Ptr mime = KMimeTypeRepository::self()->findFromContent(file.read(16384), &magicAccuracy);
             // mime can't be 0, except in case of install problems.
             // However we get magicAccuracy==0 for octet-stream, i.e. no magic match found.
             // kDebug() << "findFromContent said" << (mime?mime->name():QString()) << "with accuracy" << magicAccuracy;
@@ -288,9 +289,7 @@ KMimeType::Ptr KMimeType::findByName(const QString &fileName, int *accuracy)
 KMimeType::Ptr KMimeType::findByContent(const QByteArray &data, int *accuracy)
 {
     KMimeTypeRepository::self()->checkEssentialMimeTypes();
-    QBuffer buffer(const_cast<QByteArray *>(&data));
-    buffer.open(QIODevice::ReadOnly);
-    return KMimeTypeRepository::self()->findFromContent(&buffer, accuracy);
+    return KMimeTypeRepository::self()->findFromContent(data, accuracy);
 }
 
 QString KMimeType::extractKnownExtension(const QString &fileName)

kdecore/services/kmimetyperepository.cpp

@@ -314,26 +314,18 @@ QStringList KMimeTypeRepository::findFromFileName(const QString &fileName, QStri
     return matchingMimeTypes;
 }
 
-KMimeType::Ptr KMimeTypeRepository::findFromContent(QIODevice* device, int* accuracy)
+KMimeType::Ptr KMimeTypeRepository::findFromContent(const QByteArray &data, int* accuracy)
 {
-    Q_ASSERT(device->isOpen());
-    const qint64 deviceSize = device->size();
-    if (deviceSize == 0) {
+    if (data.size() == 0) {
         if (accuracy) {
             *accuracy = 100;
         }
         return findMimeTypeByName(QLatin1String("application/x-zerosize"), KMimeType::DontResolveAlias);
     }
-    // provide enough data for most rules (there are exceptions which require twice as much tho)
-    const qint64 dataNeeded = qMin(deviceSize, (qint64) 16384);
-    QByteArray beginning(dataNeeded, '\0');
-    if (!device->seek(0) || device->read(beginning.data(), dataNeeded) == -1) {
-        return defaultMimeTypePtr(); // don't bother detecting unreadable file
-    }
 
     // Apply magic rules
     Q_FOREACH ( const KMimeMagicRule& rule, m_magicRules ) {
-        if (rule.match(device, deviceSize, beginning)) {
+        if (rule.match( data)) {
             if (accuracy) {
                 *accuracy = rule.priority();
             }
@@ -343,7 +335,7 @@ KMimeType::Ptr KMimeTypeRepository::findFromContent(QIODevice* device, int* accu
 
     // Do fallback code so that we never return 0
     // Nothing worked, check if the file contents looks like binary or text
-    if (!KMimeType::isBufferBinaryData(beginning)) {
+    if (!KMimeType::isBufferBinaryData(data)) {
         if (accuracy) {
             *accuracy = 5;
         }

kdecore/services/kmimetyperepository_p.h

@@ -97,13 +97,13 @@ private: // only for KMimeType and unittests
     QStringList findFromFileName(const QString &filename, QString *matchingExtension = nullptr) const;
 
     /**
-     * Find a mimetype from the content of a file or buffer
-     * @param device the file or buffer. Must be open.
+     * Find a mimetype from the content of data chunk
+     * @param data chunk of data
      * @param accuracy returns the priority of the rule that matched
      *
      * This is internal API, use KMimeType::findByUrl instead.
      */
-    KMimeType::Ptr findFromContent(QIODevice *device, int *accuracy);
+    KMimeType::Ptr findFromContent(const QByteArray &data, int *accuracy);
 
     /**
      * @return true if at least one mimetype is present

kdecore/tests/kmimetypetest.cpp

@@ -842,20 +842,15 @@ void KMimeTypeTest::testParseMagicFile()
     QFETCH(QString, testData);
     // kDebug() << QTest::currentDataTag();
     QFETCH(QString, expected);
-    QBuffer testBuffer;
-    testBuffer.setData(testData.toLatin1());
-    QVERIFY(testBuffer.open(QIODevice::ReadOnly));
-    const qint64 testBufferSize = testBuffer.size();
+    QByteArray testDataBytes = testData.toLatin1();
     QString found;
-    QByteArray beginning;
     Q_FOREACH(const KMimeMagicRule& rule, m_rules) {
-        if (rule.match(&testBuffer, testBufferSize, beginning)) {
+        if (rule.match(testDataBytes)) {
             found = rule.mimetype();
             break;
         }
     }
     QCOMPARE(found, expected);
-    testBuffer.close();
 }
 
 void KMimeTypeTest::testHelperProtocols()