qnetworkaccessmanager.cpp

Go to the documentation of this file.

// Copyright (C) 2020 The Qt Company Ltd.
// SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only
// Qt-Security score:significant reason:default
 
#include <QtNetwork/private/qtnetworkglobal_p.h>
 
#include "qnetworkaccessmanager.h"
#include "qnetworkaccessmanager_p.h"
#include "qnetworkrequest.h"
#include "qnetworkreply.h"
#include "qnetworkreply_p.h"
#include "qnetworkcookie.h"
#include "qnetworkcookiejar.h"
#include "qabstractnetworkcache.h"
#include "qhstspolicy.h"
#include "qhsts_p.h"
 
#if QT_CONFIG(settings)
#include "qhstsstore_p.h"
#endif // QT_CONFIG(settings)
 
#include "qnetworkaccessfilebackend_p.h"
#include "qnetworkaccessdebugpipebackend_p.h"
#include "qnetworkaccesscachebackend_p.h"
#include "qnetworkreplydataimpl_p.h"
#include "qnetworkreplyfileimpl_p.h"
 
#include "qnetworkaccessbackend_p.h"
#include "qnetworkreplyimpl_p.h"
 
#include "QtCore/qbuffer.h"
#include "QtCore/qlist.h"
#include "QtCore/qurl.h"
#include "QtNetwork/private/qauthenticator_p.h"
#include "QtNetwork/qsslconfiguration.h"
 
#if QT_CONFIG(http)
#include "QtNetwork/private/http2protocol_p.h"
#include "qhttpmultipart.h"
#include "qhttpmultipart_p.h"
#include "qnetworkreplyhttpimpl_p.h"
#endif
 
#include "qthread.h"
 
#include <QHostInfo>
 
#include "QtCore/qapplicationstatic.h"
#include "QtCore/qloggingcategory.h"
#include <QtCore/private/qfactoryloader_p.h>
 
#if defined(Q_OS_MACOS)
#include <QtCore/private/qcore_mac_p.h>
 
#include <CoreServices/CoreServices.h>
#include <SystemConfiguration/SystemConfiguration.h>
#include <Security/Security.h>
#endif
#ifdef Q_OS_WASM
#include "qnetworkreplywasmimpl_p.h"
#include "qhttpmultipart.h"
#include "qhttpmultipart_p.h"
#endif
 
#include <mutex>
#include <utility>
 
QT_BEGIN_NAMESPACE
 
using namespace Qt::StringLiterals;
using namespace std::chrono_literals;
 
#if defined(Q_OS_MACOS)
Q_STATIC_LOGGING_CATEGORY(lcQnam, "qt.network.access.manager")
#endif
 
Q_APPLICATION_STATIC(QNetworkAccessFileBackendFactory, fileBackend)
 
#if QT_CONFIG(private_tests)
Q_GLOBAL_STATIC(QNetworkAccessDebugPipeBackendFactory, debugpipeBackend)
#endif
 
Q_APPLICATION_STATIC(QFactoryLoader, qnabfLoader, QNetworkAccessBackendFactory_iid, "/networkaccess"_L1)
 
#if defined(Q_OS_MACOS)
bool getProxyAuth(const QString& proxyHostname, const QString &scheme, QString& username, QString& password)
{
    CFStringRef protocolType = nullptr;
    if (scheme.compare("ftp"_L1, Qt::CaseInsensitive) == 0) {
        protocolType = kSecAttrProtocolFTPProxy;
    } else if (scheme.compare("http"_L1, Qt::CaseInsensitive) == 0
               || scheme.compare("preconnect-http"_L1, Qt::CaseInsensitive) == 0) {
        protocolType = kSecAttrProtocolHTTPProxy;
    } else if (scheme.compare("https"_L1,Qt::CaseInsensitive)==0
               || scheme.compare("preconnect-https"_L1, Qt::CaseInsensitive) == 0) {
        protocolType = kSecAttrProtocolHTTPSProxy;
    } else {
        qCWarning(lcQnam) << "Cannot query user name and password for a proxy, unnknown protocol:"
                          << scheme;
        return false;
    }
 
    QCFType<CFMutableDictionaryRef> query(CFDictionaryCreateMutable(kCFAllocatorDefault,
                                                                    0, nullptr, nullptr));
    Q_ASSERT(query);
 
    CFDictionaryAddValue(query, kSecClass, kSecClassInternetPassword);
    CFDictionaryAddValue(query, kSecAttrProtocol, protocolType);
 
    QCFType<CFStringRef> serverName; // Note the scope.
    if (proxyHostname.size()) {
        serverName = proxyHostname.toCFString();
        CFDictionaryAddValue(query, kSecAttrServer, serverName);
    }
 
    // This is to get the user name in the result:
    CFDictionaryAddValue(query, kSecReturnAttributes, kCFBooleanTrue);
    // This one to get the password:
    CFDictionaryAddValue(query, kSecReturnData, kCFBooleanTrue);
 
    // The default for kSecMatchLimit key is 1 (the first match only), which is fine,
    // so don't set this value explicitly.
 
    QCFType<CFTypeRef> replyData;
    if (SecItemCopyMatching(query, &replyData) != errSecSuccess) {
        qCWarning(lcQnam, "Failed to extract user name and password from the keychain.");
        return false;
    }
 
    if (!replyData || CFDictionaryGetTypeID() != CFGetTypeID(replyData)) {
        qCWarning(lcQnam, "Query returned data in unexpected format.");
        return false;
    }
 
    CFDictionaryRef accountData = replyData.as<CFDictionaryRef>();
    const void *value = CFDictionaryGetValue(accountData, kSecAttrAccount);
    if (!value || CFGetTypeID(value) != CFStringGetTypeID()) {
        qCWarning(lcQnam, "Cannot find user name or its format is unknown.");
        return false;
    }
    username = QString::fromCFString(static_cast<CFStringRef>(value));
 
    value = CFDictionaryGetValue(accountData, kSecValueData);
    if (!value || CFGetTypeID(value) != CFDataGetTypeID()) {
        qCWarning(lcQnam, "Cannot find password or its format is unknown.");
        return false;
    }
    const CFDataRef passData = static_cast<const CFDataRef>(value);
    password = QString::fromLocal8Bit(reinterpret_cast<const char *>(CFDataGetBytePtr(passData)),
                                      qsizetype(CFDataGetLength(passData)));
    return true;
}
#endif // Q_OS_MACOS
 
 
 
static void ensureInitialized()
{
#if QT_CONFIG(private_tests)
    (void) debugpipeBackend();
#endif
 
    // leave this one last since it will query the special QAbstractFileEngines
    (void) fileBackend();
}
 
/*!
    \class QNetworkAccessManager
    \brief The QNetworkAccessManager class allows the application to
    send network requests and receive replies.
    \since 4.4
 
    \ingroup network
    \inmodule QtNetwork
    \reentrant
 
    The Network Access API is constructed around one QNetworkAccessManager
    object, which holds the common configuration and settings for the requests
    it sends. It contains the proxy and cache configuration, as well as the
    signals related to such issues, and reply signals that can be used to
    monitor the progress of a network operation. One QNetworkAccessManager
    instance should be enough for the whole Qt application. Since
    QNetworkAccessManager is based on QObject, it can only be used from the
    thread it belongs to.
 
    Once a QNetworkAccessManager object has been created, the application can
    use it to send requests over the network. A group of standard functions
    is supplied that take a request and optional data, and each returns a
    QNetworkReply object. The returned object is used to obtain any data
    returned in response to the corresponding request.
 
    A simple download off the network could be accomplished with:
    \snippet code/src_network_access_qnetworkaccessmanager.cpp 0
 
    QNetworkAccessManager has an asynchronous API.
    When the \tt replyFinished slot above is called, the parameter it
    takes is the QNetworkReply object containing the downloaded data
    as well as meta-data (headers, etc.).
 
    \note After the request has finished, it is the responsibility of the user
    to delete the QNetworkReply object at an appropriate time. Do not directly
    delete it inside the slot connected to finished(). You can use the
    deleteLater() function.
 
    \note QNetworkAccessManager queues the requests it receives. The number
    of requests executed in parallel is dependent on the protocol.
    Currently, for the HTTP protocol on desktop platforms, 6 requests are
    executed in parallel for one host/port combination.
 
    \note QNetworkAccessManager doesn't handle RFC 2616 Section 8.2.2 properly,
    in that it doesn't react to incoming data until it's done writing. For
    example, the upload of a large file won't stop even if the server returns
    a status code that instructs the client to not continue.
 
    A more involved example, assuming the manager is already existent,
    can be:
    \snippet code/src_network_access_qnetworkaccessmanager.cpp 1
 
    Since Qt 6.11 the defaults of the TCP Keepalive parameters used by
    QNetworkAccessManager have been changed. With the current settings
    the connection will be terminated after 2 minutes of inactivity.
 
    These settings can be changed the individual requests, to make
    them more lenient, or even more aggressive via the QNetworkRequest API.
    \snippet http/httpwindow.cpp qnam-tcpkeepalive
 
    In the above snippet we are picking a more aggressive strategy, to
    terminate the connection after thirty seconds of inactivity. This can
    be useful, for example, in early detection of network hangs caused
    by network changes on Linux.
 
    \sa QNetworkRequest, QNetworkReply, QNetworkProxy
*/
 
/*!
    \enum QNetworkAccessManager::Operation
 
    Indicates the operation this reply is processing.
 
    \value HeadOperation        retrieve headers operation (created
    with head())
 
    \value GetOperation         retrieve headers and download contents
    (created with get())
 
    \value PutOperation         upload contents operation (created
    with put())
 
    \value PostOperation        send the contents of an HTML form for
    processing via HTTP POST (created with post())
 
    \value DeleteOperation      delete contents operation (created with
    deleteResource())
 
    \value CustomOperation      custom operation (created with
    sendCustomRequest())    \since 4.7
 
    \omitvalue UnknownOperation
 
    \sa QNetworkReply::operation()
*/
 
/*!
    \fn void QNetworkAccessManager::networkSessionConnected()
 
    \since 4.7
    \deprecated
 
    \internal
 
    This signal is emitted when the status of the network session changes into a usable (Connected)
    state. It is used to signal to QNetworkReplys to start or migrate their network operation once
    the network session has been opened or finished roaming.
*/
 
/*!
    \fn void QNetworkAccessManager::proxyAuthenticationRequired(const QNetworkProxy &proxy, QAuthenticator *authenticator)
 
    This signal is emitted whenever a proxy requests authentication
    and QNetworkAccessManager cannot find a valid, cached
    credential. The slot connected to this signal should fill in the
    credentials for the proxy \a proxy in the \a authenticator object.
 
    QNetworkAccessManager will cache the credentials internally. The
    next time the proxy requests authentication, QNetworkAccessManager
    will automatically send the same credential without emitting the
    proxyAuthenticationRequired signal again.
 
    If the proxy rejects the credentials, QNetworkAccessManager will
    emit the signal again.
 
    \sa proxy(), setProxy(), authenticationRequired()
*/
 
/*!
    \fn void QNetworkAccessManager::authenticationRequired(QNetworkReply *reply, QAuthenticator *authenticator)
 
    This signal is emitted whenever a final server requests
    authentication before it delivers the requested contents. The slot
    connected to this signal should fill the credentials for the
    contents (which can be determined by inspecting the \a reply
    object) in the \a authenticator object.
 
    QNetworkAccessManager will cache the credentials internally and
    will send the same values if the server requires authentication
    again, without emitting the authenticationRequired() signal. If it
    rejects the credentials, this signal will be emitted again.
 
    \note To have the request not send credentials you must not call
    setUser() or setPassword() on the \a authenticator object. This
    will result in the \l finished() signal being emitted with a
    \l QNetworkReply with error \l {QNetworkReply::} {AuthenticationRequiredError}.
 
    \note It is not possible to use a QueuedConnection to connect to
    this signal, as the connection will fail if the authenticator has
    not been filled in with new information when the signal returns.
 
    \sa proxyAuthenticationRequired(), QAuthenticator::setUser(), QAuthenticator::setPassword()
*/
 
/*!
    \fn void QNetworkAccessManager::finished(QNetworkReply *reply)
 
    This signal is emitted whenever a pending network reply is
    finished. The \a reply parameter will contain a pointer to the
    reply that has just finished. This signal is emitted in tandem
    with the QNetworkReply::finished() signal.
 
    See QNetworkReply::finished() for information on the status that
    the object will be in.
 
    \note Do not delete the \a reply object in the slot connected to this
    signal. Use deleteLater().
 
    \sa QNetworkReply::finished(), QNetworkReply::error()
*/
 
/*!
    \fn void QNetworkAccessManager::encrypted(QNetworkReply *reply)
    \since 5.1
 
    This signal is emitted when an SSL/TLS session has successfully
    completed the initial handshake. At this point, no user data
    has been transmitted. The signal can be used to perform
    additional checks on the certificate chain, for example to
    notify users when the certificate for a website has changed. The
    \a reply parameter specifies which network reply is responsible.
    If the reply does not match the expected criteria then it should
    be aborted by calling QNetworkReply::abort() by a slot connected
    to this signal. The SSL configuration in use can be inspected
    using the QNetworkReply::sslConfiguration() method.
 
    Internally, QNetworkAccessManager may open multiple connections
    to a server, in order to allow it process requests in parallel.
    These connections may be reused, which means that the encrypted()
    signal would not be emitted. This means that you are only
    guaranteed to receive this signal for the first connection to a
    site in the lifespan of the QNetworkAccessManager.
 
    \sa QSslSocket::encrypted()
    \sa QNetworkReply::encrypted()
*/
 
/*!
    \fn void QNetworkAccessManager::sslErrors(QNetworkReply *reply, const QList<QSslError> &errors)
 
    This signal is emitted if the SSL/TLS session encountered errors
    during the set up, including certificate verification errors. The
    \a errors parameter contains the list of errors and \a reply is
    the QNetworkReply that is encountering these errors.
 
    To indicate that the errors are not fatal and that the connection
    should proceed, the QNetworkReply::ignoreSslErrors() function should be called
    from the slot connected to this signal. If it is not called, the
    SSL session will be torn down before any data is exchanged
    (including the URL).
 
    This signal can be used to display an error message to the user
    indicating that security may be compromised and display the
    SSL settings (see sslConfiguration() to obtain it). If the user
    decides to proceed after analyzing the remote certificate, the
    slot should call ignoreSslErrors().
 
    \sa QSslSocket::sslErrors(), QNetworkReply::sslErrors(),
    QNetworkReply::sslConfiguration(), QNetworkReply::ignoreSslErrors()
*/
 
/*!
    \fn void QNetworkAccessManager::preSharedKeyAuthenticationRequired(QNetworkReply *reply, QSslPreSharedKeyAuthenticator *authenticator)
    \since 5.5
 
    This signal is emitted if the SSL/TLS handshake negotiates a PSK
    ciphersuite, and therefore a PSK authentication is then required.
    The \a reply object is the QNetworkReply that is negotiating
    such ciphersuites.
 
    When using PSK, the client must send to the server a valid identity and a
    valid pre shared key, in order for the SSL handshake to continue.
    Applications can provide this information in a slot connected to this
    signal, by filling in the passed \a authenticator object according to their
    needs.
 
    \note Ignoring this signal, or failing to provide the required credentials,
    will cause the handshake to fail, and therefore the connection to be aborted.
 
    \note The \a authenticator object is owned by the reply and must not be
    deleted by the application.
 
    \sa QSslPreSharedKeyAuthenticator
*/
 
/*!
    Constructs a QNetworkAccessManager object that is the center of
    the Network Access API and sets \a parent as the parent object.
*/
QNetworkAccessManager::QNetworkAccessManager(QObject *parent)
    : QObject(*new QNetworkAccessManagerPrivate, parent)
{
    ensureInitialized();
    d_func()->ensureBackendPluginsLoaded();
 
    qRegisterMetaType<QNetworkReply::NetworkError>();
#ifndef QT_NO_NETWORKPROXY
    qRegisterMetaType<QNetworkProxy>();
#endif
#ifndef QT_NO_SSL
    qRegisterMetaType<QList<QSslError> >();
    qRegisterMetaType<QSslConfiguration>();
    qRegisterMetaType<QSslPreSharedKeyAuthenticator *>();
#endif
    qRegisterMetaType<QList<std::pair<QByteArray, QByteArray>>>();
#if QT_CONFIG(http)
    qRegisterMetaType<QHttpNetworkRequest>();
#endif
    qRegisterMetaType<QNetworkReply::NetworkError>();
    qRegisterMetaType<QSharedPointer<char> >();
}
 
/*!
    Destroys the QNetworkAccessManager object and frees up any
    resources. Note that QNetworkReply objects that are returned from
    this class have this object set as their parents, which means that
    they will be deleted along with it if you don't call
    QObject::setParent() on them.
*/
QNetworkAccessManager::~QNetworkAccessManager()
{
#ifndef QT_NO_NETWORKPROXY
    delete d_func()->proxyFactory;
#endif
 
    // Delete the QNetworkReply children first.
    // Else a QAbstractNetworkCache might get deleted in ~QObject
    // before a QNetworkReply that accesses the QAbstractNetworkCache
    // object in its destructor.
    qDeleteAll(findChildren<QNetworkReply *>());
    // The other children will be deleted in this ~QObject
    // FIXME instead of this "hack" make the QNetworkReplyImpl
    // properly watch the cache deletion, e.g. via a QWeakPointer.
}
 
#ifndef QT_NO_NETWORKPROXY
/*!
    Returns the QNetworkProxy that the requests sent using this
    QNetworkAccessManager object will use. The default value for the
    proxy is QNetworkProxy::DefaultProxy.
 
    \sa setProxy(), setProxyFactory(), proxyAuthenticationRequired()
*/
QNetworkProxy QNetworkAccessManager::proxy() const
{
    return d_func()->proxy;
}
 
/*!
    Sets the proxy to be used in future requests to be \a proxy. This
    does not affect requests that have already been sent. The
    proxyAuthenticationRequired() signal will be emitted if the proxy
    requests authentication.
 
    A proxy set with this function will be used for all requests
    issued by QNetworkAccessManager. In some cases, it might be
    necessary to select different proxies depending on the type of
    request being sent or the destination host. If that's the case,
    you should consider using setProxyFactory().
 
    \sa proxy(), proxyAuthenticationRequired()
*/
void QNetworkAccessManager::setProxy(const QNetworkProxy &proxy)
{
    Q_D(QNetworkAccessManager);
    delete d->proxyFactory;
    d->proxy = proxy;
    d->proxyFactory = nullptr;
}
 
/*!
    \fn QNetworkProxyFactory *QNetworkAccessManager::proxyFactory() const
    \since 4.5
 
    Returns the proxy factory that this QNetworkAccessManager object
    is using to determine the proxies to be used for requests.
 
    Note that the pointer returned by this function is managed by
    QNetworkAccessManager and could be deleted at any time.
 
    \sa setProxyFactory(), proxy()
*/
QNetworkProxyFactory *QNetworkAccessManager::proxyFactory() const
{
    return d_func()->proxyFactory;
}
 
/*!
    \since 4.5
 
    Sets the proxy factory for this class to be \a factory. A proxy
    factory is used to determine a more specific list of proxies to be
    used for a given request, instead of trying to use the same proxy
    value for all requests.
 
    All queries sent by QNetworkAccessManager will have type
    QNetworkProxyQuery::UrlRequest.
 
    For example, a proxy factory could apply the following rules:
    \list
      \li if the target address is in the local network (for example,
         if the hostname contains no dots or if it's an IP address in
         the organization's range), return QNetworkProxy::NoProxy
      \li if the request is FTP, return an FTP proxy
      \li if the request is HTTP or HTTPS, then return an HTTP proxy
      \li otherwise, return a SOCKSv5 proxy server
    \endlist
 
    The lifetime of the object \a factory will be managed by
    QNetworkAccessManager. It will delete the object when necessary.
 
    \note If a specific proxy is set with setProxy(), the factory will not
    be used.
 
    \sa proxyFactory(), setProxy(), QNetworkProxyQuery
*/
void QNetworkAccessManager::setProxyFactory(QNetworkProxyFactory *factory)
{
    Q_D(QNetworkAccessManager);
    delete d->proxyFactory;
    d->proxyFactory = factory;
    d->proxy = QNetworkProxy();
}
#endif
 
/*!
    \since 4.5
 
    Returns the cache that is used to store data obtained from the network.
 
    \sa setCache()
*/
QAbstractNetworkCache *QNetworkAccessManager::cache() const
{
    Q_D(const QNetworkAccessManager);
    return d->networkCache;
}
 
/*!
    \since 4.5
 
    Sets the manager's network cache to be the \a cache specified. The cache
    is used for all requests dispatched by the manager.
 
    Use this function to set the network cache object to a class that implements
    additional features, like saving the cookies to permanent storage.
 
    \note QNetworkAccessManager takes ownership of the \a cache object.
 
    QNetworkAccessManager by default does not have a set cache.
    Qt provides a simple disk cache, QNetworkDiskCache, which can be used.
 
    \sa cache(), QNetworkRequest::CacheLoadControl
*/
void QNetworkAccessManager::setCache(QAbstractNetworkCache *cache)
{
    Q_D(QNetworkAccessManager);
    if (d->networkCache != cache) {
        delete d->networkCache;
        d->networkCache = cache;
        if (d->networkCache)
            d->networkCache->setParent(this);
    }
}
 
/*!
    Returns the QNetworkCookieJar that is used to store cookies
    obtained from the network as well as cookies that are about to be
    sent.
 
    \sa setCookieJar()
*/
QNetworkCookieJar *QNetworkAccessManager::cookieJar() const
{
    Q_D(const QNetworkAccessManager);
    if (!d->cookieJar)
        d->createCookieJar();
    return d->cookieJar;
}
 
/*!
    Sets the manager's cookie jar to be the \a cookieJar specified.
    The cookie jar is used by all requests dispatched by the manager.
 
    Use this function to set the cookie jar object to a class that
    implements additional features, like saving the cookies to permanent
    storage.
 
    \note QNetworkAccessManager takes ownership of the \a cookieJar object.
 
    If \a cookieJar is in the same thread as this QNetworkAccessManager,
    it will set the parent of the \a cookieJar
    so that the cookie jar is deleted when this
    object is deleted as well. If you want to share cookie jars
    between different QNetworkAccessManager objects, you may want to
    set the cookie jar's parent to 0 after calling this function.
 
    QNetworkAccessManager by default does not implement any cookie
    policy of its own: it accepts all cookies sent by the server, as
    long as they are well formed and meet the minimum security
    requirements (cookie domain matches the request's and cookie path
    matches the request's). In order to implement your own security
    policy, override the QNetworkCookieJar::cookiesForUrl() and
    QNetworkCookieJar::setCookiesFromUrl() virtual functions. Those
    functions are called by QNetworkAccessManager when it detects a
    new cookie.
 
    \sa cookieJar(), QNetworkCookieJar::cookiesForUrl(), QNetworkCookieJar::setCookiesFromUrl()
*/
void QNetworkAccessManager::setCookieJar(QNetworkCookieJar *cookieJar)
{
    Q_D(QNetworkAccessManager);
    d->cookieJarCreated = true;
    if (d->cookieJar != cookieJar) {
        if (d->cookieJar && d->cookieJar->parent() == this)
            delete d->cookieJar;
        d->cookieJar = cookieJar;
        if (cookieJar && thread() == cookieJar->thread())
            d->cookieJar->setParent(this);
    }
}
 
/*!
    \since 5.9
 
    If \a enabled is \c true, QNetworkAccessManager follows the HTTP Strict Transport
    Security policy (HSTS, RFC6797). When processing a request, QNetworkAccessManager
    automatically replaces the "http" scheme with "https" and uses a secure transport
    for HSTS hosts. If it's set explicitly, port 80 is replaced by port 443.
 
    When HSTS is enabled, for each HTTP response containing HSTS header and
    received over a secure transport, QNetworkAccessManager will update its HSTS
    cache, either remembering a host with a valid policy or removing a host with
    an expired or disabled HSTS policy.
 
    \sa isStrictTransportSecurityEnabled()
*/
void QNetworkAccessManager::setStrictTransportSecurityEnabled(bool enabled)
{
    Q_D(QNetworkAccessManager);
    d->stsEnabled = enabled;
}
 
/*!
    \since 5.9
 
    Returns true if HTTP Strict Transport Security (HSTS) was enabled. By default
    HSTS is disabled.
 
    \sa setStrictTransportSecurityEnabled()
*/
bool QNetworkAccessManager::isStrictTransportSecurityEnabled() const
{
    Q_D(const QNetworkAccessManager);
    return d->stsEnabled;
}
 
/*!
    \since 5.10
 
    If \a enabled is \c true, the internal HSTS cache will use a persistent store
    to read and write HSTS policies. \a storeDir defines where this store will be
    located. The default location is defined by QStandardPaths::CacheLocation.
    If there is no writable QStandartPaths::CacheLocation and \a storeDir is an
    empty string, the store will be located in the program's working directory.
 
    \note If HSTS cache already contains HSTS policies by the time persistent
    store is enabled, these policies will be preserved in the store. In case both
    cache and store contain the same known hosts, policies from cache are considered
    to be more up-to-date (and thus will overwrite the previous values in the store).
    If this behavior is undesired, enable HSTS store before enabling Strict Transport
    Security. By default, the persistent store of HSTS policies is disabled.
 
    \note The HSTS store persists policies to disk using QSettings in INI format
    without encryption or integrity protection. The store reveals which hosts the
    application has communicated with over HTTPS, since an HSTS entry is recorded
    for each host that sends a Strict-Transport-Security header. In privacy-sensitive
    applications, this connection history should be treated as confidential data.
 
    An attacker with write access to the store file could remove HSTS entries for
    specific domains, causing the application to permit plaintext HTTP connections to
    those domains on subsequent requests and potentially enabling SSL stripping attacks.
    Applications that rely on HSTS should ensure the store directory is protected by
    appropriate filesystem permissions.
 
    \sa isStrictTransportSecurityStoreEnabled(), setStrictTransportSecurityEnabled(),
    QStandardPaths::standardLocations()
*/
 
void QNetworkAccessManager::enableStrictTransportSecurityStore(bool enabled, const QString &storeDir)
{
#if QT_CONFIG(settings)
    Q_D(QNetworkAccessManager);
    d->stsStore.reset(enabled ? new QHstsStore(storeDir) : nullptr);
    d->stsCache.setStore(d->stsStore.get());
#else
    Q_UNUSED(enabled);
    Q_UNUSED(storeDir);
    qWarning("HSTS permanent store requires the feature 'settings' enabled");
#endif // QT_CONFIG(settings)
}
 
/*!
    \since 5.10
 
    Returns true if HSTS cache uses a permanent store to load and store HSTS
    policies.
 
    \sa enableStrictTransportSecurityStore()
*/
 
bool QNetworkAccessManager::isStrictTransportSecurityStoreEnabled() const
{
#if QT_CONFIG(settings)
    Q_D(const QNetworkAccessManager);
    return bool(d->stsStore);
#else
    return false;
#endif // QT_CONFIG(settings)
}
 
/*!
    \since 5.9
 
    Adds HTTP Strict Transport Security policies into HSTS cache.
    \a knownHosts contains the known hosts that have QHstsPolicy
    information.
 
    \note An expired policy will remove a known host from the cache, if previously
    present.
 
    \note While processing HTTP responses, QNetworkAccessManager can also update
    the HSTS cache, removing or updating exitsting policies or introducing new
    \a knownHosts. The current implementation thus is server-driven, client code
    can provide QNetworkAccessManager with previously known or discovered
    policies, but this information can be overridden by "Strict-Transport-Security"
    response headers.
 
    \sa strictTransportSecurityHosts(), enableStrictTransportSecurityStore(), QHstsPolicy
*/
 
void QNetworkAccessManager::addStrictTransportSecurityHosts(const QList<QHstsPolicy> &knownHosts)
{
    Q_D(QNetworkAccessManager);
    d->stsCache.updateFromPolicies(knownHosts);
}
 
/*!
    \since 5.9
 
    Returns the list of HTTP Strict Transport Security policies. This list can
    differ from what was initially set via addStrictTransportSecurityHosts() if
    HSTS cache was updated from a "Strict-Transport-Security" response header.
 
    \sa addStrictTransportSecurityHosts(), QHstsPolicy
*/
QList<QHstsPolicy> QNetworkAccessManager::strictTransportSecurityHosts() const
{
    Q_D(const QNetworkAccessManager);
    return d->stsCache.policies();
}
 
/*!
    Posts a request to obtain the network headers for \a request
    and returns a new QNetworkReply object which will contain such headers.
 
    The function is named after the HTTP request associated (HEAD).
*/
QNetworkReply *QNetworkAccessManager::head(const QNetworkRequest &request)
{
    return d_func()->postProcess(createRequest(QNetworkAccessManager::HeadOperation, request));
}
 
/*!
    Posts a request to obtain the contents of the target \a request
    and returns a new QNetworkReply object opened for reading which emits the
    \l{QIODevice::readyRead()}{readyRead()} signal whenever new data
    arrives.
 
    The contents as well as associated headers will be downloaded.
 
    \sa post(), put(), deleteResource(), sendCustomRequest()
*/
QNetworkReply *QNetworkAccessManager::get(const QNetworkRequest &request)
{
    return d_func()->postProcess(createRequest(QNetworkAccessManager::GetOperation, request));
}
 
/*!
   \since 6.7
 
   \overload
 
   \note A GET request with a message body is not cached.
 
   \note If the request is redirected, the message body will be kept only if the status code is
   308.
*/
 
QNetworkReply *QNetworkAccessManager::get(const QNetworkRequest &request, QIODevice *data)
{
    QNetworkRequest newRequest(request);
    return d_func()->postProcess(
            createRequest(QNetworkAccessManager::GetOperation, newRequest, data));
}
 
/*!
   \since 6.7
 
   \overload
 
   \note A GET request with a message body is not cached.
 
   \note If the request is redirected, the message body will be kept only if the status code is
   308.
*/
 
QNetworkReply *QNetworkAccessManager::get(const QNetworkRequest &request, const QByteArray &data)
{
    QBuffer *buffer = new QBuffer;
    buffer->setData(data);
    buffer->open(QIODevice::ReadOnly);
 
    QNetworkReply *reply = get(request, buffer);
    buffer->setParent(reply);
    return reply;
}
 
/*!
    Sends an HTTP POST request to the destination specified by \a request
    and returns a new QNetworkReply object opened for reading that will
    contain the reply sent by the server. The contents of  the \a data
    device will be uploaded to the server.
 
    \a data must be open for reading and must remain valid until the
    finished() signal is emitted for this reply.
 
    \note Sending a POST request on protocols other than HTTP and
    HTTPS is undefined and will probably fail.
 
    \sa get(), put(), deleteResource(), sendCustomRequest()
*/
QNetworkReply *QNetworkAccessManager::post(const QNetworkRequest &request, QIODevice *data)
{
    return d_func()->postProcess(createRequest(QNetworkAccessManager::PostOperation, request, data));
}
 
/*!
    \overload
 
    Sends the contents of the \a data byte array to the destination
    specified by \a request.
*/
QNetworkReply *QNetworkAccessManager::post(const QNetworkRequest &request, const QByteArray &data)
{
    QBuffer *buffer = new QBuffer;
    buffer->setData(data);
    buffer->open(QIODevice::ReadOnly);
 
    QNetworkReply *reply = post(request, buffer);
    buffer->setParent(reply);
    return reply;
}
 
/*!
    \fn QNetworkReply *QNetworkAccessManager::post(const QNetworkRequest &request, std::nullptr_t nptr)
 
    \since 6.8
 
    \overload
 
    Sends the POST request specified by \a request without a body and returns
    a new QNetworkReply object.
*/
 
#if QT_CONFIG(http) || defined(Q_OS_WASM)
/*!
    \since 4.8
 
    \overload
 
    Sends the contents of the \a multiPart message to the destination
    specified by \a request.
 
    This can be used for sending MIME multipart messages over HTTP.
 
    \sa QHttpMultiPart, QHttpPart, put()
*/
QNetworkReply *QNetworkAccessManager::post(const QNetworkRequest &request, QHttpMultiPart *multiPart)
{
    QNetworkRequest newRequest = d_func()->prepareMultipart(request, multiPart);
    QIODevice *device = multiPart->d_func()->device;
    QNetworkReply *reply = post(newRequest, device);
    return reply;
}
 
/*!
    \since 4.8
 
    \overload
 
    Sends the contents of the \a multiPart message to the destination
    specified by \a request.
 
    This can be used for sending MIME multipart messages over HTTP.
 
    \sa QHttpMultiPart, QHttpPart, post()
*/
QNetworkReply *QNetworkAccessManager::put(const QNetworkRequest &request, QHttpMultiPart *multiPart)
{
    QNetworkRequest newRequest = d_func()->prepareMultipart(request, multiPart);
    QIODevice *device = multiPart->d_func()->device;
    QNetworkReply *reply = put(newRequest, device);
    return reply;
}
#endif // QT_CONFIG(http)
 
/*!
    Uploads the contents of \a data to the destination \a request and
    returns a new QNetworkReply object that will be open for reply.
 
    \a data must be opened for reading when this function is called
    and must remain valid until the finished() signal is emitted for
    this reply.
 
    Whether anything will be available for reading from the returned
    object is protocol dependent. For HTTP, the server may send a
    small HTML page indicating the upload was successful (or not).
    Other protocols will probably have content in their replies.
 
    \note For HTTP, this request will send a PUT request, which most servers
    do not allow. Form upload mechanisms, including that of uploading
    files through HTML forms, use the POST mechanism.
 
    \sa get(), post(), deleteResource(), sendCustomRequest()
*/
QNetworkReply *QNetworkAccessManager::put(const QNetworkRequest &request, QIODevice *data)
{
    return d_func()->postProcess(createRequest(QNetworkAccessManager::PutOperation, request, data));
}
 
/*!
    \overload
 
    Sends the contents of the \a data byte array to the destination
    specified by \a request.
*/
QNetworkReply *QNetworkAccessManager::put(const QNetworkRequest &request, const QByteArray &data)
{
    QBuffer *buffer = new QBuffer;
    buffer->setData(data);
    buffer->open(QIODevice::ReadOnly);
 
    QNetworkReply *reply = put(request, buffer);
    buffer->setParent(reply);
    return reply;
}
 
/*!
    \since 6.8
 
    \overload
 
    \fn QNetworkReply *QNetworkAccessManager::put(const QNetworkRequest &request, std::nullptr_t nptr)
 
    Sends the PUT request specified by \a request without a body and returns
    a new QNetworkReply object.
*/
 
/*!
    \since 4.6
 
    Sends a request to delete the resource identified by the URL of \a request.
 
    \note This feature is currently available for HTTP only, performing an
    HTTP DELETE request.
 
    \sa get(), post(), put(), sendCustomRequest()
*/
QNetworkReply *QNetworkAccessManager::deleteResource(const QNetworkRequest &request)
{
    return d_func()->postProcess(createRequest(QNetworkAccessManager::DeleteOperation, request));
}
 
#ifndef QT_NO_SSL
/*!
    \since 5.2
 
    Initiates a connection to the host given by \a hostName at port \a port, using
    \a sslConfiguration. This function is useful to complete the TCP and SSL handshake
    to a host before the HTTPS request is made, resulting in a lower network latency.
 
    \note Preconnecting a HTTP/2 connection can be done by calling setAllowedNextProtocols()
    on \a sslConfiguration with QSslConfiguration::ALPNProtocolHTTP2 contained in
    the list of allowed protocols. When using HTTP/2, one single connection per host is
    enough, i.e. calling this method multiple times per host will not result in faster
    network transactions.
 
    \note This function has no possibility to report errors.
 
    \sa connectToHost(), get(), post(), put(), deleteResource()
*/
 
void QNetworkAccessManager::connectToHostEncrypted(const QString &hostName, quint16 port,
                                                   const QSslConfiguration &sslConfiguration)
{
    connectToHostEncrypted(hostName, port, sslConfiguration, QString());
}
 
/*!
    \since 5.13
    \overload
 
    Initiates a connection to the host given by \a hostName at port \a port, using
    \a sslConfiguration with \a peerName set to be the hostName used for certificate
    validation. This function is useful to complete the TCP and SSL handshake
    to a host before the HTTPS request is made, resulting in a lower network latency.
 
    \note Preconnecting a HTTP/2 connection can be done by calling setAllowedNextProtocols()
    on \a sslConfiguration with QSslConfiguration::ALPNProtocolHTTP2 contained in
    the list of allowed protocols. When using HTTP/2, one single connection per host is
    enough, i.e. calling this method multiple times per host will not result in faster
    network transactions.
 
    \note This function has no possibility to report errors.
 
    \sa connectToHost(), get(), post(), put(), deleteResource()
*/
 
void QNetworkAccessManager::connectToHostEncrypted(const QString &hostName, quint16 port,
                                                   const QSslConfiguration &sslConfiguration,
                                                   const QString &peerName)
{
    QUrl url;
    url.setHost(hostName);
    url.setPort(port);
    url.setScheme("preconnect-https"_L1);
    QNetworkRequest request(url);
    if (sslConfiguration != QSslConfiguration::defaultConfiguration())
        request.setSslConfiguration(sslConfiguration);
 
    // There is no way to enable HTTP2 via a request after having established the connection,
    // so we need to check the ssl configuration whether HTTP2 is allowed here.
    if (!sslConfiguration.allowedNextProtocols().contains(QSslConfiguration::ALPNProtocolHTTP2))
        request.setAttribute(QNetworkRequest::Http2AllowedAttribute, false);
 
    request.setPeerVerifyName(peerName);
    get(request);
}
#endif
 
/*!
    \since 5.2
 
    Initiates a connection to the host given by \a hostName at port \a port.
    This function is useful to complete the TCP handshake
    to a host before the HTTP request is made, resulting in a lower network latency.
 
    \note This function has no possibility to report errors.
 
    \sa connectToHostEncrypted(), get(), post(), put(), deleteResource()
*/
void QNetworkAccessManager::connectToHost(const QString &hostName, quint16 port)
{
    QUrl url;
    url.setHost(hostName);
    url.setPort(port);
    url.setScheme("preconnect-http"_L1);
    QNetworkRequest request(url);
    get(request);
}
 
/*!
    \since 5.9
 
    Sets the manager's redirect policy to be the \a policy specified. This policy
    will affect all subsequent requests created by the manager.
 
    Use this function to enable or disable HTTP redirects on the manager's level.
 
    \note When creating a request QNetworkRequest::RedirectAttributePolicy has
    the highest priority, next by priority the manager's policy.
 
    The default value is QNetworkRequest::NoLessSafeRedirectPolicy.
    Clients relying on manual redirect handling are encouraged to set
    this policy explicitly in their code.
 
    \sa redirectPolicy(), QNetworkRequest::RedirectPolicy
*/
void QNetworkAccessManager::setRedirectPolicy(QNetworkRequest::RedirectPolicy policy)
{
    Q_D(QNetworkAccessManager);
    d->redirectPolicy = policy;
}
 
/*!
    \since 5.9
 
    Returns the redirect policy that is used when creating new requests.
 
    \sa setRedirectPolicy(), QNetworkRequest::RedirectPolicy
*/
QNetworkRequest::RedirectPolicy QNetworkAccessManager::redirectPolicy() const
{
    Q_D(const QNetworkAccessManager);
    return d->redirectPolicy;
}
 
/*!
    \since 4.7
 
    Sends a custom request to the server identified by the URL of \a request.
 
    It is the user's responsibility to send a \a verb to the server that is valid
    according to the HTTP specification.
 
    This method provides means to send verbs other than the common ones provided
    via get() or post() etc., for instance sending an HTTP OPTIONS command.
 
    If \a data is not empty, the contents of the \a data
    device will be uploaded to the server; in that case, data must be open for
    reading and must remain valid until the finished() signal is emitted for this reply.
 
    \note This feature is currently available for HTTP(S) only.
 
    \sa get(), post(), put(), deleteResource()
*/
QNetworkReply *QNetworkAccessManager::sendCustomRequest(const QNetworkRequest &request, const QByteArray &verb, QIODevice *data)
{
    QNetworkRequest newRequest(request);
    newRequest.setAttribute(QNetworkRequest::CustomVerbAttribute, verb);
    return d_func()->postProcess(createRequest(QNetworkAccessManager::CustomOperation, newRequest, data));
}
 
/*!
    \fn QNetworkReply *QNetworkAccessManager::sendCustomRequest(const QNetworkRequest &request, const QByteArray &verb, std::nullptr_t)
    \since 6.13
    \overload
 
    Sends the custom request without body to the server specified by the URL of
    \a request and returns a new QNetworkReply object.
*/
 
/*!
    \since 5.8
 
    \overload
 
    Sends the contents of the \a data byte array to the destination
    specified by \a request.
*/
QNetworkReply *QNetworkAccessManager::sendCustomRequest(const QNetworkRequest &request, const QByteArray &verb, const QByteArray &data)
{
    QBuffer *buffer = new QBuffer;
    buffer->setData(data);
    buffer->open(QIODevice::ReadOnly);
 
    QNetworkReply *reply = sendCustomRequest(request, verb, buffer);
    buffer->setParent(reply);
    return reply;
}
 
#if QT_CONFIG(http) || defined(Q_OS_WASM)
/*!
    \since 5.8
 
    \overload
 
    Sends a custom request to the server identified by the URL of \a request.
 
    Sends the contents of the \a multiPart message to the destination
    specified by \a request.
 
    This can be used for sending MIME multipart messages for custom verbs.
 
    \sa QHttpMultiPart, QHttpPart, put()
*/
QNetworkReply *QNetworkAccessManager::sendCustomRequest(const QNetworkRequest &request, const QByteArray &verb, QHttpMultiPart *multiPart)
{
    QNetworkRequest newRequest = d_func()->prepareMultipart(request, multiPart);
    QIODevice *device = multiPart->d_func()->device;
    QNetworkReply *reply = sendCustomRequest(newRequest, verb, device);
    return reply;
}
#endif // QT_CONFIG(http)
 
/*!
    Returns a new QNetworkReply object to handle the operation \a op
    and request \a originalReq. The device \a outgoingData is always 0
    for Get and Head requests, but is the value passed to post() and
    put() in those operations (the QByteArray variants will pass a QBuffer
    object).
 
    The default implementation calls QNetworkCookieJar::cookiesForUrl()
    on the cookie jar set with setCookieJar() to obtain the cookies to
    be sent to the remote server.
 
    The returned object must be in an open state.
*/
QNetworkReply *QNetworkAccessManager::createRequest(QNetworkAccessManager::Operation op,
                                                    const QNetworkRequest &originalReq,
                                                    QIODevice *outgoingData)
{
    Q_D(QNetworkAccessManager);
 
    QNetworkRequest req(originalReq);
    if (redirectPolicy() != QNetworkRequest::NoLessSafeRedirectPolicy
        && req.attribute(QNetworkRequest::RedirectPolicyAttribute).isNull()) {
        req.setAttribute(QNetworkRequest::RedirectPolicyAttribute, redirectPolicy());
    }
 
#if QT_CONFIG(http) || defined (Q_OS_WASM)
    if (req.transferTimeoutAsDuration() == 0ms)
        req.setTransferTimeout(transferTimeoutAsDuration());
#endif
 
    if (autoDeleteReplies()
        && req.attribute(QNetworkRequest::AutoDeleteReplyOnFinishAttribute).isNull()) {
        req.setAttribute(QNetworkRequest::AutoDeleteReplyOnFinishAttribute, true);
    }
 
    bool isLocalFile = req.url().isLocalFile();
    QString scheme = req.url().scheme();
 
    // Remap local+http to unix+http to make further processing easier
    if (scheme == "local+http"_L1) {
        scheme = u"unix+http"_s;
        QUrl url = req.url();
        url.setScheme(scheme);
        req.setUrl(url);
    }
 
    // fast path for GET on file:// URLs
    // The QNetworkAccessFileBackend will right now only be used for PUT
    if (op == QNetworkAccessManager::GetOperation
     || op == QNetworkAccessManager::HeadOperation) {
        if (isLocalFile
#ifdef Q_OS_ANDROID
            || scheme == "assets"_L1
#endif
            || scheme == "qrc"_L1) {
            return new QNetworkReplyFileImpl(this, req, op);
        }
 
        if (scheme == "data"_L1)
            return new QNetworkReplyDataImpl(this, req, op);
 
        // A request with QNetworkRequest::AlwaysCache does not need any bearer management
        QNetworkRequest::CacheLoadControl mode =
            static_cast<QNetworkRequest::CacheLoadControl>(
                req.attribute(QNetworkRequest::CacheLoadControlAttribute,
                              QNetworkRequest::PreferNetwork).toInt());
        if (mode == QNetworkRequest::AlwaysCache) {
            // FIXME Implement a QNetworkReplyCacheImpl instead, see QTBUG-15106
            QNetworkReplyImpl *reply = new QNetworkReplyImpl(this);
            QNetworkReplyImplPrivate *priv = reply->d_func();
            priv->manager = this;
            priv->backend = new QNetworkAccessCacheBackend();
            priv->backend->setManagerPrivate(this->d_func());
            priv->backend->setParent(reply);
            priv->backend->setReplyPrivate(priv);
            priv->setup(op, req, outgoingData);
            return reply;
        }
    }
    QNetworkRequest request = req;
    auto h = request.headers();
#ifndef Q_OS_WASM // Content-length header is not allowed to be set by user in wasm
    if (!h.contains(QHttpHeaders::WellKnownHeader::ContentLength) &&
        outgoingData && !outgoingData->isSequential() && outgoingData->size()) {
        // request has no Content-Length
        // but the data that is outgoing is random-access
        h.append(QHttpHeaders::WellKnownHeader::ContentLength,
                 QByteArray::number(outgoingData->size()));
    }
#endif
    if (static_cast<QNetworkRequest::LoadControl>
        (request.attribute(QNetworkRequest::CookieLoadControlAttribute,
                           QNetworkRequest::Automatic).toInt()) == QNetworkRequest::Automatic) {
        if (d->cookieJar) {
            QList<QNetworkCookie> cookies = d->cookieJar->cookiesForUrl(request.url());
            if (!cookies.isEmpty())
                h.replaceOrAppend(QHttpHeaders::WellKnownHeader::Cookie,
                                  QNetworkHeadersPrivate::fromCookieList(cookies));
        }
    }
    request.setHeaders(std::move(h));
#ifdef Q_OS_WASM
    Q_UNUSED(isLocalFile);
    // Support http, https, and relative urls
    if (scheme == "http"_L1 || scheme == "https"_L1 || scheme.isEmpty()) {
        QNetworkReplyWasmImpl *reply = new QNetworkReplyWasmImpl(this);
        QNetworkReplyWasmImplPrivate *priv = reply->d_func();
        priv->manager = this;
        priv->setup(op, request, outgoingData);
        return reply;
    }
#endif
 
#if QT_CONFIG(http)
    constexpr char16_t httpSchemes[][17] = {
        u"http",
        u"preconnect-http",
#ifndef QT_NO_SSL
        u"https",
        u"preconnect-https",
#endif
        u"unix+http",
    };
    // Since Qt 5 we use the new QNetworkReplyHttpImpl
    if (std::find(std::begin(httpSchemes), std::end(httpSchemes), scheme) != std::end(httpSchemes)) {
 
#ifndef QT_NO_SSL
        const bool isLocalSocket = scheme.startsWith("unix"_L1);
        if (!isLocalSocket && isStrictTransportSecurityEnabled()
            && d->stsCache.isKnownHost(request.url())) {
            QUrl stsUrl(request.url());
            // RFC6797, 8.3:
            // The UA MUST replace the URI scheme with "https" [RFC2818],
            // and if the URI contains an explicit port component of "80",
            // then the UA MUST convert the port component to be "443", or
            // if the URI contains an explicit port component that is not
            // equal to "80", the port component value MUST be preserved;
            // otherwise,
            // if the URI does not contain an explicit port component, the UA
            // MUST NOT add one.
            if (stsUrl.port() == 80)
                stsUrl.setPort(443);
            stsUrl.setScheme("https"_L1);
            request.setUrl(stsUrl);
        }
#endif
        QNetworkReplyHttpImpl *reply = new QNetworkReplyHttpImpl(this, request, op, outgoingData);
        return reply;
    }
#endif // QT_CONFIG(http)
 
    // first step: create the reply
    QNetworkReplyImpl *reply = new QNetworkReplyImpl(this);
    QNetworkReplyImplPrivate *priv = reply->d_func();
    priv->manager = this;
 
    // second step: fetch cached credentials
    // This is not done for the time being, we should use signal emissions to request
    // the credentials from cache.
 
    // third step: find a backend
    priv->backend = d->findBackend(op, request);
 
    if (priv->backend) {
        priv->backend->setParent(reply);
        priv->backend->setReplyPrivate(priv);
    }
 
#ifndef QT_NO_SSL
    reply->setSslConfiguration(request.sslConfiguration());
#endif
 
    // fourth step: setup the reply
    priv->setup(op, request, outgoingData);
 
    return reply;
}
 
/*!
    \since 5.2
 
    Lists all the URL schemes supported by the access manager.
 
    Reimplement this method to provide your own supported schemes
    in a QNetworkAccessManager subclass. It is for instance necessary
    when your subclass provides support for new protocols.
*/
QStringList QNetworkAccessManager::supportedSchemes() const
{
    QStringList schemes;
    QNetworkAccessManager *self = const_cast<QNetworkAccessManager *>(this); // We know we call a const slot
    QMetaObject::invokeMethod(self, "supportedSchemesImplementation", Qt::DirectConnection,
                              Q_RETURN_ARG(QStringList, schemes));
    schemes.removeDuplicates();
    return schemes;
}
 
/*!
    \since 5.2
    \deprecated
 
    Lists all the URL schemes supported by the access manager.
 
    You should not call this function directly; use
    QNetworkAccessManager::supportedSchemes() instead.
 
    Because of binary compatibility constraints, the supportedSchemes()
    method (introduced in Qt 5.2) was not virtual in Qt 5, but now it
    is. Override the supportedSchemes method rather than this one.
 
    \sa supportedSchemes()
*/
QStringList QNetworkAccessManager::supportedSchemesImplementation() const
{
    Q_D(const QNetworkAccessManager);
 
    QStringList schemes = d->backendSupportedSchemes();
    // Those ones don't exist in backends
#if QT_CONFIG(http)
    schemes << QStringLiteral("http");
    schemes << QStringLiteral("unix+http");
    schemes << QStringLiteral("local+http");
#ifndef QT_NO_SSL
    if (QSslSocket::supportsSsl())
        schemes << QStringLiteral("https");
#endif
#endif
    schemes << QStringLiteral("data");
    return schemes;
}
 
/*!
    \since 5.0
 
    Flushes the internal cache of authentication data and network connections.
 
    This function is useful for doing auto tests.
 
    \sa clearConnectionCache()
*/
void QNetworkAccessManager::clearAccessCache()
{
    QNetworkAccessManagerPrivate::clearAuthenticationCache(this);
    QNetworkAccessManagerPrivate::clearConnectionCache(this);
}
 
/*!
    \since 5.9
 
    Flushes the internal cache of network connections.
    In contrast to clearAccessCache() the authentication data
    is preserved.
 
    \sa clearAccessCache()
*/
void QNetworkAccessManager::clearConnectionCache()
{
    QNetworkAccessManagerPrivate::clearConnectionCache(this);
}
 
 
/*!
    \since 5.14
 
    Returns the true if QNetworkAccessManager is currently configured
    to automatically delete QNetworkReplies, false otherwise.
 
    \sa setAutoDeleteReplies,
    QNetworkRequest::AutoDeleteReplyOnFinishAttribute
*/
bool QNetworkAccessManager::autoDeleteReplies() const
{
    return d_func()->autoDeleteReplies;
}
 
/*!
    \since 5.14
 
    Enables or disables automatic deletion of \l {QNetworkReply} {QNetworkReplies}.
 
    Setting \a shouldAutoDelete to true is the same as setting the
    QNetworkRequest::AutoDeleteReplyOnFinishAttribute attribute to
    true on all \e{future} \l {QNetworkRequest} {QNetworkRequests}
    passed to this instance of QNetworkAccessManager unless the
    attribute was already explicitly set on the QNetworkRequest.
 
    \sa autoDeleteReplies,
    QNetworkRequest::AutoDeleteReplyOnFinishAttribute
*/
void QNetworkAccessManager::setAutoDeleteReplies(bool shouldAutoDelete)
{
    d_func()->autoDeleteReplies = shouldAutoDelete;
}
 
/*!
    \fn int QNetworkAccessManager::transferTimeout() const
    \since 5.15
 
    Returns the timeout used for transfers, in milliseconds.
 
    \sa setTransferTimeout()
*/
 
/*!
    \fn void QNetworkAccessManager::setTransferTimeout(int timeout)
    \since 5.15
 
    Sets \a timeout as the transfer timeout in milliseconds.
 
    \sa setTransferTimeout(std::chrono::milliseconds),
        transferTimeout(), transferTimeoutAsDuration()
*/
 
/*!
    \since 6.7
 
    Returns the timeout duration after which the transfer is aborted if no
    data is exchanged.
 
    The default duration is zero, which means that the timeout is not used.
 
    \sa setTransferTimeout(std::chrono::milliseconds)
 */
std::chrono::milliseconds QNetworkAccessManager::transferTimeoutAsDuration() const
{
    return d_func()->transferTimeout;
}
 
/*!
    \since 6.7
 
    Sets the timeout \a duration to abort the transfer if no data is exchanged.
 
    Transfers are aborted if no bytes are transferred before
    the timeout expires. Zero means no timer is set. If no
    argument is provided, the timeout is
    QNetworkRequest::DefaultTransferTimeout. If this function
    is not called, the timeout is disabled and has the
    value zero. The request-specific non-zero timeouts set for
    the requests that are executed override this value. This means
    that if QNetworkAccessManager has an enabled timeout, it needs
    to be disabled to execute a request without a timeout.
 
    \sa transferTimeoutAsDuration()
 */
void QNetworkAccessManager::setTransferTimeout(std::chrono::milliseconds duration)
{
    d_func()->transferTimeout = duration;
}
 
void QNetworkAccessManagerPrivate::_q_replyFinished(QNetworkReply *reply)
{
    Q_Q(QNetworkAccessManager);
 
    emit q->finished(reply);
    if (reply->request().attribute(QNetworkRequest::AutoDeleteReplyOnFinishAttribute, false).toBool())
        QMetaObject::invokeMethod(reply, [reply] { reply->deleteLater(); }, Qt::QueuedConnection);
}
 
void QNetworkAccessManagerPrivate::_q_replyEncrypted(QNetworkReply *reply)
{
#ifndef QT_NO_SSL
    Q_Q(QNetworkAccessManager);
    emit q->encrypted(reply);
#else
    Q_UNUSED(reply);
#endif
}
 
void QNetworkAccessManagerPrivate::_q_replySslErrors(const QList<QSslError> &errors)
{
#ifndef QT_NO_SSL
    Q_Q(QNetworkAccessManager);
    QNetworkReply *reply = qobject_cast<QNetworkReply *>(q->sender());
    if (reply)
        emit q->sslErrors(reply, errors);
#else
    Q_UNUSED(errors);
#endif
}
 
#ifndef QT_NO_SSL
void QNetworkAccessManagerPrivate::_q_replyPreSharedKeyAuthenticationRequired(QSslPreSharedKeyAuthenticator *authenticator)
{
    Q_Q(QNetworkAccessManager);
    QNetworkReply *reply = qobject_cast<QNetworkReply *>(q->sender());
    if (reply)
    emit q->preSharedKeyAuthenticationRequired(reply, authenticator);
}
#endif
 
QNetworkReply *QNetworkAccessManagerPrivate::postProcess(QNetworkReply *reply)
{
    Q_Q(QNetworkAccessManager);
    QNetworkReplyPrivate::setManager(reply, q);
    q->connect(reply, &QNetworkReply::finished, reply,
               [this, reply]() { _q_replyFinished(reply); });
#ifndef QT_NO_SSL
    /* In case we're compiled without SSL support, we don't have this signal and we need to
     * avoid getting a connection error. */
    q->connect(reply, &QNetworkReply::encrypted, reply,
               [this, reply]() { _q_replyEncrypted(reply); });
    q->connect(reply, SIGNAL(sslErrors(QList<QSslError>)), SLOT(_q_replySslErrors(QList<QSslError>)));
    q->connect(reply, SIGNAL(preSharedKeyAuthenticationRequired(QSslPreSharedKeyAuthenticator*)), SLOT(_q_replyPreSharedKeyAuthenticationRequired(QSslPreSharedKeyAuthenticator*)));
#endif
 
    return reply;
}
 
void QNetworkAccessManagerPrivate::createCookieJar() const
{
    if (!cookieJarCreated) {
        // keep the ugly hack in here
        QNetworkAccessManagerPrivate *that = const_cast<QNetworkAccessManagerPrivate *>(this);
        that->cookieJar = new QNetworkCookieJar(that->q_func());
        that->cookieJarCreated = true;
    }
}
 
void QNetworkAccessManagerPrivate::authenticationRequired(QAuthenticator *authenticator,
                                                          QNetworkReply *reply,
                                                          bool synchronous,
                                                          QUrl &url,
                                                          QUrl *urlForLastAuthentication,
                                                          bool allowAuthenticationReuse)
{
    Q_Q(QNetworkAccessManager);
 
    // don't try the cache for the same URL twice in a row
    // being called twice for the same URL means the authentication failed
    // also called when last URL is empty, e.g. on first call
    if (allowAuthenticationReuse && (urlForLastAuthentication->isEmpty()
            || url != *urlForLastAuthentication)) {
        // if credentials are included in the url, then use them, unless they were already used
        if (!url.userName().isEmpty() && !url.password().isEmpty()
            && (url.userName() != authenticator->user()
                || url.password() != authenticator->password())) {
            authenticator->setUser(url.userName(QUrl::FullyDecoded));
            authenticator->setPassword(url.password(QUrl::FullyDecoded));
            *urlForLastAuthentication = url;
            authenticationManager->cacheCredentials(url, authenticator);
            return;
        }
 
        QNetworkAuthenticationCredential cred = authenticationManager->fetchCachedCredentials(url, authenticator);
        if (!cred.isNull()
            && (cred.user != authenticator->user() || cred.password != authenticator->password())) {
            authenticator->setUser(cred.user);
            authenticator->setPassword(cred.password);
            *urlForLastAuthentication = url;
            return;
        }
    }
 
    // if we emit a signal here in synchronous mode, the user might spin
    // an event loop, which might recurse and lead to problems
    if (synchronous)
        return;
 
    *urlForLastAuthentication = url;
    emit q->authenticationRequired(reply, authenticator);
    if (allowAuthenticationReuse)
        authenticationManager->cacheCredentials(url, authenticator);
}
 
#ifndef QT_NO_NETWORKPROXY
void QNetworkAccessManagerPrivate::proxyAuthenticationRequired(const QUrl &url,
                                                               const QNetworkProxy &proxy,
                                                               bool synchronous,
                                                               QAuthenticator *authenticator,
                                                               QNetworkProxy *lastProxyAuthentication)
{
    Q_Q(QNetworkAccessManager);
    QAuthenticatorPrivate *priv = QAuthenticatorPrivate::getPrivate(*authenticator);
    if (proxy != *lastProxyAuthentication && (!priv || !priv->hasFailed)) {
        QNetworkAuthenticationCredential cred = authenticationManager->fetchCachedProxyCredentials(proxy);
        if (!cred.isNull()) {
            authenticator->setUser(cred.user);
            authenticator->setPassword(cred.password);
            return;
        }
    }
 
#if defined(Q_OS_MACOS)
    //now we try to get the username and password from keychain
    //if not successful signal will be emitted
    QString username;
    QString password;
    if (getProxyAuth(proxy.hostName(), url.scheme(), username, password)) {
        // only cache the system credentials if they are correct (or if they have changed)
        // to not run into an endless loop in case they are wrong
        QNetworkAuthenticationCredential cred = authenticationManager->fetchCachedProxyCredentials(proxy);
        if (!priv->hasFailed || cred.user != username || cred.password != password) {
            authenticator->setUser(username);
            authenticator->setPassword(password);
            authenticationManager->cacheProxyCredentials(proxy, authenticator);
            return;
        }
    }
#else
    Q_UNUSED(url);
#endif
 
    // if we emit a signal here in synchronous mode, the user might spin
    // an event loop, which might recurse and lead to problems
    if (synchronous)
        return;
 
    *lastProxyAuthentication = proxy;
    emit q->proxyAuthenticationRequired(proxy, authenticator);
    authenticationManager->cacheProxyCredentials(proxy, authenticator);
}
 
QList<QNetworkProxy> QNetworkAccessManagerPrivate::queryProxy(const QNetworkProxyQuery &query)
{
    QList<QNetworkProxy> proxies;
    if (proxyFactory) {
        proxies = proxyFactory->queryProxy(query);
        if (proxies.isEmpty()) {
            qWarning("QNetworkAccessManager: factory %p has returned an empty result set",
                     proxyFactory);
            proxies << QNetworkProxy::NoProxy;
        }
    } else if (proxy.type() == QNetworkProxy::DefaultProxy) {
        // no proxy set, query the application
        return QNetworkProxyFactory::proxyForQuery(query);
    } else {
        proxies << proxy;
    }
 
    return proxies;
}
#endif
 
void QNetworkAccessManagerPrivate::clearAuthenticationCache(QNetworkAccessManager *manager)
{
    manager->d_func()->authenticationManager->clearCache();
}
 
void QNetworkAccessManagerPrivate::clearConnectionCache(QNetworkAccessManager *manager)
{
    manager->d_func()->destroyThread();
}
 
QNetworkAccessManagerPrivate::~QNetworkAccessManagerPrivate()
{
    destroyThread();
}
 
QThread * QNetworkAccessManagerPrivate::createThread()
{
    if (!thread) {
        thread = new QThread;
        thread->setObjectName(QStringLiteral("QNetworkAccessManager thread"));
        thread->start();
    }
    Q_ASSERT(thread);
    return thread;
}
 
void QNetworkAccessManagerPrivate::destroyThread()
{
    if (thread) {
        thread->quit();
        thread->wait(QDeadlineTimer(5000));
        if (thread->isFinished())
            delete thread;
        else
            QObject::connect(thread, SIGNAL(finished()), thread, SLOT(deleteLater()));
        thread = nullptr;
    }
}
 
 
#if QT_CONFIG(http) || defined(Q_OS_WASM)
 
QNetworkRequest QNetworkAccessManagerPrivate::prepareMultipart(const QNetworkRequest &request, QHttpMultiPart *multiPart)
{
    // copy the request, we probably need to add some headers
    QNetworkRequest newRequest(request);
    auto h = newRequest.headers();
 
    // add Content-Type header if not there already
    if (!h.contains(QHttpHeaders::WellKnownHeader::ContentType)) {
        QByteArray contentType;
        contentType.reserve(34 + multiPart->d_func()->boundary.size());
        contentType += "multipart/";
        switch (multiPart->d_func()->contentType) {
        case QHttpMultiPart::RelatedType:
            contentType += "related";
            break;
        case QHttpMultiPart::FormDataType:
            contentType += "form-data";
            break;
        case QHttpMultiPart::AlternativeType:
            contentType += "alternative";
            break;
        default:
            contentType += "mixed";
            break;
        }
        // putting the boundary into quotes, recommended in RFC 2046 section 5.1.1
        contentType += "; boundary=\"" + multiPart->d_func()->boundary + '"';
        h.append(QHttpHeaders::WellKnownHeader::ContentType, contentType);
    }
 
    // add MIME-Version header if not there already (we must include the header
    // if the message conforms to RFC 2045, see section 4 of that RFC)
    if (!h.contains(QHttpHeaders::WellKnownHeader::MIMEVersion))
        h.append(QHttpHeaders::WellKnownHeader::MIMEVersion, "1.0"_ba);
 
    newRequest.setHeaders(std::move(h));
 
    QIODevice *device = multiPart->d_func()->device;
    if (!device->isReadable()) {
        if (!device->isOpen()) {
            if (!device->open(QIODevice::ReadOnly))
                qWarning("could not open device for reading");
        } else {
            qWarning("device is not readable");
        }
    }
 
    return newRequest;
}
#endif // QT_CONFIG(http)
 
/*!
    \internal
    Go through the instances so the factories will be created and
    register themselves to QNetworkAccessBackendFactoryData
*/
void QNetworkAccessManagerPrivate::ensureBackendPluginsLoaded()
{
    Q_CONSTINIT static QBasicMutex mutex;
    std::unique_lock locker(mutex);
    if (!qnabfLoader())
        return;
#if QT_CONFIG(library)
    qnabfLoader->update();
#endif
    const int pluginCount = qnabfLoader->metaData().size();
    for (int i = 0; i < pluginCount; ++i)
        qnabfLoader->instance(i);
}
 
QT_END_NAMESPACE
 
#include "moc_qnetworkaccessmanager.cpp"