summaryrefslogtreecommitdiffstats
path: root/src/plugins/messageservices/imap
Commit message (Collapse)AuthorAgeFilesLines
* Add a specific logging category for IMAP stateDamien Caliste2025-07-113-7/+9
| | | | | | | | | | | | | Previously to 8e1451a8, the change of state in IMAP protocol was log with MessagingState category which was disabled. Create here a specific new category to avoid having the state changes printed out with the normal IMAP logging by default. It's now possible to get the state logging by using QT_LOGGING_RULES environment variable without recompiling. Change-Id: I7bfee95bd8dd483b2716e60b2866e778861c61ef Reviewed-by: Pekka Vuorela <[email protected]>
* Don't report credentials error after a connection being closedDamien Caliste2025-07-041-11/+11
| | | | | | | | | If the connection already breaks while waiting for the credentials to be fetched, it should not report error or proceed to login. Change-Id: I68a6f95232bdde03cc6cd52a2e0bf3cdfc989772 Reviewed-by: Pekka Vuorela <[email protected]>
* Rewrite logging with QLoggingCategoryDamien Caliste2025-06-1314-306/+394
| | | | | | | | | Replace the custom logging mechanism with QLoggingCategory. Also declare the warnings with the proper category. Change-Id: If87d53d27ea0c65abd434af9f99fe49ce634d6dd Reviewed-by: Pekka Vuorela <[email protected]>
* Add missing spaces after for/if/while/switchPekka Vuorela2025-05-097-101/+101
| | | | | | | Change-Id: I57c6ed78099b3359a16ce807da95325e755f0197 Reviewed-by: Pekka Vuorela <[email protected]> Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: <[email protected]>
* Remove trailing whitespace with sed scriptPekka Vuorela2025-05-0911-299/+299
| | | | | | | Change-Id: Iebfffe7a6d5b2db40f0508ecd7c8ccddfa6665a6 Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Declare tests in the build systemDamien Caliste2025-04-021-0/+6
| | | | | | | | | | | | | | | | | | Tests can be run with `ctest`. The `--verbose` option makes the output of each test visible. Fix also minor failures: - make qlogsystem test locale independent, - make qmailstore test properly listening to the messagePropertyUpdated() signal. IMAP, POP and SMTP tests are declared as always failing for the moment, since they try to connect to a mail server (but they are still useful for debug purposes). Change-Id: I7ae548110a8eeb37bd07b031a8433d38ea53529e Reviewed-by: Pekka Vuorela <[email protected]>
* Continue login when there is no valid push foldersDamien Caliste2025-03-262-3/+6
| | | | | | | | | | | | | | | | | Currently, login is deferred until all configured push folders have been established (or failed), when push email is enabled. This does not work for a newly created account where there is no folder registered. This patch ensures that the code continues to the login when push email is enabled but when there is no valid configured folders. Change-Id: I397d4d2616c3c6cebc9a3f4dcc03f4b5e3564ebe Reviewed-by: Guido Berhoerster <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Fix crash on imap search without supported keysPekka Vuorela2025-03-031-0/+6
| | | | | | | | | | | | | Just passing QMailMessageKey() ended up with empty search query and then crashing on QStringList::takeFirst(). Added similar error handling as CreateState::transmit() and RenameState::transmit() have. Change-Id: I6fdaa8ac11507dd880676d1f020c53d4964c85aa Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Don't continue on log-in after idle renewalDamien Caliste2025-02-281-3/+5
| | | | | | | | | Fix a regression introduced in 679b6a2d. The setIdlingForFolder() is also called on idle renewal, when the main session is in logout state. There is no need to ask for a login then. Change-Id: Iec6649a529a2d2b5b9b9d817f0ea2df5ff03f383 Reviewed-by: Pekka Vuorela <[email protected]>
* Make ImapService the only place to delay push mail restartDamien Caliste2025-02-283-65/+32
| | | | | | | | | | | | | | | | | | | | | | | The IdleProtocol was delaying its signal emission of openRequest in case of an error. Then the client emitted restartPushEmail and the service was immediately calling initiatePushEmail(). This is the case when the error happened in the idle protocol. But the service is already having a delayed machinery to restart the idle session in case an error happens in the main client session. This patch is removing the delay timer in IdleProtocol so any error happening there is immediately reported to the service, which decides to delay the call to initiatePushEmail(). This requires some safe guards to be added in the service to ensure that the delay is applied only once even when there are several push folders emitting errors. Change-Id: I2b49c0445640d58bb391088b602f152cfe45eb49 Reviewed-by: Pekka Vuorela <[email protected]>
* Synchronize idle renewal timersDamien Caliste2025-02-282-10/+20
| | | | | | | | | | | Use a common QTimer in ImapClient for every IdleProtocol, so the network accesses to renew the idling connections are all done in a close- by period of time, allowing the radio to sleep for a longer time between wakes-up. Change-Id: I7c9378a0af74015a34fd6cc98fa9517433d28392 Reviewed-by: Pekka Vuorela <[email protected]>
* Remove initial imap push delayDamien Caliste2025-02-262-24/+1
| | | | | | | | | | | | | | | | There is no need anymore to delay IMAP start on push enabled. If an error occurs and idling can't be established, there are already two exponential timers: - one in ImapService that restart idling for errors happening in the main client session. - one in IdleTransport that delays the emission of restartPushEmail signal on error in the idle connection. Change-Id: I6ccbd7fccc0d36f40d2dc2ca6ddf7b6f099f1638 Reviewed-by: Pekka Vuorela <[email protected]>
* Report client error even when establishing idle connectionsDamien Caliste2025-02-214-17/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | The errorOccurred signal is only emitted for client session, and not for idle connections (there are reporting via the client itself, which emits restartPushEmail signal for idle errors). It is thus necessary to report any error the client is emitting. If the client session was launched to establish idle connections and these sessions are not running, then one should retry to reestablish the idle connections later. Without the patch, in case of a non responding server (connection to a captive portal without network granted yet), the client error was not reported when the connection could not be established. It was accepted wrongly as succeeded when a new attempt to establish idle connections was made and Source::retrievalCompleted() was emitted when queueing the idle folder for scan. But with the retry timeout becoming larger and larger, the strategy expiry timeout finally kicked in and the service was fully restarted. Change-Id: I206e77f221d2a7c4bdfcee1952dbe1127431e6d6 Reviewed-by: Pekka Vuorela <[email protected]>
* Rework ImapService::onAccountsUpdated()Damien Caliste2025-02-211-26/+23
| | | | | | | | | | | | | | | | Avoid the case of calling disable() in the callback for any other case than when the account is disabled. In particular, don't call disable() + enabled() to apply a change. The only account change that requires to stop (and restart) a running client is a change in the idle folder list. In that case, there is no need to kill and restart the client, just restarting the idle function is enough. Change-Id: Ica39c1c09bb5b156670a0f1301bc6dc9703a7451 Reviewed-by: Pekka Vuorela <[email protected]>
* Reduce status variable number in ImapClient::monitor()Damien Caliste2025-02-212-63/+42
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Makes the idle state to rely on _monitored and _waitingForIdleFolderIds variables only. _monitored is the set of IdleProtocol listening to changes in a folder. _waitingForIdleFolderIds is a subset of _monitored, listing folder where the protocol was started, but has not reached yet the idling status. _idlesEstablished can be replaced by a combination of the two, being true when and only when some folders are monitored and all of them are in the idling state. Old variable _waitingForIdle is not needed anymore if monitor() function is modified to ensure that the list of folders given as argument is properly transformed into monitored folders, and only these folders as argument. Calling monitor on the same list is a no operation, making _waitingForIdle completely obsolete. A new closeIdleConnections() function is added to the public API of ImapClient, since there was previously no mean to close idle connections, except by killing the client itself. This is achieved internally by closing non desired idle protocols in monitor() before checking if the settings are push enabled. Change-Id: I8c62b6d5ff7150ee2c90d7dcfec3bd44f66cf7b6 Reviewed-by: Pekka Vuorela <[email protected]>
* Separate push email related code from enable/disableDamien Caliste2025-02-192-30/+38
| | | | | Change-Id: If39a2c6d106c88a0f3bc059db1d8f40957649193 Reviewed-by: Pekka Vuorela <[email protected]>
* Make configuration local in the IMAP clientDamien Caliste2025-02-196-124/+124
| | | | | | | | | | | | | | | | Avoid a complete restart of the client on setting changes. Reload of the configuration was already done at each new connection. But completion of the client task may take time and settings could be changed in-between. Let instead the client reports a failure if the settings are not correct. Doing a deletion of the client and a recreation of the client on setting changes won't allow a running task to complete anyway. Change-Id: Icf11e6f33a5574d8cd7ccd07fe2f3f06e2d5d558 Reviewed-by: Pekka Vuorela <[email protected]>
* Test account enabled on _client existing in ImapServiceDamien Caliste2025-02-192-7/+4
| | | | | | | | | | There is no need to maintain _accountWasEnabled variable, since a client is created when enabled and deleted when disabled. Testing _client can play its role. Change-Id: I37bbb1bd58368bad626994ce89544ba4dfbaa45c Reviewed-by: Pekka Vuorela <[email protected]>
* Add a couple null pointer checksPekka Vuorela2025-02-142-1/+3
| | | | | | | Change-Id: Id4cac8f496e9734f8f9b514e5d6c9234d0438508 Reviewed-by: <[email protected]> Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Remove semicolons after method bodiesPekka Vuorela2025-02-142-2/+2
| | | | | | | | | | Fixed also a copy-paste error on qmailthreadsortkey_p.h. Change-Id: Ide4bb24e69a94f64c9521f5c4cce7a33921fab11 Reviewed-by: <[email protected]> Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Switch to CamelCased Qt #includesPekka Vuorela2025-02-129-24/+33
| | | | | | | Change-Id: Ifa2dad56db9667fcbeaff4c618a8a8de75d693c4 Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]> Reviewed-by: <[email protected]>
* Move LongStream to server library, streamline the implementationPekka Vuorela2025-02-113-4/+4
| | | | | | | | | | | | | | | | | | The qmfclient was using this just to get storage info, which is somewhat out of scope for the class api anyway. And also the "errorMessage" meaning out-of-space error was bad API. Got rid of OS specific storage calculation in favor of QStorageInfo and streamlined the error string method. To me the implementation feels like trying a bit too much to check the available space, but I'll not touch that now. Change-Id: Ice4256c0cdf3fa3dc52c1354ab07a433862e66ae Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: <[email protected]> Reviewed-by: Pekka Vuorela <[email protected]>
* Setup IMAP account status in one placeDamien Caliste2025-02-032-17/+15
| | | | | | | | Centralize the status flags according to IMAP possibilities into ImapClient. Change-Id: I1019a9348f53ce98124baf297136f6e285c996dd Reviewed-by: Pekka Vuorela <[email protected]>
* Wait for credentials fetching in idle protocolDamien Caliste2025-02-032-24/+56
| | | | | | | | | | | | | | While an idle connection is setting up, the main client, with the credentials, may not yet have fetched and receive the credentials. Thus, authentication of within the idle protocol may fail. This patch delays sendLogin() for idle protocols waiting for the credentials to be ready. Change-Id: If14a4104f5943dcb32eda7fd969623a1ab362b0f Reviewed-by: Pekka Vuorela <[email protected]>
* Set accountId on construction for IETF clientsDamien Caliste2025-01-283-19/+11
| | | | | | | | | | A client is created by the corresponding service. A service is linked to an account on construction. There is no use case to change the account during the life time of a client. Change-Id: I43c60741d28b011bef9c7770433557cf54e870e6 Reviewed-by: Pekka Vuorela <[email protected]>
* Remove unused ImapProtocol::quoteString(QByteArray)Damien Caliste2025-01-172-6/+0
| | | | | | | | | This function is not used and can be a source of bugs because it assume its argument to be UTF8 encoded while it converts the output to latin1. Change-Id: Ib7209b15959741cc10ebebb2cbceefc42fb753f6 Reviewed-by: Pekka Vuorela <[email protected]>
* Add a plugin mechanism to handle credentialsDamien Caliste2025-01-156-60/+150
| | | | | | | | | | | Credentials for authentication can be username / passwords or something more complex. The storage for such data can also be done outside QMF. So this commit introduce a plugin mechanism to get such credential data. Change-Id: I26ec6e98b5e900a9f93c420b00f0253f2abad25c Reviewed-by: Pekka Vuorela <[email protected]>
* Use QMailServiceConfiguration in authenticatorDamien Caliste2024-10-314-22/+19
| | | | | | | | | | This API change allows to create the service configuration in each plugin with its object constructor instead of its name. Change-Id: I77b253a795d6b452fbeb650cf80d925e5c53e724 Reviewed-by: Pekka Vuorela <[email protected]>
* Add CMake build systemDamien Caliste2024-10-175-4/+53
| | | | | | | | | | | | | | | | Adjust source code where necessary like: - #include <private/...> cannot be used since private is the installation path, but not the path used in the sources, - fix map definition in qmailservice.xml, - had to add #include <qmailstore.h> to qmailserviceaction.h since qt_add_dbus_adaptor can only include one header, - add a method using QDBusVariant in ServiceHandler so the adaptor can call it. Change-Id: Idbecf4214dffdf523ccd8558370e8d2854b5d99a Reviewed-by: Pekka Vuorela <[email protected]>
* Remove Factory from QMailCryptographicServiceFactoryDamien Caliste2024-04-102-2/+2
| | | | | | | | | | | This class is not simply a factory creating instances of crypto plugins, but it's more like a service, providing methods on QMailMessagePart to execute cryptographic operations like signing or signature verification. Change-Id: I74f65f27137e3bfe78c0f9614728eca37a031b0c Reviewed-by: Pekka Vuorela <[email protected]>
* Remove traces of QAlignedTimerPekka Vuorela2024-04-012-21/+0
| | | | | | | | For what I can tell, this was something in Qt Mobility long time ago. Shouldn't be a thing anymore. Change-Id: Ice7bf7ffb9c561f0757c970eed7804508c1735f3 Reviewed-by: Matthew Vogt <[email protected]>
* Mark overridden methods with keywordPekka Vuorela2023-12-148-307/+321
| | | | | | | | | | | | | Avoiding a lot of compiler noise from -Wsuggest-override. Also removed the now even reduntant virtual keywords on overridden methods as per qt coding conventions. And wrapped some overly long lines touched by the changes. Change-Id: I4e7febfa9554dfa7f366a1aabba6133a9f6c5614 Reviewed-by: Matthew Vogt <[email protected]> Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: Christopher Adams <[email protected]>
* Introduce acceptUntrustedCertificates() configurationPekka Vuorela2023-11-305-21/+21
| | | | | | | | | Allows connections to be made to servers even when the TLS certficiate is not trusted (e.g. broken certificate chain). Change-Id: If912be7d7e4824f243471b40ef5c4d5acb0c31ab Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: Matthew Vogt <[email protected]>
* Do AUTHENTICATE PLAIN in two stagesValério Valério2023-11-211-10/+16
| | | | | | | | | | Some servers(e.g yandex.ru) do not support AUTHENTICATE PLAIN in a single command, according to RFC4616 this way is optional, so we always perform the authentication in two stages that is guarantee to be supported by all implementations of the protocol. Change-Id: I56ea5a4acaf446fdea14543416866442c27dc074 Reviewed-by: Pekka Vuorela <[email protected]>
* Discover authentication for SMTP/IMAPDamien Caliste2023-11-211-0/+24
| | | | | | | | | | Use the announced capabilities in SMTP/IMAP plugin to setup the authentication method to use. Add handling for error 504 on authentication in SMTP. Change-Id: I556281895e650768537fc6169816c3e8e6a81362 Reviewed-by: Pekka Vuorela <[email protected]>
* Set replied and forwarded flags on message update from serverDamien Caliste2023-11-212-0/+13
| | | | | | | | | | | | | The replied flag (\Answered in IMAP protocol) is properly set for a new mail. But it was not set on a qresync action (fetching mail updates). The forwarded flag ($Forwarded as an extension in IMAP protocol), was detected in the received flags but was not set or updated in created QMailMessages. Change-Id: I78eb16650dc706e08fee1c6aab68d601159a52e4 Reviewed-by: Pekka Vuorela <[email protected]>
* Add QMailMessage::CalendarCancellation flagDavid Llewellyn-Jones2023-11-152-1/+3
| | | | | | | | | | | | | | | | | | | | | | Introduces a CalendarCancellation flag similar to the QMailMessage::CalendarInvitation to be matched against the QMailMessage::status() bitfield. While the CalendarInvitation flag indicates that the message contains an attachment of type text/calendar with "REQUEST" method, the CalendarCancellation flag indicates that there's an attachment of type text/calendar with "CANCEL" method type. In other words, a calendar event cancellation. For reference, the "CANCEL" method of the Content Type header is described in RFC2447 (iMIP) Section 2.4 [1] in combination with the RFC2446 (iTIP) sections 3.2 and 3.2.5 [2] [1] https://tools.ietf.org/html/rfc2447#section-2.4 [2] https://tools.ietf.org/html/rfc2446#section-3.2 Change-Id: I1ef1d8a070e527552dae1ebe1677156bc07ba04d Reviewed-by: Matthew Vogt <[email protected]>
* Fix some compiler warningsPekka Vuorela2023-11-131-2/+2
| | | | | | | | | Avoid unused variable, QString conversion from ascii and move fall through note so GCC detects it. Change-Id: Id50a2b90c8c8c345a117c713849f65f45515eddb Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: Damien Caliste <[email protected]>
* Set the encryption status on IMAP receptionDamien Caliste2023-09-201-0/+3
| | | | | | | | | When the structure of the email is received in an IMAP transfer, set or not the encryption flag, as it is done for attachments. Change-Id: Ief42663d2c38c19b1cd29c21b35f67e4cfcb1fb2 Reviewed-by: Pekka Vuorela <[email protected]>
* Fix bug when detecting HEADER / BODY substringsChris Adams2021-08-301-2/+3
| | | | | | | | | | The behavior should be equivalent to old QRegExp::lastIndexIn(str). Without this fix, we never hit the "detach" which causes a variety of further issues when creating the QMailMessage. Change-Id: Ib9b91cd16e1d9cb6eeb1c255eea06ab387c2b9b4 Reviewed-by: Pekka Vuorela <[email protected]>
* Set PLUGIN_CLASS_NAME in plugin .pro filesChris Adams2021-04-091-0/+1
| | | | | | | | | Ensures that cmake files are generated appropriately. Change-Id: I2521114a527510ff7bf2e73092251147157a29d1 Reviewed-by: Damien Caliste <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]> Reviewed-by: Christopher Adams <[email protected]>
* Stop _incomingDataTimer when imapprotocol object is destroyedDavid Llewellyn-Jones2021-04-091-0/+1
| | | | | | | | | | A crash occurs in case the timer is active and the object is destroyed. Created by Valério Valério <[email protected]>, 23 Feb 2015. Change-Id: I95725448eff61a9e51eff0c5dd8fe0e33b971b08 Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: Matthew Vogt <[email protected]>
* Check if IDLE connection needs to be established after loginDavid Llewellyn-Jones2021-04-091-0/+10
| | | | | | | | | | | Some servers only advertise full capabilities after sucessful login, so we need to check if IDLE connection needs to be established in such cases. Created by Valério Valério <[email protected]>, 11 May 2015. Change-Id: I307f7be0e07b1639c80f01a3205bcff9de96525a Reviewed-by: Christopher Adams <[email protected]>
* Set new IMAP folders to inherit SynchronizationEnabled flagDavid Llewellyn-Jones2021-04-091-1/+4
| | | | | | | | | | | | | | | | | When new folders are received from the server, the client-side SynchronizationEnabled flag must be set for them. Previously they were always set to true, but if they're inside a folder that isn't being synced, this probably doesn't reflect the user's intent. This change makes it so that for IMAP any new folders received from the server will have the SynchronizationEnabled flag set to whatever value their parent has it set to. If they have no parent (i.e. are in the root folder) then the flag is set to true as before. Change-Id: If43d5e2b4469a402d203841c04cc4b1b30ffc3b1 Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: Matthew Vogt <[email protected]> Reviewed-by: Damien Caliste <[email protected]>
* Follow the synchronizationEnabled flag when folders are not specifiedDamien Caliste2021-04-093-11/+33
| | | | | | | | | | | | | | If the synchronizationEnabled flag is set to false, a variety of strategy, like search, folder listing, will not search in such folders. This is maybe a too restrictive usage of this flag. This patch change the behavior of this flag to act on synchronization strategies only, when desired. It activates it only for message retrieval in all folders. Change-Id: Iebec872984c7e2650167ae585b06bfa3d8cd6990 Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
* Fix some warningsChris Adams2021-04-081-3/+3
| | | | | | Change-Id: Ifad72f55ca3988e84749d9f8868ec6d8d6c986ec Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
* Use QmfList where required to maintain stable referencesChris Adams2021-04-084-25/+25
| | | | | | | | | | Any case where the address of an element is taken and stored or used, must be changed to use QmfList instead of QList, otherwise non-const operations will cause the reference to become invalid. Change-Id: Ic0017cb9eddcf27b93d26adb3ea9c71682c4421b Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
* Provide mock IdleNetworkSession class for IMAP pluginChris Adams2021-04-085-107/+298
| | | | | | | | | | | | | | | | | The IMAP plugin previously used QNetworkSession to detect changes in connectivity and attempt to maintain a network connection to receive IMAP IDLE messages from the server. Unfortunately, however, QNetworkSession and QNetworkConfiguration no longer exist in Qt6. This commit introduces a mock IdleNetworkSession class with the appropriate interface, to ensure that the project can build with Qt6 and unit tests pass. Clients who wish to use QMF on real devices should implement IdleNetworkSession using the appropriate platform APIs available on their target device. Change-Id: I740ebd28686f5647a327ac7a47916d15ca7eb0ac Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
* Port uses of QRegExp to QRegularExpression in IMAP pluginChris Adams2021-04-082-84/+97
| | | | | | Change-Id: Ife7188b294dc3d89e8087674825a8805c0121d7b Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
* Use QMultiMap when necessary instead of QMapDamien Caliste2021-04-082-2/+2
| | | | | | Change-Id: Iabf4dad9998e46163e4f2e2090419c662a817377 Reviewed-by: Christopher Adams <[email protected]> Reviewed-by: David Llewellyn-Jones <[email protected]>
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-21 09:18:50 +0000