Age | Commit message (Collapse) | Author |
|
|
|
|
|
* Set SameSite=lax for session cookie
* Update warning text when Matomo is installed on HTTP
* urlencode all session cookie values
|
|
login. (#14736)
|
|
I updated the link tag for all php files with the new matomo link.
Not sure if it is realistic that this will get merged but I though that someday someone has to do this(?)
|
|
* Add INI config option and add tests.
* Detect expired sessions.
* Update config docs.
* Apply review feedback including storing expiration in session fingerprint.
* fixing tests.
* fix unit tests
* fix test
|
|
* Send all session cookie params when updating session expire time.
* Use login_cookie_path INI config to init the session cookie path.
|
|
* use db sessions by default, deprecate file session handler
* trying to fix tests
|
|
|
|
(#13554)
* Update session cookie on each request if session was remembered.
* Fix unit test.
|
|
* Remove user-agent checking code in SessionAuth.
* Fixing test.
* Fixing couple more tests.
|
|
|
|
(#13279)
* Make sure Auth interface is always set even if session auth succeeds.
* Add failing test.
* Fix FrontControllerTest
* Put hash token authentication back since it is still in use in plugins.
|
|
|
|
* Modifying "cookie authentication" to be more secure.
Instead of authenticating by token auth if it exists in the cookie, validate an existing session. If the session
has the user name stored as a session var, it has been authenticated. If the request has the same IP address and
user agent as the request that created the session, the request is from the user that created the session. If
both of these are true, then the session is valid, and we don't need a token auth to authenticate.
If the session is deleted before the Piwik auth cookie expires (due to garbage collection), we attempt to
re-authenticate using a secure hash of the token auth. We don't do this on every request since password_verify()
will, at BEST, add 3.5ms to every request.
* Invalidate existing sessions after user password change.
Invalidation is accomplished w/o having to individually touch sessions by:
1. Using the password hash as the piwik_auth key secret, instead of the token auth. So when a password changes, existing piwik_auth keys are no longer valid. This affects session re-authentication.
2. Saving the session start time & the last time a user's password was modified, and checking that the session start time is always newer than the password modification time.
* Set session.gc_maxlifetime to login_cookie_expire time so session data does not disappear, remove session re-auth functionality & tie cookie hash to password modified time instead of password hash to retain automatic session invalidation on password change.
* In SessionInitializer, clear other cookie values so previously stored token auths will be removed.
* Make sure anonymous user is still default user whan authenticating.
* fixing test failures
* Remove hash checking in piwik_auth cookie.
piwik_auth cookie still required since it's presence indicates we should use SessionAuth instead of the normal authentication mechanism. Since there's always a session, even if you're not logged in, PIWIK_SESSID can't be used by itself to determine this.
* Make sure session auth doesnt break in edge case where ts_password_modified column does not exist.
* Clarify session destruction/invalidation logic in SessionAuth.
* Make UsersManagerTest slightly more comprehensive.
* Use Date::now()->getTimestampUTC() instead of time() in SessionFingerprint::initialize().
* Check getUser returns correct user info in SessionAuth for sanity.
* Add SessionInitializer::getAuthCookie() back since it is @api.
* Remove IP address from session auth info + check.
* Refactor session start changes so it is started in one place only.
* Remove SessionAuthCookieFactory & deprecate auth cookie INI config vars (still needed for SessionInitializer deprectaed method).
* Make sure user can still login if ts_password_modified column is not present in database.
* Rename ts_password_modified Update class.
* Update comment in SessionAuth to include why Piwik tries to create another session.
* Restore 3.x-dev SessionInitializer for BC (deprecated), move new SessionInitializer to core, add tests for both SessionInitializers.
* Change update to 3.5 version.
* Make sure normal auth implementation is used if sessionauth fails so anonymous user can be logged in.
* On logout clear session fingerprint so same session cannot be used to login.
* Change update name + bump version, and make sure Session::rememberMe() is called before session is started (otherwise it has no effect).
* Fixing tests.
* apply review fixes
* remove test
|
|
|
|
Sometimes, eg when module=API, we do not start a session. If any code
part during such a request uses SessionNamespace although we did not start
a session, Zend_Session_Namespace will start a Zend_Session which does not
respect our configuration. It would use default Zend Session options
and therefore probably file based sessions. Therefore, we force the
creation of a "Piwik session". If we have created the session before it
won't be started again.
|
|
code, few refactorings, all as part of our code cleanup strategy
|
|
|
|
to use free/libre instead of open source
|
|
storing the connection upon creation.
|
|
refs #4609
|
|
|
|
|
|
|
|
|
|
|
|
renaming Unzip to Uncompress,
Renaming Tracker/Referer to Tracker/Referrer,
and few more
|
|
Removed many Piwik_ functions, in Piwik 2 it is best practise to use the methods calls instead
Todo: finish converting core/ classes + convert plugins/ classes to use \Piwik\Plugin namespace + fix build + Merge master
|
|
|
|
Notice that auto refactoring has created a nested namespace. Not sure this is what we want - so we might have to edit those nested namespaces afterwards (I think they don't look so good)
|
|
Notice that auto refactoring has created a nested namespace. Not sure this is what we want - so we might have to edit those nested namespaces afterwards (I think they don't look so good)
|
|
Converting Piwik core source files, PHP, JS, TPL, CSS
More info: http://piwik.org/participate/coding-standards/
|
|
|
|
git-svn-id: http://dev.piwik.org/svn/trunk@6325 59fd770c-687e-43c8-a1e3-f5a4ff64c105
|
|
The new savehandler uses the DB adapters that have already been well-tested in Piwik, instead of using Zend_DB_Table (as Zend_Db_Adapter_Mysqli isn't as well-supported as the PDO adapters in ZF).
Note: there's a MySQL-ism which we can address later in #1368.
git-svn-id: http://dev.piwik.org/svn/trunk@4970 59fd770c-687e-43c8-a1e3-f5a4ff64c105
|
|
cleaner implementation
git-svn-id: http://dev.piwik.org/svn/trunk@4969 59fd770c-687e-43c8-a1e3-f5a4ff64c105
|
|
git-svn-id: http://dev.piwik.org/svn/trunk@4297 59fd770c-687e-43c8-a1e3-f5a4ff64c105
|
|
calls, so the wrappers around Zend_Session and Zend_Session_Namespace dummy initialize when executing from php-cli or headless php-cgi; this also addresses the hack by Parallels in their APS scripts
git-svn-id: http://dev.piwik.org/svn/trunk@3468 59fd770c-687e-43c8-a1e3-f5a4ff64c105
|