Commit Graph

184 Commits

Author SHA1 Message Date
Richard van der Hoff
1189be43a2 Factor _AccountHandler proxy out to ModuleApi
We're going to need to use this from places that aren't password auth, so let's
move it to a proper class.
2017-11-02 14:36:11 +00:00
David Baker
b19d9e2174
Merge pull request #2624 from matrix-org/rav/password_provider_notify_logout
Notify auth providers on logout
2017-11-02 10:55:17 +00:00
David Baker
1f080a6c97
Merge pull request #2623 from matrix-org/rav/callbacks_for_auth_providers
Allow password_auth_providers to return a callback
2017-11-02 10:49:03 +00:00
David Baker
04897c9dc1
Merge pull request #2622 from matrix-org/rav/db_access_for_auth_providers
Let auth providers get to the database
2017-11-02 10:41:25 +00:00
Richard van der Hoff
979eed4362 Fix user-interactive password auth
this got broken in the previous commit
2017-11-01 17:03:20 +00:00
Richard van der Hoff
bc8a5c0330 Notify auth providers on logout
Provide a hook by which auth providers can be notified of logouts.
2017-11-01 16:51:51 +00:00
Richard van der Hoff
4c8f94ac94 Allow password_auth_providers to return a callback
... so that they have a way to record access tokens.
2017-11-01 16:51:03 +00:00
Richard van der Hoff
846a94fbc9
Merge pull request #2620 from matrix-org/rav/auth_non_password
Let password auth providers handle arbitrary login types
2017-11-01 16:45:33 +00:00
Richard van der Hoff
3cd6b22c7b Let password auth providers handle arbitrary login types
Provide a hook where password auth providers can say they know about other
login types, and get passed the relevant parameters
2017-11-01 16:43:57 +00:00
David Baker
4f0488b307 Merge remote-tracking branch 'origin/develop' into rav/refactor_accesstoken_delete 2017-11-01 16:20:19 +00:00
Richard van der Hoff
dd13310fb8 Move access token deletion into auth handler
Also move duplicated deactivation code into the auth handler.

I want to add some hooks when we deactivate an access token, so let's bring it
all in here so that there's somewhere to put it.
2017-11-01 15:46:22 +00:00
Richard van der Hoff
74c56f794c Break dependency of auth_handler on device_handler
I'm going to need to make the device_handler depend on the auth_handler, so I
need to break this dependency to avoid a cycle.

It turns out that the auth_handler was only using the device_handler in one
place which was an edge case which we can more elegantly handle by throwing an
error rather than fixing it up.
2017-11-01 10:27:06 +00:00
Richard van der Hoff
3e0aaad190 Let auth providers get to the database
Somewhat open to abuse, but also somewhat unavoidable :/
2017-10-31 17:22:29 +00:00
Richard van der Hoff
1b65ae00ac Refactor some logic from LoginRestServlet into AuthHandler
I'm going to need some more flexibility in handling login types in password
auth providers, so as a first step, move some stuff from LoginRestServlet into
AuthHandler.

In particular, we pass everything other than SAML, JWT and token logins down to
the AuthHandler, which now has responsibility for checking the login type and
fishing the password out of the login dictionary, as well as qualifying the
user_id if need be. Ideally SAML, JWT and token would go that way too, but
there's no real need for it right now and I'm trying to minimise impact.

This commit *should* be non-functional.
2017-10-31 10:48:41 +00:00
Richard van der Hoff
785bd7fd75 Allow ASes to deactivate their own users 2017-10-27 00:01:00 +01:00
Richard van der Hoff
631d7b87b5 Remove pointless create() method
It just calls the constructor, so we may as well kill it rather than having
random codepaths.
2017-10-20 22:14:55 +01:00
Erik Johnston
c72058bcc6 Use an ExpiringCache for storing registration sessions
This is because pruning them was a significant performance drain on
matrix.org
2017-06-29 14:08:37 +01:00
David Baker
73a5f06652 Support registration / login with phone number
Changes from https://github.com/matrix-org/synapse/pull/1971
2017-03-13 17:27:51 +00:00
Erik Johnston
7eae6eaa2f Revert "Support registration & login with phone number" 2017-03-13 09:59:33 +00:00
David Baker
0e0aee25c4 Fix log line 2017-03-08 11:46:22 +00:00
David Baker
88df6c0c9a Factor out msisdn canonicalisation
Plus a couple of other minor fixes
2017-03-08 11:03:39 +00:00
David Baker
402a7bf63d Fix pep8 2017-03-08 09:33:40 +00:00
David Baker
ad882cd54d Just return the deferred straight off
defer.returnValue doth not maketh a generator: it would need a
yield to be a generator, and this doesn't need a yield.
2017-03-01 18:08:51 +00:00
David Baker
ce3e583d94 WIP support for msisdn 3pid proxy methods 2017-02-14 15:05:55 +00:00
Erik Johnston
51adaac953 Fix email push in pusher worker
This was broken when device list updates were implemented, as Mailer
could no longer instantiate an AuthHandler due to a dependency on
federation sending.
2017-02-02 10:53:36 +00:00
David Baker
84cf00c645 Fix another comment typo 2016-12-21 09:51:43 +00:00
David Baker
0c88ab1844 Add /account/3pid/delete endpoint
Also fix a typo in a comment
2016-12-20 18:27:30 +00:00
Matthew Hodgson
f2a5aebf98 fix ability to change password to a non-ascii one
https://github.com/vector-im/riot-web/issues/2658
2016-12-18 22:25:21 +00:00
Erik Johnston
338df4f409 Merge pull request #1649 from matrix-org/dbkr/log_ui_auth_args
Log the args that we have on UI auth completion
2016-12-05 16:40:58 +00:00
Richard van der Hoff
aa09d6b8f0 Rip out more refresh_token code
We might as well treat all refresh_tokens as invalid. Just return a 403 from
/tokenrefresh, so that we don't have a load of dead, untestable code hanging
around.

Still TODO: removing the table from the schema.
2016-11-30 17:40:18 +00:00
Richard van der Hoff
dc4b23e1a1 Merge branch 'develop' into rav/no_more_refresh_tokens 2016-11-30 17:10:04 +00:00
Richard van der Hoff
1c4f05db41 Stop putting a time caveat on access tokens
The 'time' caveat on the access tokens was something of a lie, since we weren't
enforcing it; more pertinently its presence stops us ever adding useful time
caveats.

Let's move in the right direction by not lying in our caveats.
2016-11-29 16:49:41 +00:00
Richard van der Hoff
5c4edc83b5 Stop generating refresh tokens
Since we're not doing refresh tokens any more, we should start killing off the
dead code paths. /tokenrefresh itself is a bit of a thornier subject, since
there might be apps out there using it, but we can at least not generate
refresh tokens on new logins.
2016-11-28 10:13:01 +00:00
David Baker
c9d4e7b716 Clarify that creds doesn not contain passwords. 2016-11-24 10:54:59 +00:00
David Baker
f681aab895 Log the args that we have on UI auth completion
This will be super helpful for debugging if we have more
registration woes.
2016-11-24 10:11:45 +00:00
Erik Johnston
d56c39cf24 Use external ldap auth pacakge 2016-11-15 13:03:19 +00:00
David Baker
9084720993 Don't error on non-ascii passwords 2016-11-03 10:42:14 +00:00
David Baker
df2a616c7b Convert emails to lowercase when storing
And db migration sql to convert existing addresses.
2016-10-19 11:13:55 +01:00
Erik Johnston
35e2cc8b52 Merge pull request #1155 from matrix-org/erikj/pluggable_pwd_auth
Implement pluggable password auth
2016-10-12 11:41:20 +01:00
Richard van der Hoff
fa74fcf512 Work around email-spamming Riot bug
5d9546f9 introduced a change to synapse behaviour, in that failures in the
interactive-auth process would return the flows and params data as well as an
error code (as specced in https://github.com/matrix-org/matrix-doc/pull/397).

That change exposed a bug in Riot which would make it request a new validation
token (and send a new email) each time it got a 401 with a `flows` parameter
(see https://github.com/vector-im/vector-web/issues/2447 and the fix at
https://github.com/matrix-org/matrix-react-sdk/pull/510).

To preserve compatibility with broken versions of Riot, grandfather in the old
behaviour for the email validation stage.
2016-10-11 11:34:40 +01:00
Richard van der Hoff
8681aff4f1 Merge pull request #1160 from matrix-org/rav/401_on_password_fail
Interactive Auth: Return 401 from for incorrect password
2016-10-07 10:57:43 +01:00
Richard van der Hoff
5d9546f9f4 Interactive Auth: Return 401 from for incorrect password
This requires a bit of fettling, because I want to return a helpful error
message too but we don't want to distinguish between unknown user and invalid
password. To avoid hardcoding the error message into 15 places in the code,
I've had to refactor a few methods to return None instead of throwing.

Fixes https://matrix.org/jira/browse/SYN-744
2016-10-07 00:00:00 +01:00
Erik Johnston
850b103b36 Implement pluggable password auth
Allows delegating the password auth to an external module. This also
moves the LDAP auth to using this system, allowing it to be removed from
the synapse tree entirely in the future.
2016-10-03 10:36:40 +01:00
Martin Weinelt
3027ea22b0 Restructure ldap authentication
- properly parse return values of ldap bind() calls
- externalize authentication methods
- change control flow to be more error-resilient
- unbind ldap connections in many places
- improve log messages and loglevels
2016-09-29 15:30:08 +01:00
Erik Johnston
dc3a00f24f Refactor user_delete_access_tokens. Invalidate get_user_by_access_token to slaves. 2016-08-15 17:04:39 +01:00
Daniel Ehlers
dfaf0fee31
Log the value which is observed in the first place.
The name 'result' is of bool type and has no len property,
resulting in a TypeError. Futhermore in the flow control
conn.response is observed and hence should be reported.

Signed-off-by: Daniel Ehlers <sargon@toppoint.de>
2016-08-14 16:49:05 +02:00
Daniel Ehlers
e380538b59
Fix AttributeError when bind_dn is not defined.
In case one does not define bind_dn in ldap configuration, filter
attribute is not declared. Since auth code only uses ldap_filter attribute
when according LDAP mode is selected, it is safe to only declare the
attribute in that case.

Signed-off-by: Daniel Ehlers <sargon@toppoint.de>
2016-08-14 16:48:33 +02:00
Richard van der Hoff
79ebfbe7c6 /login: Respond with a 403 when we get an invalid m.login.token 2016-08-09 16:29:28 +01:00
Richard van der Hoff
6fe6a6f029 Fix login with m.login.token
login with token (as used by CAS auth) was broken by 067596d, such that it
always returned a 401.
2016-08-08 16:40:39 +01:00
Richard van der Hoff
436bffd15f Implement deleting devices 2016-07-26 07:35:48 +01:00
David Baker
7ed58bb347 Use get to avoid KeyErrors 2016-07-22 17:18:50 +01:00
David Baker
dad2da7e54 Log the hostname the reCAPTCHA was completed on
This could be useful information to have in the logs. Also comment about how & why we don't verify the hostname.
2016-07-22 17:00:56 +01:00
Richard van der Hoff
3413f1e284 Type annotations
Add some type annotations to help PyCharm (in particular) to figure out the
types of a bunch of things.
2016-07-19 18:56:16 +01:00
Richard van der Hoff
f863a52cea Add device_id support to /login
Add a 'devices' table to the storage, as well as a 'device_id' column to
refresh_tokens.

Allow the client to pass a device_id, and initial_device_display_name, to
/login. If login is successful, then register the device in the devices table
if it wasn't known already. If no device_id was supplied, make one up.

Associate the device_id with the access token and refresh token, so that we can
get at it again later. Ensure that the device_id is copied from the refresh
token to the access_token when the token is refreshed.
2016-07-18 16:39:44 +01:00
Richard van der Hoff
dcfd71aa4c Refactor login flow
Make sure that we have the canonical user_id *before* calling
get_login_tuple_for_user_id.

Replace login_with_password with a method which just validates the password,
and have the caller call get_login_tuple_for_user_id. This brings the password
flow into line with the other flows, and will give us a place to register the
device_id if necessary.
2016-07-18 15:23:54 +01:00
Negar Fazeli
0136a522b1 Bug fix: expire invalid access tokens 2016-07-13 15:00:37 +02:00
Kent Shikama
14362bf359
Fix password config 2016-07-05 19:12:53 +09:00
Kent Shikama
1ee2584307
Fix pep8 2016-07-05 19:01:00 +09:00
Kent Shikama
8bdaf5f7af
Add pepper to password hashing
Signed-off-by: Kent Shikama <kent@kentshikama.com>
2016-07-05 02:13:52 +09:00
Martin Weinelt
0a32208e5d Rework ldap integration with ldap3
Use the pure-python ldap3 library, which eliminates the need for a
system dependency.

Offer both a `search` and `simple_bind` mode, for more sophisticated
ldap scenarios.
- `search` tries to find a matching DN within the `user_base` while
  employing the `user_filter`, then tries the bind when a single
  matching DN was found.
- `simple_bind` tries the bind against a specific DN by combining the
  localpart and `user_base`

Offer support for STARTTLS on a plain connection.

The configuration was changed to reflect these new possibilities.

Signed-off-by: Martin Weinelt <hexa@darmstadt.ccc.de>
2016-06-22 17:51:59 +02:00
Salvatore LaMendola
ed5f43a55a
Fix TypeError in call to bcrypt.hashpw
- At the very least, this TypeError caused logins to fail on my own
  running instance of Synapse, and the simple (explicit) UTF-8
  conversion resolved login errors for me.

Signed-off-by: Salvatore LaMendola <salvatore.lamendola@gmail.com>
2016-06-16 00:43:42 -04:00
David Baker
a15ad60849 Email unsubscribing that may in theory, work
Were it not for that fact that you can't use the base handler in the pusher because it pulls in the world. Comitting while I fix that on a different branch.
2016-06-02 11:44:15 +01:00
Erik Johnston
cc84f7cb8e Send down correct error response if user not found 2016-05-27 10:35:15 +01:00
Erik Johnston
99b5a2e560 Merge pull request #741 from negzi/create_user_with_expiry
Create user with expiry
2016-05-13 14:46:53 +01:00
Negi Fazeli
40aa6e8349 Create user with expiry
- Add unittests for client, api and handler

Signed-off-by: Negar Fazeli <negar.fazeli@ericsson.com>
2016-05-13 15:34:15 +02:00
Erik Johnston
1400bb1663 Correctly handle NULL password hashes from the database 2016-05-11 12:06:02 +01:00
Erik Johnston
6fd2f685fe Simplify _check_password 2016-04-15 11:17:18 +01:00
Mark Haines
3c79bdd7a0 Fix check_password rather than inverting the meaning of _check_local_password (#730) 2016-04-14 19:00:21 +01:00
David Baker
4c1b32d7e2 Fix login to error for nonexistent users
Fixes SYN-680
2016-04-14 18:28:42 +01:00
Christoph Witzany
ed4d18f516 fix check for failed authentication 2016-04-06 18:30:11 +02:00
Christoph Witzany
9c62fcdb68 remove line 2016-04-06 18:23:46 +02:00
Christoph Witzany
27a0c21c38 make tests for ldap more specific to not be fooled by Mocks 2016-04-06 18:23:46 +02:00
Christoph Witzany
3555a659ec output ldap version for info and to pacify pep8 2016-04-06 18:23:46 +02:00
Christoph Witzany
4c5e8adf8b conditionally import ldap 2016-04-06 18:23:46 +02:00
Christoph Witzany
875ed05bdc fix pep8 2016-04-06 18:23:46 +02:00
Christoph Witzany
67f3a50e9a fix exception handling 2016-04-06 18:23:46 +02:00
Christoph Witzany
afff321e9a code style 2016-04-06 18:23:46 +02:00
Christoph Witzany
823b8be4b7 add tls property and twist my head around twisted 2016-04-06 18:23:45 +02:00
Christoph Witzany
7b9319b1c8 move LDAP authentication to AuthenticationHandler 2016-04-06 18:23:45 +02:00
Mark Haines
2a37467fa1 Use google style doc strings.
pycharm supports them so there is no need to use the other format.

Might as well convert the existing strings to reduce the risk of
people accidentally cargo culting the wrong doc string format.
2016-04-01 16:12:07 +01:00
David Baker
a7daa5ae13 Make registration idempotent, part 2: be idempotent if the client specifies a username. 2016-03-16 19:36:57 +00:00
David Baker
b58d10a875 pep8 2016-03-16 16:22:20 +00:00
David Baker
3ee7d7dc7f time_msec() 2016-03-16 16:18:52 +00:00
David Baker
3176aebf9d string with symbols is a bit too symboly. 2016-03-16 15:55:49 +00:00
David Baker
9671e6750c Replace other time.time(). 2016-03-16 15:51:28 +00:00
David Baker
742b6c6d15 Use hs get_clock instead of time.time() 2016-03-16 15:42:35 +00:00
David Baker
99797947aa pep8 & remove debug logging 2016-03-16 12:51:34 +00:00
David Baker
c12b9d719a Make registration idempotent: if you specify the same session, make it give you an access token for the user that was registered on previous uses of that session. Tweak the UI auth layer to not delete sessions when their auth has completed and hence expire themn so they don't hang around until server restart. Allow server-side data to be associated with UI auth sessions. 2016-03-16 11:56:24 +00:00
David Baker
af59826a2f Make select more sensible when dseleting access tokens, rename pusher deletion to match access token deletion and make exception arg optional. 2016-03-11 14:34:09 +00:00
David Baker
f523177850 Delete old, unused methods and rename new one to just be user_delete_access_tokens with an except_token_ids argument doing what it says on the tin. 2016-03-11 14:29:01 +00:00
David Baker
57c444b3ad Dear PyCharm, please indent sensibly for me. Thx. 2016-03-11 14:25:05 +00:00
David Baker
aa11db5f11 Fix cache invalidation so deleting access tokens (which we did when changing password) actually takes effect without HS restart. Reinstate the code to avoid logging out the session that changed the password, removed in 415c2f0549 2016-03-11 13:14:18 +00:00
David Baker
ff8b87118d Stop using checkpw as it seems to have vanished from bcrypt. Use bcrypt.hashpw(password, hashed) == hashed as per the bcrypt README. 2016-03-02 18:06:45 +00:00
Daniel Wagner-Hall
cfd07aafff Allow guests to upgrade their accounts 2016-01-05 18:01:18 +00:00
Daniel Wagner-Hall
248cfd5eb3 Take a boolean not a list of lambdas 2015-11-19 15:16:25 +00:00
Steven Hammerton
2b779af10f Minor review fixes 2015-11-11 11:21:43 +00:00
Steven Hammerton
dd2eb49385 Share more code between macaroon validation 2015-11-11 11:12:35 +00:00
Steven Hammerton
414a4a71b4 Allow hs to do CAS login completely and issue the client with a login token that can be redeemed for the usual successful login response 2015-11-05 14:06:48 +00:00
Daniel Wagner-Hall
f522f50a08 Allow guests to register and call /events?room_id=
This follows the same flows-based flow as regular registration, but as
the only implemented flow has no requirements, it auto-succeeds. In the
future, other flows (e.g. captcha) may be required, so clients should
treat this like the regular registration flow choices.
2015-11-04 17:29:07 +00:00
Mark Haines
f2f031fd57 Add config for how many bcrypt rounds to use for password hashes
By default we leave it at the default value of 12. But now we can reduce
it for preparing users for loadtests or running integration tests.
2015-10-16 14:52:08 +01:00