When using MariaDB v10.5+ Foreign-Key errors were popping up because of
some changes in that version. To mitigate this on MariaDB and other
MySQL forks those errors are now catched, and instead of a replace_into
an update will happen. I have tested this as thorough as possible with
MariaDB 10.5, 10.4, 10.3 and the default MySQL on Ubuntu Focal. And
tested it again using sqlite, all seems to be ok on all tables.
resolves#1081. resolves#1065, resolves#1050
Made some small changes to the description of the config options for
SMTP. Some were a bit cryptic and missing some extra descriptions.
Also made it more clear which type of secured smtp connection is going
to used.
Currently when for example using the admin interface to send out a test e-mail just
returns `SmtpError`. This is not very helpful. What i have done.
- Match some common Lettre errors to return the error message.
- Other errors will just be passed on as before.
Some small other changes:
- Fixed a clippy warning about using clone().
- Fixed a typo where Lettere was spelled with one t.
- Allow all SMTP Auth meganisms supported by Lettre.
- The config value order is leading and values can be separated by a
comma ','
- Case doesn't matter, and invalid values are ignored.
- Warning is printed when no valid value is found at all.
General:
- Updated several dependancies
Lettre:
- Updateded lettere and the workflow
- Changed encoding to base64
- Convert unix newlines to dos newlines for e-mails.
- Created custom e-mail boundary (auto generated could cause errors)
Tested the e-mails sent using several clients (Linux, Windows, MacOS, Web).
Run msglint (https://tools.ietf.org/tools/msglint/) on the generated e-mails until all errors were gone.
Lettre has changed quite some stuff compared between alpha.1 and alpha.2, i haven't noticed any issues sending e-mails during my tests.
If org owners/admins set their org access to only include selected
collections, then ciphers from non-selected collections shouldn't
appear in "My Vault". This matches the upstream behavior.
Diesel requires the following changes:
- Separate connection and pool types per connection, the generate_connections! macro generates an enum with a variant per db type
- Separate migrations and schemas, these were always imported as one type depending on db feature, now they are all imported under different module names
- Separate model objects per connection, the db_object! macro generates one object for each connection with the diesel macros, a generic object, and methods to convert between the connection-specific and the generic ones
- Separate connection queries, the db_run! macro allows writing only one that gets compiled for all databases or multiple ones
Currently, favorites are tracked at the cipher level. For org-owned ciphers,
this means that if one user sets it as a favorite, it automatically becomes a
favorite for all other users that the cipher has been shared with.
In this implementation, the `TZ` environment variable must be set
in order for the formatted output to use a more user-friendly
time zone abbreviation (e.g., `UTC`). Otherwise, the output uses
the time zone's UTC offset (e.g., `+00:00`).
In the event of a failed DNS Resolving checking for new versions will
cause a huge delay, and in the end a timeout when loading the page.
- Check if DNS resolving failed, if that is the case, do not check for
new versions
- Changed `fn get_github_api` to make use of structs
- Added a timeout of 10 seconds for the version check requests
- Moved the "Unknown" lables to the "Latest" lable
- Updated bootstrap js and css to the latest version
- Fixed issue with small-screens where the menu overlaps the token input
- The menu now collapses to a hamburger menu
- Menu's only accessable when logedin are hidden when you are not
- Changed Users Overview to use a table to prevent small-screen issues.
Main changes:
- Splitted up settings and users into two separate pages.
- Added verified shield when the e-mail address has been verified.
- Added the amount of personal items in the database to the users overview.
- Added Organizations and Diagnostics pages.
- Shows if DNS resolving works.
- Shows if there is a posible time drift.
- Shows current versions of server and web-vault.
- Optimized logo-gray.png using optipng
Items which can be added later:
- Amount of cipher items accessible for a user, not only his personal items.
- Amount of users per Org
- Version update check in the diagnostics overview.
- Copy/Pasteable runtime config which has sensitive data changed or removed for support questions either on the forum or github issues.
- Option to delete Orgs and all its passwords (when there are no members anymore).
- Etc....
PostgreSQL updates/inserts ignored None/null values.
This is nice for new entries, but not for updates.
Added derive option to allways add these none/null values for Option<>
variables.
This solves issue #965
* Make `SIGNUPS_DOMAINS_WHITELIST` override the `SIGNUPS_ALLOWED` setting.
Otherwise, a common pitfall is to set `SIGNUPS_DOMAINS_WHITELIST` without
realizing that `SIGNUPS_ALLOWED=false` must also be set.
* Whitespace is now accepted in `SIGNUPS_DOMAINS_WHITELIST`. That is,
`foo.com, bar.com` is now equivalent to `foo.com,bar.com`.
* Add validation on `SIGNUPS_DOMAINS_WHITELIST`. For example, `foo.com,`
is rejected as containing an empty token.
Ignore a missing `id` query param; it's unclear what this ID represents,
but it wasn't being used in the existing bitwarden_rs code, and no longer
seems to be sent in the latest versions of the official clients.
The organization uuid is most of the time within the uri path as a
parameter. But sometimes it only is there as a query value.
This fix checks both, and returns the uuid when possible.
During migrations some queries are out of order regarding to foreign
keys.
Because of this the migrations fail when the sql database has this
enforced by default.
Turning of this check during the migrations will fix this and this is
only per session.
If SSL is disabled, the SMTP ClientSecurity of the lettre crate
defaults to None, that is, an insecure connection. This is changed to
Opportunistic, which uses TLS if available. If TLS is not available,
the insecure connection is used (i.e., this change is backward
compatible).
- Moved smtp test option to within the "SMTP Email" Settings block.
- Added optional option to prevent full page reload.
- SMTP Test and Backup do not reload the admin interface any more.
- Added a test button for checking the e-mail settings.
- Fixed a bug with the _post JavaScript function:
A function was overwriten with a variable and errors were not handled
correctly like a 500 for example.
panic!()'s only appear on stderr, this makes tracking down some strange
issues harder with the usage of docker since stderr does not get logged
into the bitwarden.log file. This change logs the message to stdout and
the logfile when activated.
I've checked the spots when `Invitation::new()` and `Invitation::take()`
are used and it seems like all spots are already correctly gated. So to
enable invitations via admin API even when invitations are otherwise
disabled, this check can be removed.
This was brought up today:
https://github.com/dani-garcia/bitwarden_rs/issues/752#issuecomment-586715073
I don't think it makes much sense in checking whether admin has the
right to send invitation as admin can change the setting anyway.
Removing the condition allows users to forbid regular users from
inviting new users to server while still preserving the option to do so
via the admin API.
Because of differences in how .on_conflict() works compared to .replace_into() the PostgreSQL backend wasn't correctly ensuring the unique constraint on user_uuid and atype wasn't getting violated.
This change simply issues a DELETE on the unique constraint prior to the insert to ensure uniqueness. PostgreSQL does not support multiple constraints in ON CONFLICT clauses.
Use LOG_LEVEL debug or trace to recover them.
Removed LOG_MOUNTS and bundled it with LOG_LEVEL debug and trace.
Removed duplicate error messages
Made websocket not proxied message more prominent, but only print it once.
Some sites are using base64 encoded inline images for favicons.
This will try to match those with some sane checks and return that.
These icons will have lower prio then the icons with a normal URL.
This feature can be enabled by setting SIGNUPS_ALLOWED=false and
providing a comma-separated list of whitelisted domains in
SIGNUPS_DOMAINS_WHITELIST.
Fixes#727
Now creates icon cache directory at startup.
And it also creates the directory if it went missing during runtime.
Also modified the icon_save/mark_negcache to be one.
When the icon_cache directory doesn't exists yet, and the first icon
catched is a miss this .miss file was not able to be created since the
directory was only created during a valid icon download.
