Follows #19266, #8553, Close#18553, now there are only three `Run..(&RunOpts{})` functions.
* before: `stdout, err := RunInDir(path)`
* now: `stdout, _, err := RunStdString(&git.RunOpts{Dir:path})`
This make checks in one single place so they dont differ and maintainer can not forget a check in one place while adding it to the other .... ( as it's atm )
Fix:
* The API does ignore issue dependencies where Web does not
* The API checks if "IsSignedIfRequired" where Web does not - UI probably do but nothing will some to craft custom requests
* Default merge message is crafted a bit different between API and Web if not set on specific cases ...
Adding additional usernames which are already routes, remove unused ones.
In future, avoid reserving names as much as possible, use `/-/` in path instead.
There is yet another problem with conflicted files not being reset when
the test patch resolves them.
This PR adjusts the code for checkConflicts to reset the ConflictedFiles
field immediately at the top. It also adds a reset to conflictedFiles
for the manuallyMerged and a shortcut for the empty status in
protectedfiles.
Signed-off-by: Andrew Thornton <art27@cantab.net>
The last PR about clone buttons introduced an JS error when visiting an empty repo page:
* https://github.com/go-gitea/gitea/pull/19028
* `Uncaught ReferenceError: isSSH is not defined`, because the variables are scoped and doesn't share between sub templates.
This:
1. Simplify `templates/repo/clone_buttons.tmpl` and make code clear
2. Move most JS code into `initRepoCloneLink`
3. Remove unused `CloneLink.Git`
4. Remove `ctx.Data["DisableSSH"] / ctx.Data["ExposeAnonSSH"] / ctx.Data["DisableHTTP"]`, and only set them when is is needed (eg: deploy keys / ssh keys)
5. Introduce `Data["CloneButton*"]` to provide data for clone buttons and links
6. Introduce `Data["RepoCloneLink"]` for the repo clone link (not the wiki)
7. Remove most `ctx.Data["PageIsWiki"]` because it has been set in the `/wiki` middleware
8. Remove incorrect `quickstart` class in `migrating.tmpl`
There is a bug in the system webhooks whereby the active state is not checked when
webhooks are prepared and there is a bug that deactivating webhooks do not prevent
queued deliveries.
* Only add SystemWebhooks to the prepareWebhooks list if they are active
* At the time of delivery if the underlying webhook is not active mark it
as "delivered" but with a failed delivery so it does not get delivered.
Fix#19220
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Touch mirrors on even on fail to update
If a mirror fails to be synchronised it should be pushed to the bottom of the queue
of the awaiting mirrors to be synchronised. At present if there LIMIT number of
broken mirrors they can effectively prevent all other mirrors from being synchronized
as their last_updated time will remain earlier than other mirrors.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Set the default branch for repositories generated from templates
* Allows default branch to be set through the API for repos generated from templates
* Update swagger API template
* Only set default branch to the one from the template if not specified
* Use specified default branch if it exists while generating git commits
Fix#19082
Co-authored-by: John Olheiser <john.olheiser@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
- Make a restriction on which issues can be shown based on if you the user or team has write permission to the repository.
- Fixes a issue whereby you wouldn't see any associated issues with a specific team on a organization if you wasn't a member(fixed by zeroing the User{ID} in the options).
- Resolves#18913
* Clean up protected_branches when deleting user
fixes#19094
* Clean up protected_branches when deleting teams
* fix issue
Co-authored-by: Lauris BH <lauris@nix.lv>
Storing the foreign identifier of an imported issue in the database is a prerequisite to implement idempotent migrations or mirror for issues. It is a baby step towards mirroring that introduces a new table.
At the moment when an issue is created by the Gitea uploader, it fails if the issue already exists. The Gitea uploader could be modified so that, instead of failing, it looks up the database to find an existing issue. And if it does it would update the issue instead of creating a new one. However this is not currently possible because an information is missing from the database: the foreign identifier that uniquely represents the issue being migrated is not persisted. With this change, the foreign identifier is stored in the database and the Gitea uploader will then be able to run a query to figure out if a given issue being imported already exists.
The implementation of mirroring for issues, pull requests, releases, etc. can be done in three steps:
1. Store an identifier for the element being mirrored (issue, pull request...) in the database (this is the purpose of these changes)
2. Modify the Gitea uploader to be able to update an existing repository with all it contains (issues, pull request...) instead of failing if it exists
3. Optimize the Gitea uploader to speed up the updates, when possible.
The second step creates code that does not yet exist to enable idempotent migrations with the Gitea uploader. When a migration is done for the first time, the behavior is not changed. But when a migration is done for a repository that already exists, this new code is used to update it.
The third step can use the code created in the second step to optimize and speed up migrations. For instance, when a migration is resumed, an issue that has an update time that is not more recent can be skipped and only newly created issues or updated ones will be updated. Another example of optimization could be that a webhook notifies Gitea when an issue is updated. The code triggered by the webhook would download only this issue and call the code created in the second step to update the issue, as if it was in the process of an idempotent migration.
The ForeignReferences table is added to contain local and foreign ID pairs relative to a given repository. It can later be used for pull requests and other artifacts that can be mirrored. Although the foreign id could be added as a single field in issues or pull requests, it would need to be added to all tables that represent something that can be mirrored. Creating a new table makes for a simpler and more generic design. The drawback is that it requires an extra lookup to obtain the information. However, this extra information is only required during migration or mirroring and does not impact the way Gitea currently works.
The foreign identifier of an issue or pull request is similar to the identifier of an external user, which is stored in reactions, issues, etc. as OriginalPosterID and so on. The representation of a user is however different and the ability of users to link their account to an external user at a later time is also a logic that is different from what is involved in mirroring or migrations. For these reasons, despite some commonalities, it is unclear at this time how the two tables (foreign reference and external user) could be merged together.
The ForeignID field is extracted from the issue migration context so that it can be dumped in files with dump-repo and later restored via restore-repo.
The GetAllComments downloader method is introduced to simplify the implementation and not overload the Context for the purpose of pagination. It also clarifies in which context the comments are paginated and in which context they are not.
The Context interface is no longer useful for the purpose of retrieving the LocalID and ForeignID since they are now both available from the PullRequest and Issue struct. The Reviewable and Commentable interfaces replace and serve the same purpose.
The Context data member of PullRequest and Issue becomes a DownloaderContext to clarify that its purpose is not to support in memory operations while the current downloader is acting but is not otherwise persisted. It is, for instance, used by the GitLab downloader to store the IsMergeRequest boolean and sort out issues.
---
[source](https://lab.forgefriends.org/forgefriends/forgefriends/-/merge_requests/36)
Signed-off-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: Loïc Dachary <loic@dachary.org>
* Update the webauthn_credential_id_sequence in Postgres
There is (yet) another problem with v210 in that Postgres will silently allow preset
ID insertions ... but it will not update the sequence value.
This PR simply adds a little step to the end of the v210 migration to update the
sequence number.
Users who have already migrated who find that they cannot insert new
webauthn_credentials into the DB can either run:
```bash
gitea doctor recreate-table webauthn_credential
```
or
```bash
./gitea doctor --run=check-db-consistency --fix
```
which will fix the bad sequence.
Fix#19012
Signed-off-by: Andrew Thornton <art27@cantab.net>
Yet another issue has come up where the logging from SyncMirrors does not provide
enough context. This PR adds more context to these logging events.
Related #19038
Signed-off-by: Andrew Thornton <art27@cantab.net>
The review request feature was added in https://github.com/go-gitea/gitea/pull/10756,
where the doer got explicitly excluded from available reviewers. I don't see a
functionality or security related reason to forbid this case.
As shown by GitHubs implementation, it may be useful to self-request a review,
to be reminded oneselves about reviewing, while communicating to team mates that a
review is missing.
Co-authored-by: delvh <dev.lh@web.de>
* ignore missing comment for user notifications
* instead fix bug in notifications model
* use local variable instead
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: 6543 <6543@obermui.de>
Add new feature to delete issues and pulls via API
Co-authored-by: fnetx <git@fralix.ovh>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: Gusted <williamzijl7@hotmail.com>
Co-authored-by: 6543 <6543@obermui.de>
- Add helper method to reduce redundancy
- Expand the scope from displaying days to years
- Reduce irrelevance by not displaying small units (hours, minutes, seconds) when bigger ones apply (years)
* Avoid database lookups for `DescriptionHTML`
- Don't Compose meta's for DescriptionHTML, they are only needed in
order to correctly format and show issue's but it's highly unlikely that
a repository description will refer to a local issue.
Using 125 Connections for 5 seconds: on `/explore/repos`(which is the most
noticeable usage by this function's database lookups):
Before:
Statistics Avg Stdev Max
Reqs/sec 569.41 506.05 2715.00
Latency 214.27ms 16.60ms 294.84ms
HTTP codes:
1xx - 0, 2xx - 2974, 3xx - 0, 4xx - 0, 5xx - 0
others - 0
Throughput: 27.17MB/s
After:
Statistics Avg Stdev Max
Reqs/sec 1585.04 789.84 4144.56
Latency 78.89ms 15.89ms 206.94ms
HTTP codes:
1xx - 0, 2xx - 7975, 3xx - 0, 4xx - 0, 5xx - 0
others - 0
Throughput: 73.85MB/s
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: zeripath <art27@cantab.net>
* Fix page and missing return on unadopted repos API
Page must be 1 if it's not specified and it should return after sending an internal server error.
* Allow ignore pages
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
- Fix regression caused by: f1b1472632
- Don't try to insert a email for Organisation(as they don't have one).
- Resolves#18891
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* logs: add the buffer logger to inspect logs during testing
Signed-off-by: Loïc Dachary <loic@dachary.org>
* migrations: add test for importing pull requests in gitea uploader
Signed-off-by: Loïc Dachary <loic@dachary.org>
* for each git.OpenRepositoryCtx, call Close
* Content is expected to return the content of the log
* test for errors before defer
Co-authored-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Following the merging of #17811 teams can now have differing write and readonly permissions, however the assignee list will not include teams which have mixed perms.
Further the org sidebar is no longer helpful as it can't describe these mixed permissions situations.
Fix#18572
Signed-off-by: Andrew Thornton <art27@cantab.net>
We can't depend on `latest` version of gofumpt because the output will
not be stable across versions. Lock it down to the latest version
released yesterday and run it again.
* Use email_address table to check user's email when login with email adress
* Update services/auth/signin.go
* Fix test
* Fix test
* Fix logging in with ldap username != loginname
* Fix if user does not exist yet
* Make more clear this is loginName
* Fix formatting
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
v208.go is seriously broken as it misses an ID() check. We need to no-op and remigrate all of the u2f keys.
See #18756
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Fix display time of milestones
* Move the SecToTime function
From the models/issue_stopwatch.go file to the modules/util package
* Rename the sec_to_time file
* Updated formatting
* Include copyright notice in sec_to_time.go
* Apply PR review suggestions
- Update copyright notice dates to 2022
- Change `1 day 3h 5min 7s` to `1d 3h 5m 7s`
* Rename hrs var and combine conditions
* Update unit tests to match new time pattern
Changed `1min` to `1m`
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Unfortunately credentialIDs in u2f are 255 bytes long which with base32 encoding
becomes 408 bytes. The default size of a xorm string field is only a VARCHAR(255)
This problem is not apparent on SQLite because strings get mapped to TEXT there.
Fix#18727
Signed-off-by: Andrew Thornton <art27@cantab.net>
- Don't let `TypeExternalTracker` or `TypeExternalWiki` influence the
minimal permission, as they won't be higher than read. So even if all
the other ones are write, these 2 will ensure that's not higher than
read.
- Partially resolves#18572 (Point 1,2,5?)
Co-authored-by: zeripath <art27@cantab.net>
There is no need to call UpdateRepoStats in the InsertIssues and
InsertPullRequests function. They are only called during migration by
the CreateIssues and CreateReviews methods of the gitea uploader.
The UpdateRepoStats function will be called by the Finish method of
the gitea uploader after all reviews and issues are inserted. Calling
it before is therefore redundant and the associated SQL requests are
not cheap.
The statistics tests done after inserting an issue or a pull request
are also removed. They predate the implementation of UpdateRepoStats,
back when the calculation of the statistics was an integral part of
the migration function. The UpdateRepoStats is now tested
independantly and these tests are no longer necessary.
Signed-off-by: singuliere <singuliere@autistici.org>
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
When calling DumpRepository and RestoreRepository on the same Gitea
instance, the users are preserved: all labels, issues etc. belong to
the external user who is, in this particular case, the local user.
Dead code verifying g.gitServiceType.Name() == "" (i.e. plain git) is
removed. The function is never called because the plain git downloader
does not migrate anything that is associated to a user, by definition.
Errors returned by GetUserIDByExternalUserID are no longer ignored.
The userMap is used when the external user is not kown, which is the
most common case. It was only used when the external user exists
which happens less often and, as a result, every occurence of an
unknown external user required a SQL query.
Signed-off-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
- Switch to use `CryptoRandomBytes` instead of `CryptoRandomString`, OAuth's secrets are copied pasted and don't need to avoid dubious characters etc.
- `CryptoRandomBytes` gives  `CryptoRandomString` gives  possible states.
- Add a prefix, such that code scanners can easily grep these in source code.
- 32 Bytes + prefix
* Collaborator trust model should trust collaborators
There was an unintended regression in #17917 which leads to only
repository admin commits being trusted. This PR restores the old logic.
Fix#18501
Signed-off-by: Andrew Thornton <art27@cantab.net>
* COrrect use `UserID` in `SearchTeams`
- Use `UserID` in the `SearchTeams` function, currently it was useless
to pass such information. Now it does a INNER statement to `team_user`
which obtains UserID -> TeamID data.
- Make OrgID optional.
- Resolves#18484
* Seperate searching specific user
* Add condition back
* Use correct struct type
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* add test coverage for original author conversion during migrations
And create a function to factorize a code snippet that is repeated
five times and would otherwise be more difficult to test and maintain
consistently.
Signed-off-by: Loïc Dachary <loic@dachary.org>
* fix variable scope and int64 formatting
* add missing calls to remapExternalUser and fix misplaced %d
Co-authored-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
The endpoint /{username}/{reponame}/milestone/{id} is not currently restricted to
the repo. This PR restricts the milestones to those within the repo.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Add config option to hide issue events
Adds a config option `HIDE_ISSUE_EVENTS` to hide most issue events (changed labels, milestones, projects...) on the issue detail page.
If this is true, only the following events (comment types) are shown:
* plain comments
* closed/reopned/merged
* reviews
* Make configurable using a list
* Add docs
* Add missing newline
* Fix merge issues
* Allow changes per user settings
* Fix lint
* Rm old docs
* Apply suggestions from code review
* Use bitsets
* Rm comment
* fmt
* Fix lint
* Use variable/constant to provide key
* fmt
* fix lint
* refactor
* Add a prefix for user setting key
* Add license comment
* Add license comment
* Update services/forms/user_form_hidden_comments.go
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* check len == 0
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: Gusted <williamzijl7@hotmail.com>
Co-authored-by: 6543 <6543@obermui.de>
This PR continues the work in #17125 by progressively ensuring that git
commands run within the request context.
This now means that the if there is a git repo already open in the context it will be used instead of reopening it.
Signed-off-by: Andrew Thornton <art27@cantab.net>
The CheckRepoStats function missed the following counters:
- label num_closed_issues & num_closed_pulls
- milestone num_closed_issues & num_closed_pulls
The update SQL statements for updating the repository
num_closed_issues & num_closed_pulls fields were repeated in three
functions (repo.CheckRepoStats, migrate.insertIssues and
models.Issue.updateClosedNum) and were moved to a single helper.
The UpdateRepoStats is implemented and called in the Finish migration method so that it happens immediately instead of wating for the
CheckRepoStats to run.
Signed-off-by: Loïc Dachary loic@dachary.org
---
[source](https://lab.forgefriends.org/forgefriends/forgefriends/-/merge_requests/34)
This contains some additional fixes and small nits related to #17957
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Migrate from U2F to Webauthn
Co-authored-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
we don't want reviews to count towards comments, as this needs changes
in other components as well (eg repo stats cron job, etc).
Co-authored-by: 6543 <6543@obermui.de>
* migrations: a deadline at January 1st, 1970 is valid
Do not change the deadline value if it is set to January 1st, 1970.
Setting the deadline to year 9999 when it is zero (which is equal to
January 1st, 1970) modifies a deadline set to January 1st, 1970 which
is a valid date. In addition, setting a date in year 9999 will be
converted to a null date in some cases.
Signed-off-by: Loïc Dachary <loic@dachary.org>
* tests: set milestone.deadline_unix in fixtures
The value of deadline_unix must be set to 253370764800 (i.e. 9999-01-01) in
fixtures, otherwise it will be inserted as null which leads to
unexpected errors. For instance, DumpRepository will store a null
deadline_unix as 0 (i.e. 1970-01-01) and RestoreRepository will change
it to 9999-01-01.
Signed-off-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: Loïc Dachary <loic@dachary.org>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
- Don't use `ioutil` package anymore as it doesn't anything special
anymore since Go 1.16:
```
// As of Go 1.16, the same functionality is now provided
// by package io or package os, and those implementations
// should be preferred in new code.
```
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Team permission allow different unit has different permission
* Finish the interface and the logic
* Fix lint
* Fix translation
* align center for table cell content
* Fix fixture
* merge
* Fix test
* Add deprecated
* Improve code
* Add tooltip
* Fix swagger
* Fix newline
* Fix tests
* Fix tests
* Fix test
* Fix test
* Max permission of external wiki and issues should be read
* Move team units with limited max level below units table
* Update label and column names
* Some improvements
* Fix lint
* Some improvements
* Fix template variables
* Add permission docs
* improve doc
* Fix fixture
* Fix bug
* Fix some bug
* fix
* gofumpt
* Integration test for migration (#18124)
integrations: basic test for Gitea {dump,restore}-repo
This is a first step for integration testing of DumpRepository and
RestoreRepository. It:
runs a Gitea server,
dumps a repo via DumpRepository to the filesystem,
restores the repo via RestoreRepository from the filesystem,
dumps the restored repository to the filesystem,
compares the first and second dump and expects them to be identical
The verification is trivial and the goal is to add more tests for each
topic of the dump.
Signed-off-by: Loïc Dachary <loic@dachary.org>
* Team permission allow different unit has different permission
* Finish the interface and the logic
* Fix lint
* Fix translation
* align center for table cell content
* Fix fixture
* merge
* Fix test
* Add deprecated
* Improve code
* Add tooltip
* Fix swagger
* Fix newline
* Fix tests
* Fix tests
* Fix test
* Fix test
* Max permission of external wiki and issues should be read
* Move team units with limited max level below units table
* Update label and column names
* Some improvements
* Fix lint
* Some improvements
* Fix template variables
* Add permission docs
* improve doc
* Fix fixture
* Fix bug
* Fix some bug
* Fix bug
Co-authored-by: Lauris BH <lauris@nix.lv>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Aravinth Manivannan <realaravinth@batsense.net>
- The current implementation of `RandomString` doesn't give you a most-possible unique randomness. It gives you 6*`length` instead of the possible 8*`length` bits(or as `length`x bytes) randomness. This is because `RandomString` is being limited to a max value of 63, this in order to represent the random byte as a letter/digit.
- The recommendation of pbkdf2 is to use 64+ bit salt, which the `RandomString` doesn't give with a length of 10, instead of increasing 10 to a higher number, this patch adds a new function called `RandomBytes` which does give you the guarentee of 8*`length` randomness and thus corresponding of `length`x bytes randomness.
- Use hexadecimal to store the bytes value in the database, as mentioned, it doesn't play nice in order to convert it to a string. This will always be a length of 32(with `length` being 16).
- When we detect on `Authenticate`(source: db) that a user has the old format of salt, re-hash the password such that the user will have it's password hashed with increased salt.
Thanks to @zeripath for working out the rouge edges from my first commit 😄.
Co-authored-by: lafriks <lauris@nix.lv>
Co-authored-by: zeripath <art27@cantab.net>
They were previously not covered at all, either by integration tests or unit tests.
This PR also fixes a bug where the `num_comments` field was incorrectly set to include all types of comments.
It sets num_closed_issues: 0 as default in milestone unit test fixtures. If they are not set, Incr("num_closed_issues") will be a noop because the field is null.
* Add API to get issue/pull comments and events (timeline)
Adds an API to get both comments and events in one endpoint with all required data.
Closesgo-gitea/gitea#13250
* Fix swagger
* Don't show code comments (use review api instead)
* fmt
* Fix comment
* Time -> TrackedTime
* Use var directly
* Add logger
* Fix lint
* Fix test
* Add comments
* fmt
* [test] get issue directly by ID
* Update test
* Add description for changed refs
* Fix build issues + lint
* Fix build
* Use string enums
* Update swagger
* Support `page` and `limit` params
* fmt + swagger
* Use global slices
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
This PR reworked the Find pointer files feature in Settings -> LFS page.
When a LFS object is missing from database but exists in LFS content store, admin can associate it to the repository by clicking the Associate button.
This PR is not perfect (because the LFS module itself should be improved too), it's just a nice-to-have feature to help users recover their LFS repositories (eg: database was lost / table was truncated)
The GITEA_UNIT_TESTS_VERBOSE variable is an undocumented variable
introduced in 2017 (see 1028ef2def)
whose sole purpose has been to log SQL statements when running unit
tests.
It is renamed for clarity and a warning is displayed for backward
compatibility for people and scripts that know about it.
The documentation is updated to reflect this change.
When viewing issues in sorted order, some issues are duplicated across
pages and some are missing. This is caused by the lack of tie-breakers
in database queries, making pagination inconsistent.
* Add support for ssh commit signing
* Split out ssh verification to separate file
* Show ssh key fingerprint on commit page
* Update sshsig lib
* Make sure we verify against correct namespace
* Add ssh public key verification via ssh signatures
When adding a public ssh key also validate that this user actually
owns the key by signing a token with the private key.
* Remove some gpg references and make verify key optional
* Fix spaces indentation
* Update options/locale/locale_en-US.ini
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* Update templates/user/settings/keys_ssh.tmpl
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* Update options/locale/locale_en-US.ini
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* Update options/locale/locale_en-US.ini
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* Update models/ssh_key_commit_verification.go
Co-authored-by: Gusted <williamzijl7@hotmail.com>
* Reword ssh/gpg_key_success message
* Change Badsignature to NoKeyFound
* Add sign/verify tests
* Fix upstream api changes to user_model User
* Match exact on SSH signature
* Fix code review remarks
Co-authored-by: Gusted <williamzijl7@hotmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
This PR contains multiple fixes. The most important of which is:
* Prevent hang in git cat-file if the repository is not a valid repository
Unfortunately it appears that if git cat-file is run in an invalid
repository it will hang until stdin is closed. This will result in
deadlocked /pulls pages and dangling git cat-file calls if a broken
repository is tried to be reviewed or pulls exists for a broken
repository.
Fix#14734Fix#9271Fix#16113
Otherwise there are a few small other fixes included which this PR was initially intending to fix:
* Fix panic on partial compares due to missing PullRequestWorkInProgressPrefixes
* Fix links on pulls pages due to regression from #17551 - by making most /issues routes match /pulls too - Fix#17983
* Fix links on feeds pages due to another regression from #17551 but also fix issue with syncing tags - Fix#17943
* Add missing locale entries for oauth group claims
* Prevent NPEs if ColorFormat is called on nil users, repos or teams.
Save a bit of bandwidth by only requesting 3-times the rendered avatar
size. Factor 4 is only really beneficial on a handful of mobile phones
and I don't think they are the primary device we design for.
Configurability contributed by zeripath.
Fixes: https://github.com/go-gitea/gitea/pull/17422
Fixes: https://github.com/go-gitea/gitea/issues/16287
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
* Add missing `X-Total-Count` and fix some related bugs
Adds `X-Total-Count` header to APIs that return a list but doesn't have it yet.
Fixed bugs:
* not returned after reporting error (39eb82446c/routers/api/v1/user/star.go (L70))
* crash with index out of bounds, API issue/issueSubscriptions
I also found various endpoints that return lists but do not apply/support pagination yet:
```
/repos/{owner}/{repo}/issues/{index}/labels
/repos/{owner}/{repo}/issues/comments/{id}/reactions
/repos/{owner}/{repo}/branch_protections
/repos/{owner}/{repo}/contents
/repos/{owner}/{repo}/hooks/git
/repos/{owner}/{repo}/issue_templates
/repos/{owner}/{repo}/releases/{id}/assets
/repos/{owner}/{repo}/reviewers
/repos/{owner}/{repo}/teams
/user/emails
/users/{username}/heatmap
```
If this is not expected, an new issue should be opened.
Closes#13043
* fmt
* Update routers/api/v1/repo/issue_subscription.go
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
* Use FindAndCount
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
Co-authored-by: 6543 <6543@obermui.de>
* Add setting to OAuth handlers to override local 2FA settings
This PR adds a setting to OAuth and OpenID login sources to allow the source to
override local 2FA requirements.
Fix#13939
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Fix regression from #16544
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Add scopes settings
Signed-off-by: Andrew Thornton <art27@cantab.net>
* fix trace logging in auth_openid
Signed-off-by: Andrew Thornton <art27@cantab.net>
* add required claim options
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Move UpdateExternalUser to externalaccount
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Allow OAuth2/OIDC to set Admin/Restricted status
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Allow use of the same group claim name for the prohibit login value
Signed-off-by: Andrew Thornton <art27@cantab.net>
* fixup! Move UpdateExternalUser to externalaccount
* as per wxiaoguang
Signed-off-by: Andrew Thornton <art27@cantab.net>
* add label back in
Signed-off-by: Andrew Thornton <art27@cantab.net>
* adjust localisation
Signed-off-by: Andrew Thornton <art27@cantab.net>
* placate lint
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Running `make test-backend` will delete `data/` due to reloading the configuration and resetting the appdatapath.
This PR removes this unnecessary config reload but also adds extra code in to the unittest main to prevent its cleanup from deleting the wrong directory.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Move keys to models/keys
* Rename models/keys -> models/asymkey
* change the missed package name
* Fix package alias
* Fix test
* Fix docs
* Fix test
* Fix test
* merge
* Some refactors related repository model
* Move more methods out of repository
* Move repository into models/repo
* Fix test
* Fix test
* some improvements
* Remove unnecessary function
* Fix a panic in NotifyCreateIssueComment (caused by string truncation)
* more unit tests
* refactor
* fix some edge cases
* use SplitStringAtByteN for comment content
* Refactor install page (db type)
* set correct default DB HOST for different DB TYPE
* remove legacy TiDB from documents
* unify the usage of DB TYPE, in code we only use "mysql". "MySQL" is only shown to users for friendly name.
* Gitea can use TiDB via MySQL protocol
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Check if column exist before rename if exist, just return with no error
* Also check if errors column exist
* Add comment for migration
* Fix sqlite test
* Improve install code to avoid low-level mistakes.
If a user tries to do a re-install in a Gitea database, they gets a warning and double check.
When Gitea runs, it never create empty app.ini automatically.
Also some small (related) refactoring:
* Refactor db.InitEngine related logic make it more clean (especially for the install code)
* Move some i18n strings out from setting.go to make the setting.go can be easily maintained.
* Show errors in CLI code if an incorrect app.ini is used.
* APP_DATA_PATH is created when installing, and checked when starting (no empty directory is created any more).
sshd(8) list restrict as a future-proof way to restrict feature
enabled in ssh. It is supported since OpenSSH 7.2, out since
2016-02-29.
OpenSSH will ignore unknown options (see sshauthopt_parse in
auth-options.c), so it should be safe to add the option and
no-user-rc.
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* More pleasantly handle broken or missing git repositories
In #17742 it was noted that there a completely invalid git repository underlying a
repo on gitea.com. This happened due to a problem during a migration however, it
is not beyond the realms of possibility that a corruption could occur to another
user.
This PR adds a check to RepoAssignment that will detect if a repository loading has
failed due to an absent git repository. It will then show a page suggesting the user
contacts the administrator or deletes the repository.
Fix#17742
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Update options/locale/locale_en-US.ini
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* Remove unnecessary functions of User struct
* Move more database methods out of user struct
* Move more database methods out of user struct
* Fix template failure
* Fix bug
* Remove finished FIXME
* remove unnecessary code
* Improvements to content history
* initialize content history when making an edit to an old item created before the introduction of content history
* show edit history for code comments on pull request files tab
* Fix a flaw in keepLimitedContentHistory
Fix a flaw in keepLimitedContentHistory, the first and the last should never be deleted
* Remove obsolete eager initialization of content history
Use hostmacher to replace matchlist.
And we introduce a better DialContext to do a full host/IP check, otherwise the attackers can still bypass the allow/block list by a 302 redirection.
* Use a standalone struct name for Organization
* recover unnecessary change
* make the code readable
* Fix template failure
* Fix template failure
* Move HasMemberWithUserID to org
* Fix test
* Remove unnecessary user type check
* Fix test
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
- The code will get the first and second character `link[{0,1]]`.
However in a rare case the `link` could have 1 character and thus the
`link[1]` will create a panic.
There are multiple places where Gitea does not properly escape URLs that it is building and there are multiple places where it builds urls when there is already a simpler function available to use this.
This is an extensive PR attempting to fix these issues.
1. The first commit in this PR looks through all href, src and links in the Gitea codebase and has attempted to catch all the places where there is potentially incomplete escaping.
2. Whilst doing this we will prefer to use functions that create URLs over recreating them by hand.
3. All uses of strings should be directly escaped - even if they are not currently expected to contain escaping characters. The main benefit to doing this will be that we can consider relaxing the constraints on user names and reponames in future.
4. The next commit looks at escaping in the wiki and re-considers the urls that are used there. Using the improved escaping here wiki files containing '/'. (This implementation will currently still place all of the wiki files the root directory of the repo but this would not be difficult to change.)
5. The title generation in feeds is now properly escaped.
6. EscapePound is no longer needed - urls should be PathEscaped / QueryEscaped as necessary but then re-escaped with Escape when creating html with locales Signed-off-by: Andrew Thornton <art27@cantab.net>
Signed-off-by: Andrew Thornton <art27@cantab.net>
* feat: Allow multiple tags on comments
- Allow for multiples tags(Currently Poster + {Owner, Writer}).
- Utilize the Poster tag within the commentTag function and remove the
checking from templates.
- Use bitwise on CommentTags to enable specific tags.
- Don't show poster tag(view_content.tmpl) on the initial issue comment.
* Change parameters naming
* Change function name
* refactor variable wording
* Merge 'master' branch into 'tags-comments' branch
* Change naming
* `tag` -> `role`
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
* Fix 500 when a comment was deleted which has a notification
* Tolerate missing Comment in other places too
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
This change enables the usage of U2F without being forced to enroll an TOTP authenticator.
The `/user/auth/u2f` has been changed to hide the "use TOTP instead" bar if TOTP is not enrolled.
Fixes#5410Fixes#17495
* Fix stat chunks searching
- Fixes a issue whereby the given chunk of issueIDs wasn't respected and
thus the returned results where not the correct results.
* Add tests
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
- This will only allow `GetDeletedBranchByID` to return deletedBranch
which are on the repo, and thus don't return a deletedBranch from
another repo.
- This just should prevent possible bugs in the futher when a code is
passing the wrong ID into this function.
* Simplify Gothic to use our session store instead of creating a different store
We have been using xormstore to provide a separate session store for our OAuth2 logins
however, this relies on using gorilla context and some doubling of our session storing.
We can however, simplify and simply use our own chi-based session store. Thus removing
a cookie and some of the weirdness with missing contexts.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* as per review
Signed-off-by: Andrew Thornton <art27@cantab.net>
* as per review
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Handle MaxTokenLength
Signed-off-by: Andrew Thornton <art27@cantab.net>
* oops
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Co-authored-by: Lauris BH <lauris@nix.lv>
CountOrphanedObjects needs to quote the table it is joining with as this table may
be `user`.
Fix#17485
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Run Migrate in Install rather than just SyncTables
The underlying problem in #17328 appears to be that users are re-running the install
page during upgrades. The function that tests and creates the db did not intend for
this and thus instead the migration scripts being run - a simple sync tables occurs.
This then causes a weird partially migrated DB which causes, in this release cycle,
the duplicate column in task table error. It is likely the cause of some weird
partial migration errors in other cycles too.
This PR simply ensures that the migration scripts are also run at this point too.
Fix#17328
Signed-off-by: Andrew Thornton <art27@cantab.net>
There is a small bug in the way that repo access is checked in
repoAssignment: Accessibility is checked by checking if the user has a
marked access to the repository instead of checking if the user has any
team granted access.
This PR changes this permissions check to use HasAccess() which does the
correct test. There is also a fix in the release api ListReleases where
it should return draft releases if the user is a member of a team with
write access to the releases.
The PR also adds a testcase.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Improve: make diff result better, make the HTML element fit the full height in the content history diff dialog
* Bug fix: when edit the main issue, the poster is wrongly set to the issue poster
We have the `AppState` module now, it can store app related data easily. We do not need to create separate tables for each feature.
So the update checker can use `AppState` instead of a one-row dedicate table.
And the code of update checker is moved from `models` to `modules`.
Gitea writes its own AppPath into git hook scripts. If Gitea's AppPath changes, then the git push will fail.
This PR:
* Introduce an AppState module, it can persist app states into database
* During GlobalInit, Gitea will check if the current AppPath is the same as last one. If they don't match, Gitea will sync git hooks.
* Refactor some code to make them more clear.
* Also, "Detect if gitea binary's name changed" #11341 is related, we call models.RewriteAllPublicKeys to update ssh authorized_keys file
* Don't panic if we fail to parse a U2FRegistration data
Downgrade logging statement from Fatal to Error so that errors parsing
U2FRegistration data does not panic; instead, the invalid key will be
skipped and we will attempt to parse the next one, if available.
Signed-off-by: David Jimenez <dvejmz@sgfault.com>
* Ensure that git daemon export ok is created for mirrors
There is an issue with #16508 where it appears that create repo requires that the
repo does not exist. This causes #17241 where an error is reported because of this.
This PR fixes this and also runs update-server-info for mirrors and generated repos.
Fix#17241
Signed-off-by: Andrew Thornton <art27@cantab.net>
It makes Admin's life easier to filter users by various status.
* introduce window.config.PageData to pass template data to javascript module and small refactor
move legacy window.ActivityTopAuthors to window.config.PageData.ActivityTopAuthors
make HTML structure more IDE-friendly in footer.tmpl and head.tmpl
remove incorrect <style class="list-search-style"></style> in head.tmpl
use log.Error instead of log.Critical in admin user search
* use LEFT JOIN instead of SubQuery when admin filters users by 2fa. revert non-en locale.
* use OptionalBool instead of status map
* refactor SearchUserOptions.toConds to SearchUserOptions.toSearchQueryBase
* add unit test for user search
* only allow admin to use filters to search users
* issue content history
* Use timeutil.TimeStampNow() for content history time instead of issue/comment.UpdatedUnix (which are not updated in time)
* i18n for frontend
* refactor
* clean up
* fix refactor
* re-format
* temp refactor
* follow db refactor
* rename IssueContentHistory to ContentHistory, remove empty model tags
* fix html
* use avatar refactor to generate avatar url
* add unit test, keep at most 20 history revisions.
* re-format
* syntax nit
* Add issue content history table
* Update models/migrations/v197.go
Co-authored-by: 6543 <6543@obermui.de>
* fix merge
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Lauris BH <lauris@nix.lv>
- Update default branch if needed
- Update protected branch if needed
- Update all not merged pull request base branch name
- Rename git branch
- Record this rename work and auto redirect for old branch on ui
Signed-off-by: a1012112796 <1012112796@qq.com>
Co-authored-by: delvh <dev.lh@web.de>
It is possible that a keyring can contain duplicate keys on a keyring due to jpegs or
other layers. This currently leads to a confusing error for the user - where we report
a duplicate key insertion.
This PR simply coalesces keys into one key if there are duplicates.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Why this refactor
The goal is to move most files from `models` package to `models.xxx` package. Many models depend on avatar model, so just move this first.
And the existing logic is not clear, there are too many function like `AvatarLink`, `RelAvatarLink`, `SizedRelAvatarLink`, `SizedAvatarLink`, `MakeFinalAvatarURL`, `HashedAvatarLink`, etc. This refactor make everything clear:
* user.AvatarLink()
* user.AvatarLinkWithSize(size)
* avatars.GenerateEmailAvatarFastLink(email, size)
* avatars.GenerateEmailAvatarFinalLink(email, size)
And many duplicated code are deleted in route handler, the handler and the model share the same avatar logic now.
* Nicely handle missing user in collaborations
It is possible to have a collaboration in a repository which refers to a no-longer
existing user. This causes the repository transfer to fail with an unusual error.
This PR makes `repo.getCollaborators()` nicely handle the missing user by ghosting
the collaboration but also adds consistency check. It also adds an
Access consistency check.
Fix#17044
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
There was a serious issue with the `gitea dump` command in 1.14.3-1.14.6 which led to corruption of the `config` field of the `repo_unit` table.
This PR adds a doctor command to attempt to fix the broken repo_units. Users affected by #16961 should run:
```
gitea doctor --fix --run fix-broken-repo-units
```
Fix#16961
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Allow LDAP Sources to provide Avatars
Add setting to LDAP source to allow it to provide an Avatar.
Currently this is required to point to the image bytes.
Fix#4144
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Rename as Avatar Attribute (drop JPEG)
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Always synchronize avatar if there is change
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Actually get the avatar from the ldap
Signed-off-by: Andrew Thornton <art27@cantab.net>
* clean-up
Signed-off-by: Andrew Thornton <art27@cantab.net>
* use len()>0 rather than != ""
Signed-off-by: Andrew Thornton <art27@cantab.net>
* slight shortcut in IsUploadAvatarChanged
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* DBContext is just a Context
This PR removes some of the specialness from the DBContext and makes it context
This allows us to simplify the GetEngine code to wrap around any context in future
and means that we can change our loadRepo(e Engine) functions to simply take contexts.
Signed-off-by: Andrew Thornton <art27@cantab.net>
* fix unit tests
Signed-off-by: Andrew Thornton <art27@cantab.net>
* another place that needs to set the initial context
Signed-off-by: Andrew Thornton <art27@cantab.net>
* avoid race
Signed-off-by: Andrew Thornton <art27@cantab.net>
* change attachment error
Signed-off-by: Andrew Thornton <art27@cantab.net>
* Fix commit status index problem
* remove unused functions
* Add fixture and test for migration
* Fix lint
* Fix fixture
* Fix lint
* Fix test
* Fix bug
* Fix bug
The io/ioutil package has been deprecated as of Go 1.16, see
https://golang.org/doc/go1.16#ioutil. This commit replaces the existing
io/ioutil functions with their new definitions in io and os packages.
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
This PR adds a `ListOptions` type which is not paged but uses absolute values. It is implemented as discussed in Discord.
Extracted from #16510 to clean that PR.
When converting repositories from forks to normal the root NumFork needs to be
decremented too.
Fix#17026
Signed-off-by: Andrew Thornton <art27@cantab.net>
The rollback functionality in
services/repository/repository.go:ForkRepository is incorrect and could
lead to a deadlock as it uses DeleteRepository to delete the rolled-back
repository - a function which creates its own transaction.
This PR adjusts the rollback function to only use RemoveAll as any
database changes will be automatically rolled-back. It also handles
panics and adjusts the Close within WithTx to ensure that if there is a
panic the session will always be closed.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Fixes#16381
Note that changes to unprotected files via the web editor still cannot be pushed directly to the protected branch. I could easily add such support for edits and deletes if needed. But for adding, uploading or renaming unprotected files, it is not trivial.
* Extract & Move GetAffectedFiles to modules/git
if AllowedUserVisibilityModes allow only public & limited, and orgs can be private, a user can create a repo to that organisation whitch will result in an update of the user. On this call the user is validaten and will be rejected since private is not allowed, but its not an user its an valid org ...
Co-authored-by: Alexey 〒erentyev <axifnx@gmail.com>
When create a new issue or comment and paste/upload an attachment/image, it will not assign an issue id before submit. So if user give up the creating, the attachments will lost key feature and become dirty content. We don't know if we need to delete the attachment even if the repository deleted.
This PR add a repo_id in attachment table so that even if a new upload attachment with no issue_id or release_id but should have repo_id. When deleting a repository, they could also be deleted.
Co-authored-by: 6543 <6543@obermui.de>
Calculate and return the number of Repositories on the dashboard
Organization list.
This PR restores some of the logic that was removed in #14032 to
calculate the number of repos on the dashboard orgs list.
Fix#16648
Replaces #16799
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
* Fix dump and restore
* return different error message for get commit
* Fix missing delete release attachment when deleting repository
* Fix ci and add some comments
Co-authored-by: zeripath <art27@cantab.net>
* Refactor the fork service slightly to take ForkRepoOptions
This reduces the number of places we need to change if we want to add other
options during fork time.
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
* Fix integrations and tests after ForkRepository refactor
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
* Update OldRepo -> BaseRepo
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
* gofmt pass
Signed-off-by: Kyle Evans <kevans@FreeBSD.org>
#16831 has occurred because of a missed regression. This PR adds a simple test to
try to prevent this occuring again.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Make the group_id a primary key in issue_index. This already has an unique index
and therefore is a good candidate for becoming a primary key.
This PR also changes all other uses of this table to add the group_id as the
primary key.
Fix#16802
Signed-off-by: Andrew Thornton <art27@cantab.net>
The MySQL indexes are not being renamed at the same time as RENAME table despite the
CASCADE. Therefore it is probably better to just recreate the indexes instead.
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: 6543 <6543@obermui.de>
One of the issues holding back performance of the API is the problem of hashing.
Whilst banning BASIC authentication with passwords will help, the API Token scheme
still requires a PBKDF2 hash - which means that heavy API use (using Tokens) can
still cause enormous numbers of hash computations.
A slight solution to this whilst we consider moving to using JWT based tokens and/or
a session orientated solution is to simply cache the successful tokens. This has some
security issues but this should be balanced by the security issues of load from
hashing.
Related #14668
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
