Backport (#21708)
This addresses #21707 and adds a second package test case for a
non-semver compatible version (this might be overkill though since you
could also edit the old package version to have an epoch in front and
see the error, this just seemed more flexible for the future).
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
Backport #21646
A bug was introduced in #17865 where filepath.Join is used to join
putative unadopted repository owner and names together. This is
incorrect as these names are then used as repository names - which shoud
have the '/' separator. This means that adoption will not work on
Windows servers.
Fix#21632
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #21619 .
A patch to #17335.
Just like AppPath, Gitea writes its own CustomConf into git hook scripts
too. If Gitea's CustomConf changes, then the git push may fail.
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Backport #21540
At the moment, If admin disable Packages, still show the Packages on the
admin dashboard.
This patch added a check to hide the Packages entry.
Backports #21459
When actions besides "delete" are performed on issues, the milestone
counter is updated. However, since deleting issues goes through a
different code path, the associated milestone's count wasn't being
updated, resulting in inaccurate counts until another issue in the same
milestone had a non-delete action performed on it.
I verified this change fixes the inaccurate counts using a local docker
build.
Co-authored-by: 6543 <6543@obermui.de>
Backport #21520
Added checks for logged user token.
Some builds fail at unrelated tests, due to missing token.
Co-authored-by: Vladimir Yakovlev <nagos@inbox.ru>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
When a PR reviewer reviewed a file on a commit that was later gc'ed,
they would always get a `500` response from then on when loading the PR.
This PR simply ignores that error and instead marks all files as
unchanged.
This approach was chosen as the only feasible option without diving into
**a lot** of error handling.
Fixes#21392
Backport of #21487
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #21479
For normal commits the notification url was wrong because oldCommitID is
received from the shrinked commits list.
This PR moves the commits list shrinking after the oldCommitID
assignment.
Backport #21401
When merge was changed to run in the background context, the db updates
were still running in request context. This means that the merge could
be successful but the db not be updated.
This PR changes both these to run in the hammer context, this is not
complete rollback protection but it's much better.
Fix#21332
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #21404
Currently `repository.Num{Issues,Pulls}` weren't checked and could
become out-of-consistency. Adds these two checks to `CheckRepoStats`.
Fix incorrect SQL query for `repository.NumClosedPulls`, the check
should be for `repo_num_pulls`.
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #21064
We should only log CheckPath errors if they are not simply due to
context cancellation - and we should add a little more context to the
error message.
Fix#20709
Signed-off-by: Andrew Thornton <art27@cantab.net>
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #21351
This fixes error "unauthorized_client: invalid client secret" when
client includes secret in Authorization header rather than request body.
OAuth spec permits both:
https://www.rfc-editor.org/rfc/rfc6749#section-2.3.1
Clients in possession of a client password MAY use the HTTP Basic
authentication scheme ... Alternatively, the authorization server MAY
support including the client credentials in the request-body
Sanity validation that client id and client secret in request are
consistent with Authorization header.
Improve error descriptions. Error codes remain the same.
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: zeripath <art27@cantab.net>
Backport #21225, fix for #21128 (also in 1.17.3), close#21224
The indent was incorrect before, so this PR did some formatting work.
Bypass Golang's template bug for JS string interpolation. And since
there are JS lint rules for templates, so the string interpolation is
also a must.
Backport of #20873
When REQUIRE_SIGNIN_VIEW = true, even with public repositories, you can only see them after you login. The packages should not be accessed without login.
Co-authored-by: Lauris BH <lauris@nix.lv>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Backport of #21230
The problem was that many PR review components loaded by `Show more`
received the same ID as previous batches, which confuses browsers (when
clicked). All such occurrences should now be fixed.
Additionally improved the background of the `viewed` checkbox.
Fixes#21228.
Fixes#20681.
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
The images used by Gitea's drone pipeline were upgraded to Go 1.19.x
It causes the lint fails because Go 1.19 uses new code format.
This PR partially backport #20758 (including the emoji-data sync),
partially fix the format manually.
Backport #21195
Git uses 040000 for tree object, but some users may get 040755 for
unknown reasons, fix#21190
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
backport #21210, fix#21206
If user and viewer are equal the method should return true.
Also the common organization check was wrong as count can never be less then 0.
Tests are on main branch.
