In the `TestDatabaseMissingABranch` testcase, make sure that the
branches are in sync between the db and git before deleting a branch via
git, then compare the branch count from the web UI, making sure that it
returns an out-of-sync value first, and the correct one after another
sync.
This is currently tested by scraping the UI, and relies on the fact that
the branch counter is out of date before syncing. If that issue gets
resolved, we'll have to adjust the test to verify the sync another way.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
This tests the scenario reported in Codeberg/Community#1408: a branch
that is recorded in the database, but missing on disk was causing
internal server errors. With recent changes, that is no longer the case,
the error is logged and then ignored.
This test case tests this behaviour, that the repo's branches page on
the web UI functions even if the git branch is missing.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
When loading branches, if loading one fails, log an error, and ignore
the branch, rather than returning and causing an internal server error.
Ideally, we would only ignore the error if it was caused by a missing
branch, and do it silently, like the respective API endpoint does.
However, veryfing that at this place is not very practical, so for the
time being, ignore any and all branch loading errors.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
If we can't count the number of diverging commits for one reason or
another (such as the branch being in the database, but missing from
disk), rather than logging an error and continuing into a crash (because
`divergence` will be nil), return an error instead.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
- Remove `container` to remove unnecessary margins being added to the
whole page.
- Specify max width for the 404 image to avoid overflow of the image.
Expose the repository flags feature over the API, so the flags can be
managed by a site administrator without using the web API.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
This implements "repository flags", a way for instance administrators to
assign custom flags to repositories. The idea is that custom templates
can look at these flags, and display banners based on them, Forgejo does
not provide anything built on top of it, just the foundation. The
feature is optional, and disabled by default. To enable it, set
`[repository].ENABLE_FLAGS = true`.
On the UI side, instance administrators will see a new "Manage flags"
tab on repositories, and a list of enabled tags (if any) on the
repository home page. The "Manage flags" page allows them to remove
existing flags, or add any new ones that are listed in
`[repository].SETTABLE_FLAGS`.
The model does not enforce that only the `SETTABLE_FLAGS` are present.
If the setting is changed, old flags may remain present in the database,
and anything that uses them, will still work. The repository flag
management page will allow an instance administrator to remove them, but
not set them, once removed.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
Files can have an RSS feed, but those only make sense when taken in the
context of a branch. There is no history to make a feed of on a tag or a
commit: they're static. Forgejo does not provide a feed for them for
this reason.
However, the file view on the web UI was offering a link to these
non-existent feeds. With this patch, it does that no longer, and only
provides a link when viewing the file in the context of a branch.
Fixes#2102.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
When trying to find a `README.md` in a `.profile` repo, do so case
insensitively. This change does not make it possible to render readmes
in formats other than Markdown, it just removes the hard-coded
"README.md".
Also adds a few tests to make sure the change works.
Fixes#1494.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit edd219d8e9)
- It's possible that `PageIsDiff` is set but not `Commit` resulting in a
NPE in the template. This can happen when the requested commit doesn't exist.
- Regression of c802c46a9b &
5743d7cb5b
- Added 'hacky' integration test.
(cherry picked from commit 8db2d5e4a7)
Adds a new `/{username}/{repo}/badges` family of routes, which redirect
to various shields.io badges. The goal is to not reimplement badge
generation, and delegate it to shields.io (or a similar service), which
are already used by many. This way, we get all the goodies that come
with it: different styles, colors, logos, you name it.
So these routes are just thin wrappers around shields.io that make it
easier to display the information we want. The URL is configurable via
`app.ini`, and is templatable, allowing to use alternative badge
generator services with slightly different URL patterns.
Additionally, for compatibility with GitHub, there's an
`/{username}/{repo}/actions/workflows/{workflow_file}/badge.svg` route
that works much the same way as on GitHub. Change the hostname in the
URL, and done.
Fixes gitea#5633, gitea#23688, and also fixes#126.
Work sponsored by Codeberg e.V.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit fcd0f61212)
When displaying the repo home view, do not redirect to unit types that
can't be defaults (which, at the moment, are the external wiki and issue
tracker unit types).
If we'd redirect to those, that would mean that a repository with the
Code unit disabled, and an external issue tracker would immediately
redirect to the external issue tracker, making it harder to reach other,
non-external units of the repo.
Fixes#1965.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 44078e5460)
- The transaction in combination with Git push was causing deadlocks if
you had the `push_update` queue set to `immediate`. This was the root
cause of slow integration tests in CI.
- Remove the sync branch code as this is already being done in the Git
post-receive hook.
- Add tests to proof the branch models are in sync even with this code
removed.
(cherry picked from commit 90110e1f44)
(cherry picked from commit a064065cb9)
(cherry picked from commit 7713e558eb)
Conflicts:
services/repository/branch.go
https://codeberg.org/forgejo/forgejo/pulls/2068
(cherry picked from commit 3bb73e0813)
Adds `[repository].DOWNLOAD_OR_CLONE_METHODS` (defaulting to
"download-zip,download-targz,download-bundle,vscode-clone"), which lets
an instance administrator override the additional clone methods
displayed on the repository home view.
This is purely display-only, the clone methods not listed here are still
available, unless disabled elsewhere. They're just not displayed.
Fixes#710.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 2aadcf4946)
(cherry picked from commit 42ac34fbf9)
- Add condition to ensure doer isn't nil when using it.
- Added unit test.
- Resolves#2055
(cherry picked from commit 8f1a74fb29)
(cherry picked from commit 60ac881776)
- Rewrite `UpdateCommentsMigrationsByType` to not use `WHERE IN` as
that's a performance diaster for MariaDB, it now use batching to query
the the relevant comment IDs via JOINs (which is not possible in a
UPDATE query for SQLite) and then update them in a seperate query.
- Add unit test.
- Resolves https://codeberg.org/forgejo/forgejo/issues/1856
(cherry picked from commit 8098ca9d2e)
Conflicts:
models/issues/comment.go
https://codeberg.org/forgejo/forgejo/pulls/2075
(cherry picked from commit ca65deba1c)
- The endpoint was moved from being an API endpoint to an web endpoint
with JSON result. However the API context isn't the same as the web
context, for example the `ctx.Error` only takes in the first two
arguments into consideration and doesn't do logging, which is not the
same behavior as the API context where there's three arguments and does
do logging and only reveal the function + error if the user is admin.
- Remove any details in the error message and do the logging seperatly,
this is somewhat consistent with how other API endpoints behave.
- Ref: https://codeberg.org/forgejo/forgejo/issues/1998
(cherry picked from commit fe71e32ace)
(cherry picked from commit c89e0735fa)
This is largely based on gitea#6312 by @ashimokawa, with updates and
fixes by myself, and incorporates the review feedback given in that pull
request, and more.
What this patch does is add a new "default_permissions" column to the
`repo_units` table (defaulting to read permission), adjusts the
permission checking code to take this into consideration, and then
exposes a setting that lets a repo administrator enable any user on a
Forgejo instance to edit the repo's wiki (effectively giving the wiki
unit of the repo "write" permissions by default).
By default, wikis will remain restricted to collaborators, but with the
new setting exposed, they can be turned into globally editable wikis.
FixesCodeberg/Community#28.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 4b74439922)
(cherry picked from commit 337cf62c10)
- If the session doesn't exist, it shouldn't be expected that the
variable is non-nil. Define the session variable instead and insert that.
- Add unit tests to test the behavior of the database sessions code .
- Regression caused by dd30d9d5c0.
- Resolves https://codeberg.org/forgejo/forgejo/issues/2042
(cherry picked from commit 90307ad200)
(cherry picked from commit 874ef1978d)
The cancelation of jobs is taken care of by handleWorkflows which is
called right after handleSchedules with the same event.
(cherry picked from commit ad1af2e436)
(cherry picked from commit e21252f147)
(cherry picked from commit 08195803eb)
do not reuse the payload of the event that triggered the creation of
the scheduled event. Create a new one instead that contains no other
information than the event name in the action field ("schedule").
(cherry picked from commit 0b40ca1ea5)
(cherry picked from commit f86487432b)
(cherry picked from commit 4bd5d2e9d0)
handleSchedules() is called every time an event is received and will
check the content of the main branch to (re)create scheduled events.
There is no reason why intput.Event will be relevant when the schedule
workflow runs.
(cherry picked from commit 9a712bb276)
(cherry picked from commit 41af36da81)
(cherry picked from commit bb83604fa2)
- When the user is not found in `reloadparam`, early return when the
user is not found to avoid calling `IsUserVisibleToViewer` which in turn
avoids causing a NPE.
- This fixes the case that a 500 error and 404 error is shown on the
same page.
- Add integration test for non-existant user RSS.
- Regression by c6366089df
(cherry picked from commit f0e0696278)
(cherry picked from commit 75d8066908)
(cherry picked from commit 4d0a1e0637)
It shows warnings although the setting is not set, this will surely be
fixed later but there is no sense in spaming the users right now. This
revert can be discarded when another fix lands in v1.21.
su -c "forgejo admin user generate-access-token -u root --raw --scopes 'all,sudo'" git
2023/12/12 15:54:45 .../setting/security.go:166:loadSecurityFrom() [W] Enabling Query API Auth tokens is not recommended. DISABLE_QUERY_AUTH_TOKEN will default to true in gitea 1.23 and will be removed in gitea 1.24.
This reverts commit 4e879fed90.
Conflicts:
routers/api/v1/api.go
trivial context conflict
(cherry picked from commit dec6f80559)
(cherry picked from commit 807b6d9066)
(cherry picked from commit 37e21a0c17)
- This adds coverage to the most common and the edge cases of what the
footnote implementation should be capable of. This was partly done to
ensure no hidden surprises when changing the implementation, as markdown
rendering is one of the more important features of Forgejo.
(cherry picked from commit 16ecdb4170)
(cherry picked from commit 19dc5ef5e5)
(cherry picked from commit d5955efc0a)
(cherry picked from commit 2cdaf10836)
During registration, one may be required to give their email address, to
be verified and activated later. However, if one makes a mistake, a
typo, they may end up with an account that cannot be activated due to
having a wrong email address.
They can still log in, but not change the email address, thus, no way to
activate it without help from an administrator.
To remedy this issue, lets allow changing the email address for logged
in, but not activated users.
This fixes gitea#17785.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit aaaece28e4)
(cherry picked from commit 639dafabec)
(cherry picked from commit d699c12ceb)
[GITEA] Allow changing the email address before activation (squash) cache is always active
This needs to be revisited because the MailResendLimit is not enforced
and turns out to not be tested.
See e7cb8da2a8 * Always enable caches (#28527)
(cherry picked from commit 43ded8ee30)
Rate limit pre-activation email change separately
Changing the email address before any email address is activated should
be subject to a different rate limit than the normal activation email
resending. If there's only one rate limit for both, then if a newly
signed up quickly discovers they gave a wrong email address, they'd have
to wait three minutes to change it.
With the two separate limits, they don't - but they'll have to wait
three minutes before they can change the email address again.
The downside of this setup is that a malicious actor can alternate
between resending and changing the email address (to something like
`user+$idx@domain`, delivered to the same inbox) to effectively halving
the rate limit. I do not think there's a better solution, and this feels
like such a small attack surface that I'd deem it acceptable.
The way the code works after this change is that `ActivatePost` will now
check the `MailChangeLimit_user` key rather than `MailResendLimit_user`,
and if we're within the limit, it will set `MailChangedJustNow_user`. The
`Activate` method - which sends the activation email, whether it is a
normal resend, or one following an email change - will check
`MailChangedJustNow_user`, and if it is set, it will check the rate
limit against `MailChangedLimit_user`, otherwise against
`MailResendLimit_user`, and then will delete the
`MailChangedJustNow_user` key from the cache.
Fixes#2040.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit e35d2af2e5)
(cherry picked from commit 03989418a7)
It will determine how anchors are created and will break existing
links otherwise.
Adapted from Revert "Make `user-content-* ` consistent with github (#26388)
(cherry picked from commit 1666fba8f5)
(cherry picked from commit 48f38280e8)
(cherry picked from commit 03adb3a2b4)
(cherry picked from commit a0ad36f0ad)
Similar to how some other parts of the web UI support a `/latest` path
to directly go to the latest of a certain thing, let the Actions web UI
do the same: `/{owner}/{repo}/actions/runs/latest` will redirect to the
latest run, if there's one available.
Fixes gitea#27991.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit f67ccef1dd)
Code cleanup in the actions.ViewLatest route handler
Based on feedback received after the feature was merged, use
`ctx.NotFound` and `ctx.ServerError`, and drop the use of the
unnecessary `ctx.Written()`.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 74e42da563)
(cherry picked from commit f7535a1cef)
(cherry picked from commit 1a90cd37c3)
(cherry picked from commit d86d71340a)
(cherry picked from commit 9e5cce1afc)
- Switch the supported schemas for the Swagger API around, such that
https is the first one listed. This ensures that when the Swagger API is
used it will default to the https schema, which is likely the schema you
want to use in the majority of the cases.
- Resolves#1895
BREAKING CHANGE NOTICE:
If you are using the Swagger API JSON directly to communicate with the
Forgejo API, the library you are using may be using the first schema
defined in the JSON file (e.g. https://code.forgejo.org/swagger.v1.json)
to construct the request url, this used to be `http` but has now changed
to `https`. This can cause failures if you want to send the swagger
request over `http` (and there is no HTTPS redirection configured).
(cherry picked from commit 81e5f43886)
(cherry picked from commit d847469ea2)
(cherry picked from commit 96e75e1d5c)
(cherry picked from commit 65baa64261)
(cherry picked from commit cd3e0a74e6)
Without this change, I get the error:
> Error creating test engine: sqlite3 requires: import _ "github.com/mattn/go-sqlite3" or -tags sqlite,sqlite_unlock_notify
(cherry picked from commit 15fa4cf98e)
(cherry picked from commit 2993203093)
(cherry picked from commit f5054e4883)
(cherry picked from commit e14c5f934c)
(cherry picked from commit ed53157084)
This adds a new `doctor` check: `fix-push-mirrors-without-git-remote`. The new check looks for push mirrors that do not have their remotes configured in git. If automatic fixing is enabled, it will remove these push mirrors from the database.
The check is not run by default, and thus, must be invoked manually. It should be usable in a half-migrated state, too, and as such, fixes#1800.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/1853
Co-authored-by: Gergely Nagy <forgejo@gergo.csillger.hu>
Co-committed-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 9038e07ef3)
(cherry picked from commit b15bafcbc7)
(cherry picked from commit 93ba05a2dd)
(cherry picked from commit e418ea8082)
(cherry picked from commit 321790a91e)
(cherry picked from commit f4e19d3323)
- Consider executable files as a valid case when returning a downloadURL
for them. They are just regular files with the difference being the
executable permission bit being set.
- Not integration testing as it's not possible without adding adjusting
the existing repositories to have a executable file.
- Resolves https://codeberg.org/forgejo/forgejo/issues/1825
(cherry picked from commit ca32891d54)
(cherry picked from commit 72c9df8e45)
(cherry picked from commit 0eae22d429)
(cherry picked from commit d37d0773bc)
(cherry picked from commit de4532a967)
(cherry picked from commit f5b41300a8)
- Currently the parser will look for `\[` and `$$` to detect when Latex
code starts, it will look for `\]` and `$$` respectively in order to
determine the end of the code. However if no end is found the parser
assumes the rest of the input is part of the Latex code.
- Adjust the parser's behavior to not allow the case to assume the rest
of the input is part of the Latex code and requires in order to
determine if some input is Latex code that the end sequence is also
specified.
- Example: `\[hello]` would no longer be detected as Latex code with
this patch.
- Added unit tests.
- Resolves https://codeberg.org/forgejo/forgejo/issues/1817
(cherry picked from commit 452aef1bb1)
(cherry picked from commit 8a857c24b0)
(cherry picked from commit acd1456db9)
(cherry picked from commit 6523b45073)
(cherry picked from commit e2e1a8afe7)
(cherry picked from commit a46ef652eb)
- Add a dropdown to the web interface for changing files to select which
Email should be used for the commit. It only shows (and verifies) that a
activated mail can be used, while this isn't necessary, it's better to
have this already in place.
- Added integration testing.
- Resolves https://codeberg.org/forgejo/forgejo/issues/281
(cherry picked from commit 564e701f40)
(cherry picked from commit de8f2e03cc)
(cherry picked from commit 0182cff12e)
(cherry picked from commit 9c74254d46)
(cherry picked from commit 2f0b68f821)
(cherry picked from commit 079b995d49)
