This updates the repo index/file view endpoints so annex files match the way
LFS files are rendered, making annexed files accessible via the web instead of
being black boxes only accessible by git clone.
This mostly just duplicates the existing LFS logic. It doesn't try to combine itself
with the existing logic, to make merging with upstream easier. If upstream ever
decides to accept, I would like to try to merge the redundant logic.
The one bit that doesn't directly copy LFS is my choice to hide annex-symlinks.
LFS files are always _pointer files_ and therefore always render with the "file"
icon and no special label, but annex files come in two flavours: symlinks or
pointer files. I've conflated both kinds to try to give a consistent experience.
The tests in here ensure the correct download link (/media, from the last PR)
renders in both the toolbar and, if a binary file (like most annexed files will be),
in the main pane, but it also adds quite a bit of code to make sure text files
that happen to be annexed are dug out and rendered inline like LFS files are.
Previously, Gitea's LFS support allowed direct-downloads of LFS content,
via http://$HOSTNAME:$PORT/$USER/$REPO/media/branch/$BRANCH/$FILE
Expand that grace to git-annex too. Now /media should provide the
relevant *content* from the .git/annex/objects/ folder.
This adds tests too. And expands the tests to try symlink-based annexing,
since /media implicitly supports both that and pointer-file-based annexing.
Usage of `path` was replaced by `path/filepath` in upstream forgejo, and
it made sense to use that as well where `path` was previously used. The
`setHeaderCacheForever` function and the `sendFile` method had their
signature changed.
This makes HTTP symmetric with SSH clone URLs.
This gives us the fancy feature of _anonymous_ downloads,
so people can access datasets without having to set up an
account or manage ssh keys.
Previously, to access "open access" data shared this way,
users would need to:
1. Create an account on gitea.example.com
2. Create ssh keys
3. Upload ssh keys (and make sure to find and upload the correct file)
4. `git clone git@gitea.example.com:user/dataset.git`
5. `cd dataset`
6. `git annex get`
This cuts that down to just the last three steps:
1. `git clone https://gitea.example.com/user/dataset.git`
2. `cd dataset`
3. `git annex get`
This is significantly simpler for downstream users, especially for those
unfamiliar with the command line.
Unfortunately there's no uploading. While git-annex supports uploading
over HTTP to S3 and some other special remotes, it seems to fail on a
_plain_ HTTP remote. See https://github.com/neuropoly/gitea/issues/7
and https://git-annex.branchable.com/forum/HTTP_uploads/#comment-ce28adc128fdefe4c4c49628174d9b92.
This is not a major loss since no one wants uploading to be anonymous anyway.
To support private repos, I had to hunt down and patch a secret extra security
corner that Gitea only applies to HTTP for some reason (services/auth/basic.go).
This was guided by https://git-annex.branchable.com/tips/setup_a_public_repository_on_a_web_site/
Fixes https://github.com/neuropoly/gitea/issues/3
Co-authored-by: Mathieu Guay-Paquet <mathieu.guaypaquet@polymtl.ca>
[git-annex](https://git-annex.branchable.com/) is a more complicated cousin to
git-lfs, storing large files in an optional-download side content. Unlike lfs,
it allows mixing and matching storage remotes, so the content remote(s) doesn't
need to be on the same server as the git remote, making it feasible to scatter
a collection across cloud storage, old harddrives, or anywhere else storage can
be scavenged. Since this can get complicated, fast, it has a content-tracking
database (`git annex whereis`) to help find everything later.
The use-case we imagine for including it in Gitea is just the simple case, where
we're primarily emulating git-lfs: each repo has its large content at the same URL.
Our motivation is so we can self-host https://www.datalad.org/ datasets, which
currently are only hostable by fragilely scrounging together cloud storage --
and having to manage all the credentials associated with all the pieces -- or at
https://openneuro.org which is fragile in its own ways.
Supporting git-annex also allows multiple Gitea instance to be annex remotes for
each other, mirroring the content or otherwise collaborating the split up the
hosting costs.
Enabling
--------
TODO
HTTP
----
TODO
Permission Checking
-------------------
This tweaks the API in routers/private/serv.go to expose the calling user's
computed permission, instead of just returning HTTP 403.
This doesn't fit in super well. It's the opposite from how the git-lfs support is
done, where there's a complete list of possible subcommands and their matching
permission levels, and then the API compares the requested with the actual level
and returns HTTP 403 if the check fails.
But it's necessary. The main git-annex verbs, 'git-annex-shell configlist' and
'git-annex-shell p2pstdio' are both either read-only or read-write operations,
depending on the state on disk on either end of the connection and what the user
asked it to ask for, with no way to know before git-annex examines the situation.
So tell the level via GIT_ANNEX_READONLY and trust it to handle itself.
In the older Gogs version, the permission was directly read in cmd/serv.go:
```
mode, err = db.UserAccessMode(user.ID, repo)
```
- 966e925cf3/internal/cmd/serv.go (L334)
but in Gitea permission enforcement has been centralized in the API layer.
(perhaps so the cmd layer can avoid making direct DB connections?)
Deletion
--------
git-annex has this "lockdown" feature where it tries
really quite very hard to prevent you deleting its
data, to the point that even an rm -rf won't do it:
each file in annex/objects/ is nested inside a
folder with read-only permissions.
The recommended workaround is to run chmod -R +w when
you're sure you actually want to delete a repo. See
https://git-annex.branchable.com/internals/lockdown
So we edit util.RemoveAll() to do just that, so now
it's `chmod -R +w && rm -rf` instead of just `rm -rf`.
Backport #31188
Fixes issue when running `choco info pkgname` where `pkgname` is also a
substring of another package Id.
Relates to #31168
---
This might fix the issue linked, but I'd like to test it with more choco
commands before closing the issue in case I find other problems if
that's ok.
I'm pretty inexperienced with Go, so feel free to nitpick things.
Not sure I handled
[this](70f87e11b5/routers/api/packages/nuget/nuget.go (L135-L137))
in the best way, so looking for feedback on if I should fix the
underlying issue (`nil` might be a better default for `Value`?).
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
(cherry picked from commit ca414a7ccf5e26272662e360c44ac50221a0f2d4)
Do not try to create a new authorization grant when one exists already,
thus preventing a DB-related authorization issue.
Fix https://github.com/go-gitea/gitea/pull/30790#issuecomment-2118812426
---------
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
(cherry picked from commit 9c8c9ff6d10b35de8d2d7eae0fc2646ad9bbe94a)
(cherry picked from commit 07fe5a8b13)
This commit forces the resource owner (user) to always approve OAuth 2.0
authorization requests if the client is public (e.g. native
applications).
As detailed in [RFC 6749 Section 10.2](https://www.rfc-editor.org/rfc/rfc6749.html#section-10.2),
> The authorization server SHOULD NOT process repeated authorization
requests automatically (without active resource owner interaction)
without authenticating the client or relying on other measures to ensure
that the repeated request comes from the original client and not an
impersonator.
With the implementation prior to this patch, attackers with access to
the redirect URI (e.g., the loopback interface for
`git-credential-oauth`) can get access to the user account without any
user interaction if they can redirect the user to the
`/login/oauth/authorize` endpoint somehow (e.g., with `xdg-open` on
Linux).
Fixes#25061.
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
(cherry picked from commit 5c542ca94caa3587329167cfe9e949357ca15cf1)
(cherry picked from commit 1b088fade6)
These changes were missed when cherry-picking the following
c9d0e63c202827756c637d9ca7bbde685c1984b7 Remove unnecessary "Str2html" modifier from templates (#29319)
Fixes: https://codeberg.org/forgejo/forgejo/issues/3623
(cherry picked from commit caadd1815a)
The right to force merge is uses the wrong predicate and
applies to instance admins:
ctx.user.IsAdmin
It must apply to repository admins and use the following predicate:
ctx.userPerm.IsAdmin()
This regression is from the ApplyToAdmins implementation in
79b7089360.
Fixes: https://codeberg.org/forgejo/forgejo/issues/3780
(cherry picked from commit 09f3518069)
This exception existed for both instance admins and repo admins
before ApplyToAdmins was introduced in
79b7089360.
It should have been kept for instance admins only because they are not
subject to permission checks.
(cherry picked from commit 05f0007437)
This allows `nix flake metadata` and nix in general to lock a *branch*
tarball link in a manner that causes it to fetch the correct commit even
if the branch is updated with a newer version.
For further context, Nix flakes are a feature that, among other things,
allows for "inputs" that are "github:someuser/somerepo",
"https://some-tarball-service/some-tarball.tar.gz",
"sourcehut:~meow/nya" or similar. This feature allows our users to fetch
tarballs of git-based inputs to their builds rather than using git to
fetch them, saving significant download time.
There is presently no gitea or forgejo specific fetcher in Nix, and we
don't particularly wish to have one. Ideally (as a developer on a Nix
implementation myself) we could just use the generic tarball fetcher and
not add specific forgejo support, but to do so, we need additional
metadata to know which commit a given *branch* tarball represents, which
is the purpose of the Link header added here.
The result of this patch is that a Nix user can specify `inputs.something.url =
"https://forgejo-host/some/project/archive/main.tar.gz"` in flake.nix
and get a link to some concrete tarball for the actual commit in the
lock file, then when they run `nix flake update` in the future, they
will get the latest commit in that branch.
Example of it working locally:
» nix flake metadata --refresh 'http://localhost:3000/api/v1/repos/jade/cats/archive/main.tar.gz?dir=configs/nix'
Resolved URL: http://localhost:3000/api/v1/repos/jade/cats/archive/main.tar.gz?dir=configs/nix
Locked URL: 804ede182b.tar.gz?dir=configs
/nix&narHash=sha256-yP7KkDVfuixZzs0fsqhSETXFC0y8m6nmPLw2GrAMxKQ%3D
Description: Computers with the nixos
Path: /nix/store/s856c6yqghyan4v0zy6jj19ksv0q22nx-source
Revision: 804ede182b6b66469b23ea4d21eece52766b7a06
Last modified: 2024-05-02 00:48:32
For details on the header value, see:
56763ff918/doc/manual/src/protocols/tarball-fetcher.md
(cherry picked from commit 6631f56ebf)
Fix#30921
(cherry picked from commit 821d2fc2a3cc897f21d707455850177077b72410)
(cherry picked from commit 50b4e7f26e)
Equivalent to the Gitea v1.22 commit
(cherry picked from commit 8eac16de217978c1f7034f8e360f54d8d638e95e)
Fix#30923
(cherry picked from commit effb405cae88474c27f5c8322a2627019af1cf64)
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
Conflicts:
- modules/setting/oauth2.go
Conflicted due to different ways of logging. Since the log
message is removed anyway, resolved by removing it.
- modules/setting/oauth2_test.go
Manually copied the test added by Gitea.
- routers/install/install.go
Not a conflict per se, but adjusted to use NewJwtSecret().
(cherry picked from commit 193ac67176)
Equivalent to the Gitea v1.22 commit
(cherry picked from commit 5b7e54f72f7b85b3394d7af20b27152d26e26256)
Previously, if no branch was explicitly specified for a workflow, it
defaulted to the default branch of the repo. This worked fine for
workflows that were triggered on push, but it prevented showing badges
for workflows that only run on tags, or on schedule - since they do not
run on a specific branch.
Thus, relax the conditions, and if no branch is specified, just return
the latest run of the given workflow. If one is specified, *then*
restrict it to said branch.
Fixes#3487.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit d6915f4d5f)
Backport #27821 by @lunny
When a user logout and then login another user, the reverseproxy auth
should be checked before session otherwise the old user is still login.
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
(cherry picked from commit 14dc00ae01c35ad51a3982a5725777bbeabe2b78)
The previous implementation will start multiple POST requests from the
frontend when moving a column and another bug is moving the default
column will never be remembered in fact.
- [x] This PR will allow the default column to move to a non-first
position
- [x] And it also uses one request instead of multiple requests when
moving the columns
- [x] Use a star instead of a pin as the icon for setting the default
column action
- [x] Inserted new column will be append to the end
- [x] Fix#30701 the newly added issue will be append to the end of the
default column
- [x] Fix when deleting a column, all issues in it will be displayed
from UI but database records exist.
- [x] Add a limitation for columns in a project to 20. So the sorting
will not be overflow because it's int8.
---------
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
(cherry picked from commit a303c973e0264dab45a787c4afa200e183e0d953)
Conflicts:
routers/web/web.go
e91733468ef726fc9365aa4820cdd5f2ddfdaa23 Add missing database transaction for new issue (#29490) was not cherry-picked
services/issue/issue.go
fe6792dff3 Enable/disable owner and repo projects independently (#28805) was not cherry-picked
(cherry picked from commit 7d3ca90dfe)
(cherry picked from commit 084bec89ed7ae0816fc2d8db6784ad22523d1fc4)
Merging PR may fail because of various problems. The pull request may
have a dirty state because there is no transaction when merging a pull
request. ref
https://github.com/go-gitea/gitea/pull/25741#issuecomment-2074126393
This PR moves all database update operations to post-receive handler for
merging a pull request and having a database transaction. That means if
database operations fail, then the git merging will fail, the git client
will get a fail result.
There are already many tests for pull request merging, so we don't need
to add a new one.
---------
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
(cherry picked from commit ebf0c969403d91ed80745ff5bd7dfbdb08174fc7)
Conflicts:
modules/private/hook.go
routers/private/hook_post_receive.go
trivial conflicts because
263a716cb5 * Performance optimization for git push (#30104)
was not cherry-picked and because of
998a431747 Do not update PRs based on events that happened before they existed
(cherry picked from commit eb792d9f8a)
(cherry picked from commit ec3f5f9992d7ff8250c044a4467524d53bd50210)
similar to #30784 but only for the repo explore page
is covered by #30876 for the main branch
(cherry picked from commit d410e2acce22e5b3518a9bf64a9152b32a91fe18)
Before explaining the fix itself, lets look at the `action` table, and
how it is populated. Data is only ever inserted into it via
`activities_model.NotifyWatchers`, which will:
- Insert a row for each activity with `UserID` set to the acting user's
ID - this is the original activity, and is always inserted if anything
is to be inserted at all.
- It will insert a copy of each activity with the `UserID` set to the
repo's owner, if the owner is an Organization, and isn't the acting
user.
- It will insert a copy of each activity for every watcher of the repo,
as long as the watcher in question has read permission to the repo
unit the activity is about.
This means that if a repository belongs to an organizations, for most
activities, it will have at least two rows in the table. For
repositories watched by people other than their owner, an additional row
for each watcher.
These are useful duplicates, because they record which activities are
relevant for a particular user. However, for cases where we wish to see
the activities that happen around a repository, without limiting the
results to a particular user, we're *not* interested in the duplicates
stored for the watchers and the org. We only need the originals.
And this is what this change does: it introduces an additional option to
`GetFeedsOptions`: `OnlyPerformedByActor`. When this option is set,
`activities.GetFeeds()` will only return the original activities, where
the user id and the acting user id are the same. As these are *always*
inserted, we're not missing out on any activities. We're just getting
rid of the duplicates. As this is an additional `AND` condition, it can
never introduce items that would not have been included in the result
set before, it can only reduce, not extend.
These duplicates were only affecting call sites where `RequestedRepo`
was set, but `RequestedUser` and `RequestedTeam` were not. Both of those
call sites were updated to set `OnlyPerformedByActor`. As a result,
repository RSS feeds, and the `/repos/{owner}/{repo}/activities/feeds`
API end points no longer return dupes, only the original activities.
Rather than hardcoding this behaviour into `GetFeeds()` itself, I chose
to implement it as an explicit option, for the sake of clarity.
FixesCodeberg/Community#684, and addresses gitea#20986.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
(cherry picked from commit 9cb2aa989a)
Fix#30807
reuse functions in services
(cherry picked from commit a50026e2f30897904704895362da0fb12c7e5b26)
Conflicts:
models/issues/issue_update.go
routers/api/v1/repo/issue.go
trivial context conflict because of 'allow setting the update date on issues and comments'
(cherry picked from commit 6a4bc0289d)
We should be listing all repositories by default.
Fixes#28483.
(cherry picked from commit 9f0ef3621a3b63ccbe93f302a446b67dc54ad725)
Conflict:
- if ctx.IsSigned && ctx.Doer.IsAdmin || permission.UnitAccessMode(unit_model.TypeCode) >= perm.AccessModeRead {
+ if ctx.IsSigned && ctx.Doer.IsAdmin || permission.HasAccess() {
because of https://codeberg.org/forgejo/forgejo/pulls/2001
(cherry picked from commit e388822e9d)
Backport #30790 by archer-321
This commit forces the resource owner (user) to always approve OAuth 2.0
authorization requests if the client is public (e.g. native
applications).
As detailed in [RFC 6749 Section
10.2](https://www.rfc-editor.org/rfc/rfc6749.html#section-10.2),
> The authorization server SHOULD NOT process repeated authorization
requests automatically (without active resource owner interaction)
without authenticating the client or relying on other measures to ensure
that the repeated request comes from the original client and not an
impersonator.
With the implementation prior to this patch, attackers with access to
the redirect URI (e.g., the loopback interface for
`git-credential-oauth`) can get access to the user account without any
user interaction if they can redirect the user to the
`/login/oauth/authorize` endpoint somehow (e.g., with `xdg-open` on
Linux).
Fixes#25061.
Co-authored-by: Archer <archer@beezig.eu>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
(cherry picked from commit 6d83f5eddc0f394f6386e80b86a3221f6f4925ff)
Before, we would just throw 500 if a user passes an attachment that is
not an allowed type. This commit catches this error and throws a 422
instead since this should be considered a validation error.
(cherry picked from commit 872caa17c0a30d95f85ab75c068d606e07bd10b3)
Conflicts:
tests/integration/api_comment_attachment_test.go
tests/integration/api_issue_attachment_test.go
trivial context conflict because of 'allow setting the update date on issues and comments'
(cherry picked from commit 9cd0441cd3)
When subscribing or unsubscribing to/from an issue on the web ui, the
request was posted to a route handled by `repo.IssueWatch`. This
function used `ctx.Req.PostForm.Get()`, erroneously.
`request.PostForm` is *only* available if `request.ParseForm()` has been
called before it. The function in question did not do that. Under some
circumstances, something, somewhere did end up calling `ParseForm()`,
but not in every scenario.
Since we do not need to check for multiple values, the easiest fix here
is to use `ctx.Req.PostFormValue`, which will call `ParseForm()` if
necessary.
Fixes#3516.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
When all repository units are deactivated except for the code unit, the activity tab will not be shown.
Since the activities tab also shows contributing stats, it would be good to show the activities tab also when only code is active.
This commit changes the behavior when the activities tab is shown.
Previous it would only be shown when Issues, Pull-Requests or Releases are activated. Now it would additionally be shown when the code unit is activated.
Refs: #3429
| Before (Code + Issues - Owner) | Before (Code - Viewer) | After (Code + Issues - Owner) | After (Code - Viewer) |
| -- | -- | -- | -- |
|  |  |  |  |
| | `/activity` returns 404 for everyone |  | - |
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/3455
Reviewed-by: Otto <otto@codeberg.org>
Co-authored-by: Beowulf <beowulf@beocode.eu>
Co-committed-by: Beowulf <beowulf@beocode.eu>
(cherry picked from commit c54896ba70)
Backport #30291 by @edwardzhanged
Add some logic in `convert.ToBranchProtection` to return only the names
associated with readAccess instead of returning all names. This will
ensure consistency in behavior between the frontend and backend.
Fixes: #27694
Co-authored-by: Edward Zhang <45360012+edwardzhanged@users.noreply.github.com>
Co-authored-by: techknowlogick <techknowlogick@gitea.com>
Co-authored-by: wenzhuo.zhang <wenzhuo.zhang@geely.com>
(cherry picked from commit d88958bb99eabc07dead6965e396755e7b6d947f)
Backport #30511 by @yp05327
Fix#30508
ps: if `sourceID` is not set, `LoginName` will be ignored
Co-authored-by: yp05327 <576951401@qq.com>
(cherry picked from commit f9a025f6a358aa6e34408743a59f5081f397d47c)
It is possible to change some repo settings (its visibility, and
template status) via `git push` options: `-o repo.private=true`, `-o
repo.template=true`.
Previously, there weren't sufficient permission checks on these, and
anyone who could `git push` to a repository - including via an AGit
workflow! - was able to change either of these settings. To guard
against this, the pre-receive hook will now check if either of these
options are present, and if so, will perform additional permission
checks to ensure that these can only be set by a repository owner or
an administrator. Additionally, changing these settings is disabled for
forks, even for the fork's owner.
There's still a case where the owner of a repository can change the
visibility of it, and it will not propagate to forks (it propagates to
forks when changing the visibility via the API), but that's an
inconsistency, not a security issue.
Signed-off-by: Gergely Nagy <forgejo@gergo.csillger.hu>
Signed-off-by: Earl Warren <contact@earl-warren.org>
(cherry picked from commit 8eba631f8d)
Fix#30378
(cherry picked from commit 0fe9f93eb4c94d55e43b18b9c3cc6d513a34c0b5)
Conflicts:
- models/organization/org.go
- services/repository/delete.go
- services/user/delete.go
In all three cases, conflicts were resolved by manually adding
the lines added by the Gitea patch, keeping the Forgejo code
surrounding them.
