Backport #22392
Fix #22386
`GetDirectorySize` moved as `getDirectorySize` because it becomes a
special function which should not be put in `util`.
Co-authored-by: Jason Song <i@wolfogre.com>
Backport #22454
There is a mistake in the code for SearchRepositoryCondition where it
tests topics as a string. This is incorrect for postgres where topics is
cast and stored as json. topics needs to be cast to text for this to
work. (For some reason JSON_ARRAY_LENGTH does not work, so I have taken
the simplest solution of casting to text and doing a string comparison.)
Ref https://github.com/go-gitea/gitea/pull/21962#issuecomment-1379584057
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22453
In #22447 it was noticed that display environments were not working
correctly. This was due to the setting displayMode not being set.
Further it was noticed that the error was not being displayed correctly.
This PR fixes both of these issues by forcibly setting the displayMode
setting and corrects an error in displayError.
Fix #22447
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22467
Fix #22370 and more.
Before Go 1.19, the `netgo` tag for Windows does nothing.
But Go 1.19 rewrite the net package code for Windows DNS, and there is a
bug:
* https://github.com/golang/go/issues/57757
This PR just removes the `netgo` tag for Windows build, then the Gitea
for Windows can have the old DNS behavior.
Backport #22457
The API endpoints for "git" can panic if they are called on an empty
repo. We can simply allow empty repos for these endpoints without worry
as they should just work.
Fix #22452
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22092
As described in the linked issue (#22091), semi-transparent UI elements
would result in JS errors due to the fact that the CSS `backgroundColor`
element was being matched by the pattern
`^rgb\((\d+),\s*(\d+),\s*(\d+)\)$`, which does not take the alpha
channel into account.
I changed the pattern to `^rgba?\((\d+),\s*(\d+),\s*(\d+).*\)$`. This
new pattern accepts both `rgb` and `rgba` tuples, and ignores the alpha
channel (that little `.*` at the end) from the sorting criteria. The
reason why I chose to ignore alpha is because when it comes to kanban
colour sorting, only the hue is important; the order of the panels
should stay the same, even if some of them are transparent.
Alternative solutions were discussed in the bug report and are included
here for completeness:
1. Change the regex from ^rgb\((\d+),\s*(\d+),\s*(\d+)\)$ to
^rgba?\((\d+),\s*(\d+),\s*(\d+)(,\s*(\d+(\.\d+)?))?\)$ (alpha channel is
a float or NaN on 5th group) and include the alpha channel in the
sorting criteria.
2. Rethink on why you're reading colours out of the CSS in the first
place, then reformat this sorting procedure.
Fix #22091
Co-authored-by: MisterCavespider <deler.urist@tutanota.de>
Backport #21762
Wechatwork webhook is sending the following string for pull request
reviews:
``` markdown
>
```
This commit fixes this problem.
Co-authored-by: Jim Kirisame <jim@lotlab.org>
Backport #22422
The current code propagates all errors up to the iteration step meaning
that a single malformed repo will prevent GC of other repos.
This PR simply stops that propagation.
Fix #21605
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22146
There are a few places in FlushQueueWithContext which make an incorrect
assumption about how `select` on multiple channels works.
The problem is best expressed by looking at the following example:
```go
package main
import "fmt"
func main() {
closedChan := make(chan struct{})
close(closedChan)
toClose := make(chan struct{})
count := 0
for {
select {
case <-closedChan:
count++
fmt.Println(count)
if count == 2 {
close(toClose)
}
case <-toClose:
return
}
}
}
```
This PR double-checks that the contexts are closed outside of checking
if there is data in the dataChan. It also rationalises the WorkerPool
FlushWithContext because the previous implementation failed to handle
pausing correctly. This will probably fix the underlying problem in
#22145
Fix #22145
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22399
There was a serious regression in #21012 which broke the Show More
button on the diff page, and the show more button was also broken on the
file tree too.
This PR fixes this by resetting the pageData.diffFiles as the vue
watched value and reattachs a function to the show more button outside
of the file tree view.
Fix #22380
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22374
Fix #22286
When timetracking is disabled, the stop watch top bar icon should be
hidden. When the stop watch recording popup, it should be allowed to
hide with some operation. Now click any place on this page will hide the
popup window.
backport #22261
A drawback is the previous generated template has been cached, so you
cannot get error in the UI but only from log
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
Co-authored-by: delvh <dev.lh@web.de>
Backport #22302
Previously, the last minute of the chosen date caused bad timezone
rendering.
For example, I chose January 4th, 2023.
### Before
```html
<time data-format="date" datetime="Wed, 04 Jan 2023 23:59:59 +0000">January 5, 2023</time>
```
### After
```html
<time data-format="date" datetime="2023-01-04">January 4, 2023</time>
```
---
Closes #21999
Signed-off-by: Yarden Shoham <hrsi88@gmail.com>
- Backport of #22229
- When the GPG key contains an error, such as an invalid signature or an
email address that does not match the user.A page will be shown that
says you must provide a signature for the token.
- This page had two errors: one had the wrong translation key and the
other tried to use an undefined variable
[`.PaddedKeyID`](e81ccc406b/models/asymkey/gpg_key.go (L65-L72)),
which is a function implemented on the `GPGKey` struct, given that we
don't have that, we use
[`KeyID`](e81ccc406b/routers/web/user/setting/keys.go (L102))
which is [the fingerprint of the
publickey](https://pkg.go.dev/golang.org/x/crypto/openpgp/packet#PublicKey.KeyIdString)
and is a valid way for opengpg to refer to a key.
backport #22246
Since #20218 introduced internal watching template, template watching
should be removed from `air`. This will prevent restart the whole server
once the template files changed to speed up developing when using `make
watch`.
To ensure `make watch` will reuse template watching, this PR introduced
a new ENV `GITEA_RUN_MODE` to make sure `make watch` will always run in
a dev mode of Gitea so that template watching will open.
This PR also added more exclude testdata directories.
Co-authored-by: 6543 <6543@obermui.de>
backport #22119
This PR changed the Auth interface signature from `Verify(http
*http.Request, w http.ResponseWriter, store DataStore, sess
SessionStore) *user_model.User`
to
`Verify(http *http.Request, w http.ResponseWriter, store DataStore, sess
SessionStore) (*user_model.User, error)`.
There is a new return argument `error` which means the verification
condition matched but verify process failed, we should stop the auth
process.
Before this PR, when return a `nil` user, we don't know the reason why
it returned `nil`. If the match condition is not satisfied or it
verified failure? For these two different results, we should have
different handler. If the match condition is not satisfied, we should
try next auth method and if there is no more auth method, it's an
anonymous user. If the condition matched but verify failed, the auth
process should be stop and return immediately.
This will fix #20563
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
Co-authored-by: Jason Song <i@wolfogre.com>
backport from #22219
Since we changed the /api/v1/ routes to disallow session authentication
we also removed their reliance on CSRF. However, we left the
ReverseProxy authentication here - but this means that POSTs to the API
are no longer protected by CSRF.
Now, ReverseProxy authentication is a kind of session authentication,
and is therefore inconsistent with the removal of session from the API.
This PR proposes that we simply remove the ReverseProxy authentication
from the API and therefore users of the API must explicitly use tokens
or basic authentication.
Replace #22077
Close #22221
Close #22077
Signed-off-by: Andrew Thornton <art27@cantab.net>
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: zeripath <art27@cantab.net>
Backport #22031
There is an unaligned atomic field in zapx 15.3.5 which should have been
fixed in a subsequent patch
This bug causes issues on 32bit builds.
Update bleve and zapx to account for this.
Fix #21957
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22130
For a long time Gitea has tested PR patches using a git apply --check
method, and in fact prior to the introduction of a read-tree assisted
three-way merge in #18004, this was the only way of checking patches.
Since #18004, the git apply --check method has been a fallback method,
only used when the read-tree three-way merge method has detected a
conflict. The read-tree assisted three-way merge method is much faster
and less resource intensive method of detecting conflicts. #18004 kept
the git apply method around because it was thought possible that this
fallback might be able to rectify conflicts that the read-tree three-way
merge detected. I am not certain if this could ever be the case.
Given the uncertainty here and the now relative stability of the
read-tree method - this PR makes using this fallback optional but
enables it by default. A `log.Critical` has been added which will alert
if the `git apply --check` method was successful at checking a PR that
`read-tree` failed on.
The hope is that none of these log.Critical messages will be found and
there will be no significant difference in conflict detection. Thus we
will be able to remove the git apply fallback in future, and/or improve
the read-tree three-way merge method to catch any conflicts that git
apply method might have been able to fix.
An additional benefit for anyone who disables the check method is that
patch checking should be significantly less resource intensive and much
quicker.
(See
https://github.com/go-gitea/gitea/issues/22083\#issuecomment-1347961737)
Ref #22083
Signed-off-by: Andrew Thornton <art27@cantab.net>
<!--
Please check the following:
1. Make sure you are targeting the `main` branch, pull requests on
release branches are only allowed for bug fixes.
2. Read contributing guidelines:
https://github.com/go-gitea/gitea/blob/main/CONTRIBUTING.md
3. Describe what your pull request does and which issue you're targeting
(if any)
-->
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: KN4CK3R <admin@oldschoolhack.me>
- Backport of #22171
- Currently, the 'IsZero' function for 'TimeStamp' just checks if the
unix time is zero, which is not the behavior of 'Time.IsZero()', but
Gitea is using this method in accordance with the behavior of
'Time.IsZero()'.
- Adds a new condition to check for the zero time instant.
- Fixes a bug where non-expiring GPG keys where shown as they expired on
Jan 01, 0001.
- Related https://codeberg.org/Codeberg/Community/issues/791
Backport #22017
As recognised in #21841 the rendering of plain text files is somewhat
incorrect when there are ambiguous characters as the html code is double
escaped. In fact there are several more problems here.
We have a residual isRenderedHTML which is actually simply escaping the
file - not rendering it. This is badly named and gives the wrong
impression.
There is also unusual behaviour whether the file is called a Readme or
not and there is no way to get to the source code if the file is called
README.
In reality what should happen is different depending on whether the file
is being rendered a README at the bottom of the directory view or not.
1. If it is rendered as a README on a directory - it should simply be
escaped and rendered as `<pre>` text.
2. If it is rendered as a file then it should be rendered as source
code.
This PR therefore does:
1. Rename IsRenderedHTML to IsPlainText
2. Readme files rendered at the bottom of the directory are rendered
without line numbers
3. Otherwise plain text files are rendered as source code.
Replace #21841
Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Backport #22034
This fixes a bug where, when searching unadopted repositories, active
repositories will be listed as well. This is because the size of the
array of repository names to check is larger by one than the
`IterateBufferSize`.
For an `IterateBufferSize` of 50, the original code will pass 51
repository names but set the query to `LIMIT 50`. If all repositories in
the query are active (i.e. not unadopted) one of them will be omitted
from the result. Due to the `ORDER BY` clause it will be the oldest (or
least recently modified) one.
Co-authored-by: Christian Ullrich <christian.ullrich@traditionsa.lu>
Backport #22162
The PR #21198 introduced a probable security vulnerability which
resulted in making all storage files be marked as executable.
This PR ensures that these are forcibly marked as non-executable.
Fix #22161
Signed-off-by: Andrew Thornton <art27@cantab.net>
Signed-off-by: Andrew Thornton <art27@cantab.net>
Backport #22157. vue3-calendar-heatmap has the behaviour that the first
and second colors are mapped to values null and 0, meaning the second
color was not used as intended for values > 0. I think this is a
behaviour change from previous vue2 version that was missed during the
upgrade.
This change makes first and second values the same, so the heatmap can
now use one additional color for meaningful values.
