- The root cause is described in b4f1988a35
- Move to a fork of `github.com/gliderlabs/ssh` that exposes the
permissions that was chosen by `x/crypto/ssh` after succesfully
authenticating, this is the recommended mitigation by the Golang
security team. The fork exposes this, since `gliderlabs/ssh` instead
relies on context values to do so, which is vulnerable to the same
attack, although partially mitigated by the fix in `x/crypto/ssh` it
would not be good practice and defense deep to rely on it.
- Existing tests covers that the functionality is preserved.
- No tests are added to ensure it fixes the described security, the
exploit relies on non-standard SSH behavior it would be too hard to
craft SSH packets to exploit this.
## Overview
Hi all, I'm a first-time contributor to Forgejo. I was looking for something interesting to contribute and the first thing that caught my attention was https://codeberg.org/forgejo/forgejo/issues/6043, a request for an enhancement to include "issue previews" when publishing links to social media platforms. As a bit of background, the way these platforms work is that they search for meta tags in the posted link's content, and if they find a meta `og:image` (along with other meta tags) they'll pull the image to include in the social media post. Forgejo currently provides an `og:image` tag but it just renders the repository or repository-owner's avatar.
This PR will render `og:image` for an issue or PR into a link to `{...}/summary-card`, which is a dynamically generated image that contains a summary of the issue.
## Design Notes
### Rendering / Rasterization
The tricky part of solving this problem is rendering an image that combines some text, some images, and some layout elements. To address this, I've created a `card` module which allows for a handful of operations:
- Create a new rendered image (a "Card")
- Add a margin to a card
- Split the card, horizontally or vertically, into two pieces with a proportional layout (eg. 70%/30%, as desired), each of which are "Cards" that render into the same root image
- Render text into a card, with line-wrapping and text-alignment capabilities
- Render an image onto a card
- Fetches an external image as safely as possible (for server-side fetch of Gravatar, etc.)
The card module can be reused to create `og:image` summary cards for any object in the future, although obviously it's capabilities are limited. The current implementation is on issues/PRs.
I considered a few alternative approaches before taking this approach, and here's why I rejected those options:
- Provide the summary card as an SVG object which could be rendered much more easily with a template file -- however, support for SVG isn't defined as positive for OpenGraph, and a quick look through some existing implementations suggest that it is not widely supported, if at all
- Rendering as HTML/CSS, or SVG, and then using an external tool to convert into a PNG (or other static) image -- this would be much nicer and easier to implement, but would require tying in some very heavy-weight dependencies
- Rendering using a more sophisticated graphics library, eg. cairo -- also would be nicer and easier to implement, but again a heavy dependency for a small functionality
As a result of the limited capabilities of the new card module, summary cards don't have icons on them (which would require SVG rasterization) or pretty status badges with colors and rounded rects. In the future if better drawing capabilities were added, the graphics could be improved, but it doesn't seem too important.
### External Avatars
In order to rasterize a user's avatar onto the summary card, it might have to be retrieved by the server from the external source (eg. Gravatar). A `fetchExternalImage` routine attempts to do this in the safest way possible to protect the server from any possible security exposure from this; (a) verifying that the content-types are acceptable, (b) ensuring that the file-size and image-size are within the safe bounds that are used for custom avatars, (c) using a very-short timeout to avoid stalling the server if an external dependency is offline.
### Caching
Summary cards are cached after rendered. This has the downside of causing updates to statuses, avatars, titles, etc. being stale on the summary card for the cache TTL. However, during testing I found that some social media engines like Mastodon will cause the summary card to be accessed a significant number of times after being referenced by a post, causing a mini-tornado of requests. The cache compensates for this to avoid server load in this situation.
### Scope
I'm considering out-of-scope:
- Summary cards on other objects (eg. repos, users) can be left for future implementation
## Checklist
The [contributor guide](https://forgejo.org/docs/next/contributor/) contains information that will be helpful to first time contributors. There also are a few [conditions for merging Pull Requests in Forgejo repositories](https://codeberg.org/forgejo/governance/src/branch/main/PullRequestsAgreement.md). You are also welcome to join the [Forgejo development chatroom](https://matrix.to/#/#forgejo-development:matrix.org).
### Tests
- I added test coverage for Go changes...
- [x] in their respective `*_test.go` for unit tests.
- [x] in the `tests/integration` directory if it involves interactions with a live Forgejo server.
- ~~I added test coverage for JavaScript changes...~~ n/a, no JS changes
- [x] ~~in `web_src/js/*.test.js` if it can be unit tested.~~
- [x] ~~in `tests/e2e/*.test.e2e.js` if it requires interactions with a live Forgejo server (see also the [developer guide for JavaScript testing](https://codeberg.org/forgejo/forgejo/src/branch/forgejo/tests/e2e/README.md#end-to-end-tests)).~~
- Manual testing
- [x] Access & attach screenshots of both an issue and a pull-request's summary card; see below
- [x] Ensure reasonable (non-crash) behavior of rendering text with glyphs outside the font -- correctly rendered as replacement unicode chars
- [x] Using a public test instance, verify that og:image behavior looks good on platforms like Mastodon and BlueSky
- [x] Bluesky: ✅
- [x] Mastodon: ✅ (Note that the summary card will be requested many times as the post is federated; either each server, or each client, will fetch it itself)
- [x] OpenGraph test site (https://www.opengraph.xyz/): ✅
- [x] Discord: Looks OK ✅; needs "twitter:card" to be set to "summary_large_image" to display the large-scale image, but (a) that's probably annoying to use, (b) probably wrong because it doesn't match Twitter Card's spec for a "photographic image", and (c) don't want to encourage/continue use of vendor-specific tag
- [x] Verify cases with user avatar missing (or autogen), and repo avatar missing (falls back to repo owner avatar)
Pull request summary card:
Issue summary card:
(images to the right are the custom repo avatar, w/ fallback to the repo owner avatar)
### Documentation
- [ ] I created a pull request [to the documentation](https://codeberg.org/forgejo/docs) to explain to Forgejo users how to use this change.
- [x] I did not document these changes and I do not expect someone else to do it.
- OpenGraph capabilities are expected to work in the background without user awareness, and so there is no need for documentation to explain the capabilities for users.
### Release notes
- [ ] I do not want this change to show in the release notes.
- [x] I want the title to show in the release notes with a link to this pull request.
- [ ] I want the content of the `release-notes/<pull request number>.md` to be be used for the release notes instead of the title.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/6053
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Co-authored-by: Mathieu Fenniak <mathieu@fenniak.net>
Co-committed-by: Mathieu Fenniak <mathieu@fenniak.net>
- This uses a forked version of https://github.com/goccy/go-json, that
has [this pull request](https://github.com/goccy/go-json/pull/490)
applied. It reduces the heap memory usage by 8MiB (idle heap usage from
startup: 40126.59kB -> 32073.56kB). This should be generally safe to
replace as goccy/go-json doesn't see frequent updates and the other user
of this fork is grafana which is another big Go project.
- The only user of this library is minio, but having a configuration
with minio is not a common setup, AFAIK, so this is essentialy wasted
memory for most Forgejo instances. Having it lazy-loaded solves that
problem.
- Use the forked [binding](https://code.forgejo.org/go-chi/binding)
library. This library has two benefits, it removes the usage of
`github.com/goccy/go-json` (has no benefit as the minimo library is also
using it). It adds the `TrimSpace` feature, which will during the
binding part trim the spaces around the value it got from the form, this
is done before validation.
- This is in the spirit of #5090.
- Move to a fork of gitea.com/go-chi/cache,
code.forgejo.org/go-chi/cache. It removes unused code (a lot of
adapters, that can't be used by Forgejo) and unused dependencies (see
go.sum). Also updates existing dependencies.
8c64f1a362..main
- This is a fork of https://github.com/dchest/captcha, as
https://gitea.com/go-chi/captcha is a fork of
github.com/go-macaron/captcha which is a fork (although not properly
credited) of a older version of https://github.com/dchest/captcha. Hence
why I've just forked the original.
- The fork includes some QoL improvements (uses standard library for
determistic RNG instead of rolling your own crypto), and removal of
audio support (500KiB unused data that bloated the binary otherwise).
Flips the image over the x-asis.
47270f2b55..main
- This move is needed for the next commit, because
gitea.com/go-chi/captcha included the gitea.com/go-chi/cache dependency.
- Moves to a fork of gitea.com/go-chi/session that removed support for
couchbase (and ledis, but that was never made available in Forgejo)
along with other code improvements.
f8ce677595..main
- The rationale for removing Couchbase is quite simple. Its not licensed
under FOSS
license (https://www.couchbase.com/blog/couchbase-adopts-bsl-license/)
and therefore cannot be tested by Forgejo and shouldn't be supported.
This is a similair vein to the removal of MSSQL
support (https://codeberg.org/forgejo/discussions/issues/122)
- A additional benefit is that this reduces the Forgejo binary by ~600Kb.
* support changing label colors
* support changing issue state
* use helpers to keep type conversions DRY
* drop the x/exp license because it is no longer used
The tests are performed by the gof3 compliance suite
Support compression for Actions logs to save storage space and
bandwidth. Inspired by
https://github.com/go-gitea/gitea/issues/24256#issuecomment-1521153015
The biggest challenge is that the compression format should support
[seekable](https://github.com/facebook/zstd/blob/dev/contrib/seekable_format/zstd_seekable_compression_format.md).
So when users are viewing a part of the log lines, Gitea doesn't need to
download the whole compressed file and decompress it.
That means gzip cannot help here. And I did research, there aren't too
many choices, like bgzip and xz, but I think zstd is the most popular
one. It has an implementation in Golang with
[zstd](https://github.com/klauspost/compress/tree/master/zstd) and
[zstd-seekable-format-go](https://github.com/SaveTheRbtz/zstd-seekable-format-go),
and what is better is that it has good compatibility: a seekable format
zstd file can be read by a regular zstd reader.
This PR introduces a new package `zstd` to combine and wrap the two
packages, to provide a unified and easy-to-use API.
And a new setting `LOG_COMPRESSION` is added to the config, although I
don't see any reason why not to use compression, I think's it's a good
idea to keep the default with `none` to be consistent with old versions.
`LOG_COMPRESSION` takes effect for only new log files, it adds `.zst` as
an extension to the file name, so Gitea can determine if it needs
decompression according to the file name when reading. Old files will
keep the format since it's not worth converting them, as they will be
cleared after #31735.
<img width="541" alt="image"
src="https://github.com/user-attachments/assets/e9598764-a4e0-4b68-8c2b-f769265183c9">
(cherry picked from commit 33cc5837a655ad544b936d4d040ca36d74092588)
Conflicts:
assets/go-licenses.json
go.mod
go.sum
resolved with make tidy
Earl Warren