Commit Graph
94 Commits
Author SHA1 Message Date
Timothy Jaeryang Baek 6fce92aa12 chore: format 2026-06-01 13:56:55 -07:00
HW bddadafa07 fix(images): pass content_type=None to r.json() to accept non-standard MIME types (#24838)
aiohttp's ClientResponse.json() validates the Content-Type header against
'application/json' by default and raises ContentTypeError for any other
value — including 'application/x-ndjson', which Ollama returns for its
OpenAI-compatible /v1/images/generations endpoint.

Pass content_type=None to skip this check while keeping all other parsing
behaviour unchanged.  The fix covers image generation (openai, gemini,
automatic1111 engines) and image editing (openai, gemini engines).
2026-06-01 10:33:24 -07:00
Timothy Jaeryang Baek 8aa2a42dc7 refac 2026-05-14 02:40:20 +09:00
Timothy Jaeryang Baek 7dcd932ad7 refac 2026-05-14 02:34:19 +09:00
Timothy Jaeryang Baek 6d0295588e refac: modernize type annotations (PEP 604 / PEP 585) 2026-05-12 17:10:15 +09:00
Classic298andbrodmart 2d9939ed49 chore: add validate_url() to get_image_data() for cohort consistency hardening (#24518)
* chore: add validate_url() to get_image_data() for cohort consistency hardening

`get_image_data()` in `backend/open_webui/routers/images.py` fetches the
URL returned by the configured image generation API directly via
`session.get(data)` without first calling `validate_url()`. The sibling
`load_url_image()` in the same file (called from /images/edit) calls
`validate_url(data)` first — that gate was added under
GHSA-jgx9-jr5x-mvpv. The two functions handle structurally identical
input (an attacker-or-server-supplied URL string) and should enforce the
same SSRF gate as a matter of code hygiene.

In the current call graph, the URL passed to `get_image_data()` comes
from the admin-configured image generation API's response, so an
exploitable SSRF chain additionally requires admin-side trust delegation
(misconfigured/untrusted upstream image API, or a custom
OpenAI-compatible server that reflects user input into response URLs).
That makes the missing call a defense-in-depth gap rather than a
vulnerability per SECURITY.md Rule 9 — a position the GHSA-h7cc-wwjp-5xqh
advisory is being closed under. This change is hardening: it brings the
two image-fetch helpers into alignment so any future caller that begins
passing user-influenced URLs into `get_image_data()` is gated by the
same private-IP / loopback / metadata-IP filter the rest of the
codebase enforces.

Surface raised by brodmart in GHSA-h7cc-wwjp-5xqh.

Co-authored-by: brodmart <brodmart@users.noreply.github.com>

* chore: trim comment

---------

Co-authored-by: brodmart <brodmart@users.noreply.github.com>
2026-05-11 00:59:32 +09:00
Timothy Jaeryang Baek 2e71b3fbb8 chore: format 2026-05-09 21:07:08 +09:00
Timothy Jaeryang Baek df42d96c95 refac 2026-05-09 21:05:49 +09:00
Classic298 8854541508 fix: prevent redirect-based SSRF in web-fetch and image-load call sites (#24491)
validate_url() in retrieval/web/utils.py only validates the initial URL.
The HTTP clients used downstream (sync requests, sync requests via the
parent WebBaseLoader._scrape, aiohttp via load_url_image) followed 3xx
redirects by default and did not re-validate the redirect target against
the private-IP / metadata-IP block list. An authenticated user could
submit a public URL that 302-redirected to an internal address (RFC1918,
127.0.0.1, 169.254.169.254, etc.) and the redirected response was returned
to them, enabling SSRF reads of internal services and cloud metadata.

Three call sites needed allow_redirects=False to match the policy already
enforced on the async _fetch() path:

- SafeWebBaseLoader: override requests_kwargs in __init__ so that the
  inherited synchronous _scrape() path passes allow_redirects=False to
  self.session.get() (the parent WebBaseLoader uses requests' default
  allow_redirects=True).
- get_content_from_url (retrieval/utils.py): pass allow_redirects=False
  on the streamed requests.get(...) call.
- load_url_image (routers/images.py, image-edits endpoint): pass
  allow_redirects=False on the aiohttp session.get(...) call.

Reports consolidated under GHSA-rh5x-h6pp-cjj6:
- GHSA-rh5x-h6pp-cjj6 (tenbbughunters / Tenable) - sync _scrape
- GHSA-5vxg-6gmv-m2qr (YLChen-007) - load_url_image
- GHSA-hf76-c83f-63w2 (tempcollab) - aiohttp _fetch (already fixed)
- GHSA-h55f-h5fh-mvm4 (sneaXOR) - get_content_from_url
2026-05-09 21:01:45 +09:00
Timothy Jaeryang Baek f1ef09ddc8 refac 2026-04-17 11:48:17 +09:00
Timothy Jaeryang Baek 128cf41fce refac 2026-04-17 11:12:42 +09:00
Timothy Jaeryang Baek 5dae600ce7 chore: format 2026-04-14 17:27:31 -05:00
Timothy Jaeryang Baek 869cf9e848 refac 2026-04-13 14:33:23 -05:00
Timothy Jaeryang Baek 2ddcb30b9a refac 2026-04-13 14:29:27 -05:00
Timothy Jaeryang Baek de27a12151 refac 2026-04-12 14:39:23 -05:00
Timothy Jaeryang Baek 27169124f2 refac: async db 2026-04-12 14:22:11 -05:00
Skyzi000 09dccdd42f fix: use unique client_id per ComfyUI WebSocket connection (#23592)
Using user.id as client_id causes WebSocket deadlocks when the same
user generates images concurrently (e.g., multi-model chat). ComfyUI
routes messages by clientId, so shared IDs mean only one connection
receives the completion — others hang forever.

Generate a unique UUID per request, matching ComfyUI's own examples.
2026-04-11 14:55:04 -06:00
Timothy Jaeryang Baek ade617efa8 refac 2026-03-24 04:49:48 -05:00
Timothy Jaeryang Baek de3317e26b refac 2026-03-17 17:58:01 -05:00
Timothy Jaeryang Baek f6bd54fb1f refac 2026-02-22 18:55:24 -06:00
Timothy Jaeryang Baek ef036529b5 chore: format 2026-02-17 01:11:56 -06:00
Timothy Jaeryang Baek 9be45f49e4 refac 2026-02-16 14:13:49 -06:00
Timothy Jaeryang Baek f376d4f378 chore: format 2026-02-11 16:24:11 -06:00
Timothy Jaeryang Baek ecf3fa2feb refac 2026-02-03 23:36:15 -06:00
Timothy Jaeryang Baek ce50d9bac4 refac 2026-01-28 01:14:22 +04:00
7. Sun 33020d826f perf: parallelize image loading in image_edits endpoint (#20911)
Use asyncio.gather() to load multiple images concurrently instead of
sequentially, significantly reducing latency for multi-image edit
operations.
2026-01-28 00:35:25 +04:00
G30 4b4743b497 feat: enforce permissions in backend (#20471)
* feat: enforce image generation permissions in backend

* feat: enforce web search permissions in backend

* feat: enforce audio (tts/stt) permissions in backend
2026-01-08 02:48:35 +04:00
Timothy Jaeryang Baek 927a765641 refac 2026-01-06 03:24:08 +04:00
G30 bf42535d12 feat: add support for passing 'steps' in image generation requests (#20337) 2026-01-03 18:11:13 +04:00
Timothy Jaeryang Baek 5d1459df16 refac 2025-12-29 01:20:04 +04:00
Timothy Jaeryang Baek 0a78323406 fix/refac: image action button 2025-12-22 09:56:44 +04:00
Timothy Jaeryang Baek 446cc0ac60 refac 2025-12-22 00:39:05 +04:00
Timothy Jaeryang Baek 4c2e5c93e9 refac: gpt-image support 2025-12-20 17:56:46 +04:00
823b9a6dd9 chore/perf: Remove old SRC level log env vars with no impact (#20045)
* Update openai.py

* Update env.py

* Merge pull request open-webui#19030 from open-webui/dev (#119)

Co-authored-by: Tim Baek <tim@openwebui.com>
Co-authored-by: Claude <noreply@anthropic.com>

---------

Co-authored-by: Tim Baek <tim@openwebui.com>
Co-authored-by: Claude <noreply@anthropic.com>
2025-12-20 08:16:14 -05:00
Alexandr Promakh a7b611c0e5 fix: "No connection adapters were found" routers/images.py (#19435) 2025-11-24 14:51:52 -05:00
Timothy Jaeryang Baek 0c18cd67d5 refac/fix: openai edit multiple images 2025-11-21 04:13:59 -05:00
Timothy Jaeryang Baek 557170c0b6 refac/enh: dedicated enable image edit toggle 2025-11-20 19:52:31 -05:00
Timothy Jaeryang Baek bbd48b3638 enh: images openai api params 2025-11-19 01:04:42 -05:00
Timothy Jaeryang Baek 9370b263f5 refac/fix: automatic1111 params 2025-11-15 15:43:23 -05:00
Timothy Jaeryang Baek 62073d3b7f fix: images edit openai base url/key save issue 2025-11-09 22:36:00 -05:00
Timothy Jaeryang Baek 63e8ab7a05 feat: comfyui image edit support 2025-11-06 03:43:59 -05:00
Timothy Jaeryang Baek 5c1ba23026 refac 2025-11-05 03:42:34 -05:00
Timothy Jaeryang Baek 72f8539fd2 feat: image edit support 2025-11-05 03:31:37 -05:00
Timothy Jaeryang Baek 8d34fcb586 enh: gemini flash image generation support 2025-11-05 01:59:16 -05:00
Timothy Jaeryang Baek 75efa4f931 refac 2025-11-04 17:58:51 -05:00
Timothy Jaeryang Baek 8e5690aab4 refac: images 2025-11-04 13:30:59 -05:00
Timothy Jaeryang Baek 199f371f58 refac 2025-09-24 07:17:41 -05:00
Andreas Fuerer df0d29c81c style: fix formatting issues 2025-09-02 19:16:02 +02:00
Andreas Fuerer bc6afc9057 feature: Azure OpenAI image generation support
The image generation API used on Azure OpenAI requires to specify the API
version by appending an `api-version` query parameter to the endpoint URL.
Added the environment variable `IMAGES_OPENAI_API_VERSION` with
configuration functionality in the administration UI.
2025-09-02 15:51:45 +02:00
Timothy Jaeryang Baek 37a3de0703 fix 2025-08-22 17:19:57 +04:00