Fix: Rendering of vnc to not assume png #365
Conversation
|
Code Climate has analyzed commit 1f45dab and detected 0 issues on this pull request. View more on Code Climate. |
| def get_screenshot_path(inhash: str, service: str = "HTTP"): | ||
| ext_map = {"HTTP": ".png", "HTTPS": ".png", "VNC": ".jpg"} | ||
|
|
||
| return f"{inhash[0:2]}/{inhash[2:4]}/{inhash}{ext_map[service]}" |
There was a problem hiding this comment.
I understand this is a short-term fix since you mentioned you were going to add a better strategy, but we shouldn't make assumptions like this in the rendering logic just in case we change formats.
There was a problem hiding this comment.
Agreed, this just brings the rendering logic in line with the logic in screenshot processing:
natlas/natlas-server/app/api/processing/screenshot.py
Lines 29 to 33 in b715caa
With the agent rearchitecture we'll probably lose backwards compatibility and it's the perfect time to make big changes to the data model too for a 0.7.0 release.
Now that agents can actually take screenshots of vnc again, they are being stored as jpg but the (poorly named) hashpath function hardcoded the extension to
.pngeven though.jpgis the known format that vncsnapshot produces.In the future we should update the screenshot stored in elastic to also contain the file extension, that way we could change the code to convert jpg to png or whatever else and the new documents would still be renderable instead of relying on the name of the service to map to extensions.
But this is just fixing the rendering for now. Trying not to make dramatic changes as we approach another tagged release.