Adds the necessary configuration and the `npm run bench` script, as well
as the first benchmark to verify everything is working. The first
benchmark is permissionUtils. This is important to benchmark because
permission strings are used often and there is some non-trivial logic
such as escaping of full-colon characters. Other overheads like database
access will be much more significant, but it may be useful to know how
much more significant in quantifiable terms.
* Revert "fix: together ai image prices (#2130)"
This reverts commit 04c8b06616.
* Revert "feat: single image ai endpoint (#2129)"
This reverts commit 161ca74aa3.
* feat: refactor ai to have single entry point and follow defined model structure
* fix: missing openrouter registration
* fix: dedupe models
* feat: provide usage in stream mode 🚀
We weren't able to see any custom spans, only the builtin
instrumentations for node.js. This commit downgrades opentelemetry so
that we can see custom spans in tools like jaeger as before.
- A new stt driver in `AIInterfaceService` and `PuterAIModule`.
- added methods for audio transcription and translation in the speech-to-text interface.
- updated cost mapping for stt models in `openAiCostMap.ts`.
- Updated permissions and interfaces to support new speech-to-text features.
* Update OpenAI package to version 6.7.0 and add video generation capabilities
- Updated `openai` dependency in `package-lock.json` and `package.json`.
- Introduced new permissions for video generation in `hardcoded-permissions.js`.
- Added video generation service in `AIInterfaceService.js` and registered it in `PuterAIModule.js`.
- Implemented coercion for video streams in `CoercionService.js`.
- Updated driver service to include video generation.
- Added `txt2video` method and options in `AI.js` and corresponding TypeScript definitions in `index.d.ts`.
* Update openaiVideoCostMap.ts
* Revert "Update openaiVideoCostMap.ts"
This reverts commit 9e66f93605.
* change `txt2video` to `txt2vid`
This commit implements uploading Puter files through ai.chat using
Claude's beta "file inputs" feature. To use this, simply specify
`puter_path` in a content part, and that content part will be
transformed into the correct type of content part for the specified file
and the file will be uploaded.
* experimental Cloudflare Workers for Platforms support
* add support for the puter driver
* stop hardcoding api.puter.localhost
* support destroy from express route as well
* initial worker support
* xhrshim + fixes (incomplete)
* change order of readyState + load event
* remove some debug logs
* change worker/puterUtils into a cjs module
* Cloudflare workers eventtarget workaround
* worker preamble webpack
* edit worker readme to reflect reality
* allow a way to code in api endpoint instead of hardcoding it to api.puter.com
* move cloudflare eventtarget fix to puter-portable template
This is so it gets run before the rest of puter-js initializes
* remove express route for worker
Backend had tests written in mocha as well as a test written for jest
(created by an AI utility, wasn't compatible with mocha tests). The AI
generated test had merit so it was migrated to vitest along with all the
mocha tests, which supports conventions from both frameworks.
Vitest also has an excellent watcher cli.
The root repo package.json now runs unit tests in backend. Before it was
only running tests defined in the _test method of backend services.
This adds an experimental custom token authentication endpoint to Puter.
This is disabled unless the firebase-auth service is explicitly
configured, and is not yet ready for production use.
* Added Revis distributed cash to enhance our Captcha Verification system so that we prevent our system from replay attacks
* Fix: There was an error with the implementation of Redis, so I reverted to our previous version that uses in memory storage
* Integrated the captcha verification system into our sign in Form. The captcha verification system now works on both login and sign int
* Remove test files from captcha module
* Update src/backend/src/modules/captcha/middleware/captcha-middleware.js
Co-authored-by: Eric Dubé <eric.alex.dube@gmail.com>
* Update src/backend/src/modules/captcha/middleware/captcha-middleware.js
Co-authored-by: Eric Dubé <eric.alex.dube@gmail.com>
* Now the captcha can be requested on condition, this llaows extenstions to control wether a captcha should be required,
I fixed the code in CaptchaModule to use config
and got rid of the lines that made captcha middleware available since it wasn't used anywhre
* I split the middleware into two distinct parts, so that the frontend can now determine captach requirements. PuterHomePageService can set GUI parameters for captcha requirements. The /whoarewe endpoint provides captcha requirement information and the extensuo system integration is maintained
* Fix security issues with password handling in URL query parameters
* Made sure that the enter key, submits the login request instead of refreshing the captcha
* In development we can now disable the Captcha verification system by running it with CAPTCHA_ENABLED=false npm start
* Went back and modified checkCaptcha so that it checks at the start to check what CAPTCHA_ENABLED is equal to
* Refactor captcha system to use configuration values instead of environment variables
* Fix captcha verification and align with project standards
* Update src/backend/src/modules/captcha/README.md
Co-authored-by: Eric Dubé <eric.alex.dube@gmail.com>
* fix: incorrect service name
* dev: use Endpoint for captcha endpoints
Use Endpoint class, which uses eggspress behind the scenes, which handles
async errors in handlers automatically.
* dev: add extension support and simplify captcha
- removed extra error handling
- removed dormant code
- no distinction between login and signup (for now)
* clean: remove local files
* fix: undefined edge case
---------
Co-authored-by: Eric Dubé <eric.alex.dube@gmail.com>
There were a few things which made this rather difficult:
- AWS SDK doesn't appear to have a utility to validate message
signatures.
- The only available node.js module that does this is known to be
unreliable.
- SNS sends 'text/plain' MIME type even though the data is in JSON
format, so a middleware had to be added to account for this.
- We don't accept POST requests with no Origin header. Since SNS doesn't
send the Origin header, an exception had to be made for this.
- The endpoint needs to be public and SNS doesn't seem to have a
proxying utility for developers such as what Stripe has.
- Because of the above point, debugging time was affected by deployment
time to the staging server.
Neal Shah