Commit graph

10 commits

Author SHA1 Message Date
Richard R
cd530a365d refactor(whisper): migrate default ONNX model references to q4 variant
Update all configuration, manifest, and documentation references to use the q4
Whisper ONNX model files instead of int8. Adjust expected file names, hashes,
and environment variable descriptions to reflect this new default. This aligns
runtime, deployment, and developer documentation with the updated model artifact
expectations for improved consistency.
2026-05-26 19:22:07 -06:00
Richard R
12c586d7be refactor(core,server): migrate contracts to api-contracts and modularize job handlers
Move shared type contracts from `contracts` to new `api-contracts` module in `compute/core` for clearer API boundaries. Remove legacy `contracts`, `pdf`, and `whisper` index files. Update all imports to use `api-contracts`. Refactor server job logic by replacing `parsePdfJob.ts` with modular `user-pdf-layout-job.ts` and `user-whisper-align-job.ts`, updating all relevant API routes and compute integration. This improves maintainability and separation of concerns across compute and server layers.
2026-05-24 11:39:25 -06:00
Richard R
ad82aede73 feat(worker,whisper): improve logging for operation lifecycle and remove align mutex
Enhance worker operation handling by adding detailed logging for operation
acceptance, reuse, replacement, and stale detection, including opKey hashes and
reasons for replacement. Remove the alignMutex from Whisper alignment logic to
simplify concurrency control and reduce unnecessary serialization of alignment
requests. These changes improve observability and throughput for alignment and
layout parsing jobs.
2026-05-22 12:49:33 -06:00
Richard R
50c4330ab1 refactor(core,config,tests): update ONNX model manifest loading and test imports
- Replace static JSON imports with runtime manifest loading in pdf and whisper model modules for compatibility with ESM and bundlers
- Refactor Next.js config to clarify compute mode logic and improve worker bundling conditions
- Update test imports to use package entrypoints instead of relative paths
- Remove redundant whisper alignment/model tests now covered elsewhere or by integration
- No breaking changes to public API or model handling logic
2026-05-21 23:49:56 -06:00
Richard R
55acf05ad6 refactor(compute-core): rename pdf and whisper modules to clean capability layout 2026-05-21 21:41:37 -06:00
Richard R
3a2c6ffa42 refactor(compute-core): split runtime into config/platform and move contracts barrel 2026-05-21 21:40:16 -06:00
Richard R
37d11bf9b8 refactor(compute): remove legacy app whisper duplicate and unify core whisper boundary 2026-05-21 21:15:10 -06:00
Richard R
661af71fa4 refactor(whisper): implement dynamic timeout configuration for alignment and decoding
Replace static timeout constants with functions that derive alignment and ffmpeg
decode timeouts from compute timeout config. This enables adaptive timeout
behavior based on runtime configuration, improving robustness under varying
resource constraints. Update error handling and internal APIs to propagate
timeout values accordingly.

Remove unused Node.js engine constraint from package.json for broader compatibility.
2026-05-21 19:29:03 -06:00
Richard R
f8672b073e refactor(compute): unify job concurrency and CPU thread allocation logic
Standardize compute job concurrency configuration by introducing a shared
COMPUTE_JOB_CONCURRENCY setting, replacing separate PDF and Whisper concurrency
controls. Add cross-platform CPU core detection and thread budgeting utilities,
and update both ONNX model execution and concurrency limiters to use dynamic
thread allocation per job. Refactor environment variable docs and examples to
reflect unified concurrency management. Streamline PDF.js font path resolution
for improved reliability across environments.
2026-05-21 15:58:29 -06:00
Richard R
f1aa1c3e3b feat(compute): add external compute worker backend and integration
Introduce support for external compute worker mode (`COMPUTE_MODE=worker`) using a new `WorkerComputeBackend`. This enables offloading heavy ONNX Whisper alignment and PDF layout parsing to a standalone worker service (Redis + BullMQ), improving scalability and compatibility with serverless/limited environments.

- Add `@openreader/compute-core` as a shared package for ONNX inference and PDF parsing logic.
- Implement `WorkerComputeBackend` and worker contract/types for remote job execution.
- Update compute backend selection logic and remove previous worker mode guards.
- Extend `WhisperAlignInput` and `PdfLayoutInput` types to support object keys for remote data access.
- Refactor local compute backend to use `@openreader/compute-core` and support both buffer and object key inputs.
- Update job runner, TTS segment alignment, and PDF layout parsing flows to use new compute backend APIs.
- Add scripts, Docker workflow, and documentation for deploying and running the compute worker.
- Update environment variable docs and examples for worker mode, including storage requirements and configuration.
- Document published images and stack changes to reflect the new compute worker architecture.

BREAKING CHANGE: `COMPUTE_MODE=worker` now requires an external compute worker service and S3-compatible object storage. Embedded SeaweedFS (`weed mini`) is not supported in worker mode. See the new documentation for deployment and configuration details.
2026-05-19 15:21:25 -06:00