core.hypermedia Roadmap

Living task list for core.hypermedia.

Current Sprint

• [ ] Set up job queue infrastructure — Redis queues (images, documents, ocr, media, bulk), job claim pattern, visibility timeout


• [ ] Build base worker image — Docker image with PHP CLI, FFmpeg, ImageMagick, Ghostscript, LibreOffice, PaddleOCR


• [ ] Implement job lifecycle — create → queue → claim → process → complete/fail, with retry logic


• [ ] Create processing_profiles table — seed with initial profiles (image-thumbnail, pdf-preview, ocr-standard)

Backlog

Phase 1: Core Processing (VPS only)

• [ ] Image pipeline — thumbnail, compress, convert, multi-size generation


• [ ] Document pipeline — PDF preview, text extraction, page rendering


• [ ] OpenXML pipeline — DOCX/XLSX/PPTX text extraction, LibreOffice conversion


• [ ] Job status API — poll for progress, get outputs


• [ ] Quota integration — credit estimation, charging, refunding via core.auth

Phase 2: OCR

• [ ] Classical OCR integration — PaddleOCR PP-Structure setup


• [ ] Routing logic — feature extraction, threshold-based escalation


• [ ] AI OCR escalation — integration with core.ai for complex pages


• [ ] Per-page routing — mixed classical/AI within same document

Phase 3: Scaling

• [ ] Scheduler implementation — queue monitoring, scaling rules evaluation


• [ ] Cloud provider adapters — AWS, Google, IONOS instance provisioning


• [ ] Spot instance support — for bulk/non-urgent workloads


• [ ] Cost tracking and limits — spend monitoring, hard caps

Phase 4: Media

• [ ] FFmpeg pipeline — video/audio transcode profiles


• [ ] Progress reporting — parse FFmpeg stderr, update job status


• [ ] Checkpointing for long jobs — resume interrupted transcodes

Completed

<!-- Move completed tasks here -->