How zhimaai/qiweidoc works
Project type and purpose
• From the README (“芝麻会话存档私有化部署”) and folder layout, this is a full-stack, private-cloud chat-archiving solution.
• It ingests real‐time chat content (WeChat/企业微信), stores messages, media, and metadata in a dedicated environment, and exposes both a user-facing management UI and background processing pipelines.
File structure
Top level:
• docker/ – all containers (httpbatch, main, minio, nats, postgres, redis, wxfinance)
• golang/ – Go microservices (httpbatch, master orchestrator, wxfinance bridge)
• php/ – PHP application using RoadRunner + Yii to serve HTTP APIs and run consumers (modules under php/modules/*)
• static/ – front-end assets:
– static/home – marketing site
– static/management – Vue.js SPA for admin
– static/vitepress – documentation site
Key patterns:
• Docker Compose ties everything together.
• Event-driven, plugin-based Go “master” with NATS as message bus.
• PHP back end handles user HTTP API, integrates directly with DB, Redis, and NATS.
• Storage: Postgres (relational data), Redis (cache/state), MinIO (binary/media).
README
• Confirms one-click local deployment, open-source license, high-privacy, self-hosted.
• Use cases: real-time capture (text, media, deletions/recalls), multi-angle views, sensitive-word alerts, statistics.
Detail
• For Go, illustrate plugin sub-components inside the master orchestrator.
• Lifecycle: from HTTP/webhook → NATS → worker → DB/MinIO → back to UI or alert channel.