PHI-servis (Python / FastAPI)

• Bu nima: ~/development/phi umbrella papkasi bo'lib, Healthcare CRM platform ichida himoyalangan tibbiy ma'lumotlar (PHI) bilan ishlaydigan yagona healthtech mikroservisning ikki implementatsiyasini o'z ichiga oladi (Saudiya/KSA — Vault vt.ksa.healthcare-platform.com manzilida):
• phi-service (Python/FastAPI) — dastlabki implementatsiya (bu fayl asosan shunga bag'ishlangan).
• phi-elysia (TypeScript/Elysia/Bun) — migratsiya/qayta yozish (MIGRATION_PLAN.md bilan); alohida task #45'da yoritilgan.
• Turi / holati / roli: api (tibbiy mikroservis) · active · lead — foydalanuvchi asosiy muallif (phi-service ~79 kommit, phi-elysia 45); jamoa: Alisher Mukhtorov (phi-service'da ~28), Anna Shevtsova/aniashev, Harsh Manwani.
• Faollik davri: phi-service 2025-09-18 → 2026-01-12 (109 kommit); phi-elysia 2026-01-05 → 2026-02-12 (57). Python→Bun/Elysia stek migratsiyasi bilan faol rivojlanish.

• phi-service (Python): Python 3.12+, FastAPI + uvicorn, SQLAlchemy 2.0 async + asyncpg + Alembic, Postgres, kalitlar uchun cryptography + HashiCorp Vault (hvac), python-jose (JWT), Temporal (temporalio — durable vorkflowlar), MinIO (object storage), HL7 (hl7), PDF lab-hisobotlarini parsing (pymupdf, pdf2image, Pillow), Groq/OpenAI (AI talqin), structlog + python-json-logger + prometheus-client, slowapi (rate-limit), tenacity (retry), httpx. UV (paket menejeri), Docker (+ Dockerfile.worker), Makefile, pytest+xdist+cov, ruff/mypy/bandit.
• phi-elysia (TypeScript, → task #45): Elysia (Bun) + @elysiajs/{jwt,bearer,cors,swagger,opentelemetry}, Temporal (client/worker/workflow) + Inngest (event-driven), Drizzle ORM, MinIO, jose, AWS Secrets Manager, OpenTelemetry + LogTape, Biome.
• Infra/deploy: Docker Compose (dev/prod/inngest variantlari), alohida worker konteyneri, alembic migratsiyalar (Python) / drizzle (TS), shifrlash kalitlari uchun Vault/Secrets Manager.
• Ma'lumotlar: PostgreSQL (shifrlangan tibbiy ma'lumotlar), MinIO (lab-hisobot fayllari/PDF'lar), Vault (kalitlar). PII Service (psevdonimlashtirish) va Healthcare CRM platform bilan integratsiya.

• Xavfsiz PHI saqlash: ikki bosqichli shifrlash (AES-256-GCM + HMAC-SHA256), Vault orqali kalitlarni boshqarish, psevdonimlashtirish (PHI↔PII bog'lanishi alohida PII Service orqali).
• Lab natijalari: PDF natijalarini parsing (PyMuPDF/pdf2image), HL7 almashinuvi, src/labs/{converters,workflows,repositories,services,rest} ichidagi konvertorlar/vorkflowlar.
• AI talqin: Groq/OpenAI orqali versiyalangan natija talqinlari.
• Audit: tibbiy ma'lumotlar ustidagi to'liq kirish jurnali.
• Domen modullari: labs, medical_profile, health_metrics, chat, ai, partner, authentication, security, external, facades (Onion Architecture).
• Python→Elysia migratsiyasi (phi-elysia): Bun/Elysia + Temporal + Inngest'ga port qilish, test rejalari (TESTING_PLAN*.md) va IMPLEMENTATION_SUMMARY bilan.
• Hajmi: ~5 oy davomida 166 ta jami kommit (~124 tasi foydalanuvchidan).

README/CLAUDE/pyproject/struktura bilan tasdiqlangan:

• Tibbiy ma'lumotlarni ikki bosqichli shifrlash + Vault (src/security/): AES-256-GCM (maxfiylik) + HMAC-SHA256 (yaxlitlik/haqiqiylik), kalitlar env'dan emas, HashiCorp Vault'dan olinadi. → Tibbiy compliance uchun jiddiy kriptografiya.
• PHI/PII ajratish orqali psevdonimlashtirish (src/external/, PII Service'ga bog'lanish): tibbiy ma'lumotlar identifikatsiya ma'lumotlaridan alohida saqlanadi va psevdonimlar orqali bog'lanadi — bu health-privacy patterni (HIPAA/GDPR darajasidagi). → Puxta privacy arxitekturasi.
• Onion Architecture (REST → Facade → Domain Service → Repository): aniq qatlamlash, domenlararo facadelar, o'zining modellari/schemalari/dtolari/repozitoriyalari/servislari bilan domen modullari. → Katta servis uchun yetuk struktura.
• HL7 + lab natijalarini AI talqin qilish (src/labs, src/ai): PDF qabul → parsing (PyMuPDF) → HL7 → versiyalangan AI talqin (Groq/OpenAI). → Nozik tibbiy ma'lumot pipeline'i.
• Temporal'da durable vorkflowlar (ikkala implementatsiyada ham): uzoq davom etadigan lab-natija qayta ishlashi Temporal vorkflowlari sifatida (alohida worker konteyneri) — nosozliklarga chidamlilik. phi-elysia event-driven ishlov berish uchun Inngest qo'shadi.
• Production darajasidagi montaj: Prometheus metrikalari, structlog (JSON loglar), slowapi rate-limit, tenacity retrylar, bandit security skanerlash, mypy strict, xdist/coverage bilan pytest. → Operatsion va sifat yetukligi.

• Topilgan sessiyalar: 2 ta papka — biri phi-service, biri phi-elysia uchun (subdir shakli; ikkala implementatsiya ham Claude Code bilan qurilgan). .claude/ umbrella papkada ham mavjud.
• AI bilan nima qilingan: CLAUDE.md hajmi (phi-service'da 314 qator!) + migratsiya/test rejalari (phi-elysia'da MIGRATION_PLAN, TESTING_PLAN, IMPLEMENTATION_SUMMARY) AI yordamidagi ishlab chiqish va spec-first jarayonga ishora qiladi, ayniqsa Python→Elysia qayta yozish davrida.
• AI vorkflow patternlari: hujjatga yo'naltirilgan ishlab chiqish (batafsil CLAUDE.md agent qo'llanmasi sifatida), AI yordamida stek migratsiyasi, bosqichli test rejalari.

• Bir xil tibbiy mikroservisning 2 ta to'liq implementatsiyasi (Python + TypeScript), stek migratsiyasi.
• ~10 ta domen moduli (labs, medical_profile, health_metrics, chat, ai, security, external, authentication, partner, facades).
• Tibbiy darajadagi kriptografiya (AES-256-GCM + HMAC, Vault), HL7, audit.
• ~5 oy davomida 166 kommit, 3–4 kishilik jamoa, foydalanuvchi asosiy muallif.
• To'liq CI toolchain (ruff/mypy/bandit/pytest-cov), Prometheus/structlog observability.