# Novya Database Architecture

This document implements the approved scalable database architecture for Novya using a Prisma-ready MySQL model.

## Scope

- Backend target: Node.js + Hono
- Database target: MySQL (InnoDB, utf8mb4, UTC timestamps)
- ORM target: Prisma
- AI provider target: GapGPT via OpenAI-compatible runtime

## Implemented Domains

- Identity and access
- Assessment engine (PCQ-24 ready, versioned)
- Program, lessons, and exercises
- User journey and progress tracking
- AI analysis and lesson-scoped chat
- Notifications, behavior telemetry, and admin snapshots

The concrete model is implemented in [`prisma/schema.prisma`](../prisma/schema.prisma).

## Domain Glossary and Lifecycle States

### Assessment

- `AttemptPhase`: `pre`, `post`
- `AttemptStatus`: `in_progress`, `submitted`, `scored`, `canceled`

### Learning Journey

- `EnrollmentStatus`: `active`, `paused`, `completed`, `dropped`
- `UnlockReason`: `schedule`, `manual`, `prerequisite_completed`, `retry`
- `LessonProgressStatus`: `not_started`, `in_progress`, `completed`, `skipped`
- `SubmissionStatus`: `draft`, `submitted`, `reviewed`

### AI Runtime

- `AiRunType`: `assessment_analysis`, `final_report`, `exercise_feedback`, `lesson_chat`
- `AiRunStatus`: `queued`, `running`, `succeeded`, `failed`, `canceled`
- `ChatRole`: `system`, `user`, `assistant`, `tool`

## Tables by Domain

### Identity and Access

- `organizations`
- `users`
- `auth_identities`
- `auth_sessions`
- `roles`
- `user_roles`
- `user_profiles`
- `consents`
- `user_consents`

### Assessment Domain

- `assessment_scales`
- `assessment_scale_versions`
- `assessment_questions`
- `assessment_options`
- `assessment_attempts`
- `assessment_answers`
- `assessment_scores`
- `assessment_comparisons`

### Program Content

- `programs`
- `program_versions`
- `lessons`
- `lesson_contents`
- `exercises`
- `exercise_prompts`

### User Journey and Progress

- `user_program_enrollments`
- `user_lesson_unlocks`
- `user_lesson_progress`
- `user_exercise_submissions`
- `user_exercise_answers`
- `user_journey_events`
- `daily_activity_summaries`

### AI Runtime

- `ai_providers`
- `ai_models`
- `ai_prompts`
- `ai_analysis_runs`
- `ai_analysis_outputs`
- `chat_sessions`
- `chat_messages`
- `ai_feedback`

### Notifications and Operations

- `notification_templates`
- `notifications`
- `behavior_logs`
- `audit_logs`
- `admin_report_snapshots`

## Relationship Map

- `users 1-1 user_profiles`
- `users 1-N auth_sessions`
- `users N-N roles` via `user_roles`
- `assessment_scales 1-N assessment_scale_versions 1-N assessment_questions 1-N assessment_options`
- `users 1-N assessment_attempts 1-N assessment_answers`
- `assessment_attempts 1-N assessment_scores`
- `assessment_comparisons` references both pre and post attempts
- `programs 1-N program_versions 1-N lessons 1-N exercises 1-N exercise_prompts`
- `users 1-N user_program_enrollments`
- `user_program_enrollments 1-N user_lesson_unlocks` and `1-N user_lesson_progress`
- `chat_sessions` references `(user, enrollment, lesson)` and `chat_messages N-1 chat_sessions`
- `ai_analysis_runs` references `(user, enrollment, provider, model, prompt)` and `ai_analysis_outputs N-1 ai_analysis_runs`
- `ai_feedback` references user and either lesson progress or exercise submission

## ER Diagram (Text)

```mermaid
erDiagram
  users ||--|| user_profiles : has
  users ||--o{ auth_identities : links
  users ||--o{ auth_sessions : owns
  users ||--o{ user_roles : assigned
  roles ||--o{ user_roles : grants

  assessment_scales ||--o{ assessment_scale_versions : versions
  assessment_scale_versions ||--o{ assessment_questions : contains
  assessment_questions ||--o{ assessment_options : offers
  users ||--o{ assessment_attempts : submits
  assessment_attempts ||--o{ assessment_answers : has
  assessment_attempts ||--o{ assessment_scores : produces
  assessment_attempts ||--o| assessment_comparisons : preAttempt
  assessment_attempts ||--o| assessment_comparisons : postAttempt

  programs ||--o{ program_versions : versions
  program_versions ||--o{ lessons : contains
  lessons ||--o{ lesson_contents : renders
  lessons ||--o{ exercises : includes
  exercises ||--o{ exercise_prompts : asks

  users ||--o{ user_program_enrollments : enrolls
  user_program_enrollments ||--o{ user_lesson_unlocks : unlocks
  user_program_enrollments ||--o{ user_lesson_progress : tracks
  user_lesson_progress ||--o{ chat_sessions : scopes
  chat_sessions ||--o{ chat_messages : stores

  user_program_enrollments ||--o{ user_exercise_submissions : submits
  user_exercise_submissions ||--o{ user_exercise_answers : contains

  ai_providers ||--o{ ai_models : exposes
  ai_prompts ||--o{ ai_analysis_runs : usedBy
  ai_models ||--o{ ai_analysis_runs : runsOn
  users ||--o{ ai_analysis_runs : requestedFor
  ai_analysis_runs ||--o{ ai_analysis_outputs : emits
  user_exercise_submissions ||--o{ ai_feedback : receives

  users ||--o{ notifications : receives
  users ||--o{ behavior_logs : generates
  users ||--o{ audit_logs : affects
```

## Implementation Notes

- High-write tables include indexed temporal access paths:
  - `chat_messages(session_id, created_at)`
  - `behavior_logs(user_id, created_at)`
  - `assessment_attempts(user_id, phase, created_at)`
- Content and assessments are versioned to preserve research reproducibility.
- AI output storage is append-only (`ai_analysis_runs`, `ai_analysis_outputs`) to support auditable research workflows.
- The schema keeps tenant-readiness via optional `organization_id`.