sample_data.py

"""
Sample data
===========

Seed documents for each role on first launch. Delete `data/` and
`chroma_db/` to re-seed.
"""

from __future__ import annotations

import logging

from vector_store import VectorDocumentStore

logger = logging.getLogger(__name__)


FINANCE_DOCS = [
    {
        "title": "Quarterly Financial Report Q3 2024",
        "content": """
# Q3 2024 Financial Report

## Executive Summary
Q3 2024 revenue reached $5.2M, up 12% QoQ. Net profit grew 24% to $1.4M.

## Key Financial Metrics
- Revenue: $5.2M (up 12% QoQ)
- Operating Expenses: $3.1M (up 5% QoQ)
- Net Profit: $1.4M (up 24% QoQ)
- Cash Reserves: $8.7M

## Department Breakdown
- Sales: $3.2M (up 15%)
- Services: $1.5M (up 8%)
- Licensing: $0.5M (up 5%)

## Projections for Q4
Q4 revenue is projected at $5.8M with continued growth in Sales.
""",
    },
    {
        "title": "2025 Budget Planning Guidelines",
        "content": """
# Budget Planning Guidelines for 2025

## Overview
Process and requirements for departmental budget submissions for FY 2025.

## Timeline
- Budget templates distributed: October 1
- Initial submissions due: October 31
- Review meetings: November 7-18
- Final approvals: December 15

## Budget Constraints
- Total budget increase capped at 8% over 2024
- New headcount limited to critical roles only
- Capital expenditures require ROI analysis for amounts over $25,000

## Required Documentation
1. Completed budget template
2. Headcount justification form (if applicable)
3. Capital expenditure requests with ROI analysis
4. Revenue projections (for revenue-generating departments)

## Approval Process
All budgets require approval from the department head, finance director, and appropriate VP.
""",
    },
]


ENGINEERING_DOCS = [
    {
        "title": "System Architecture Overview",
        "content": """
# System Architecture Overview

## Core Components

1. Frontend Layer
   - React.js SPA
   - Responsive design using Material-UI
   - Client-side state management with Redux

2. API Gateway
   - AWS API Gateway
   - Authentication and request routing
   - Rate limiting and caching

3. Microservices
   - User Service (Node.js)
   - Product Service (Python / Django)
   - Order Service (Java / Spring Boot)
   - Notification Service (Go)

4. Data Layer
   - Primary database: PostgreSQL
   - Caching: Redis
   - Search: Elasticsearch
   - Data warehouse: Snowflake

5. Infrastructure
   - AWS Cloud (primary)
   - Kubernetes for container orchestration
   - CI/CD through GitHub Actions
   - Terraform for infrastructure as code

## Communication Patterns
- Synchronous: REST APIs, gRPC
- Asynchronous: Kafka for event streaming

## Security Measures
- JWT-based authentication
- HTTPS everywhere
- WAF for API endpoints
- Regular security audits
""",
    },
    {
        "title": "Development Workflow Guidelines",
        "content": """
# Development Workflow Guidelines

## Git Workflow
Modified Git Flow with branches:
- main: production code
- develop: integration branch for features
- feature/*: new features
- bugfix/*: bug fixes
- release/*: release candidates
- hotfix/*: production fixes

## Pull Request Process
1. Create a feature/bugfix branch from develop
2. Implement your changes with appropriate tests
3. Open a PR to develop with a clear description, linked issues, passing tests, and coverage met
4. Obtain at least two code reviews
5. Address review comments
6. Merge when approved and CI passes

## Coding Standards
- JavaScript: Airbnb style guide
- Python: PEP 8
- Java: Google Java Style
- Document public APIs
- Write unit tests for all new code
- Maintain minimum 80% code coverage

## Deployment Process
1. Changes merged to develop are auto-deployed to staging
2. QA performs testing in staging
3. Release branch created for production deployment
4. Final testing on pre-production
5. Release manager approves and merges to main
6. CI/CD pipeline deploys to production
""",
    },
]


ADMIN_DOCS = [
    {
        "title": "Admin Onboarding Runbook",
        "content": """
# Admin Onboarding Runbook

## New Admin Checklist
1. Create account with role=admin via the Admin tab
2. Verify access to all role-scoped collections (finance, engineering, admin)
3. Rotate default passwords for seeded accounts
4. Review audit log every Monday

## User Management
- New employee: add_user(username, password, role)
- Role change: delete account and recreate (keeps audit history clean)
- Departure: delete account; sessions auto-expire within the configured window

## Document Management
- Use the Admin tab to upload new documents to the correct role collection
- Document title must be unique within a role; re-uploading replaces in place
- Keep documents under ~8000 characters for best retrieval results

## Incident Response
- Account lockout clears automatically after 10 minutes
- To unlock immediately: edit data/users.json, set failed_attempts=0
- Rate limit spikes: review data/audit.jsonl (if Challenge 2 is implemented)
""",
    },
]


def add_sample_documents(doc_store: VectorDocumentStore) -> None:
    for doc in FINANCE_DOCS:
        doc_store.add_document("finance", doc["title"], doc["content"])
    for doc in ENGINEERING_DOCS:
        doc_store.add_document("engineering", doc["title"], doc["content"])
    for doc in ADMIN_DOCS:
        doc_store.add_document("admin", doc["title"], doc["content"])
    logger.info("Seeded sample documents for all three roles")