kailash-dataflow 0.5.1

Workflow-native database framework for Kailash SDK

Kailash DataFlow

Multi-Database Alpha Framework - Django simplicity meets enterprise-grade production quality with PostgreSQL and SQLite support.

⚠️ ALPHA RELEASE: DataFlow supports PostgreSQL (full features) and SQLite (near-complete parity). MySQL support coming in beta.

✅ v0.4.7+ Context-Aware Improvements: String IDs preserved, multi-instance isolation, deferred schema operations

🚀 Quick Start

Prerequisites

• PostgreSQL 12+ (recommended for production) OR SQLite 3.x (development/testing)
• Python 3.8+

Installation

pip install kailash-dataflow
# or
pip install kailash[dataflow]

Basic Usage

from dataflow import DataFlow

# PostgreSQL (production) or SQLite (development)
db = DataFlow("postgresql://user:pass@localhost/dbname")
# db = DataFlow("sqlite:///app.db")  # SQLite alternative

# Define your model
@db.model
class User:
    id: str  # String IDs now preserved! (v0.4.7+)
    name: str
    email: str

# DataFlow automatically creates:
# ✅ Database schema with migrations (PostgreSQL)
# ✅ 9 workflow nodes per model (CRUD + bulk ops)
# ✅ Real SQL operations with injection protection
# ✅ Connection pooling and transaction management
# ✅ MongoDB-style query builder
# ✅ Concurrent access protection with locking
# ✅ Schema state management with rollback

🎯 What Makes DataFlow Different?

Multi-Database Support

# Production PostgreSQL
db = DataFlow("postgresql://user:pass@localhost/dbname")

# Development SQLite
db = DataFlow("sqlite:///app.db")

# Environment-based configuration
# DATABASE_URL=postgresql://... or sqlite:///...
db = DataFlow()  # Reads from DATABASE_URL

# Advanced features (both databases)
db = DataFlow(
    "postgresql://...",  # or "sqlite:///..."
    pool_size=50,
    auto_migrate=True,
    monitoring=True
)

Real Database Operations (Currently Available)

# Traditional ORMs: Imperative code
User.objects.create(name="Alice")  # Django
user = User(name="Alice"); session.add(user)  # SQLAlchemy

# DataFlow: Workflow-native database operations
workflow = WorkflowBuilder()
workflow.add_node("UserCreateNode", "create_user", {
    "name": "Alice",
    "email": "[e-mail]"
})
workflow.add_node("UserListNode", "find_users", {
    "limit": 10,
    "offset": 0
})

# Real SQL is executed: INSERT INTO users (name, email) VALUES ($1, $2)

MongoDB-Style Query Builder (NEW!)

# Get QueryBuilder from any model
builder = User.query_builder()

# MongoDB-style operators
builder.where("age", "$gte", 18)
builder.where("status", "$in", ["active", "premium"])
builder.where("email", "$regex", "^[a-z]+@company\.com$")
builder.order_by("created_at", "DESC")
builder.limit(10)

# Generates optimized SQL for your database
sql, params = builder.build_select()
# PostgreSQL: SELECT * FROM "users" WHERE "age" >= $1 AND "status" IN ($2, $3) AND "email" ~ $4 ORDER BY "created_at" DESC LIMIT 10

# Works seamlessly with ListNode
workflow.add_node("UserListNode", "search", {
    "filter": {
        "age": {"$gte": 18},
        "status": {"$in": ["active", "premium"]},
        "email": {"$regex": "^admin"}
    }
})

Database Support Status

# PostgreSQL: Full feature support
db = DataFlow(database_url="postgresql://user:pass@localhost/db")

# SQLite: Near-complete parity (missing only schema discovery)
db = DataFlow(database_url="sqlite:///app.db")

# Both support full workflow execution
runtime = LocalRuntime()
results, run_id = runtime.execute(workflow.build())  # ✅ Works with both databases

# Only limitation: Real schema discovery (PostgreSQL only)
schema = db.discover_schema(use_real_inspection=True)  # PostgreSQL only

Database Operations as Workflow Nodes

# Traditional ORMs: Imperative code
user = User.objects.create(name="Alice")  # Django
user = User(name="Alice"); session.add(user)  # SQLAlchemy

# DataFlow: Workflow-native (9 nodes per model!)
workflow = WorkflowBuilder()
workflow.add_node("UserCreateNode", "create_user", {
    "name": "Alice",
    "email": "[e-mail]"
})
workflow.add_node("UserListNode", "find_users", {
    "filter": {"name": {"$like": "A%"}}
})

Enterprise Configuration

# Multi-tenancy configuration (query modification planned)
db = DataFlow(multi_tenant=True)

# Real SQL generation with security
db = DataFlow(
    database_url="postgresql://user:pass@localhost/db",
    pool_size=20,
    pool_max_overflow=30,
    monitoring=True,
    echo=False  # No SQL logging in production
)

# All generated nodes use parameterized queries for security
# INSERT INTO users (name, email) VALUES ($1, $2)  -- Safe from SQL injection

🔧 Context-Aware Improvements (v0.4.7+)

String ID Preservation

# String IDs are now preserved without forced integer conversion
@db.model
class Session:
    id: str  # Explicitly string - no more PostgreSQL type errors!
    user_id: str
    token: str

# String IDs work correctly in all operations
workflow.add_node("SessionCreateNode", "create", {
    "id": "sess-uuid-12345",  # Preserved as string
    "user_id": "user-uuid-67890",
    "token": "token-abc-def"
})

Multi-Instance Isolation

# Each DataFlow instance maintains separate context
dev_db = DataFlow("sqlite:///dev.db")
prod_db = DataFlow("postgresql://prod...")

# Models registered to specific instances
@dev_db.model
class User:
    name: str

@prod_db.model
class User:  # Same name, different instance - works!
    name: str
    email: str

# Nodes bound to correct instance automatically

Deferred Schema Operations

• Synchronous registration: Models register immediately with @db.model
• Async table creation: Tables created on first use, not registration
• Better performance: No blocking during model definition phase

🚦 Implementation Status

✅ Currently Available (Production-Ready)

• Database Schema Generation: Complete CREATE TABLE for PostgreSQL, MySQL, SQLite
• Auto-Migration System: PostgreSQL-only, production-ready automatic schema synchronization
• Real Database Operations: All 9 CRUD + bulk nodes execute actual SQL
• SQL Security: Parameterized queries prevent SQL injection
• Connection Management: Connection pooling, DDL execution, error handling
• Workflow Integration: Full compatibility with WorkflowBuilder/LocalRuntime
• Configuration System: Zero-config to enterprise patterns
• MongoDB-Style Query Builder: Complete with all operators ($eq, $gt, $in, $regex, etc.)
• Concurrent Access Protection: Migration locking and atomic operations
• Schema State Management: Change detection, caching, and rollback capabilities
• String ID Preservation: String/UUID IDs preserved without forced conversion (v0.4.7+)
• Multi-Instance Isolation: Separate contexts for different DataFlow instances (v0.4.7+)
• Deferred Schema Operations: Better performance with lazy table creation (v0.4.7+)

⚠️ ALPHA LIMITATIONS

• Schema Discovery: Real database introspection (discover_schema(use_real_inspection=True)) is PostgreSQL-only
• MySQL Support: Not available in alpha release
• Complex Migrations: Some SQLite migration operations limited by ALTER TABLE syntax
• Production Use: Alpha software - thorough testing recommended for production deployments

🔄 Planned Features (Roadmap)

• Redis Query Caching: User.cached_query() with automatic invalidation
• Multi-Database Runtime: SQLite/MySQL execution support
• Advanced Multi-Tenancy: Automatic query modification for tenant isolation

📚 Documentation

Getting Started

• 5-Minute Tutorial - Build your first app
• Core Concepts - Understand DataFlow
• Examples - Complete applications

Development

• Models - Define your schema
• CRUD Operations - Basic operations
• Relationships - Model associations

Production

• Deployment - Go to production
• Performance - Optimization guide
• Monitoring - Observability

💡 Real-World Examples

E-Commerce Platform

# Define your models
@db.model
class Product:
    id: int
    name: str
    price: float
    stock: int

@db.model
class Order:
    id: int
    user_id: int
    total: float
    status: str

# Use in workflows
workflow = WorkflowBuilder()

# Check inventory
workflow.add_node("ProductGetNode", "check_stock", {
    "id": "{product_id}"
})

# Create order with transaction
workflow.add_node("TransactionContextNode", "tx_start")
workflow.add_node("OrderCreateNode", "create_order", {
    "user_id": "{user_id}",
    "total": "{total}"
})
workflow.add_node("ProductUpdateNode", "update_stock", {
    "id": "{product_id}",
    "stock": "{new_stock}"
})

Multi-Tenant SaaS (Current Implementation)

# Enable multi-tenancy configuration
db = DataFlow(
    database_url="postgresql://user:pass@localhost/db",
    multi_tenant=True
)

# Multi-tenant models get tenant_id field automatically
@db.model
class User:
    name: str
    email: str
    # tenant_id: str automatically added

# Use in workflows with real database operations
workflow.add_node("UserCreateNode", "create_user", {
    "name": "Alice",
    "email": "[e-mail]"
})
workflow.add_node("UserListNode", "list_users", {
    "limit": 10,
    "filter": {}
})

High-Performance ETL (Current Implementation)

# Bulk operations with real database execution
workflow.add_node("UserBulkCreateNode", "import_users", {
    "data": users_data,  # List of user records
    "batch_size": 1000,
    "conflict_resolution": "skip"
})

# Real bulk INSERT operations executed
# Uses parameterized queries for security
# Processes data in configurable batches

# List operations with filters
workflow.add_node("UserListNode", "active_users", {
    "limit": 1000,
    "offset": 0,
    "order_by": ["created_at"],
    "filter": {"active": True}
})

🏗️ Architecture

DataFlow seamlessly integrates with Kailash's workflow architecture:

┌─────────────────────────────────────────────────────┐
│                 Your Application                     │
├─────────────────────────────────────────────────────┤
│                    DataFlow                          │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐         │
│  │  Models  │  │   Nodes  │  │ Migrations│         │
│  └────┬─────┘  └────┬─────┘  └────┬─────┘         │
│       └──────────────┴──────────────┘               │
│                Core Features                         │
│  QueryBuilder │ QueryCache │ Monitoring │ Multi-tenant │
│  ┌──────────┐  ┌──────────┐  ┌──────────┐         │
│  │MongoDB-  │  │Redis     │  │Pattern   │         │
│  │style     │  │Caching   │  │Invalidate│         │
│  └──────────┘  └──────────┘  └──────────┘         │
├─────────────────────────────────────────────────────┤
│               Kailash SDK                           │
│         Workflows │ Nodes │ Runtime                 │
└─────────────────────────────────────────────────────┘

🧪 Testing

DataFlow includes comprehensive testing support:

# Test with in-memory database
def test_user_creation():
    db = DataFlow(testing=True)

    @db.model
    class User:
        id: int
        name: str

    # Automatic test isolation
    user = db.test_create(User, name="Test User")
    assert user.name == "Test User"

🤝 Contributing

We welcome contributions! DataFlow follows Kailash SDK patterns:

1. Use SDK components and patterns
2. Maintain zero-config philosophy
3. Write comprehensive tests
4. Update documentation

See CONTRIBUTING.md for details.

📊 Performance & Testing Status

Current Performance (PostgreSQL Alpha)

• Real SQL execution with parameterized queries
• Connection pooling with configurable pool sizes
• Bulk operations with batching for large datasets
• 95% unit test pass rate (615/648 tests passing)

Recent Test Improvements

• 100% NO MOCKING compliance in Tier 2-3 tests
• Real infrastructure testing with PostgreSQL
• 167 test files covering all scenarios
• 3-tier testing strategy (Unit/Integration/E2E)
• Fixed critical bugs: checksum tracking, field type serialization

Alpha Testing Requirements

• PostgreSQL 12+ required for all testing
• Performance benchmarks available for PostgreSQL only
• Advanced caching and query optimization features planned for beta

⚡ Why DataFlow?

• Real Database Operations: Actual SQL execution, not mocks
• Workflow-Native: Database ops as first-class nodes
• Production-Ready: PostgreSQL support with connection pooling
• Progressive: Simple to start, enterprise features available
• 100% Kailash: Built on proven SDK components

---

Built with Kailash SDK | Parent Project | SDK Docs