string-schema 0.1.4

A simple, LLM-friendly schema definition library for converting string syntax to structured schemas

String Schema

A simple, LLM-friendly schema definition library for Python that converts intuitive string syntax into structured data schemas.

🚀 Quick Start

Installation

pip install string-schema

30-Second Example

from string_schema import string_to_json_schema

# Define a schema with simple, readable syntax
schema = string_to_json_schema("name:string, email:email, age:int?")

# Get a complete JSON Schema ready for validation
print(schema)
# {
#   "type": "object",
#   "properties": {
#     "name": {"type": "string"},
#     "email": {"type": "string", "format": "email"},
#     "age": {"type": "integer"}
#   },
#   "required": ["name", "email"]
# }

Create Pydantic Models Directly

from string_schema import string_to_model

# Create Pydantic model from string syntax
UserModel = string_to_model("name:string, email:email, age:int?")

# Use immediately
user = UserModel(name="Alice", email="[e-mail]")
print(user.model_dump_json())  # {"name": "Alice", "email": "[e-mail]"}

🎯 What String Schema Does

String Schema takes human-readable text descriptions and converts them into structured schemas for data validation, extraction, and API documentation. Perfect for LLM data extraction, API development, and configuration validation.

Input: Human-readable string syntax Output: JSON Schema, Pydantic models, or OpenAPI specifications

🚀 Core Functions & Use Cases

📝 Schema Conversion Matrix

Forward Conversions (Source → Target):

Function	Input	Output	Use Case
string_to_json_schema()	String syntax	JSON Schema dict	Main conversion - string to JSON Schema
string_to_model()	String syntax	Pydantic model class	Direct path - string to Pydantic model
string_to_model_code()	String syntax	Python code string	Code generation - for templates
string_to_openapi()	String syntax	OpenAPI schema dict	Direct path - string to OpenAPI
json_schema_to_model()	JSON Schema dict	Pydantic model class	When you already have JSON Schema
json_schema_to_openapi()	JSON Schema dict	OpenAPI schema dict	When you already have JSON Schema

Reverse Conversions (Target → Source):

Function	Input	Output	Use Case
model_to_string()	Pydantic model	String syntax	Schema introspection - model to string
model_to_json_schema()	Pydantic model	JSON Schema dict	Export - model to JSON Schema
json_schema_to_string()	JSON Schema dict	String syntax	Migration - JSON Schema to string
openapi_to_string()	OpenAPI schema	String syntax	Import - OpenAPI to string
openapi_to_json_schema()	OpenAPI schema	JSON Schema dict	Conversion - OpenAPI to JSON Schema

🔍 Data Validation Functions

Function	Input	Output	Use Case
validate_to_dict()	Data + schema	Validated dict	API responses - clean dicts
validate_to_model()	Data + schema	Pydantic model	Business logic - typed models
validate_string_syntax()	String syntax	Validation result	Check syntax and get feedback

🎨 Function Decorators

Decorator	Purpose	Returns	Use Case
@returns_dict()	Auto-validate to dict	Validated dict	API endpoints
@returns_model()	Auto-validate to model	Pydantic model	Business logic

🔧 Utility Functions

Function	Purpose	Returns	Use Case
get_model_info()	Model introspection	Model details dict	Debugging & analysis
validate_schema_compatibility()	Schema validation	Compatibility info	Schema validation

🎯 Key Scenarios

• 🤖 LLM Data Extraction: Define extraction schemas that LLMs can easily follow
• 🔧 API Development: Generate Pydantic models and OpenAPI docs from simple syntax
• ✅ Data Validation: Create robust validation schemas with minimal code
• 📋 Configuration: Define and validate application configuration schemas
• 🔄 Data Transformation: Convert between different schema formats

✅ Validate Data

from string_schema import validate_to_dict, validate_to_model

# Example raw data (from API, user input, etc.)
raw_data = {
    "name": "John Doe",
    "email": "[e-mail]",
    "age": "25",  # String that needs conversion
    "extra_field": "ignored"  # Will be filtered out
}

# Validate to clean dictionaries (perfect for API responses)
user_dict = validate_to_dict(raw_data, "name:string, email:email, age:int?")
print(user_dict)  # {"name": "John Doe", "email": "[e-mail]", "age": 25}

# Validate to typed models (perfect for business logic)
user_model = validate_to_model(raw_data, "name:string, email:email, age:int?")
print(user_model.name)  # "John Doe" - Full type safety
print(user_model.age)   # 25 - Converted to int

🎨 Function Decorators

from string_schema import returns_dict, returns_model
import uuid

# Auto-validate function returns to dicts
@returns_dict("id:string, name:string, active:bool")
def create_user(name):
    # Input: "Alice"
    # Function returns unvalidated dict
    return {"id": str(uuid.uuid4()), "name": name, "active": True, "extra": "ignored"}
    # Output: {"id": "123e4567-...", "name": "Alice", "active": True}
    # Note: 'extra' field filtered out, types validated

# Auto-validate function returns to models
@returns_model("name:string, email:string")
def process_user(raw_input):
    # Input: {"name": "Bob", "email": "[e-mail]", "junk": "data"}
    # Function returns unvalidated dict
    return {"name": raw_input["name"], "email": raw_input["email"], "junk": "data"}
    # Output: UserModel(name="Bob", email="[e-mail]")
    # Note: Returns typed Pydantic model, 'junk' field filtered out

# Usage examples:
user_dict = create_user("Alice")  # Returns validated dict
user_model = process_user({"name": "Bob", "email": "[e-mail]"})  # Returns Pydantic model
print(user_model.name)  # "Bob" - Full type safety

🌐 FastAPI Integration

from string_schema import string_to_model, returns_dict

# Create models for FastAPI
UserRequest = string_to_model("name:string, email:email")

@app.post("/users")
@returns_dict("id:int, name:string, email:string")
def create_user_endpoint(user: UserRequest):
    return {"id": 123, "name": user.name, "email": user.email}

Features: Arrays [{name:string}], nested objects {profile:{bio:text?}}, enums, constraints, decorators.

📖 Complete Documentation

🎓 More Examples

🌱 Simple Example - Basic User Data

from string_schema import string_to_json_schema

# Define schema using intuitive string syntax
schema = string_to_json_schema("""
name:string(min=1, max=100),
email:email,
age:int(0, 120)?,
active:bool
""")

# Result: JSON Schema dictionary ready for validation
print(schema)
# {
#   "type": "object",
#   "properties": {
#     "name": {"type": "string", "minLength": 1, "maxLength": 100},
#     "email": {"type": "string", "format": "email"},
#     "age": {"type": "integer", "minimum": 0, "maximum": 120},
#     "active": {"type": "boolean"}
#   },
#   "required": ["name", "email", "active"]
# }

🌳 Moderately Complex Example - Product Data

# Product with arrays and nested objects
schema = string_to_json_schema("""
{
    id:uuid,
    name:string,
    price:number(min=0),
    tags:[string]?,
    inventory:{
        in_stock:bool,
        quantity:int?
    }
}
""")

# Result: Complete JSON Schema with nested objects and arrays
print(schema)  # Full JSON Schema ready for validation

📚 For more examples and advanced syntax, see our detailed documentation

🔄 Output Formats & Results

🐍 Pydantic Models (Python Classes)

from string_schema import string_to_model

# Direct conversion: String syntax → Pydantic model
UserModel = string_to_model("name:string, email:email, active:bool")

# Use the model for validation
user = UserModel(name="John Doe", email="[e-mail]", active=True)
print(user.model_dump_json())  # {"name": "John Doe", "email": "[e-mail]", "active": true}

🔧 Code Generation (For Templates & Tools)

from string_schema import string_to_model_code

# Generate Pydantic model code as a string
code = string_to_model_code("User", "name:string, email:email, active:bool")
print(code)

# Output:
# from pydantic import BaseModel
# from typing import Optional
#
# class User(BaseModel):
#     name: str
#     email: str
#     active: bool

# Perfect for code generators, templates, or saving to files
with open('models.py', 'w') as f:
    f.write(code)

🌐 OpenAPI Schemas (API Documentation)

from string_schema import string_to_openapi

# Direct conversion: String syntax → OpenAPI schema
openapi_schema = string_to_openapi("name:string, email:email")
print(openapi_schema)
# {
#   "type": "object",
#   "properties": {
#     "name": {"type": "string"},
#     "email": {"type": "string", "format": "email"}
#   },
#   "required": ["name", "email"]
# }

🔄 Reverse Conversions (Universal Schema Converter)

String Schema provides complete bidirectional conversion between all schema formats!

⚠️ Information Loss Notice: Reverse conversions (from JSON Schema/OpenAPI/Pydantic back to string syntax) may lose some information due to format differences. However, the resulting schemas are designed to cover the most common use cases and maintain functional equivalence for typical validation scenarios.

🔍 Schema Introspection

from string_schema import model_to_string, string_to_model

# Create a model first
UserModel = string_to_model("name:string, email:email, active:bool")

# Reverse engineer it back to string syntax
schema_string = model_to_string(UserModel)
print(f"Model schema: {schema_string}")
# Output: "name:string, email:email, active:bool"

📦 Migration & Import

from string_schema import json_schema_to_string

# Example: Convert existing JSON Schema to String Schema syntax
json_schema = {
    "type": "object",
    "properties": {
        "name": {"type": "string"},
        "email": {"type": "string", "format": "email"},
        "age": {"type": "integer"}
    },
    "required": ["name", "email"]
}

simple_syntax = json_schema_to_string(json_schema)
print(f"Converted: {simple_syntax}")
# Output: "name:string, email:email, age:int?"

🔧 Schema Comparison & Analysis

from string_schema import string_to_model, model_to_string

# Create two versions of a model
UserV1 = string_to_model("name:string, email:email")
UserV2 = string_to_model("name:string, email:email, active:bool")

# Compare them
v1_str = model_to_string(UserV1)
v2_str = model_to_string(UserV2)

print("Schema changes:")
print(f"V1: {v1_str}")  # "name:string, email:email"
print(f"V2: {v2_str}")  # "name:string, email:email, active:bool"

🎨 String Syntax Reference

Basic Types

• string, int, number, bool → Basic data types
• email, url, datetime, date, uuid, phone → Special validated types

Field Modifiers

• field_name:type → Required field
• field_name:type? → Optional field
• field_name:type(constraints) → Field with validation

Common Patterns

• string(min=1, max=100) → Length constraints
• int(0, 120) → Range constraints
• [string] → Simple arrays
• [{name:string, email:email}] → Object arrays
• status:enum(active, inactive) → Enum values
• id:string|uuid → Union types

📖 Complete syntax guide: See docs/string-syntax.md for full reference

✅ Validation

from string_schema import validate_string_syntax

# Validate your schema syntax
result = validate_string_syntax("name:string, email:email, age:int?")

print(f"Valid: {result['valid']}")  # True
print(f"Features used: {result['features_used']}")  # ['basic_types', 'optional_fields']
print(f"Field count: {len(result['parsed_fields'])}")  # 3

# Example with invalid syntax
bad_result = validate_string_syntax("name:invalid_type")
print(f"Valid: {bad_result['valid']}")  # False
print(f"Errors: {bad_result['errors']}")  # ['Unknown type: invalid_type']

🏗️ Common Use Cases

🤖 LLM Data Extraction

from string_schema import string_to_json_schema

# Define what data to extract
schema = string_to_json_schema("company:string, employees:[{name:string, email:email}], founded:int?")
# Use schema in LLM prompts for structured data extraction

🔧 FastAPI Development

from string_schema import string_to_model

# Create model for FastAPI
UserModel = string_to_model("name:string, email:email")

# Use in FastAPI endpoints
@app.post("/users/")
async def create_user(user: UserModel):
    return {"id": 123, "name": user.name, "email": user.email}

🏗️ Code Generation & Templates

from string_schema import string_to_model_code

# Generate model code
code = string_to_model_code("User", "name:string, email:email")
print(code)
# Output: Complete Pydantic model class as string

# Save to file
with open('user_model.py', 'w') as f:
    f.write(code)

📋 Configuration Validation

from string_schema import string_to_json_schema

# Validate app configuration
config_schema = string_to_json_schema("database:{host:string, port:int}, debug:bool")
# Use for validating config files

📚 Documentation

• Getting Started - Quick start guide and installation
• String Syntax Guide - Complete syntax reference
• API Reference - Full API documentation
• Examples - Practical examples and patterns

📋 Example Schemas

Ready-to-use schema examples are available in the examples/ directory:

# Import example schemas if needed
from string_schema.examples.presets import user_schema, product_schema
from string_schema.examples.recipes import create_ecommerce_product_schema

# Or better yet, use string syntax directly:
user_schema = string_to_json_schema("name:string, email:email, age:int?")

🧪 Testing

The library includes comprehensive tests covering all functionality:

# Run tests (requires pytest)
pip install pytest
pytest tests/

# Results: 66 tests passed, 3 skipped (Pydantic tests when not installed)

🤝 Contributing

Contributions are welcome! The codebase is well-organized and documented:

string_schema/
├── core/          # Core functionality (fields, builders, validators)
├── parsing/       # String parsing and syntax
├── integrations/  # Pydantic, JSON Schema, OpenAPI
└── examples/      # Built-in schemas and recipes

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

String Schema - Making data validation simple, intuitive, and LLM-friendly! 🚀