Code Generation Tools

Code Generation Tools

Implementing code generation tools — from template-based scaffolding through AST manipulation, DSL parsing, and AI-assisted code generation.

When to Use

• Auto-generating boilerplate code for repetitive patterns
• Building project scaffolding generators
• Creating domain-specific language (DSL) compilers
• Implementing code transformation and migration tools

Code Generation Patterns

import ast, json

class CodeGenerator:
    """Generate code from templates and models."""
    
    def generate_from_template(self, template: str, variables: Dict) -> str:
        for k, v in variables.items():
            template = template.replace(f'{{{{ {k} }}}}', v)
        return template
    
    def generate_class_from_schema(self, schema: Dict) -> str:
        imports = []
        fields = []
        for field_name, field_type in schema.get('fields', {}).items():
            fields.append(f"    {field_name}: {field_type}")
            if '=' in field_type:
                import_name = field_type.split('=')[0].strip()
                if import_name not in imports:
                    imports.append(import_name)
        
        code = 'from dataclasses import dataclass\n'
        if imports: code += f'from typing import {", ".join(set(imports))}\n'
        code += f'\n\n@dataclass\nclass {schema["name"]}:\n'
        code += '\n'.join(fields) if fields else '    pass'
        return code

Verification Checklist

• Template variables properly escaped and substituted
• Generated code is syntactically valid
• Code formatting matches project standards
• Generators are deterministic (same input → same output)
• Error handling for missing variables or invalid schemas