llm-json-fixer 1.0.1

A simple utility to fix malformed JSON from LLM outputs

JSON Fixer

A simple, lightweight Python utility to fix malformed JSON strings commonly generated by Large Language Models (LLMs) like ChatGPT, Claude, and others.

🚀 Features

• Zero Dependencies: Uses only Python standard library
• Simple API: One main function to fix JSON
• Handles Common Issues: Fixes quotes, brackets, commas, Python booleans, and more
• Extracts JSON: Can extract JSON from surrounding text
• Lightweight: Less than 200 lines of code
• Fast: Optimized for performance

📦 Installation

pip install llm_json_fixer

🔧 Usage

Basic Usage

from llm_json_fixer import fix_json

# Fix single quotes
result = fix_json("{'name': 'John', 'age': 30}")
print(result)  # {'name': 'John', 'age': 30}

# Fix trailing commas
result = fix_json('{"name": "John", "age": 30,}')
print(result)  # {'name': 'John', 'age': 30}

# Fix Python booleans
result = fix_json('{"active": True, "data": None}')
print(result)  # {'active': True, 'data': None}

# Extract JSON from text
result = fix_json('Here is the JSON: {"name": "John"} and that\'s it.')
print(result)  # {'name': 'John'}

Detailed Usage

from llm_json_fixer import fix_json

# Get detailed information about fixes applied
result = fix_json('{"name": "John", "age": 30,}', return_dict=True)
print(result)
# {
#     'success': True,
#     'data': {'name': 'John', 'age': 30},
#     'error': None,
#     'fixes_applied': ['Applied common JSON fixes']
# }

Validation

from llm_json_fixer import is_valid_json

print(is_valid_json('{"name": "John"}'))  # True
print(is_valid_json("{'name': 'John'}"))  # False

🛠️ What It Fixes

• Single quotes → Double quotes
• Trailing commas → Removed
• Missing commas → Added
• Unquoted keys → Quoted
• Python booleans (True, False, None) → JSON equivalents
• Unbalanced brackets → Balanced
• Markdown code blocks → Removed
• Multiple commas → Single commas
• Unclosed strings → Closed
• Extracts JSON from surrounding text

📋 Examples

Common LLM JSON Issues

from llm_json_fixer import fix_json

# Single quotes (Python style)
fix_json("{'name': 'John', 'city': 'New York'}")
# → {'name': 'John', 'city': 'New York'}

# Trailing commas
fix_json('{"name": "John", "age": 30,}')
# → {'name': 'John', 'age': 30}

# Unquoted keys
fix_json('{name: "John", age: 30}')
# → {'name': 'John', 'age': 30}

# Python booleans
fix_json('{"active": True, "deleted": False, "data": None}')
# → {'active': True, 'deleted': False, 'data': None}

# Missing closing bracket
fix_json('{"name": "John", "age": 30')
# → {'name': 'John', 'age': 30}

# JSON in markdown
fix_json('```json\n{"name": "John"}\n```')
# → {'name': 'John'}

# JSON with surrounding text
fix_json('The result is: {"name": "John", "age": 30} as you can see.')
# → {'name': 'John', 'age': 30}

🔄 Return Values

Simple Mode (default)

result = fix_json(json_string)
# Returns: Parsed JSON object or None if fixing failed

Detailed Mode

result = fix_json(json_string, return_dict=True)
# Returns: {
#     'success': bool,      # Whether fixing was successful
#     'data': Any,          # Parsed JSON object (None if failed)
#     'error': str,         # Error message (None if successful)
#     'fixes_applied': list # List of fixes that were applied
# }

🎯 Use Cases

• LLM Integration: Clean up JSON responses from ChatGPT, Claude, etc.
• API Development: Handle malformed JSON from various sources
• Data Processing: Fix JSON in log files or data dumps
• Web Scraping: Clean up JSON extracted from web pages
• Configuration Files: Fix manually edited JSON configs

📄 License

MIT License - see LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📝 Changelog

1.0.1

• Initial release
• Basic JSON fixing functionality
• Support for common LLM JSON issues
• Zero dependencies
• Simple API

🆘 Support

If you encounter any issues or have questions, please file an issue on the GitHub repository.