# LLM Static Assert
LLM Static Assert is a powerful Python library that leverages Language Models (LLMs) to perform static assertions on code. This innovative approach combines the flexibility of natural language with the rigor of static analysis, offering a unique solution for code quality and correctness verification.
## Why LLM Static Assert?
Traditional static analysis tools often struggle with complex, context-dependent code properties. LLM Static Assert bridges this gap by utilizing the advanced reasoning capabilities of large language models to evaluate code against natural language expectations.
### Benefits
1. **Natural Language Assertions**: Express complex code properties and expectations using plain English, making it easier to write and understand assertions.
2. **Flexibility**: Evaluate a wide range of code properties that might be challenging for traditional static analysis tools.
3. **Context-Aware**: LLMs can consider broader context and nuanced relationships within the code.
4. **Customizable**: Adjust the assertion process with options like quorum size and model selection.
### Drawbacks
1. **Computational Overhead**: LLM inference can be more resource-intensive than traditional static analysis.
2. **Potential for Ambiguity**: Natural language assertions may sometimes lead to ambiguous interpretations.
3. **Dependence on LLM Quality**: The effectiveness of assertions relies on the capabilities of the underlying language model.
## When to Use LLM Static Assert
- When you need to verify complex, context-dependent code properties.
- For catching subtle logical errors that might escape traditional static analysis.
- In projects where code correctness is critical, and you want an additional layer of verification.
- When you want to express code expectations in a more natural, readable format.
## When Not to Use LLM Static Assert
- For simple, straightforward assertions that can be easily handled by traditional unit tests or type checkers.
- In environments where computational resources are severely constrained.
- When you need deterministic, 100% reproducible results (due to the potential variability in LLM outputs).
## Installation
You can install LLM Static Assert using pip:
```bash
pip install llm-static-assert
```
Or if you're using Poetry:
```bash
poetry add llm-static-assert
```
## Usage
Here's a basic example of how to use LLM Static Assert:
```python
from llm_static_assert import LLMStaticAssert, LLMStaticAssertOptions
# Create an instance of LLMStaticAssert with custom options
options = LLMStaticAssertOptions(quorum_size=3, model="gpt-4o-mini")
lsa = LLMStaticAssert(options)
# Define a class or function you want to assert about
class MyClass:
def some_method(self):
pass
# Perform a static assertion
lsa.static_assert(
"{class} should have a method named 'some_method'",
{"class": MyClass}
)
```
In this example, we're asserting that `MyClass` has a method named `some_method`. The LLM will analyze the provided class and evaluate whether it meets this expectation.
### Advanced Usage
You can customize the assertion process by adjusting the `LLMStaticAssertOptions`:
```python
options = LLMStaticAssertOptions(
quorum_size=5, # Perform 5 inferences and use majority voting
model="gpt-4" # Use a more advanced model for complex assertions
)
lsa = LLMStaticAssert(options)
# You can also provide custom options for a specific assertion
lsa.static_assert(
"{class} should implement proper error handling in all methods",
{"class": MyComplexClass},
options=LLMStaticAssertOptions(quorum_size=7, model="gpt-4")
)
```
## Contributing
Contributions to LLM Static Assert are welcome! Please refer to the project's issues page on GitHub for areas where help is needed or to suggest improvements.
## License
[MIT License](LICENSE)
## Acknowledgements
This project uses [LiteLLM](https://github.com/BerriAI/litellm) for LLM integration.
---
LLM Static Assert is an experimental tool and should be used in conjunction with, not as a replacement for, traditional testing and static analysis methods. Always validate its results and use it responsibly.
Raw data
{
"_id": null,
"home_page": "https://github.com/kdunee/llm-static-assert",
"name": "llm-static-assert",
"maintainer": null,
"docs_url": null,
"requires_python": "<4.0,>=3.10",
"maintainer_email": null,
"keywords": "llm, static-analysis, assertions, code-quality, natural-language",
"author": "Kosma Dunikowski",
"author_email": "kosmadunikowski@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/d5/80/8b197296287cf3a6ab1fddadbbbf87165ea3f74c16450222db0fff4224d5/llm_static_assert-0.1.0.tar.gz",
"platform": null,
"description": "# LLM Static Assert\n\nLLM Static Assert is a powerful Python library that leverages Language Models (LLMs) to perform static assertions on code. This innovative approach combines the flexibility of natural language with the rigor of static analysis, offering a unique solution for code quality and correctness verification.\n\n## Why LLM Static Assert?\n\nTraditional static analysis tools often struggle with complex, context-dependent code properties. LLM Static Assert bridges this gap by utilizing the advanced reasoning capabilities of large language models to evaluate code against natural language expectations.\n\n### Benefits\n\n1. **Natural Language Assertions**: Express complex code properties and expectations using plain English, making it easier to write and understand assertions.\n2. **Flexibility**: Evaluate a wide range of code properties that might be challenging for traditional static analysis tools.\n3. **Context-Aware**: LLMs can consider broader context and nuanced relationships within the code.\n4. **Customizable**: Adjust the assertion process with options like quorum size and model selection.\n\n### Drawbacks\n\n1. **Computational Overhead**: LLM inference can be more resource-intensive than traditional static analysis.\n2. **Potential for Ambiguity**: Natural language assertions may sometimes lead to ambiguous interpretations.\n3. **Dependence on LLM Quality**: The effectiveness of assertions relies on the capabilities of the underlying language model.\n\n## When to Use LLM Static Assert\n\n- When you need to verify complex, context-dependent code properties.\n- For catching subtle logical errors that might escape traditional static analysis.\n- In projects where code correctness is critical, and you want an additional layer of verification.\n- When you want to express code expectations in a more natural, readable format.\n\n## When Not to Use LLM Static Assert\n\n- For simple, straightforward assertions that can be easily handled by traditional unit tests or type checkers.\n- In environments where computational resources are severely constrained.\n- When you need deterministic, 100% reproducible results (due to the potential variability in LLM outputs).\n\n## Installation\n\nYou can install LLM Static Assert using pip:\n\n```bash\npip install llm-static-assert\n```\n\nOr if you're using Poetry:\n\n```bash\npoetry add llm-static-assert\n```\n\n## Usage\n\nHere's a basic example of how to use LLM Static Assert:\n\n```python\nfrom llm_static_assert import LLMStaticAssert, LLMStaticAssertOptions\n\n# Create an instance of LLMStaticAssert with custom options\noptions = LLMStaticAssertOptions(quorum_size=3, model=\"gpt-4o-mini\")\nlsa = LLMStaticAssert(options)\n\n# Define a class or function you want to assert about\nclass MyClass:\n def some_method(self):\n pass\n\n# Perform a static assertion\nlsa.static_assert(\n \"{class} should have a method named 'some_method'\",\n {\"class\": MyClass}\n)\n```\n\nIn this example, we're asserting that `MyClass` has a method named `some_method`. The LLM will analyze the provided class and evaluate whether it meets this expectation.\n\n### Advanced Usage\n\nYou can customize the assertion process by adjusting the `LLMStaticAssertOptions`:\n\n```python\noptions = LLMStaticAssertOptions(\n quorum_size=5, # Perform 5 inferences and use majority voting\n model=\"gpt-4\" # Use a more advanced model for complex assertions\n)\nlsa = LLMStaticAssert(options)\n\n# You can also provide custom options for a specific assertion\nlsa.static_assert(\n \"{class} should implement proper error handling in all methods\",\n {\"class\": MyComplexClass},\n options=LLMStaticAssertOptions(quorum_size=7, model=\"gpt-4\")\n)\n```\n\n## Contributing\n\nContributions to LLM Static Assert are welcome! Please refer to the project's issues page on GitHub for areas where help is needed or to suggest improvements.\n\n## License\n\n[MIT License](LICENSE)\n\n## Acknowledgements\n\nThis project uses [LiteLLM](https://github.com/BerriAI/litellm) for LLM integration.\n\n---\n\nLLM Static Assert is an experimental tool and should be used in conjunction with, not as a replacement for, traditional testing and static analysis methods. Always validate its results and use it responsibly.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "A Python library that leverages Language Models (LLMs) to perform static assertions on code, combining natural language flexibility with static analysis rigor.",
"version": "0.1.0",
"project_urls": {
"Homepage": "https://github.com/kdunee/llm-static-assert"
},
"split_keywords": [
"llm",
" static-analysis",
" assertions",
" code-quality",
" natural-language"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "c8824d62b64592fa866829c3410df67ba4f4ad208867d88f52a00f0ab584870c",
"md5": "7a239b8c86eabd463eb0b611a2cb1948",
"sha256": "fbd5d9c982c5ee57e225cd88e9d98c4d972097057ca5c37ca8abd5d0941f8b74"
},
"downloads": -1,
"filename": "llm_static_assert-0.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "7a239b8c86eabd463eb0b611a2cb1948",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<4.0,>=3.10",
"size": 7256,
"upload_time": "2024-10-16T17:39:09",
"upload_time_iso_8601": "2024-10-16T17:39:09.556819Z",
"url": "https://files.pythonhosted.org/packages/c8/82/4d62b64592fa866829c3410df67ba4f4ad208867d88f52a00f0ab584870c/llm_static_assert-0.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "d5808b197296287cf3a6ab1fddadbbbf87165ea3f74c16450222db0fff4224d5",
"md5": "f3131e25f7fd243f6166f8ca505df5fb",
"sha256": "6b995542ccf3d1e79dd2d1fc1542d886cc35827c73625697ec090f1d05b0b463"
},
"downloads": -1,
"filename": "llm_static_assert-0.1.0.tar.gz",
"has_sig": false,
"md5_digest": "f3131e25f7fd243f6166f8ca505df5fb",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<4.0,>=3.10",
"size": 5851,
"upload_time": "2024-10-16T17:39:12",
"upload_time_iso_8601": "2024-10-16T17:39:12.489955Z",
"url": "https://files.pythonhosted.org/packages/d5/80/8b197296287cf3a6ab1fddadbbbf87165ea3f74c16450222db0fff4224d5/llm_static_assert-0.1.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-10-16 17:39:12",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "kdunee",
"github_project": "llm-static-assert",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "llm-static-assert"
}
JSON
