AutoFix — Python Static & Runtime Analysis Tool

AutoFix is a command-line tool that accelerates debugging and test failure triage by combining AST-based static analysis and runtime traceback parsing. It scans your Python project, indexes all functions/classes for O(1) lookups, parses Pytest failures, and applies pattern-matching logic to automatically suggest likely fixes.

---

🚀 Features

• AST-based Symbol Indexing
  Parses all .py files to build an in-memory index of functions, classes, methods, and attributes for O(1) lookups.

• Runtime Traceback Parsing
  Extracts structured frame data and exception summaries from raw pytest output or plain Python tracebacks.

• AI-like Suggestion Engine
  Applies heuristic and pattern-based rules to detect common issues and suggest targeted fixes.

• Automatic Pytest Integration
  Runs pytest, captures failures, parses tracebacks, and ranks possible solutions—all in one command.

• Extensible Rule Engine
  Add your own fix patterns for custom project needs. Rules are modular and isolated for easy extension.

• Human & Machine Readable Reports
  Outputs clean human-readable text reports or structured JSON for integration with CI/CD dashboards.

---

📦 Installation

You can clone and use directly:

git clone https://github.com/yourusername/autofix.git
cd autofix
python3 autofix.py --help

No dependencies beyond Python’s standard library (requires Python 3.8+). You only need Pytest installed and accessible on your PATH.

---

⚙️ Usage

1. Analyze Codebase (Static Index)

Build an AST index for your project:

python autofix.py analyze src/ tests/

This scans your code and prints how many functions, classes, and modules were indexed.

2. Run Pytest + Auto Analysis

Run tests, parse traceback output, and get fix suggestions in one go:

python autofix.py run --pytest-args "-q -x" src/ tests/

Example output:

================================================================================
Failure #1: NameError: name 'DataFram' is not defined
Frames:
  • src/utils/data_loader.py:22 in load_dataset
Suggestions (ranked):
  - Add missing import for 'DataFrame'  [score=0.90]
      fix: from pandas import DataFrame

You can also export structured results for machine consumption:

python autofix.py run --json results.json src/ tests/

3. Analyze Saved Traceback Files

If you have saved a traceback or pytest output to a file (e.g., tb.txt):

python autofix.py suggest --traceback tb.txt src/

Export JSON output:

python autofix.py suggest --traceback tb.txt src/ --json suggestions.json

---

🧠 How It Works

🔍 1. Static Indexing via AST

AutoFix walks your source tree and parses .py files using the ast module. It collects:

• Function and method signatures (args, defaults, keywords)
• Class names, attributes, and methods
• Module-level imports

These are stored in dictionaries for O(1) lookups, enabling near-instant fix suggestions.

⚙️ 2. Runtime Traceback Parsing

AutoFix reads your pytest output or saved traceback text and extracts:

• Frame info: file, line number, function name
• Exception type & message

🤖 3. Rule-Based Suggestion Engine

Each rule is a small heuristic that matches specific error patterns. For example:

Rule Name	Target Error	Suggestion
NameErrorMissingImport	NameError	Suggests imports for known symbols
AttributeErrorMaybeTypo	AttributeError	Detects likely typos (via Levenshtein distance)
TypeErrorWrongArity	TypeError	Points to mismatched function signatures
ImportErrorModule	ImportError	Proposes pip install or relative import fix
KeyErrorDict	KeyError	Recommends .get() or membership checks
IndexErrorBound	IndexError	Suggests bound checking
AssertionErrorEquality	AssertionError	Reminds to use pytest.approx for floats

---

🧩 Example Workflow

1. Run your tests:

   pytest -q --tb=short > tb.txt

2. Feed traceback into AutoFix:

   python autofix.py suggest -t tb.txt src/

3. Get clear, actionable suggestions in seconds.

---

🧱 Architecture Overview

+------------------------+
|   AutoFix CLI (argparse)   |
+------------------------+
           |
           v
+-----------------------+
|  SymbolIndex (AST)    |
|  → functions/classes  |
+-----------------------+
           |
           v
+-----------------------+
|  TracebackParser      |
|  → extract frames & errors |
+-----------------------+
           |
           v
+-----------------------+
|  RuleEngine           |
|  → apply fix rules    |
+-----------------------+
           |
           v
+-----------------------+
|  Reporter (text/json) |
+-----------------------+

---

🧰 Extending AutoFix

You can add new rules easily:

def rule_custom(self, tb: ParsedTraceback):
    if 'ConnectionError' in tb.etype:
        return [Suggestion(title='Network Issue', detail='Check API connectivity.')]

Then register it inside RuleEngine.__init__().

---

📊 Example JSON Output

{
  "failures": [
    {
      "case": 1,
      "etype": "NameError",
      "message": "name 'DatFram' is not defined",
      "suggestions": [
        {
          "title": "Add missing import for 'DataFrame'",
          "detail": "`DataFrame` exists elsewhere. Consider importing it.",
          "fixes": ["from pandas import DataFrame"],
          "score": 0.9
        }
      ]
    }
  ]
}

---

🧩 Integration Ideas

• CI/CD pipelines: Automatically parse test failures and upload suggestions to build logs.
• IDE plugins: Hook into linting or test runners to suggest quick-fixes in real time.
• LLM copilots: Feed --json output into an LLM to generate automated PR patches.

---

🧑‍💻 Requirements

• Python 3.8+
• pytest installed on PATH

---

---

🤝 Contributions

Pull requests are welcome — feel free to extend the rule set or improve the parsing accuracy.
If you build something cool on top of AutoFix (e.g., a VSCode integration), let’s connect!

---