README.md

# numpy/numpy — Uncaught tokenize.TokenError in np.load()

![Status: READY TO SUBMIT](https://img.shields.io/badge/status-ready_to_submit-green)
![Severity: High](https://img.shields.io/badge/severity-High-orange)
![CWE-248](https://img.shields.io/badge/CWE-248-red)
![Platform: huntr.com](https://img.shields.io/badge/platform-huntr.com-blue)

## Summary

Loading a crafted 1,214-byte NPY file causes `np.load()` to raise `tokenize.TokenError`
instead of the documented `ValueError`. The exception escapes uncaught from all numpy
handlers, crashing any application that catches only `ValueError` per numpy's documented API.

## Target Info

| Field | Value |
|---|---|
| Repository | numpy/numpy |
| Affected file | `numpy/lib/_format_impl.py` |
| Affected function | `_read_array_header()` line 661 |
| Confirmed version | 2.3.5 (installed), main branch commit b6e8ad8 |
| Python version | 3.12+ (C tokenizer enforces nesting limit) |
| NPY format | v1.0 and v2.0 (NOT v3.0) |
| Platform | huntr.com |

## Root Cause (3 lines)

`_read_array_header()` calls `_filter_header()` when `ast.literal_eval()` raises `SyntaxError`.
For headers with 100+ nested parentheses, `tokenize.generate_tokens()` inside `_filter_header()`
raises `tokenize.TokenError` — which is not a subclass of `ValueError` and is never caught.

```python
# numpy/lib/_format_impl.py line 661
except SyntaxError as e:
    if version <= (2, 0):
        header = _filter_header(header)  # <-- raises TokenError, not caught!
```

## Trigger Path

```
crafted NPY v1.0 (100-level nested descr)
  → np.load()
  → format.read_array()
  → _read_array_header()
  → ast.literal_eval()  raises SyntaxError
  → _filter_header()
  → tokenize.generate_tokens()  raises TokenError
  → [UNCAUGHT] propagates out of np.load()
```

## Reproduction

```bash
# 1. Install numpy (tested on 2.3.5 + Python 3.13)
pip install numpy

# 2. Run PoC
python3 poc_numpy.py

# Expected output:
# [+] CONFIRMED: tokenize.TokenError raised!
# [+] CONFIRMED: TokenError ESCAPED ValueError handler!
# depth=99: ValueError raised (safe)
# depth=100: [+] TokenError raised (VULNERABLE)
```

## PoC File Stats

| Property | Value |
|---|---|
| Malicious file size | 1,214 bytes |
| Header length | 1,147 chars (limit: 10,000) |
| Nesting depth | 100 levels |
| User interaction required | Load the file (implicit in ML pipelines) |

## Suggested Fix

```python
# In _read_array_header(), wrap _filter_header() call:
from tokenize import TokenError as _TokenError

try:
    header = _filter_header(header)
except _TokenError as te:
    raise ValueError(
        f"Cannot parse header (tokenizer error): {header!r}"
    ) from te
```

## Files

| File | Purpose |
|---|---|
| `poc_numpy.py` | Working PoC — generates malicious NPY and demonstrates TokenError escape |
| `report.md` | Full huntr-formatted report |
| `poc-evidence.html` | Self-contained HTML evidence page with terminal output |
| `README.md` | This file |