atabet-token 1.1.0

Token validation utilities for Atabet systems

atabet-token

Token validation utilities packaged with Cython.

Installation

Prerequisites

1. Conda environment (recommended):

   conda env create -f environment.yml
   conda activate atabet_token_env
   

2. C++ Compiler (required for Windows):

   • MinGW (MSYS2) - Recommended if you have MSYS2 installed
   • Microsoft C++ Build Tools - Alternative option

Quick Start

Using MinGW (if you have MSYS2):

conda activate atabet_token_env

# Configure MinGW (one-time setup)
python configure_mingw.py

# Build and install
python setup.py build_ext --inplace --compiler=mingw32
pip install -e .[dev]

Or use the automated script:

build_with_mingw.bat

Using MSVC (default):

conda activate atabet_token_env
pip install -e .[dev]

Run Tests

pytest

GitHub Actions: wheels and PyPI

Workflow: .github/workflows/build-wheels.yml. Wheels are built with cibuildwheel; options live in pyproject.toml under [tool.cibuildwheel].

When jobs run (push): the Linux / Windows / macOS matrix runs only if the head commit message includes one of:

Tag	Effect
[build ci] (or [Build CI], [BUILD CI])	Build wheels and upload per-OS artifacts; no publish.
[pub pypi] (or [Pub PyPI], [PUB PYPI])	Build wheels, then publish merged wheels to PyPI (production).
[pub test.pypi] (or [Pub test.pypi], [PUB TEST.PYPI])	Build wheels, then publish to TestPyPI.

Plain pushes (no tag above) skip the matrix so CI does not run on every commit.

Manual runs: Actions → Build wheels → Run workflow runs the build matrix only; publishing is by push with [pub pypi] or [pub test.pypi] in the commit message.

Authentication (publish): the workflow uses Trusted Publishing (OIDC) by default — no API token in GitHub secrets. Register the GitHub repo and workflow file on each index (project Manage → Publishing). See the comment block at the top of build-wheels.yml for OIDC vs API-token options.

Private repositories are supported for OIDC the same as public ones.

Troubleshooting

MinGW Compilation Issues

SIZEOF_VOID_P Enum Error (Fixed)

Problem: When compiling with MinGW, you may encounter:

error: enumerator value for '__pyx_check_sizeof_voidp' is not an integer constant

Root Cause: MinGW's GCC is stricter about constant expressions in enum initializers. The Cython-generated code includes:

enum { __pyx_check_sizeof_voidp = 1 / (int)(SIZEOF_VOID_P == sizeof(void*)) };

MinGW doesn't always evaluate sizeof(void*) as a compile-time constant in this context, even though both values are 8.

Solution: This has been fixed in setup.py by adding the -DMS_WIN64 compiler flag, which explicitly tells GCC we're targeting 64-bit Windows. This ensures sizeof(void*) is treated as a compile-time constant.

The fix is already applied - no action needed. If you still see this error:

1. Ensure your Python is 64-bit: python -c "import sys; print('64-bit' if sys.maxsize > 2**32 else '32-bit')"
2. Ensure your MinGW GCC is 64-bit: gcc -v should show x86_64
3. Both must be consistently 64-bit (no 32/64-bit mismatch)

References:

• Cython Issue #3405
• Cython Issue #4492

Other MinGW Issues

"mingw32 compiler not found":

• Make sure MinGW is in PATH: gcc --version
• Run: python configure_mingw.py
• Check that pydistutils.cfg exists in your home directory

"gcc: command not found":

• Add MinGW bin directory to PATH:

  set PATH=C:\msys64\mingw64\bin;%PATH%
  

MSVC Issues

"Microsoft Visual C++ 14.0 or greater is required":

• Install Microsoft C++ Build Tools
• Select "Desktop development with C++" workload
• Restart your terminal after installation

General Issues

"No module named 'atabet_token'":

• Run: pip install -e .[dev] to install the package

Build succeeds but import fails:

• Make sure you ran: pip install -e .[dev]
• Check that .pyd file exists in atabet_token directory
• Try: python -c "import atabet_token.token"

Compiler Setup

Option 1: MinGW with MSYS2 (Recommended if you have MSYS2)

Pros:

• Lightweight (~100MB)
• Works well with Cython
• Already installed if you have MSYS2

Setup:

# Run configuration script
python configure_mingw.py

# Or manually add to PATH
set PATH=C:\msys64\mingw64\bin;%PATH%

Option 2: Microsoft C++ Build Tools

Pros:

• Official Microsoft compiler (best compatibility)
• Works out of the box
• Faster compilation

Setup:

1. Download from: https://visualstudio.microsoft.com/visual-cpp-build-tools/
2. Install "Desktop development with C++" workload
3. Restart terminal

Project Structure

atabet-token/
├── atabet_token/              # Source code (Cython modules)
│   ├── __init__.py        # Package initializer
│   ├── token.pyx          # Cython source (compiled to .pyd)
│   └── licensing.py       # Python source (compiled to .pyd)
├── tests/                 # Test suite
│   ├── conftest.py        # Pytest configuration
│   ├── test_token.py      # Tests for original validation
│   └── test_licensing.py  # Tests for licensing validation
├── .github/
│   └── workflows/
│       └── build_wheels.yml  # CI/CD workflow
├── setup.py               # Build configuration (compiles all sources)
├── build_wheel.py         # Wheel protection script
├── pyproject.toml         # Modern Python packaging config
├── environment.yml        # Conda environment definition
├── MANIFEST.in            # Package manifest
├── configure_mingw.py     # MinGW configuration helper
├── build_with_mingw.bat    # MinGW build script
├── build_and_test.bat     # Build and test script
├── cleanup.bat            # Cleanup script
└── README.md              # This file

Development

Building from Source

conda activate atabet_token_env

# With MinGW
python setup.py build_ext --inplace --compiler=mingw32
pip install -e .[dev]

# With MSVC (default)
pip install -e .[dev]

Running Tests

pytest -v

Token Validation Methods

atabet-token supports two validation methods:

1. Licensing Server Validation (Default)

Validates tokens through a licensing server API. This is the default validation method.

Usage:

from atabet_token import TokenCheckMeta

# Default: uses licensing validation
class MyClass(metaclass=TokenCheckMeta(application_id="my-app-id")):
    def __init__(self, token: str):
        self.token = token

# Or explicitly specify
class MyClass(metaclass=TokenCheckMeta(
    validation_method='licensing',
    application_id="my-app-id"
)):
    def __init__(self, token: str):
        self.token = token

Configuration:

• Set LICENSING_SERVER_URL environment variable (default: http://39.101.65.167:5001/; production reference: http://api.atabet.com:5001)
• Provide application_id in TokenCheckMeta
• Optional: client_identifier, session_id, licensing_server_url

Requirements:

• Licensing server must be running and accessible
• Token must be a valid JWT token issued by the licensing server
• requests package (automatically installed)

2. Original HMAC-SHA256 Validation

Validates tokens using the original HMAC-SHA256 method (monthly token rotation).

Usage:

from atabet_token import TokenCheckMeta

# Explicitly use original validation
class MyClass(metaclass=TokenCheckMeta(
    validation_method='original',
    token_key="my-token-key"
)):
    def __init__(self, token: str):
        self.token = token

Configuration:

• Provide token_key in TokenCheckMeta
• Optional: current_time (for testing only)

Note: The original validation method is still fully supported and works exactly as before. Use validation_method='original' to explicitly enable it.

Source Code Protection

atabet-token implements comprehensive source code protection to prevent reverse engineering. All source files are compiled and removed from distribution wheels.

File Type Differences

.pyc Files (Python Bytecode)

• What they are: Compiled Python bytecode files
• Security: ⚠️ NOT SECURE - Can be easily decompiled back to Python source
• Tools to decompile: uncompyle6, decompyle3, pycdc
• When used: Created from .py files by Python's standard compiler

.pyd Files (Compiled C Extensions)

• What they are: Compiled C extensions (Windows) / .so files (Linux/Mac)
• Security: ✅ SECURE - Much harder to reverse engineer (requires disassembling machine code)
• Tools to decompile: Very difficult, requires advanced reverse engineering skills
• When used: Created from .pyx files (Cython source) or .py files compiled with Cython

Protection Strategy

Compilation Behavior:

• .pyx files → Compiled to .pyd/.so (C extension) - SECURE
• .py files (including __init__.py) → Compiled to .pyd/.so (C extension) - SECURE
• All .pyc files are removed - They can be decompiled, so they are not included in distributions

Note on __init__.py:

• __init__.py files are compiled to .pyd/.so via Cython (same as other .py files)
• All .pyc files are removed from the distribution for security
• Python's import system works with compiled extensions (.pyd/.so) just as well as with source files

Source File Removal:

• After compilation, all source files (.py, .pyx, .c) are removed from wheel distributions
• All .pyc files are also removed (they can be decompiled back to source code)
• Only compiled C extensions (.pyd/.so) remain - these are secure and cannot be easily decompiled
• Package functions identically - users can import and use it normally
• Source code is completely hidden from end users

Security Comparison

Format	Security Level	Decompilation Difficulty
.py (source)	❌ None	N/A
.pyc (bytecode)	⚠️ Low	Easy (tools available)
.pyd (C extension)	✅ High	Very Hard (requires reverse engineering)

Implementation

The protection is implemented in:

1. setup.py: Compiles .py files as Cython extensions
2. build_wheel.py: Post-processes wheels to remove source files
3. GitHub Actions: Automatically protects wheels during CI/CD

Verification

To verify protection:

1. Build a wheel: python setup.py bdist_wheel
2. Extract the wheel: unzip dist/atabet_token-*.whl
3. Check contents: Should see only .pyd/.so files, no .py, .pyx, .pyc, or __pycache__ directories

License

Proprietary - All Rights Reserved