Skip to main content

High Performance Test Engine - Rust + Python

Project description

TurboPlex (tpx) — The Test Orchestration Engine for the AI Era

English | Español

Rust Python License

Why TurboPlex?

4x faster than Pytest for test execution. The orchestration engine built for the AI era.

Value Proposition

Feature Description
4x Speed Parallel execution with smart caching (12s → 3s)
🦀 Rust Core Static analysis and memory management with minimal overhead
🤖 M2M Protocol Generates .tplex_report.json with AI-actionable errors
👀 Watch Mode Auto-reload when you save .py files

Installation

# Install from PyPI
pip install turboplex

# Verify the installation
tpx --help
# Clone the repository (dev)
cd turboplex

# Install in development mode
pip install -e .

# Verify the installation
tpx --help

Quick Start

Basic Execution

# Single test
tpx --path tests/test_simple.py

# Multiple directories
tpx --path tests/ --path tests/integration/

# Auto-discover tests
tpx

Watch Mode (TDD Development)

# Run and watch changes in real time
tpx --watch --path tests/

Pytest Compatibility Mode (--compat)

Use this when your suite relies on complex pytest fixtures (conftest.py, client, db, etc.).

tpx --compat --path tests/

Light Collect Mode (--light)

For projects with heavy conftest.py initialization (e.g., database migrations, schema creation), use --light to skip conftest loading during test discovery. This can reduce discovery time from minutes to seconds.

# Fast discovery for projects with heavy conftest.py
tpx --compat --light

# Or set the environment variable
export TPX_MCP_LIGHT_COLLECT=1  # Linux/Mac
set TPX_MCP_LIGHT_COLLECT=1     # Windows
tpx --compat

When to use:

  • Your conftest.py initializes databases or runs migrations on import
  • Test discovery takes more than 10 seconds
  • You're using TurboPlex MCP with an IDE (Windsurf, Cursor, etc.)

MCP Server (IDE Integration)

tpx mcp

Notes:

  • For robustness, TurboPlex isolates tool payloads in JSON files (--out-json) to prevent stdout pollution.
  • Subprocesses enforce UTF-8 (PYTHONUTF8=1, PYTHONIOENCODING=utf-8, PYTHONUNBUFFERED=1) to avoid Windows encoding issues.
  • Timeouts are structured and configurable via environment variables:
    • TPX_PYTHON_EXE - Force specific Python interpreter (useful for venvs on Windows).
    • TPX_MCP_LIGHT_COLLECT=1 - Skip conftest.py during discovery (for projects with heavy DB initialization).
    • TPX_MCP_DEBUG=1 - Enable debug tracing to stderr (shows commands, Python used, etc.).
    • TPX_MCP_STDOUT_MODE - redirect (default) or failfast for JSON-RPC guarding.
    • TPX_MCP_TURBOPLEX_COLLECT_TIMEOUT_S (default 120)
    • TPX_MCP_TURBOPLEX_RUN_TIMEOUT_S (default 60)
    • TPX_MCP_PYTEST_COLLECT_TIMEOUT_S or TPX_PYTEST_COLLECT_TIMEOUT_S (default 120)
    • TPX_MCP_PYTEST_RUN_TIMEOUT_S or TPX_PYTEST_RUN_TIMEOUT_S (default 60)

Example MCP Configuration (Windsurf/Cursor):

{
  "mcpServers": {
    "tpx": {
      "command": "tpx",
      "args": ["mcp"],
      "env": {
        "TPX_PYTHON_EXE": "/path/to/venv/bin/python",
        "TPX_MCP_LIGHT_COLLECT": "1",
        "TPX_MCP_DEBUG": "1"
      }
    }
  }
}

Tools exposed over MCP:

  • ping
  • turboplex_version
  • discover(paths=None, compat=False)
  • run(selection, compat=False, max_workers=None)
  • get_report(path=None)

Integration with AI Agents

# INSTRUCTION FOR YOUR AI AGENT:
# If you detect a .tplex_report.json file in the project,
# read it immediately and propose an automatic fix.

# The file contains:
# - Exact error
# - Failure line
# - Code context (5 lines before and after)

Benchmarks

Speedrun: 2 Tests

Tool Time
pytest 12.69s
tpx (cached) ~3s
pytest:     ████████████████████████████ 12.69s
tpx:        █████ 3s (cached)

Per-Test Comparison

Metric pytest tpx
Time per test ~6s ~1.5s
Caching No Yes (SHA-256)
M2M Report No Yes (.tplex_report.json)

Configuration

turbo_config.toml File

[execution]
max_workers = 8
default_timeout_ms = 30000
cache_enabled = true

[python]
enabled = true
interpreter = "python"
module = "turboplex_py"
test_paths = ["tests"]
project_path = "."

Cache

The cache is stored in .turboplex_cache/ and is automatically invalidated when test files change (SHA-256 hash) or when the runtime fingerprint changes (Python version, dependency lock hash, PYTHONPATH, execution flags).

API for AI Agents

.tplex_report.json Format

{
  "timestamp": "2026-03-28 14:17:49",
  "total_tests": 1,
  "failed_count": 1,
  "failures": [
    {
      "test": "test_fiscal_year_close_logic",
      "file": "tests/test_accounting_close.py",
      "line": 42,
      "error": "parameter 'db' has no @fixture and no default",
      "context": [
        "    38: def test_fiscal_year_close_logic(db):",
        "    39:     # Arrange",
        "    40:     year = 2024",
        ">>> 41:     result = close_year(db, year)",
        "    42:     assert result.success"
      ]
    }
  ]
}

Commands

Command Description
tpx Auto-discover and run tests
tpx --path ./tests Run tests in a directory
tpx --watch Watch mode with auto-reload
tpx --compat Delegate discovery/execution to pytest for fixture-heavy suites
tpx --compat --light Fast discovery mode (skips conftest.py loading)
tpx mcp Start the MCP server over stdio for IDE integration
tpx --help Show help

Architecture

┌─────────────────────────────────────────────────────┐
│                    tpx (Rust)                        │
├─────────────────────────────────────────────────────┤
│  • Test discovery                                  │
│  • SHA-256 caching                                 │
│  • Parallel execution (Rayon)                      │
│  • Watch mode (notify)                             │
│  • M2M report (.tplex_report.json)                 │
└─────────────────────────────────────────────────────┘
                          │
                          ▼
┌─────────────────────────────────────────────────────┐
│              turboplex_py (Python)                  │
├─────────────────────────────────────────────────────┤
│  • collector.py - Test discovery                   │
│  • runner.py - Test runner                         │
│  • fixtures.py - @fixture system                   │
│  • markers.py - skip, skipif                       │
└─────────────────────────────────────────────────────┘

Files Excluded from the Repository (.gitignore)

This project ignores generated files and local configuration to keep the repository lightweight, reproducible, and free of sensitive data.

  • Build artifacts and caches (e.g., target/, **/target/, .cache/)
  • Temporary files and logs (*.tmp, *.log, *.swp)
  • Local IDE/OS configuration (e.g., .vscode/, .idea/, Thumbs.db, .DS_Store)
  • Python local environments and metadata (e.g., .venv/, __pycache__/, *.egg-info/)
  • Environment files with secrets or local configuration (.env, .env.*)
  • Web tooling dependencies and outputs if applicable (node_modules/, dist/, build/)
  • TurboPlex-generated caches and reports (.turboplex_cache/, .tplex_report.json)

License

MIT License - See LICENSE

Authors

Keita_Izumi

TurboPlex Team - @turbo plexus


🚀 The future of testing is here

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

turboplex-0.2.10.tar.gz (1.8 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

turboplex-0.2.10-py3-none-win_amd64.whl (779.4 kB view details)

Uploaded Python 3Windows x86-64

File details

Details for the file turboplex-0.2.10.tar.gz.

File metadata

  • Download URL: turboplex-0.2.10.tar.gz
  • Upload date:
  • Size: 1.8 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: maturin/1.12.6

File hashes

Hashes for turboplex-0.2.10.tar.gz
Algorithm Hash digest
SHA256 ab6bc9708c7a9d7c5683643f82480c87cdf7214ff92a33f6ad92c98b22e7d8fd
MD5 6d8eab1e3122756a013d27f4dc4c2af1
BLAKE2b-256 a432bf5d208d3215057c6f0997a75e1cfa7fb92aa8b39ae8800a2541b02b0836

See more details on using hashes here.

File details

Details for the file turboplex-0.2.10-py3-none-win_amd64.whl.

File metadata

File hashes

Hashes for turboplex-0.2.10-py3-none-win_amd64.whl
Algorithm Hash digest
SHA256 2155dffea3a75015de25777735557c257871c5d49910d396726f64666475d24a
MD5 288d05e85d278fd907f128752ae4e1d7
BLAKE2b-256 ba61a6dc21bd6d3d9a8aaab6ad71a4eb42815973b5dc2de1ea44c718d08ed643

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page