cascache-lib 0.1.0

Content-addressable storage (CAS) caching library for Python with local and remote backend support

cascache_lib

Content-addressable storage (CAS) caching library for Python

cascache_lib is a flexible, high-performance caching library that brings content-addressed caching to Python applications. It provides both local filesystem caching and remote CAS backend support via gRPC, making it perfect for build systems, CI/CD pipelines, and any workflow that benefits from intelligent caching.

Features

• 🚀 Multiple Backends: Local filesystem, remote CAS (cascache server), or hybrid (both)
• 🔐 Content-Addressed: Cache keys based on SHA256 hashes of inputs
• ⚡ Async/Await: Non-blocking I/O with asyncio support
• 🔄 Automatic Retry: Exponential backoff for transient network errors
• 📦 Compression: Efficient tar.gz compression for cached artifacts
• 🛡️ Graceful Degradation: Automatic fallback to local cache when remote unavailable
• 🎯 Type-Safe: Full type hints with Python 3.11+ support
• 🧪 Well-Tested: Comprehensive test suite with >85% coverage

Installation

# From PyPI (once published)
pip install cascache-lib

# From source
pip install git+https://gitlab.com/cascascade/cascache-lib.git

# With development dependencies
pip install cascache-lib[dev]

Quick Start

Local Caching

from pathlib import Path
from cascache_lib import LocalCache, compute_cache_key

# Create a local cache
cache = LocalCache(Path(".cache"))

# Compute cache key from inputs
cache_key = compute_cache_key(
    command="python build.py",
    inputs=[Path("src/main.py"), Path("pyproject.toml")],
    env={"PYTHON_VERSION": "3.13"},
)

# Check if artifacts are cached
if await cache.exists(cache_key):
    print("Cache hit!")
    await cache.get(cache_key, [Path("dist/")])
else:
    print("Cache miss - building...")
    # Run your build process here
    # ...
    # Cache the outputs
    await cache.put(cache_key, [Path("dist/")])

Remote CAS Caching

from cascache_lib import RemoteCache

# Connect to cascache server
cache = RemoteCache(
    cas_url="grpc://cache.example.com:50051",
    token="your-auth-token",
    timeout=30.0,
    max_retries=3,
)

# Use exactly like LocalCache
cache_key = compute_cache_key("make build", [Path("src/")])
if not await cache.exists(cache_key):
    # Build and cache
    await cache.put(cache_key, [Path("build/")])

Hybrid Caching (Recommended)

from cascache_lib import HybridCache, LocalCache, RemoteCache

# Create hybrid cache: local + remote
local = LocalCache(Path(".cache/local"))
remote = RemoteCache("grpc://cache.example.com:50051", token="...")

cache = HybridCache(
    local_cache=local,
    remote_cache=remote,
    auto_upload=True,  # Automatically sync to remote
)

# Automatic behavior:
# - get(): Check local first (fast), then remote, populate local on hit
# - put(): Store in local AND upload to remote
# - Graceful fallback to local-only on remote errors

cache_key = compute_cache_key("cargo build --release", [Path("src/")])
if await cache.get(cache_key, [Path("target/release/")]):
    print("Restored from cache (local or remote)")
else:
    # Build...
    await cache.put(cache_key, [Path("target/release/")])

Configuration-Based Setup

from cascache_lib import create_cache
from cascache_lib.config import CacheConfig

# Define configuration (or load from YAML/JSON)
config = CacheConfig(
    local={
        "enabled": True,
        "path": ".cache/cascache",
    },
    remote={
        "enabled": True,
        "url": "grpc://localhost:50051",
        "token_file": "~/.cache/cascache/token",
        "upload": True,
        "download": True,
        "timeout": 30.0,
        "max_retries": 3,
    },
)

# Create cache from config
cache = create_cache(config)

# Use the cache (automatically hybrid if both local and remote enabled)

Use Cases

Build Systems

# Cache compiled artifacts
cache_key = compute_cache_key(
    command="gcc -o myapp main.c",
    inputs=[Path("main.c"), Path("config.h")],
)

if not await cache.exists(cache_key):
    subprocess.run(["gcc", "-o", "myapp", "main.c"])
    await cache.put(cache_key, [Path("myapp")])

CI/CD Pipelines

# Share build artifacts across CI jobs
cache = HybridCache(
    local_cache=LocalCache(Path("/tmp/ci-cache")),
    remote_cache=RemoteCache(
        cas_url=os.environ["CACHE_SERVER_URL"],
        token=os.environ["CACHE_TOKEN"],
    ),
)

# First job caches, subsequent jobs restore
cache_key = compute_cache_key("npm run build", [Path("package.json"), Path("src/")])
if await cache.get(cache_key, [Path("dist/")]):
    print("Skipped build - restored from cache")

Test Frameworks

# Cache test fixtures or test results
cache_key = compute_cache_key(
    command="pytest tests/",
    inputs=expand_globs(["tests/**/*.py", "src/**/*.py"]),
)

# Cache test database fixtures
await cache.put(cache_key, [Path("tests/fixtures/test.db")])

API Reference

CacheBackend (Abstract Base)

All cache implementations inherit from CacheBackend:

class CacheBackend(ABC):
    async def exists(cache_key: str) -> bool
    async def get(cache_key: str, output_paths: list[Path]) -> bool
    async def put(cache_key: str, output_paths: list[Path]) -> bool
    async def clear() -> None

LocalCache

Filesystem-based cache with tar.gz compression:

cache = LocalCache(cache_dir: Path | None = None)

• cache_dir: Cache directory (default: .cache)

RemoteCache

Remote cache using cascache server:

cache = RemoteCache(
    cas_url: str,
    token: str | None = None,
    timeout: float = 30.0,
    max_retries: int = 3,
    initial_backoff: float = 0.1,
)

• cas_url: Server URL (format: grpc://host:port)
• token: Authentication token (optional)
• timeout: Request timeout in seconds
• max_retries: Max retry attempts for transient errors
• initial_backoff: Initial retry delay in seconds

HybridCache

Hybrid cache combining local + remote:

cache = HybridCache(
    local_cache: LocalCache,
    remote_cache: RemoteCache | None = None,
    auto_upload: bool = True,
)

• local_cache: Local cache instance (required)
• remote_cache: Remote cache instance (optional)
• auto_upload: Auto-upload to remote on put()

Methods:

• get_stats() -> dict: Get cache hit/miss statistics
• reset_stats() -> None: Reset statistics

Utility Functions

compute_cache_key

Compute deterministic SHA256 hash for cache keys:

cache_key = compute_cache_key(
    command: str,
    inputs: list[Path],
    env: dict[str, str] | None = None,
) -> str

Returns 64-character hex string (SHA256 digest).

expand_globs

Expand glob patterns to file paths:

files = expand_globs(
    patterns: list[str],
    base_dir: Path | None = None,
) -> list[Path]

Supports ** for recursive matching.

Configuration

Pydantic Models

from cascache_lib.config import (
    CacheConfig,
    LocalCacheConfig,
    RemoteCacheConfig,
)

config = CacheConfig(
    local=LocalCacheConfig(
        enabled=True,
        path=".cache",
    ),
    remote=RemoteCacheConfig(
        enabled=True,
        type="cas",
        url="grpc://localhost:50051",
        token_file=None,
        upload=True,
        download=True,
        timeout=30.0,
        max_retries=3,
        initial_backoff=0.1,
    ),
)

YAML Configuration

local:
  enabled: true
  path: .cache

remote:
  enabled: true
  type: cas
  url: grpc://cache.example.com:50051
  token_file: ~/.cache/cascache/token
  upload: true
  download: true
  timeout: 30.0
  max_retries: 3

Load with:

import yaml
from cascache_lib.config import CacheConfig

with open("cache-config.yaml") as f:
    config_dict = yaml.safe_load(f)
    config = CacheConfig(**config_dict)

Error Handling

The library handles errors gracefully:

• Network errors: Automatic retry with exponential backoff
• Timeout errors: Configurable timeout per request
• Authentication errors: Clear error messages, no retry
• Remote unavailable: Automatic fallback to local cache (CacheManager)

try:
    cache = CASCache("grpc://cache.example.com:50051")
    await cache.get(cache_key, outputs)
except grpc.RpcError as e:
    # Handle gRPC errors
    print(f"Cache error: {e}")

Development

Setup

# Clone repository
git clone https://gitlab.com/cascascade/cascache-lib.git
cd cascache-lib

# Install with dev dependencies
pip install -e ".[dev]"

Running Tests

# Run all tests
pytest

# With coverage
pytest --cov=cascache_lib --cov-report=html

# Run specific test file
pytest tests/unit/test_local_cache.py -v

Code Quality

# Type checking
pyright

# Linting
ruff check src/

# Formatting
ruff format src/

Regenerating Protobuf Code

python -m grpc_tools.protoc \
    -I src/cascache_lib/api/protos \
    --python_out=src/cascache_lib/api/generated \
    --grpc_python_out=src/cascache_lib/api/generated \
    --pyi_out=src/cascache_lib/api/generated \
    src/cascache_lib/api/protos/*.proto

Architecture

cascache_lib/
├── cache/
│   ├── backend.py      # Abstract CacheBackend interface
│   ├── local.py        # LocalCache (filesystem)
│   ├── cas.py          # CASCache (remote gRPC)
│   ├── manager.py      # CacheManager (hierarchical)
│   ├── hash.py         # Cache key computation
│   └── factory.py      # Factory function
├── api/
│   ├── protos/         # Protobuf definitions (.proto)
│   └── generated/      # Generated Python code
└── config.py           # Pydantic configuration models

Compatibility

• Python: 3.11, 3.12, 3.13+
• Platforms: Linux, macOS, Windows
• CAS Server: Compatible with cascache

Related Projects

• cascache - CAS server implementation (Python)
• cascade - Workflow orchestration tool using cascache_lib

License

MIT License - see LICENSE file for details.

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Add tests for new functionality
4. Ensure all tests pass
5. Submit a merge request

Support

• Issues: https://gitlab.com/cascascade/cascache-lib/-/issues
• Documentation: https://gitlab.com/cascascade/cascache-lib/-/blob/main/README.md

Changelog

See CHANGELOG.md for version history.

---

Built with ❤️ for better caching