fsspeckit 0.4.0

Enhanced utilities and extensions for fsspec, storage_options and obstore with multi-format I/O support.

fsspeckit

Enhanced utilities and extensions for fsspec filesystems with multi-format I/O support.

Overview

fsspeckit is a comprehensive toolkit that extends fsspec with:

• Multi-cloud storage configuration - Easy setup for AWS S3, Google Cloud Storage, Azure Storage, GitHub, and GitLab
• Enhanced caching - Improved caching filesystem with monitoring and path preservation
• Extended I/O operations - Read/write operations for JSON, CSV, Parquet with Polars/PyArrow integration
• Domain-specific packages - Organized into logical packages for better discoverability

Package Structure

fsspeckit is organized into domain-specific packages:

• fsspeckit.core - Core filesystem APIs and backend-neutral planning logic
• fsspeckit.storage_options - Multi-cloud storage configuration classes
• fsspeckit.datasets - Dataset-level operations (DuckDB & PyArrow helpers)
• fsspeckit.sql - SQL-to-filter translation helpers
• fsspeckit.common - Cross-cutting utilities (logging, parallelism, type conversion)
• fsspeckit.utils - Backwards-compatible façade that re-exports from domain packages

Note: The fsspeckit.utils module is maintained for backwards compatibility. New code should import directly from the domain packages for better discoverability.

Installation

# Basic installation
pip install fsspeckit

# Specific cloud providers
pip install "fsspeckit[aws]"     # AWS S3 support
pip install "fsspeckit[gcp]"     # Google Cloud Storage
pip install "fsspeckit[azure]"   # Azure Storage

# Multiple cloud providers
pip install "fsspeckit[aws,gcp,azure]"

Quick Start

Basic Filesystem Operations

from fsspeckit import filesystem

# Local filesystem
fs = filesystem("file")
files = fs.ls("/path/to/data")

# S3 with caching
fs = filesystem("s3://my-bucket/", cached=True)
data = fs.cat("data/file.txt")

Storage Configuration

from fsspeckit.storage_options import AwsStorageOptions

# Configure S3 access
options = AwsStorageOptions(
    region="us-west-2",
    access_key_id="YOUR_KEY",
    secret_access_key="YOUR_SECRET"
)

fs = filesystem("s3", storage_options=options, cached=True)

Environment-based Configuration

from fsspeckit.storage_options import AwsStorageOptions

# Load from environment variables
options = AwsStorageOptions.from_env()
fs = filesystem("s3", storage_options=options)

# Load with anonymous access from environment
# Set AWS_S3_ANONYMOUS=true in environment
options = AwsStorageOptions.from_env()
fs = filesystem("s3", storage_options=options)

DuckDB Parquet Maintenance

from fsspeckit.datasets import DuckDBParquetHandler

with DuckDBParquetHandler() as handler:
    # Inspect fragmentation without writing
    dry_stats = handler.compact_parquet_dataset(
        path="/data/events/",
        target_mb_per_file=256,
        dry_run=True,
    )

    # Compact tiny files and recompress with zstd
    handler.compact_parquet_dataset(
        path="/data/events/",
        target_rows_per_file=500_000,
        compression="zstd",
    )

    # Recluster partitions with z-order style ordering
    handler.optimize_parquet_dataset(
        path="/data/events/",
        zorder_columns=["user_id", "event_date"],
        partition_filter=["date=2025-11-10"],
    )

Multiple Cloud Providers

from fsspeckit.storage_options import (
    AwsStorageOptions, 
    GcsStorageOptions,
    GitHubStorageOptions
)

# AWS S3
s3_fs = filesystem("s3", storage_options=AwsStorageOptions.from_env())

# Google Cloud Storage  
gcs_fs = filesystem("gs", storage_options=GcsStorageOptions.from_env())

# GitHub repository
github_fs = filesystem("github", storage_options=GitHubStorageOptions(
    org="microsoft",
    repo="vscode", 
    token="ghp_xxxx"
))

Storage Options

AWS S3

from fsspeckit.storage_options import AwsStorageOptions

# Basic credentials
options = AwsStorageOptions(
    access_key_id="AKIAXXXXXXXX",
    secret_access_key="SECRET",
    region="us-east-1"
)

# From AWS profile
options = AwsStorageOptions.create(profile="dev")

# S3-compatible service (MinIO)
options = AwsStorageOptions(
    endpoint_url="http://localhost:9000",
    access_key_id="minioadmin",
    secret_access_key="minioadmin",
    allow_http=True
)

# Anonymous access for public buckets
options = AwsStorageOptions(anonymous=True)

Google Cloud Storage

from fsspeckit.storage_options import GcsStorageOptions

# Service account
options = GcsStorageOptions(
    token="path/to/service-account.json",
    project="my-project-123"
)

# From environment
options = GcsStorageOptions.from_env()

Azure Storage

from fsspeckit.storage_options import AzureStorageOptions

# Account key
options = AzureStorageOptions(
    protocol="az",
    account_name="mystorageacct",
    account_key="key123..."
)

# Connection string
options = AzureStorageOptions(
    protocol="az",
    connection_string="DefaultEndpoints..."
)

GitHub

from fsspeckit.storage_options import GitHubStorageOptions

# Public repository
options = GitHubStorageOptions(
    org="microsoft",
    repo="vscode",
    ref="main"
)

# Private repository
options = GitHubStorageOptions(
    org="myorg",
    repo="private-repo",
    token="ghp_xxxx",
    ref="develop"
)

GitLab

from fsspeckit.storage_options import GitLabStorageOptions

# Public project
options = GitLabStorageOptions(
    project_name="group/project",
    ref="main"
)

# Private project with token
options = GitLabStorageOptions(
    project_id=12345,
    token="glpat_xxxx",
    ref="develop"
)

Enhanced Caching

from fsspeckit import filesystem

# Enable caching with monitoring
fs = filesystem(
    "s3://my-bucket/",
    cached=True,
    cache_storage="/tmp/my_cache",
    verbose=True
)

# Cache preserves directory structure
data = fs.cat("deep/nested/path/file.txt")
# Cached at: /tmp/my_cache/deep/nested/path/file.txt

Utilities

Parallel Processing

from fsspeckit.common import run_parallel

# Run function in parallel
def process_file(path, multiplier=1):
    return len(path) * multiplier

results = run_parallel(
    process_file,
    ["/path1", "/path2", "/path3"],
    multiplier=2,
    n_jobs=4,
    verbose=True
)

Type Conversion

from fsspeckit.common.types import dict_to_dataframe, to_pyarrow_table

# Convert dict to DataFrame
data = {"col1": [1, 2, 3], "col2": [4, 5, 6]}
df = dict_to_dataframe(data)

# Convert to PyArrow table
table = to_pyarrow_table(df)

Logging

from fsspeckit.common.logging import setup_logging

# Configure logging
setup_logging(level="DEBUG", format_string="{time} | {level} | {message}")

Migration Guide

The package structure was refactored in version X.X.0 to improve discoverability and organization.

For new code, use the canonical imports from domain packages:

• Dataset operations: from fsspeckit.datasets import ...
• SQL helpers: from fsspeckit.sql import ...
• Common utilities: from fsspeckit.common import ...

For existing code, all fsspeckit.utils imports continue to work unchanged.

For detailed migration instructions, see the Migration Guide.

Dependencies

Core Dependencies

• fsspec>=2023.1.0 - Filesystem interface
• msgspec>=0.18.0 - Serialization
• pyyaml>=6.0 - YAML support
• requests>=2.25.0 - HTTP requests
• loguru>=0.7.0 - Logging

Optional Dependencies

• orjson>=3.8.0 - Fast JSON processing
• polars>=0.19.0 - Fast DataFrames
• pyarrow>=10.0.0 - Columnar data
• pandas>=1.5.0 - Data analysis
• joblib>=1.3.0 - Parallel processing
• rich>=13.0.0 - Progress bars

Cloud Provider Dependencies

• boto3>=1.26.0, s3fs>=2023.1.0 - AWS S3
• gcsfs>=2023.1.0 - Google Cloud Storage
• adlfs>=2023.1.0 - Azure Storage

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.