Snakemake Storage Plugin: Cached HTTP

A Snakemake storage plugin for downloading files via HTTP with local caching, checksum verification, and adaptive rate limiting.

Supported sources:

• zenodo.org - Zenodo data repository (checksum from API)
• data.pypsa.org - PyPSA data repository (checksum from manifest.yaml)

Features

• Local caching: Downloads are cached to avoid redundant transfers (can be disabled)
• Checksum verification: Automatically verifies checksums (from Zenodo API or data.pypsa.org manifests)
• Rate limit handling: Automatically respects Zenodo's rate limits using X-RateLimit-* headers with exponential backoff retry
• Concurrent download control: Limits simultaneous downloads to prevent overwhelming Zenodo
• Progress bars: Shows download progress with tqdm
• Immutable URLs: Returns mtime=0 since Zenodo URLs are persistent
• Environment variable support: Configure via environment variables for CI/CD workflows

Installation

From the pypsa-eur repository root:

pip install -e plugins/snakemake-storage-plugin-cached-http

Configuration

The Zenodo storage plugin works alongside other storage providers (like HTTP). Snakemake automatically routes URLs to the correct provider based on the URL pattern.

Register additional settings in your Snakefile if you want to customize the defaults:

# Optional: Configure cached HTTP storage with custom settings
# This extends the existing storage configuration (e.g., for HTTP)
storage cached_http:
    provider="cached-http",
    cache="~/.cache/snakemake-pypsa-eur",  # Default location
    max_concurrent_downloads=3,  # Download max 3 files at once

If you don't explicitly configure it, the plugin will use default settings automatically.

Settings

• cache (optional): Cache directory for downloaded files

  • Default: Platform-dependent user cache directory (via platformdirs.user_cache_dir("snakemake-pypsa-eur"))
  • Set to "" (empty string) to disable caching
  • Files are cached here to avoid re-downloading
  • Environment variable: SNAKEMAKE_STORAGE_CACHED_HTTP_CACHE

• skip_remote_checks (optional): Skip metadata checking with remote API

  • Default: False (perform checks)
  • Set to True or "1" to skip remote existence/size checks (useful for CI/CD)
  • Environment variable: SNAKEMAKE_STORAGE_CACHED_HTTP_SKIP_REMOTE_CHECKS

• max_concurrent_downloads (optional): Maximum concurrent downloads

  • Default: 3
  • Controls how many files can be downloaded simultaneously
  • No environment variable support

Usage

Use Zenodo or data.pypsa.org URLs directly in your rules. Snakemake automatically detects supported URLs and routes them to this plugin:

rule download_zenodo:
    input:
        storage("https://zenodo.org/records/3520874/files/natura.tiff"),
    output:
        "resources/natura.tiff"
    shell:
        "cp {input} {output}"

rule download_pypsa:
    input:
        storage("https://data.pypsa.org/workflows/eur/eez/v12_20231025/World_EEZ_v12_20231025_LR.zip"),
    output:
        "resources/eez.zip"
    shell:
        "cp {input} {output}"

Or if you configured a tagged storage entity:

rule download_data:
    input:
        storage.cached_http(
            "https://zenodo.org/records/3520874/files/natura.tiff"
        ),
    output:
        "resources/natura.tiff"
    shell:
        "cp {input} {output}"

The plugin will:

1. Check if the file exists in the cache (if caching is enabled)
2. If cached, copy from cache (fast)
3. If not cached, download with:
   • Progress bar showing download status
   • Automatic rate limit handling with exponential backoff retry
   • Concurrent download limiting
   • Checksum verification (from Zenodo API or data.pypsa.org manifest)
4. Store in cache for future use (if caching is enabled)

Example: CI/CD Configuration

For continuous integration environments where you want to skip caching and remote checks:

# GitHub Actions example
- name: Run snakemake workflows
  env:
    SNAKEMAKE_STORAGE_CACHED_HTTP_CACHE: ""
    SNAKEMAKE_STORAGE_CACHED_HTTP_SKIP_REMOTE_CHECKS: "1"
  run: |
    snakemake --cores all

Rate Limiting and Retry

Zenodo API limits:

• Guest users: 60 requests/minute
• Authenticated users: 100 requests/minute

The plugin automatically:

• Monitors X-RateLimit-Remaining header
• Waits when rate limit is reached
• Uses X-RateLimit-Reset to calculate wait time
• Retries failed requests with exponential backoff (up to 5 attempts)
• Handles transient errors: HTTP errors, timeouts, checksum mismatches, and network issues

URL Handling

• Handles URLs from zenodo.org, sandbox.zenodo.org, and data.pypsa.org
• Other HTTP(S) URLs are handled by the standard snakemake-storage-plugin-http
• Both plugins can coexist in the same workflow

Plugin Priority

When using storage() without specifying a plugin name, Snakemake checks all installed plugins:

• Cached HTTP plugin: Only accepts zenodo.org and data.pypsa.org URLs
• HTTP plugin: Accepts all HTTP/HTTPS URLs (including zenodo.org)

If both plugins are installed, supported URLs would be ambiguous - both plugins accept them. Typically snakemake would raise an error: "Multiple suitable storage providers found" if you try to use storage() without specifying which plugin to use, ie. one needs to explicitly call the Cached HTTP provider using storage.cached_http(url) instead of storage(url), but we monkey-patch the http plugin to refuse zenodo.org and data.pypsa.org URLs.

License

MIT License