KG-Hub Downloader

| Documentation | Repository | PyPI |

Overview

This is a configuration based file caching downloader with initial support for http requests & queries against elasticsearch.

Installation

KGHub Downloader is available to install via pip:

pip install kghub-downloader

Configure

The downloader requires a YAML file which contains a list of target URLs to download, and local names to save those downloads.
For an example, see example/download.yaml

Available options are:

• *url: The URL to download from. Currently supported:

  • http(s)
  • ftp
    • with glob: option to download files with specific extensions (only with ftp as of now and looks recursively).
  • Google Cloud Storage (gs://)
  • Google Drive (gdrive:// or https://drive.google.com/...). The file must be publicly accessible.
  • Amazon AWS S3 bucket (s3://)
  • GitHub Release Assets (git://RepositoryOwner/RepositoryName)

  If the URL includes a name in {CURLY_BRACES}, it will be expanded from environment variables.

• glob: An optional glob pattern to limit downloading files (FTP only)

• local_name: The name to save the file as locally

• tag: A tag to use to filter downloads

• api: The API to use to download the file. Currently supported: elasticsearch

• elastic search options

  • query_file: The file containing the query to run against the index
  • index: The elastic search index for query

* Note:
Google Cloud Storage URLs require that you have set up your credentials as described here. You must:

• create a service account
• add the service account to the relevant bucket and
• download a JSON key for that service account.
  Then, set the GOOGLE_APPLICATION_CREDENTIALS environment variable to point to that file.

Mirorring local files to Amazon AWS S3 bucket requires the following:

• Create an AWS account
• Create an IAM user in AWS: This enables getting the AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY needed for authentication. These two should be stored as environment variables in the user's system.
• Create an S3 bucket: This will be the destination for pushing local files.

You can also include any secrets like API keys you have set as environment variables using {VARIABLE_NAME}, for example:

---
- url: "https://example.com/myfancyfile.json?key={YOUR_SECRET}"
  localname: myfancyfile.json

Note: YOUR_SECRET MUST as an environment variable, and be sure to include the {curly braces} in the url string.

Usage

Downloader can be used directly in Python or via command line

In Python

from kghub_downloader.download_utils import download_from_yaml

download_from_yaml(yaml_file="download.yaml", output_dir="data")

Command Line

To download files listed in a download.yaml file:

$ downloader [OPTIONS] [YAML_FILE]

Arguments:

• [YAML_FILE]: List of files to download in YAML format [default: download.yaml]

Options:

• --output-dir TEXT: Path to output directory [default: .]
• --ignore-cache / --no-ignore-cache: Ignoring already downloaded files and download again [default: no-ignore-cache]
• --progress / --no-progress: Show progress for individual downloads [default: progress]
• --fail-on-error / --no-fail-on-error: Do not attempt to download more files if one raises an error [default: no-fail-on-error]
• --snippet-only / --no-snippet-only: Only download a snippet of the file. [HTTP(S) resources only. [default: no-snippet-only]
• --verbose / --no-verbose: Show verbose output [default: no-verbose]
• --tags TEXT: Optional list of tags to limit downloading to
• --mirror TEXT: Optional remote storage URL to mirror download to. Supported buckets: Google Cloud Storage

Examples:

$ downloader --output_dir example_output --tags zfin_gene_to_phenotype example.yaml
$ downloader --output_dir example_output --mirror gs://your-bucket/desired/directory

# Note that if your YAML file is named `download.yaml`,
# the argument can be omitted from the CLI call.
$ downloader --output_dir example_output

Development

Install

git clone https://github.com/monarch-initiative/kghub-downloader.git
cd kghub-downloader
poetry install

Run tests

poetry run pytest

NOTE: The tests require gcloud credentials to be set up as described above, using the Monarch github actions service account.