Get "mulled-v2-{hash}" for multi-package containers (https://github.com/BioContainers/multi-package-containers) comma-delimited multi-package definitions (like in the hash.tsv file)

multi_container_mulled_hash.py

#!/usr/bin/env python

import argparse
import hashlib
import sys
from collections import namedtuple
from typing import List, Dict, Tuple, Union

# from https://github.com/galaxyproject/galaxy/blob/f12ee5ce6d602cd4c8b4cfc2112988b84a4f255e/lib/galaxy/tool_util/deps/mulled/util.py#L185
Target = namedtuple("Target", ["package_name", "version", "build", "package"])


# from https://github.com/galaxyproject/galaxy/blob/f12ee5ce6d602cd4c8b4cfc2112988b84a4f255e/lib/galaxy/tool_util/deps/mulled/util.py#L188
def build_target(package_name, version=None, build=None, tag=None):
    """Use supplied arguments to build a :class:`Target` object."""
    if tag is not None:
        assert version is None
        assert build is None
        version, build = split_tag(tag)

    # conda package and quay image names are lowercase
    return Target(package_name.lower(), version, build, package_name)


# from https://github.com/galaxyproject/galaxy/blob/f12ee5ce6d602cd4c8b4cfc2112988b84a4f255e/lib/galaxy/tool_util/deps/mulled/util.py#L210
def _simple_image_name(targets, image_build=None):
    target = targets[0]
    suffix = ""
    if target.version is not None:
        build = target.build
        if build is None and image_build is not None and image_build != "0":
            # Special case image_build == "0", which has been built without a suffix
            print("WARNING: Hard-coding image build instead of using Conda build - this is not recommended.")
            build = image_build
        suffix += f":{target.version}"
        if build is not None:
            suffix += f"--{build}"
    return f"{target.package_name}{suffix}"


# from https://github.com/galaxyproject/galaxy/blob/f12ee5ce6d602cd4c8b4cfc2112988b84a4f255e/lib/galaxy/tool_util/deps/mulled/util.py#L264
def v2_image_name(targets, image_build=None, name_override=None):
    """Generate mulled hash version 2 container identifier for supplied arguments.

    If a single target is specified, simply use the supplied name and version as
    the repository name and tag respectively. If multiple targets are supplied,
    hash the package names as the repository name and hash the package versions (if set)
    as the tag.

    >>> single_targets = [build_target("samtools", version="1.3.1")]
    >>> v2_image_name(single_targets)
    'samtools:1.3.1'
    >>> single_targets = [build_target("samtools", version="1.3.1", build="py_1")]
    >>> v2_image_name(single_targets)
    'samtools:1.3.1--py_1'
    >>> single_targets = [build_target("samtools", version="1.3.1")]
    >>> v2_image_name(single_targets, image_build="0")
    'samtools:1.3.1'
    >>> single_targets = [build_target("samtools", version="1.3.1", build="py_1")]
    >>> v2_image_name(single_targets, image_build="0")
    'samtools:1.3.1--py_1'
    >>> multi_targets = [build_target("samtools", version="1.3.1"), build_target("bwa", version="0.7.13")]
    >>> v2_image_name(multi_targets)
    'mulled-v2-fe8faa35dbf6dc65a0f7f5d4ea12e31a79f73e40:4d0535c94ef45be8459f429561f0894c3fe0ebcf'
    >>> multi_targets_on_versionless = [build_target("samtools", version="1.3.1"), build_target("bwa")]
    >>> v2_image_name(multi_targets_on_versionless)
    'mulled-v2-fe8faa35dbf6dc65a0f7f5d4ea12e31a79f73e40:b0c847e4fb89c343b04036e33b2daa19c4152cf5'
    >>> multi_targets_versionless = [build_target("samtools"), build_target("bwa")]
    >>> v2_image_name(multi_targets_versionless)
    'mulled-v2-fe8faa35dbf6dc65a0f7f5d4ea12e31a79f73e40'
    """
    if name_override is not None:
        print(
            "WARNING: Overriding mulled image name, auto-detection of 'mulled' package attributes will fail to detect result."
        )
        return name_override

    targets = list(targets)
    if len(targets) == 1:
        return _simple_image_name(targets, image_build=image_build)
    else:
        targets_order = sorted(targets, key=lambda t: t.package_name)
        package_name_buffer = "\n".join(map(lambda t: t.package_name, targets_order))
        package_hash = hashlib.sha1()
        package_hash.update(package_name_buffer.encode())

        versions = map(lambda t: t.version, targets_order)
        if any(versions):
            # Only hash versions if at least one package has versions...
            version_name_buffer = "\n".join(map(lambda t: t.version or "null", targets_order))
            version_hash = hashlib.sha1()
            version_hash.update(version_name_buffer.encode())
            version_hash_str = version_hash.hexdigest()
        else:
            version_hash_str = ""

        if not image_build:
            build_suffix = ""
        elif version_hash_str:
            # tagged verson is <version_hash>-<build>
            build_suffix = f"-{image_build}"
        else:
            # tagged version is simply the build
            build_suffix = image_build
        suffix = ""
        if version_hash_str or build_suffix:
            suffix = f":{version_hash_str}{build_suffix}"
        return f"mulled-v2-{package_hash.hexdigest()}{suffix}"


def combo_line_to_build_targets(s: str) -> List[Target]:
    """List of Target tuples from multi-package comma-delimited definition

    See https://github.com/BioContainers/multi-package-containers/blob/master/combinations/hash.tsv for multi-package definitions built into multi-package containers
    """
    out = []
    package = None
    version = None
    build = None
    for package_version in s.split(','):
        if '=' in package_version:
            sp = package_version.split('=')
            if len(sp) == 3:
                package, version, build = sp
            elif len(sp) == 2:
                package, version = sp
            else:
                raise Exception(f'Could not parse package/version/build definition from "{package_version}" in "{s}"')
        else:
            package = package_version
            version = None
        out.append(
            build_target(
                package_name=package,
                version=version,
                build=build,
            )
        )
    return out


def main():
    parser = argparse.ArgumentParser(
        description='Get "mulled-v2-{hash}" for multi-package containers (https://github.com/BioContainers/multi-package-containers) comma-delimited multi-package definitions (like in the hash.tsv file)',
    )
    parser.add_argument('packages', metavar='P', type=str, nargs='+',
                        help='Multi-package string (e.g. "minimap2=2.18,samtools=1.12")')
    args = parser.parse_args()
    if not args.packages:
        print('No multi-package containers specified!')
        sys.exit(1)
    for p in args.packages:
        targets = combo_line_to_build_targets(p)
        print(f'Packages                   : {p}')
        print(f'# packages                 : {len(targets)}')
        mulled_name = v2_image_name(targets)
        print(f'Mulled name                : {mulled_name}')
        if ':' in mulled_name:
            packages_hash, versions_hash = mulled_name.split(':')
        else:
            packages_hash, versions_hash = mulled_name, ''
        print(f'Image Name                 : {packages_hash}')
        print(f'Version/Tag Name           : {versions_hash}')
        print(f'Quay.io URL                : https://quay.io/repository/biocontainers/{packages_hash}')
        print(f'Docker pull command (quess): quay.io/biocontainers/{mulled_name}-0')
        print()


if __name__ == '__main__':
    main()