Hack for updating Galaxy tool_dependencies.xml files with depot.galaxyproject.org URLs

update_tool_dependencies.py

# Copyright 2016 Peter Cock, James Hutton Institute.
# All Rights Reserved.
# Released as open source under the MIT license.
"""Update URLs in Galaxy tool_dependencies.xml files

Assumes have a copy of the urls.tsv file from
https://github.com/galaxyproject/cargo-port/blob/master/urls.tsv

This file defines the https://depot.galaxyproject.org/software/
URL naming used for caching Galaxy dependencies.

The script will check or insert the SHA256 checksum, and will
preserve the download filename (since the depot renames things).

Sample usage:

$ python update_tool_dependencies.py ../galaxy_blast/packages/*/tool_dependencies.xml

WARNING: Edits files in place!
"""
import os
import sys

def load_urls(filename):
    mapping = dict()
    for line in open(filename):
        if line.startswith("#"):
            continue
        parts = line.rstrip("\n").split("\t")
        id, version, platform, arch, old_url, ext, sha256sum, _ = line.rstrip("\n").split("\t", 7)
        if ext:
            assert ext.startswith("."), line
        new_url = "https://depot.galaxyproject.org/software/%s/%s_%s_%s_%s%s" % (id, id, version, platform, arch, ext)
        mapping[old_url] = new_url, sha256sum
    return mapping

def update_tool_dependency(filename, mapping):
    print("Updating %s" % filename)
    lines = list(open(filename))

    with open(filename, "w") as handle:
        for line in lines:
            if "<action " in line and "download" in line:
                indent = " " * line.index("<action ")
                assert not indent.strip(), line
                for old, (new, checksum) in mapping.items():
                    if old in line:
                        base = os.path.basename(old)
                        # TODO - escape URLs, e.g. ampersand
                        assert '>%s</action>' % old in line
                        if 'target_filename=' not in line and not new.endswith("/" + base):
                            # Preserve the original filename
                            line = line.replace('>%s</action>' % old,
                                                ' target_filename="%s">%s</action>' % (base, old))
                        if 'sha256sum=' in line:
                            # Check checksum matches
                            assert 'sha256sum="%s"' % checksum in line
                        else:
                            # Record the checksum
                            line = line.replace('>%s</action>' % old,
                                                ' sha256sum="%s">%s</action>' % (checksum, old))
                        # Finally, replace the URL
                        line = line.replace('>%s</action>' % old,
                                            '>%s</action>' % new)
                        print("Updated %s --> %s" % (old, new))
                        handle.write("%s<!-- Original URL %s -->\n" % (indent, old))
            handle.write(line)

mapping = load_urls("urls.tsv")
for filename in sys.argv[1:]:
    update_tool_dependency(filename, mapping)