def test_filter_entries():
"""Test filter_entries."""
config = NgdConfig()
with open(_get_file('assembly_status.txt'), 'r') as fh:
entries = list(core.parse_summary(fh))
assert core.filter_entries(entries, config) == entries
expected = entries[-1:]
config.assembly_accessions = "GCF_000203835.1"
assert core.filter_entries(entries, config) == expected
def prepare_create_downloadjob(req,
tmpdir,
format_map=NgdConfig._FORMATS,
human_readable=False,
create_local_file=False):
# Set up test env
entry = {
'assembly_accession': 'FAKE0.1',
'organism_name': 'Example species',
'infraspecific_name': 'strain=ABC 1234',
'ftp_path': 'https://fake/genomes/FAKE0.1'
}
config = NgdConfig()
outdir = tmpdir.mkdir('output')
download_jobs = []
config.output = str(outdir)
config.human_readable = human_readable
checksum_file_content = ''
for key, val in format_map.items():
seqfile = tmpdir.join('fake{}'.format(val))
seqfile.write(key)
checksum = core.md5sum(str(seqfile))
filename = path.basename(str(seqfile))
full_url = 'https://fake/genomes/FAKE0.1/{}'.format(filename)
local_file = outdir.join('refseq', 'bacteria', 'FAKE0.1', filename)
if create_local_file:
local_file.write(seqfile.read(), ensure=True)
symlink_path = None
if human_readable:
symlink_path = str(
outdir.join('human_readable', 'refseq', 'bacteria', 'Example',
'species', 'ABC_1234', filename))
download_jobs.append(
core.DownloadJob(full_url, str(local_file), checksum,
symlink_path))
checksum_file_content += '{}\t./{}\n'.format(checksum, filename)
req.get(full_url, text=seqfile.read())
req.get('https://fake/genomes/FAKE0.1/md5checksums.txt',
text=checksum_file_content)
return entry, config, download_jobs