def test_create_downloadjob_symlink_only(req, tmpdir):
entry, config, joblist = prepare_create_downloadjob(req, tmpdir, human_readable=True,
create_local_file=True)
jobs = core.create_downloadjob(entry, 'bacteria', config)
expected = [core.DownloadJob(None, j.local_file, None, j.symlink_path)
for j in joblist if j.local_file.endswith('_genomic.gbff.gz')]
assert jobs == expected
def test_download_entry_symlink_only(req, tmpdir):
entry, outdir, joblist = prepare_download_entry(req, tmpdir, human_readable=True,
create_local_file=True)
jobs = core.download_entry(entry, 'refseq', 'bacteria', str(outdir), 'genbank', True)
expected = [(core.DownloadJob(None, j[0].local_file, None, j[0].symlink_path), None)
for j in joblist if j[0].local_file.endswith('_genomic.gbff.gz')]
assert jobs == expected
def prepare_download_entry(req, tmpdir, format_map=core.FORMAT_NAME_MAP):
# Set up test env
entry = {
'assembly_accession': 'FAKE0.1',
'ftp_path': 'http://fake/genomes/FAKE0.1'
}
outdir = tmpdir.mkdir('output')
download_jobs = []
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 = 'http://fake/genomes/FAKE0.1/{}'.format(filename)
local_file = str(outdir.join('refseq', 'bacteria', 'FAKE0.1',
filename))
download_jobs.append(core.DownloadJob(full_url, local_file, checksum))
checksum_file_content += '{}\t./{}\n'.format(checksum, filename)
req.get(full_url, text=seqfile.read())
req.get('http://fake/genomes/FAKE0.1/md5checksums.txt',
text=checksum_file_content)
return entry, outdir, download_jobs
def test_download_defaults(monkeypatch, mocker):
"""Test _download is called an appropriate number of times."""
worker_mock = mocker.MagicMock()
_download_mock = mocker.MagicMock(
return_value=[core.DownloadJob(None, None, None, None)])
monkeypatch.setattr(core, '_download', _download_mock)
monkeypatch.setattr(core, 'worker', worker_mock)
assert core.download() == 0
assert _download_mock.call_count == len(core.SUPPORTED_TAXONOMIC_GROUPS)
def test_download_metadata(monkeypatch, mocker, req, tmpdir):
"""Test creating the metadata file works."""
metadata_file = tmpdir.join('metadata.tsv')
summary_contents = open(_get_file('partial_summary.txt'), 'r').read()
req.get('https://ftp.ncbi.nih.gov/genomes/refseq/bacteria/assembly_summary.txt',
text=summary_contents)
mocker.spy(core, 'get_summary')
mocker.spy(core, 'parse_summary')
mocker.patch('ncbi_genome_download.core.create_downloadjob', return_value=[core.DownloadJob(None, None, None, None)])
core.download(group='bacteria', output='/tmp/fake', metadata_table=str(metadata_file))
assert core.get_summary.call_count == 1
assert core.parse_summary.call_count == 1
assert core.create_downloadjob.call_count == 4
assert metadata_file.check()
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
def test_args_download_defaults(monkeypatch, mocker):
"""Test args_download does the correct thing."""
entry = {
'assembly_accession': 'FAKE0.1',
'organism_name': 'Example species',
'infraspecific_name': 'strain=ABC 1234',
'ftp_path': 'https://fake/genomes/FAKE0.1'
}
worker_mock = mocker.MagicMock()
select_candidates_mock = mocker.MagicMock(return_value=[(entry, 'bacteria')])
create_downloadjob_mock = mocker.MagicMock(return_value=[core.DownloadJob(None, None, None, None)])
monkeypatch.setattr(core, 'select_candidates', select_candidates_mock)
monkeypatch.setattr(core, 'create_downloadjob', create_downloadjob_mock)
monkeypatch.setattr(core, 'worker', worker_mock)
assert core.args_download(Namespace()) == 0
assert select_candidates_mock.call_args_list[0][0][0].group == SUPPORTED_TAXONOMIC_GROUPS
assert create_downloadjob_mock.call_args_list[0][0][0] == entry
def test_download_dry_run(monkeypatch, mocker):
"""Test _download is not called for a dry run."""
entry = {
'assembly_accession': 'FAKE0.1',
'organism_name': 'Example species',
'infraspecific_name': 'strain=ABC 1234',
'ftp_path': 'https://fake/genomes/FAKE0.1'
}
worker_mock = mocker.MagicMock()
select_candidates_mock = mocker.MagicMock(return_value=[(entry, 'bacteria')])
create_downloadjob_mock = mocker.MagicMock(return_value=[core.DownloadJob(None, None, None, None)])
monkeypatch.setattr(core, 'select_candidates', select_candidates_mock)
monkeypatch.setattr(core, 'create_downloadjob', create_downloadjob_mock)
monkeypatch.setattr(core, 'worker', worker_mock)
assert core.download(dry_run=True) == 0
assert select_candidates_mock.call_count == 1
assert create_downloadjob_mock.call_count == 0
assert worker_mock.call_count == 0