def test_run_all_success(self):
temp_dir = tempfile.mkdtemp()
try:
fakeftp = FtpFileTransfer(None)
mftp = D3RParameters()
fakeftp.set_connection(mftp)
fakeftp.set_remote_dir('/foo2')
mftp.get = Mock()
params = D3RParameters()
params.pdbfileurl = 'file://' + temp_dir
params.compinchi = 'file://' + temp_dir
make_blast = MakeBlastDBTask(temp_dir, params)
make_blast.create_dir()
open(os.path.join(make_blast.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
task = DataImportTask(temp_dir, params)
task.set_file_transfer(fakeftp)
task._retrysleep = 0
open(os.path.join(temp_dir, task.NONPOLYMER_TSV), 'a').close()
open(os.path.join(temp_dir, task.SEQUENCE_TSV), 'a').close()
open(os.path.join(temp_dir, task.OLDSEQUENCE_TSV), 'a').close()
open(os.path.join(temp_dir, task.CRYSTALPH_TSV), 'a').close()
open(os.path.join(temp_dir, task.COMPINCHI_ICH), 'a').close()
task.run()
self.assertEquals(task.get_error(), None)
# check line count is 1 now which indicates
# standard was added
self.assertEqual(
util.get_file_line_count(task.get_nonpolymer_tsv()), 1)
self.assertEqual(util.get_file_line_count(task.get_sequence_tsv()),
1)
self.assertEqual(
util.get_file_line_count(task.get_oldsequence_tsv()), 1)
self.assertEqual(
util.get_file_line_count(task.get_crystalph_tsv()), 1)
mftp.get.assert_called_with('/foo2/' +
DataImportTask.PARTICIPANT_LIST_CSV,
local=task.get_participant_list_csv())
finally:
shutil.rmtree(temp_dir)
def test_run_all_success_except_participant_download_fails(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.pdbfileurl = 'file://' + temp_dir
params.compinchi = 'file://' + temp_dir
make_blast = MakeBlastDBTask(temp_dir, params)
make_blast.create_dir()
open(os.path.join(make_blast.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
task = DataImportTask(temp_dir, params)
task._retrysleep = 0
open(os.path.join(temp_dir, task.NONPOLYMER_TSV), 'a').close()
open(os.path.join(temp_dir, task.SEQUENCE_TSV), 'a').close()
open(os.path.join(temp_dir, task.OLDSEQUENCE_TSV), 'a').close()
open(os.path.join(temp_dir, task.CRYSTALPH_TSV), 'a').close()
open(os.path.join(temp_dir, task.COMPINCHI_ICH), 'a').close()
task.run()
self.assertEquals(task.get_error(), None)
# check line count is 1 now which indicates
# standard was added
self.assertEqual(
util.get_file_line_count(task.get_nonpolymer_tsv()), 1)
self.assertEqual(util.get_file_line_count(task.get_sequence_tsv()),
1)
self.assertEqual(
util.get_file_line_count(task.get_oldsequence_tsv()), 1)
self.assertEqual(
util.get_file_line_count(task.get_crystalph_tsv()), 1)
self.assertTrue(task.get_email_log().startswith(
'\nWARNING: Unable to download'))
finally:
shutil.rmtree(temp_dir)
def run(self):
"""Runs blastnfilter task after verifying dataimport was good
Method requires can_run() to be called before hand with
successful outcome
Otherwise method invokes D3RTask.start then this method
creates a directory and invokes blastnfilter script and
postanalysis script. Upon completion results are
analyzed and success or error status is set
appropriately and D3RTask.end is invoked
"""
super(BlastNFilterTask, self).run()
if self._can_run is False:
logger.debug(self.get_dir_name() +
' cannot run cause _can_run flag '
'is False')
return
data_import = DataImportTask(self._path, self._args)
make_blastdb = MakeBlastDBTask(self._path, self._args)
try:
loglevel = self.get_args().loglevel
except AttributeError:
logger.debug('No log level set in arguments using WARNING')
loglevel = 'WARNING'
# verify sequence.tsv file exists on filesystem.
# if not fall back to oldsequence.tsv file
sequencetsv = data_import.get_sequence_tsv()
if not os.path.isfile(sequencetsv):
logger.warning(sequencetsv + ' file not found. falling '
'back to old file')
self.append_to_email_log('\n ' + sequencetsv + ' file not found ' +
'falling back to ' +
data_import.get_oldsequence_tsv() + '\n')
sequencetsv = data_import.get_oldsequence_tsv()
cmd_to_run = (self.get_args().blastnfilter + ' --nonpolymertsv ' +
data_import.get_nonpolymer_tsv() + ' --sequencetsv ' +
sequencetsv + ' --pdbblastdb ' + make_blastdb.get_dir() +
' --compinchi ' +
data_import.get_components_inchi_file() +
' --crystalpH ' + data_import.get_crystalph_tsv() +
' --pdbdb ' + self.get_args().pdbdb + ' --log ' +
loglevel + ' --outdir ' + self.get_dir())
blastnfilter_name = os.path.basename(self.get_args().blastnfilter)
self.run_external_command(
blastnfilter_name,
cmd_to_run,
False,
)
self.set_status(D3RTask.COMPLETE_STATUS)
cmd_to_run = (self.get_args().postanalysis + ' --compinchi ' +
data_import.get_components_inchi_file() + ' ' +
self.get_dir())
postanalysis_name = os.path.basename(self.get_args().postanalysis)
self.run_external_command(postanalysis_name, cmd_to_run, False)
try:
# examine output to get candidate hit count DR-12
hit_stats = self._parse_blastnfilter_output_for_hit_stats()
if hit_stats is not None:
self.append_to_email_log(hit_stats)
except Exception:
logger.exception("Error caught exception")
# assess the result
self.end()
def test_run_with_blast_success_useoldseq_and_postanalysis_fail(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.blastnfilter = '/bin/echo'
params.postanalysis = os.path.join(temp_dir, 'foo.py')
params.pdbdb = '/pdbdb'
blasttask = BlastNFilterTask(temp_dir, params)
blasttask._can_run = True
txt_file = os.path.join(blasttask.get_dir(), 'summary.txt')
txt_contents = ('INPUT SUMMARY\\n' + ' sequences: 177\\n' +
' complexes: 149\\n')
# create fake blastnfilter script that makes csv files
f = open(params.postanalysis, 'w')
f.write('#! /usr/bin/env python\n\n')
f.write('f = open(\'' + txt_file + '\', \'w\')\n')
f.write('f.write(\'' + txt_contents + '\\n\')\n')
f.write('f.flush()\nf.close()\n')
f.flush()
f.close()
os.chmod(params.postanalysis, stat.S_IRWXU)
blasttask.run()
self.assertEqual(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEqual(blasttask.get_error(), None)
complete_file = os.path.join(blasttask.get_dir(),
D3RTask.COMPLETE_FILE)
self.assertEqual(os.path.isfile(complete_file), True)
std_err_file = os.path.join(blasttask.get_dir(), 'echo.stderr')
self.assertEqual(os.path.isfile(std_err_file), True)
std_out_file = os.path.join(blasttask.get_dir(), 'echo.stdout')
dataimport = DataImportTask(temp_dir, params)
makeblast = MakeBlastDBTask(temp_dir, params)
f = open(std_out_file, 'r')
echo_out = f.read().replace('\n', '')
echo_out.index('--nonpolymertsv ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.NONPOLYMER_TSV))
echo_out.index(' --sequencetsv ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.OLDSEQUENCE_TSV))
echo_out.index(' --pdbblastdb ' +
os.path.join(temp_dir, makeblast.get_dir_name()))
echo_out.index(' --compinchi ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.COMPINCHI_ICH))
echo_out.index(' --outdir ' +
os.path.join(temp_dir, blasttask.get_dir_name()))
echo_out.index(' --crystalpH ' +
os.path.join(temp_dir, dataimport.get_dir_name(),
DataImportTask.CRYSTALPH_TSV))
echo_out.index(' --pdbdb /pdbdb ')
f.close()
self.assertEqual(os.path.isfile(std_out_file), True)
self.assertEquals(blasttask.get_status(), D3RTask.COMPLETE_STATUS)
self.assertEquals(
os.path.exists(
os.path.join(blasttask.get_dir(), 'foo.py.stderr')), True)
self.assertEquals(
os.path.exists(
os.path.join(blasttask.get_dir(), 'foo.py.stdout')), True)
res = blasttask.get_email_log().rstrip('\n')
res.index('/bin/echo')
res.index('# txt files found: 0')
res.index('Output from summary.txt')
res.index(' sequences: 177')
res.index(' complexes: 149')
res.index(dataimport.get_sequence_tsv() +
' file not found falling back to ' +
dataimport.get_oldsequence_tsv())
finally:
shutil.rmtree(temp_dir)