def test_generate_post_evaluation_email_body(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.version = '1.2.3'
weekyear = os.path.join(temp_dir, '2017', 'dataset.week.12')
os.makedirs(weekyear, mode=0o755)
task = PostEvaluationTask(weekyear, params)
task.create_dir()
f = open(task.get_summary_txt(), 'w')
f.write('hello')
f.flush()
f.close()
pee = PostEvaluationEmailer(None, None)
subject, msg = pee._generate_post_evaluation_email_body(task)
self.assertEqual(
subject, D3RTask.SUBJECT_LINE_PREFIX +
'2017 week 12 post evaluation summary report')
self.assertEqual(
msg, 'Dear CELPP Admin,\n\nHere is the post '
'evaluation summary reports for CELPP '
'week 12\n\n\n\nhello\n\n\nSincerely,\n\n'
'CELPP Automation 1.2.3')
finally:
shutil.rmtree(temp_dir)
def test_get_summary_txt(self):
params = D3RParameters()
task = PostEvaluationTask('/foo', params)
self.assertEqual(
task.get_summary_txt(),
os.path.join(task.get_dir(), PostEvaluationTask.SUMMARY_TXT))
def test_run_fails_one_evaluation_and_no_emailer(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.postevaluation = 'false'
etask = EvaluationTask(
temp_dir, 'foo_dock.' + EvaluationTaskFactory.SCORING_SUFFIX,
None, params)
etask.create_dir()
task = PostEvaluationTask(temp_dir, params)
task.run()
self.assertEqual(
task.get_error(), 'Non zero exit code: 1 received. '
'Standard out: Standard error: ')
self.assertEqual(task.get_status(), D3RTask.ERROR_STATUS)
self.assertTrue(
os.path.isfile(os.path.join(task.get_dir(),
D3RTask.ERROR_FILE)))
self.assertTrue('Caught exception trying to '
'send evaluation email' in task.get_email_log())
stdout_file = os.path.join(task.get_dir(),
'false' + D3RTask.STDOUT_SUFFIX)
self.assertTrue(os.path.isfile(stdout_file))
stderr_file = os.path.join(task.get_dir(),
'false' + D3RTask.STDERR_SUFFIX)
self.assertTrue(os.path.isfile(stderr_file))
finally:
shutil.rmtree(temp_dir)
def test_get_all_evaluation_tasks(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
# no evaluation tasks
task = PostEvaluationTask(temp_dir, params)
res = task.get_all_evaluation_tasks()
self.assertEqual(len(res), 0)
# directory in path that does NOT match suffix
task.create_dir()
os.makedirs(os.path.join(temp_dir, 'stage.x.1'), mode=0o755)
res = task.get_all_evaluation_tasks()
self.assertEqual(len(res), 0)
# one task, but is file (weird case)
weirdtaskfile = os.path.join(
temp_dir, 'stage.7.hi.' + EvaluationTaskFactory.SCORING_SUFFIX)
open(weirdtaskfile, 'a').close()
res = task.get_all_evaluation_tasks()
self.assertEqual(len(res), 0)
self.assertEqual(
task.get_email_log(),
'Just a note, found a task with valid name, but '
'it is not a directory ' + weirdtaskfile)
# one task
etask = EvaluationTask(temp_dir, 'foo.evaluation', None, params)
etask.create_dir()
res = task.get_all_evaluation_tasks()
self.assertEqual(len(res), 1)
self.assertTrue(res[0].get_dir_name(), etask.get_dir_name())
# three tasks two with non complete status
etask2 = EvaluationTask(temp_dir,
'bla_dock.extsubmission.evaluation', None,
params)
etask2.create_dir()
open(os.path.join(etask2.get_dir(), D3RTask.ERROR_FILE),
'a').close()
etask3 = EvaluationTask(temp_dir, '12_x.extsubmission.evaluation',
None, params)
etask3.create_dir()
open(os.path.join(etask3.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
res = task.get_all_evaluation_tasks()
self.assertEqual(len(res), 3)
d_name_list = []
for entry in res:
d_name_list.append(entry.get_dir_name())
self.assertTrue(etask.get_dir_name() in d_name_list)
self.assertTrue(etask2.get_dir_name() in d_name_list)
self.assertTrue(etask3.get_dir_name() in d_name_list)
finally:
shutil.rmtree(temp_dir)
def test_send_postevaluation_email_exception(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
weekyear = os.path.join(temp_dir, '2017', 'dataset.week.12')
os.makedirs(weekyear, mode=0o755)
task = PostEvaluationTask(weekyear, params)
task.create_dir()
csvfile = os.path.join(task.get_dir(), 'foo.csv')
open(csvfile, 'a').close()
smtpemailer = SmtpEmailer()
mockserver = D3RParameters()
mockserver.sendmail = Mock(side_effect=IOError('ha'))
mockserver.quit = Mock()
smtpemailer.set_alternate_smtp_server(mockserver)
pee = PostEvaluationEmailer(['*****@*****.**'], smtpemailer)
task.set_evaluation_emailer(pee)
f = open(task.get_summary_txt(), 'w')
f.write('hello')
f.flush()
f.close()
pee.send_postevaluation_email(task)
mockserver.quit.assert_any_call()
self.assertEqual(
pee.get_message_log(), '\nCaught exception trying to email :'
' [email protected] : Caught exception ha\n')
self.assertEqual(mockserver.sendmail.call_count, 1)
finally:
shutil.rmtree(temp_dir)
def test_send_postevaluation_email_success(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
weekyear = os.path.join(temp_dir, '2017', 'dataset.week.12')
os.makedirs(weekyear, mode=0o755)
task = PostEvaluationTask(weekyear, params)
task.create_dir()
smtpemailer = SmtpEmailer()
mockserver = D3RParameters()
mockserver.sendmail = Mock()
mockserver.quit = Mock()
smtpemailer.set_alternate_smtp_server(mockserver)
pee = PostEvaluationEmailer(['*****@*****.**'], smtpemailer)
task.set_evaluation_emailer(pee)
f = open(task.get_summary_txt(), 'w')
f.write('hello')
f.flush()
f.close()
pee.send_postevaluation_email(task)
mockserver.quit.assert_any_call()
self.assertEqual(pee.get_message_log(),
'\nSent post evaluation email to: [email protected]\n')
self.assertEqual(mockserver.sendmail.call_count, 1)
finally:
shutil.rmtree(temp_dir)
def test_get_evaluationdir_args(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
# no evaluation tasks
task = PostEvaluationTask(temp_dir, params)
res = task._get_evaluationdir_args()
self.assertEqual(res, '')
# one task
etask = EvaluationTask(temp_dir, 'foo.evaluation', None, params)
etask.create_dir()
res = task._get_evaluationdir_args()
self.assertEqual(
res, ' ' + PostEvaluationTask.EVALUATIONDIR_ARG + ' ' +
etask.get_dir())
# three tasks
etask2 = EvaluationTask(temp_dir,
'bla_dock.extsubmission.evaluation', None,
params)
etask2.create_dir()
open(os.path.join(etask2.get_dir(), D3RTask.ERROR_FILE),
'a').close()
etask3 = EvaluationTask(temp_dir, '12_x.extsubmission.evaluation',
None, params)
etask3.create_dir()
cfile = os.path.join(etask3.get_dir(), D3RTask.COMPLETE_FILE)
open(cfile, 'a').close()
res = task._get_evaluationdir_args()
self.assertTrue(' ' + PostEvaluationTask.EVALUATIONDIR_ARG + ' ' +
etask.get_dir() in res)
self.assertTrue(' ' + PostEvaluationTask.EVALUATIONDIR_ARG + ' ' +
etask2.get_dir() in res)
self.assertTrue(' ' + PostEvaluationTask.EVALUATIONDIR_ARG + ' ' +
etask3.get_dir() in res)
finally:
shutil.rmtree(temp_dir)
def test_run_fails_cause_can_run_is_false(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
task = PostEvaluationTask(temp_dir, params)
task.create_dir()
task.run()
self.assertEqual(
task.get_error(),
task.get_dir_name() + ' already exists and status is ' +
D3RTask.UNKNOWN_STATUS)
finally:
shutil.rmtree(temp_dir)
def test_run_fails_cause_postevaluation_not_set(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
task = PostEvaluationTask(temp_dir, params)
task.run()
self.assertEqual(task.get_error(), 'postevaluation not set')
# test files get created
self.assertEqual(os.path.isdir(task.get_dir()), True)
errfile = os.path.join(task.get_dir(), D3RTask.ERROR_FILE)
self.assertEqual(os.path.isfile(errfile), True)
finally:
shutil.rmtree(temp_dir)
def test_run_succeeds_one_evaluation_and_no_emailer(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.postevaluation = 'echo'
etask = EvaluationTask(
temp_dir, 'foo_dock.' + EvaluationTaskFactory.SCORING_SUFFIX,
None, params)
etask.create_dir()
task = PostEvaluationTask(temp_dir, params)
ctask = ChallengeDataTask(temp_dir, params)
task.run()
self.assertEqual(task.get_error(), None)
self.assertEqual(task.get_status(), D3RTask.COMPLETE_STATUS)
cfile = os.path.join(task.get_dir(), D3RTask.COMPLETE_FILE)
self.assertTrue(os.path.isfile(cfile))
stdout_file = os.path.join(task.get_dir(),
'echo' + D3RTask.STDOUT_SUFFIX)
self.assertTrue(os.path.isfile(stdout_file))
f = open(stdout_file, 'r')
data = f.read()
f.close()
ctaskdir = os.path.join(ctask.get_dir(),
ctask.get_celpp_challenge_data_dir_name())
self.assertTrue(' --evaluationdir ' + etask.get_dir() in data)
self.assertTrue(' --challengedir ' + ctaskdir in data)
self.assertTrue('--stageprefix stage.7. --evaluationsuffix ' +
'.extsubmission.evaluation$|.evaluation$' in data)
stderr_file = os.path.join(task.get_dir(),
'echo' + D3RTask.STDERR_SUFFIX)
self.assertTrue(os.path.isfile(stderr_file))
finally:
shutil.rmtree(temp_dir)
def test_run_fail_no_evaluations_and_no_emailer(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
params.postevaluation = 'true'
task = PostEvaluationTask(temp_dir, params)
task.run()
self.assertEqual(task.get_error(),
'Did not find any Evaluation tasks to summarize')
self.assertEqual(task.get_status(), D3RTask.ERROR_STATUS)
self.assertTrue(
os.path.isfile(os.path.join(task.get_dir(),
D3RTask.ERROR_FILE)))
finally:
shutil.rmtree(temp_dir)
def main(args):
"""Main entry into celpprunner
:param args: should be set to sys.argv which is a list of arguments
starting with script name as the first argument
"""
blasttask = BlastNFilterTask('', p)
dataimport = DataImportTask('', p)
challenge = ChallengeDataTask('', p)
glide = GlideTask('', p)
makedb = MakeBlastDBTask('', p)
prot = ProteinLigPrepTask('', p)
vina = AutoDockVinaTask('', p)
chimeraprep = ChimeraProteinLigPrepTask('', p)
postevalstage = PostEvaluationTask('', p)
desc = """
Version {version}
Runs the 11 stages (makedb, import, blast, challengedata,
proteinligprep, {chimeraprep}, extsubmission, glide, vina,
evaluation, & {postevaluation}) of CELPP processing pipeline
(http://www.drugdesigndata.org)
CELPP processing pipeline relies on a set of directories
with specific structure. The pipeline runs a set of stages
Each stage has a numerical value and a name. The numerical
value denotes order and the stage name identifies separate
tasks to run in the stage.
The filesystem structure of the stage is:
stage.<stage number>.<task name>
The stage(s) run are defined via the required {stageflag} flag.
To run multiple stages serially just pass a comma delimited
list to the {stageflag} flag. Example: {stageflag} import,blast
NOTE: When running multiple stages serially the program will
not run subsequent stages if a task in a stage fails.
Also note order matters, ie putting blast,import will
cause celpprunner.py to run blast stage first.
This program drops a pid lockfile
(celpprunner.<stage>.lockpid) in celppdir to prevent duplicate
invocation.
When run, this program will examine the stage and see
if work can be done. If stage is complete or previous
steps have not completed, the program will exit silently.
If previous steps have failed or current stage already
exists in an error or uncomplete state then program will
report the error via email using addresses set in --email
and in --summaryemail flag. Errors will also be reported
via stderr/stdout. The program will also exit with nonzero
exit code.
This program utilizes simple token files to denote stage
completion. If within the stage directory there is a:
'{complete}' file - then stage is done and no other
checking is done.
'error' file - then stage failed.
'start' file - then stage is running.
Notification of stage start and end will be sent to
addresses set via --email flag and only errors will
be sent to addresses in --summaryemail.
Emails are sent via smtp which can be configured via
{smtpconfig} parameter which takes a configuration
file in the following format:
[{smtp_section}]
{smtp_host} = smtp.foo.com
{smtp_port} = 1234
{smtp_user} = bob
{smtp_pass} = somepass
{smtp_from_address} = [email protected]
{smtp_replyto_address} = [email protected]
Unless --customweekdir is set, this program will
examine the 'celppdir' (last argument passed on
commandline) to find the latest directory with this path:
<year>/dataset.week.#
The program will find the latest <year> and within
that year the dataset.week.# with highest #. The output
directories created will be put within this directory.
Setting --customweekdir will cause program to use 'celppdir'
path.
Setting the --createweekdir flag will instruct this
program to create a new directory for the current
celpp week/year before running any stage processing.
NOTE: CELPP weeks start on Friday and end on Thursday
and week # follows ISO8601 rules so week numbers
at the end and start of the year are a bit
wonky.
Breakdown of behavior of program is defined by
value passed with {stageflag} flag:
If {stageflag} '{createchallenge}'
This is NOT a stage, but has the same effect as
calling {stageflag} makedb,import,blast,challengedata
The four stages that need to run to generate the challenge
data package.
If {stageflag} 'makedb'
In this stage the file {pdb_seqres} is downloaded from
an ftp site set by --pdbsequrl.
This file is then gunzipped and NCBI makeblastdb
(set by --makeblastdb) is run on it to create a blast
database. The files are stored in {makeblastdb_dirname}
If {stageflag} 'import'
In this stage 4 files are downloaded from urls specified
by --compinchi and --pdbfileurl flags on the commandline
into {dataimport_dirname} directory.
The tsv files are (--pdbfileurl flag sets url to
download these files from):
{nonpolymer_tsv}
{sequence_tsv}
{crystal_tsv}
The Components ich file is (--compinchi flag sets base url to
download this file from):
{compinchi_ich}
This stage will just wait and retry if any of the tsv files
have NOT been updated since the start of the current
celpp week as determined by a HEAD request. To bypass
this delay add --skipimportwait flag. --importsleep denotes
the time to wait before re-examining the update time of the
tsv files and --importretry sets number of times to retry
before giving up.
If {stageflag} 'blast'
Verifies {dataimport_dirname} exists and has '{complete}'
file. Also verifies {makeblastdb_dirname} exists and has
'{complete}' file. If both conditions are met then the
'blast' stage is run which invokes script set by
--blastnfilter flag and output stored in
{blast_dirname}.
Requires --pdbdb to be set to a directory with valid PDB
database files.
Note: --blastnfilter script is killed after time set with
--blastnfiltertimeout flag.
If {stageflag} 'challengedata'
Verifies {blast_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --genchallenge flag to create a challenge dataset
file. The --pdbdb flag must also be set when calling this
stage. If --ftpconfig is set with {challengepath} field then
this stage will also upload the challenge dataset tarfile
to the ftp server with path set by {challengepath}. The
code will also upload a {latest_txt} file containing name
of the tarfile to the same destination overwriting any
{latest_txt} file that already exists.
Example file for --ftpconfig:
{host} some.ftp.com
{user} bob
{passn} mypass
{path} /celpp
{challengepath} /challenge
{submissionpath} /submissions
If {stageflag} '{chimeraprep}'
Verifies {challenge_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --chimeraprep flag to prepare pdb and inchi files
storing output in {chimeraprep_dirname}. --pdbdb flag
must also be set when calling this stage.
If {stageflag} 'proteinligprep'
Verifies {challenge_dirname} exists and has '{complete}'
file. If complete, this stage runs which invokes program
set in --proteinligprep flag to prepare pdb and inchi files
storing output in {proteinligprep_dirname}. --pdbdb flag
must also be set when calling this stage.
If {stageflag} 'extsubmission'
Connects to server specified by --ftpconfig and downloads
external docking submissions from {submissionpath} on remote
server.
Submissions should be named:
celpp_weekXX_YYYY_dockedresults_ZZZZ.tar.gz as documented here:
https://github.com/drugdata/d3r/wiki/Proposed-challenge-docked\
-results-file-structure
For each submission a directory named stage.X.ZZZZ.extsubmission
will be created and uncompressed contents of package will be
stored in that directory. If data does not conform properly
'error' file will be placed in directory denoting failure
If {stageflag} 'glide'
Verifies {proteinligprep_dirname} exists and has a '{complete}'
file within it. If complete, this stage runs which invokes
program set in --glide flag to perform docking via glide
storing output in {glide_dirname}
If {stageflag} 'vina'
Verifies {proteinligprep_dirname} exists and has a '{complete}'
file within it. If complete, this stage runs which invokes
program set in --vina flag to perform docking via AutoDock Vina
storing output in {vina_dirname}
If {stageflag} 'evaluation'
Finds all stage.{dockstage}.<algo> directories with '{complete}'
files in them which do not end in name '{webdata}' and runs
script set via --evaluation parameter storing the result of
the script into stage.{evalstage}.<algo>.evaluation. --pdbdb flag
must also be set when calling this stage.
If {stageflag} '{postevaluation}'
Finds all stage.{evalstage}.<algo>.evaluation directories and
runs script set via --postevaluation parameter storing a
summary of found docking evaluations into
{postevalstage} into a file named summary.txt and a set
of files with name Overall_RMSD_<candidate_type>.csv.
In addition, results will be emailed to people in
--summaryemail and --email lists.
""".format(makeblastdb_dirname=makedb.get_dir_name(),
dataimport_dirname=dataimport.get_dir_name(),
blast_dirname=blasttask.get_dir_name(),
challenge_dirname=challenge.get_dir_name(),
createchallenge=CREATE_CHALLENGE,
proteinligprep_dirname=prot.get_dir_name(),
glide_dirname=glide.get_dir_name(),
vina_dirname=vina.get_dir_name(),
dockstage=str(glide.get_stage()),
evalstage=str(glide.get_stage() + 1),
complete=blasttask.COMPLETE_FILE,
chimeraprep_dirname=chimeraprep.get_dir_name(),
chimeraprep=CHIMERA_PREP,
compinchi_ich=DataImportTask.COMPINCHI_ICH,
pdb_seqres=MakeBlastDBTask.PDB_SEQRES_TXT_GZ,
nonpolymer_tsv=DataImportTask.NONPOLYMER_TSV,
sequence_tsv=DataImportTask.SEQUENCE_TSV,
crystal_tsv=DataImportTask.CRYSTALPH_TSV,
webdata=EvaluationTaskFactory.WEB_DATA_SUFFIX,
latest_txt=ChallengeDataTask.LATEST_TXT,
host=FtpFileTransfer.HOST,
user=FtpFileTransfer.USER,
passn=FtpFileTransfer.PASS,
path=FtpFileTransfer.PATH,
challengepath=FtpFileTransfer.CHALLENGEPATH,
submissionpath=FtpFileTransfer.SUBMISSIONPATH,
version=d3r.__version__,
stageflag=STAGE_FLAG,
postevaluation=POST_EVAL,
postevalstage=postevalstage.get_dir_name(),
smtpconfig=SMTPCONFIG_FLAG,
smtp_section=SmtpConfig.DEFAULT,
smtp_host=SmtpConfig.SMTP_HOST,
smtp_port=SmtpConfig.SMTP_PORT,
smtp_user=SmtpConfig.SMTP_USER,
smtp_pass=SmtpConfig.SMTP_PASS,
smtp_from_address=SmtpConfig.SMTP_FROM_ADDRESS,
smtp_replyto_address=SmtpConfig.SMTP_REPLYTO_ADDRESS)
theargs = _parse_arguments(desc, args[1:])
theargs.program = args[0]
theargs.version = d3r.__version__
util.setup_logging(theargs)
try:
return run_stages(theargs)
except Exception:
logger.exception("Error caught exception")
return 2
def test_postevaluationtask_constructor(self):
params = D3RParameters()
task = PostEvaluationTask('/blah', params)
self.assertEquals(task.get_name(), PostEvaluationTask.TASK_NAME)
self.assertEquals(task.get_stage(), 8)
self.assertEqual(task.get_status(), D3RTask.UNKNOWN_STATUS)
def test_get_all_csv_files_in_dir(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
task = PostEvaluationTask(temp_dir, params)
task.create_dir()
# test empty dir
res = task.get_all_csv_files_in_dir()
self.assertEqual(len(res), 0)
# add directory with csv ending
csvdir = os.path.join(task.get_dir(),
'hi ' + PostEvaluationTask.CSV_SUFFIX)
os.makedirs(csvdir, mode=0o755)
res = task.get_all_csv_files_in_dir()
self.assertEqual(len(res), 0)
# 1 csv
onecsv = os.path.join(
task.get_dir(),
'Overall_RMSD_hi' + PostEvaluationTask.CSV_SUFFIX)
open(onecsv, 'a').close()
res = task.get_all_csv_files_in_dir()
self.assertEqual(len(res), 1)
self.assertTrue(onecsv in res)
# 3 csv
twocsv = os.path.join(
task.get_dir(),
'Overall_RMSD_bye' + PostEvaluationTask.CSV_SUFFIX)
open(twocsv, 'a').close()
threecsv = os.path.join(task.get_dir(),
'some' + PostEvaluationTask.CSV_SUFFIX)
open(threecsv, 'a').close()
res = task.get_all_csv_files_in_dir()
self.assertEqual(len(res), 3)
self.assertTrue(onecsv in res)
self.assertTrue(twocsv in res)
self.assertTrue(threecsv in res)
finally:
shutil.rmtree(temp_dir)
def test_get_postevaluation_summary(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
task = PostEvaluationTask(temp_dir, params)
task.create_dir()
# no file
res = task.get_postevaluation_summary()
self.assertEqual(
res, '\n\nNo ' + task.get_summary_txt() + ' file found\n')
# empty file
s_file = os.path.join(task.get_dir(),
PostEvaluationTask.SUMMARY_TXT)
open(s_file, 'a').close()
res = task.get_postevaluation_summary()
self.assertEqual(res, '\n\n\n')
os.chmod(task.get_summary_txt(), 0)
res = task.get_postevaluation_summary()
self.assertTrue('\n\nUnable to generate postevaluation' in res)
os.chmod(task.get_summary_txt(), stat.S_IRWXU)
# some data in the file
f = open(s_file, 'w')
f.write('hello\nhi\n')
f.flush()
f.close()
res = task.get_postevaluation_summary()
self.assertEqual(res, '\n\nhello\nhi\n\n')
finally:
shutil.rmtree(temp_dir)
def test_run_succeeds_one_evaluation_with_emailer(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
foo_script = os.path.join(temp_dir, 'foo.py')
f = open(foo_script, 'w')
f.write('#! /usr/bin/env python\n')
f.write('import sys\nimport os\n')
f.write('outdir = sys.argv[1]\n')
f.write('sys.stdout.write(" ".join(sys.argv))\n')
f.write('f = open(os.path.join(outdir, "summary.txt"),"w")\n')
f.write('f.write("summary")\n')
f.write('f.flush()\nf.close()\n')
f.write('f = open(os.path.join(outdir, "yo.csv"), "a")\n')
f.write('f.write("hi")\nf.flush()\nf.close()\n')
f.write('sys.exit(0)\n')
f.flush()
f.close()
os.chmod(foo_script, stat.S_IRWXU)
params.postevaluation = foo_script
etask = EvaluationTask(
temp_dir, 'foo_dock.' + EvaluationTaskFactory.SCORING_SUFFIX,
None, params)
etask.create_dir()
task = PostEvaluationTask(temp_dir, params)
# setup emailer
smtpemailer = SmtpEmailer()
mockserver = D3RParameters()
mockserver.sendmail = Mock()
mockserver.quit = Mock()
smtpemailer.set_alternate_smtp_server(mockserver)
pee = PostEvaluationEmailer(['*****@*****.**'], smtpemailer)
task.set_evaluation_emailer(pee)
ctask = ChallengeDataTask(temp_dir, params)
task.run()
self.assertEqual(task.get_error(), None)
self.assertEqual(task.get_status(), D3RTask.COMPLETE_STATUS)
cfile = os.path.join(task.get_dir(), D3RTask.COMPLETE_FILE)
self.assertTrue(os.path.isfile(cfile))
stdout_file = os.path.join(task.get_dir(),
'foo.py' + D3RTask.STDOUT_SUFFIX)
self.assertTrue(os.path.isfile(stdout_file))
f = open(stdout_file, 'r')
data = f.read()
f.close()
ctaskdir = os.path.join(ctask.get_dir(),
ctask.get_celpp_challenge_data_dir_name())
self.assertTrue(' --evaluationdir ' + etask.get_dir() in data)
self.assertTrue(' --challengedir ' + ctaskdir in data)
self.assertTrue('--stageprefix stage.7. --evaluationsuffix ' +
'.extsubmission.evaluation$|.evaluation$' in data)
stderr_file = os.path.join(task.get_dir(),
'foo.py' + D3RTask.STDERR_SUFFIX)
f = open(stdout_file, 'r')
data = f.read()
f.close()
self.assertTrue(' --evaluationdir ' + etask.get_dir() in data)
self.assertTrue(os.path.isfile(stderr_file))
self.assertEqual(task.get_error(), None)
self.assertTrue('Sent post evaluation email to: '
'*****@*****.**' in task.get_email_log())
finally:
shutil.rmtree(temp_dir)
def test_get_uploadable_files(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
task = PostEvaluationTask(temp_dir, params)
# try with no dir
self.assertEqual(task.get_uploadable_files(), [])
# try with empty dir
task.create_dir()
self.assertEqual(task.get_uploadable_files(), [])
# try with final log
final_log = os.path.join(task.get_dir(), EvaluationTask.FINAL_LOG)
open(final_log, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 1)
self.assertTrue(final_log in flist)
# try with csv file
csvfile = os.path.join(
task.get_dir(),
'Overall_RMSD_foo' + PostEvaluationTask.CSV_SUFFIX)
open(csvfile, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 2)
self.assertTrue(csvfile in flist)
self.assertTrue(final_log in flist)
# try with summary.txt file
open(task.get_summary_txt(), 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 3)
self.assertTrue(task.get_summary_txt() in flist)
self.assertTrue(final_log in flist)
# try with 2 csv file
csvfile2 = os.path.join(
task.get_dir(),
'Overall_RMSD_foo2' + PostEvaluationTask.CSV_SUFFIX)
open(csvfile2, 'a').close()
flist = task.get_uploadable_files()
self.assertEqual(len(flist), 4)
self.assertTrue(csvfile in flist)
self.assertTrue(csvfile2 in flist)
self.assertTrue(task.get_summary_txt() in flist)
self.assertTrue(final_log in flist)
finally:
shutil.rmtree(temp_dir)
def test_can_run(self):
temp_dir = tempfile.mkdtemp()
try:
params = D3RParameters()
# post evaluation task exists already
task = PostEvaluationTask(temp_dir, params)
task.create_dir()
self.assertEqual(task.can_run(), False)
self.assertEqual(
task.get_error(),
task.get_dir_name() + ' already exists and status is unknown')
# post evaluation task already complete
task = PostEvaluationTask(temp_dir, params)
open(os.path.join(task.get_dir(), D3RTask.COMPLETE_FILE),
'a').close()
self.assertEqual(task.can_run(), False)
self.assertEqual(task.get_error(), None)
finally:
shutil.rmtree(temp_dir)
def get_task_list_for_stage(theargs, stage_name):
"""Factory method that generates a list of tasks for given stage
Using stage_name get the list of tasks that need to
be run.
:param theargs: parameters set via commandline along with
``theargs.latest_weekly`` which should be set to
to base directory where stages will be run
:param stage_name: Name of stage to run
"""
if stage_name is None:
raise NotImplementedError('stage_name is None')
task_list = []
logger.debug('Getting task list for ' + stage_name)
if stage_name == CREATE_CHALLENGE:
task_list.append(MakeBlastDBTask(theargs.latest_weekly, theargs))
task_list.append(DataImportTask(theargs.latest_weekly, theargs))
task_list.append(BlastNFilterTask(theargs.latest_weekly, theargs))
task_list.append(ChallengeDataTask(theargs.latest_weekly, theargs))
if stage_name == 'makedb':
task_list.append(MakeBlastDBTask(theargs.latest_weekly, theargs))
if stage_name == 'import':
task_list.append(DataImportTask(theargs.latest_weekly, theargs))
if stage_name == 'blast':
task_list.append(BlastNFilterTask(theargs.latest_weekly, theargs))
if stage_name == 'challengedata':
task_list.append(ChallengeDataTask(theargs.latest_weekly, theargs))
if stage_name == 'proteinligprep':
task_list.append(ProteinLigPrepTask(theargs.latest_weekly, theargs))
if stage_name == 'glide':
task_list.append(GlideTask(theargs.latest_weekly, theargs))
if stage_name == 'vina':
task_list.append(AutoDockVinaTask(theargs.latest_weekly, theargs))
if stage_name == CHIMERA_PREP:
task_list.append(
ChimeraProteinLigPrepTask(theargs.latest_weekly, theargs))
if stage_name == 'extsubmission':
extfac = ExternalDataSubmissionFactory(theargs.latest_weekly, theargs)
task_list.extend(extfac.get_external_data_submissions())
if stage_name == 'evaluation':
# use util function call to get all evaluation tasks
# append them to the task_list
eval_task_factory = EvaluationTaskFactory(theargs.latest_weekly,
theargs)
task_list.extend(eval_task_factory.get_evaluation_tasks())
if stage_name == POST_EVAL:
# create PostEvaluationEmailer object
ptask = PostEvaluationTask(theargs.latest_weekly, theargs)
a_list = _get_set_of_email_address_from_email_flags(theargs)
efac = SmtpEmailerFactory(theargs)
pmailer = PostEvaluationEmailer(a_list, efac.get_smtp_emailer())
ptask.set_evaluation_emailer(pmailer)
task_list.append(ptask)
if len(task_list) is 0:
raise NotImplementedError('uh oh no tasks for ' + stage_name +
' stage')
return task_list