#!/usr/bin/env python
+import logging
+
+logger = logging.getLogger('run-command')
+log_handler = logging.StreamHandler()
+log_handler.setFormatter(logging.Formatter("run-command: %(message)s"))
+logger.addHandler(log_handler)
+logger.setLevel(logging.INFO)
+
import arvados
import re
import os
import subprocess
import sys
import shutil
-import subst
+import crunchutil.subst as subst
import time
import arvados.commands.put as put
import signal
import traceback
import pprint
import multiprocessing
-import logging
-import robust_put
-import vwd
+import crunchutil.robust_put as robust_put
+import crunchutil.vwd as vwd
os.umask(0077)
-logging.basicConfig(format="run-command: %(message)s")
t = arvados.current_task().tmpdir
fn = subst.do_substitution(p, value)
mode = os.stat(fn).st_mode
prefix = fn[len(os.environ['TASK_KEEPMOUNT'])+1:]
- if mode != None:
+ if mode is not None:
if stat.S_ISDIR(mode):
items = ["$(dir %s/%s/)" % (prefix, l) for l in os.listdir(fn)]
elif stat.S_ISREG(mode):
stdoutname = None
stdoutfile = None
+stdinname = None
+stdinfile = None
rcode = 1
try:
if arvados.current_task()['sequence'] == 0:
var = jobp["task.foreach"]
items = get_items(jobp, jobp[var])
- logging.info("parallelizing on %s with items %s" % (var, items))
- if items != None:
+ logger.info("parallelizing on %s with items %s" % (var, items))
+ if items is not None:
for i in items:
params = copy.copy(jobp)
params[var] = i
else:
taskp = jobp
- cmd = expand_list(taskp, taskp["command"])
-
if "task.vwd" in taskp:
# Populate output directory with symlinks to files in collection
vwd.checkout(subst.do_substitution(taskp, taskp["task.vwd"]), outdir)
- if "save.stdout" in taskp:
- stdoutname = subst.do_substitution(taskp, taskp["save.stdout"])
- stdoutfile = open(stdoutname, "wb")
-
if "task.cwd" in taskp:
os.chdir(subst.do_substitution(taskp, taskp["task.cwd"]))
- logging.info("{}{}".format(' '.join(cmd), (" > " + stdoutname) if stdoutname != None else ""))
+ cmd = expand_list(taskp, taskp["command"])
+
+ if "task.stdin" in taskp:
+ stdinname = subst.do_substitution(taskp, taskp["task.stdin"])
+ stdinfile = open(stdinname, "rb")
+ if "task.stdout" in taskp:
+ stdoutname = subst.do_substitution(taskp, taskp["task.stdout"])
+ stdoutfile = open(stdoutname, "wb")
+
+ logger.info("{}{}{}".format(' '.join(cmd), (" < " + stdinname) if stdinname is not None else "", (" > " + stdoutname) if stdoutname is not None else ""))
+except subst.SubstitutionError as e:
+ logger.error(str(e))
+ logger.error("task parameters were:")
+ logger.error(pprint.pformat(taskp))
+ sys.exit(1)
except Exception as e:
- logging.exception("caught exception")
- logging.error("task parameters was:")
- logging.error(pprint.pformat(taskp))
+ logger.exception("caught exception")
+ logger.error("task parameters were:")
+ logger.error(pprint.pformat(taskp))
sys.exit(1)
try:
- sp = subprocess.Popen(cmd, shell=False, stdout=stdoutfile)
+ sp = subprocess.Popen(cmd, shell=False, stdin=stdinfile, stdout=stdoutfile)
sig = SigHandler()
# forward signals to the process.
# wait for process to complete.
rcode = sp.wait()
- if sig.sig != None:
- logging.critical("terminating on signal %s" % sig.sig)
+ if sig.sig is not None:
+ logger.critical("terminating on signal %s" % sig.sig)
sys.exit(2)
else:
- logging.info("completed with exit code %i (%s)" % (rcode, "success" if rcode == 0 else "failed"))
+ logger.info("completed with exit code %i (%s)" % (rcode, "success" if rcode == 0 else "failed"))
except Exception as e:
- logging.exception("caught exception")
+ logger.exception("caught exception")
# restore default signal handlers.
signal.signal(signal.SIGINT, signal.SIG_DFL)
for l in links:
os.unlink(l)
-logging.info("the following output files will be saved to keep:")
+logger.info("the following output files will be saved to keep:")
subprocess.call(["find", ".", "-type", "f", "-printf", "run-command: %12.12s %h/%f\\n"], stdout=sys.stderr)
-logging.info("start writing output to keep")
+logger.info("start writing output to keep")
if "task.vwd" in taskp:
if "task.foreach" in jobp:
# Just a single task, so do merge with the original collection
outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=True).manifest_text()
else:
- outcollection = robust_put.upload(outdir)
+ outcollection = robust_put.upload(outdir, logger)
api.job_tasks().update(uuid=arvados.current_task()['uuid'],
body={