5 logger = logging.getLogger('run-command')
6 log_handler = logging.StreamHandler()
7 log_handler.setFormatter(logging.Formatter("run-command: %(message)s"))
8 logger.addHandler(log_handler)
9 logger.setLevel(logging.INFO)
17 import crunchutil.subst as subst
19 import arvados.commands.put as put
25 import multiprocessing
26 import crunchutil.robust_put as robust_put
27 import crunchutil.vwd as vwd
31 t = arvados.current_task().tmpdir
33 api = arvados.api('v1')
35 os.chdir(arvados.current_task().tmpdir)
44 jobp = arvados.current_job()['script_parameters']
45 if len(arvados.current_task()['parameters']) > 0:
46 taskp = arvados.current_task()['parameters']
51 return os.path.join(arvados.current_task().tmpdir, 'tmpdir')
57 return str(multiprocessing.cpu_count())
60 return os.environ['JOB_UUID']
63 return os.environ['TASK_UUID']
66 return os.environ['CRUNCH_SRC']
68 subst.default_subs["task.tmpdir"] = sub_tmpdir
69 subst.default_subs["task.outdir"] = sub_outdir
70 subst.default_subs["job.srcdir"] = sub_jobsrc
71 subst.default_subs["node.cores"] = sub_cores
72 subst.default_subs["job.uuid"] = sub_jobid
73 subst.default_subs["task.uuid"] = sub_taskid
75 class SigHandler(object):
79 def send_signal(self, sp, signum):
80 sp.send_signal(signum)
83 def expand_item(p, c):
84 if isinstance(c, dict):
85 if "foreach" in c and "command" in c:
87 items = get_items(p, p[var])
92 r.extend(expand_list(params, c["command"]))
94 elif isinstance(c, list):
95 return expand_list(p, c)
96 elif isinstance(c, str) or isinstance(c, unicode):
97 return [subst.do_substitution(p, c)]
101 def expand_list(p, l):
102 return [exp for arg in l for exp in expand_item(p, arg)]
104 def add_to_group(gr, match):
105 m = ('^_^').join(match.groups())
108 gr[m].append(match.group(0))
110 def get_items(p, value):
111 if isinstance(value, dict):
112 if "filter" in value and "regex" in value:
113 pattern = re.compile(value["regex"])
114 items = get_items(p, value["group"])
115 return [i for i in items if pattern.match(i)]
117 if "group" in value and "regex" in value:
118 pattern = re.compile(value["regex"])
119 items = get_items(p, value["group"])
124 add_to_group(groups, p)
126 return [groups[k] for k in groups]
128 if isinstance(value, list):
129 return expand_list(p, value)
131 fn = subst.do_substitution(p, value)
132 mode = os.stat(fn).st_mode
133 prefix = fn[len(os.environ['TASK_KEEPMOUNT'])+1:]
135 if stat.S_ISDIR(mode):
136 items = ["$(dir %s/%s/)" % (prefix, l) for l in os.listdir(fn)]
137 elif stat.S_ISREG(mode):
139 items = [line.rstrip("\r\n") for line in f]
150 def recursive_foreach(params, fvars):
153 items = get_items(params, params[var])
154 logger.info("parallelizing on %s with items %s" % (var, items))
155 if items is not None:
157 params = copy.copy(params)
160 recursive_foreach(params, fvars)
162 arvados.api().job_tasks().create(body={
163 'job_uuid': arvados.current_job()['uuid'],
164 'created_by_job_task_uuid': arvados.current_task()['uuid'],
170 logger.error("parameter %s with value %s in task.foreach yielded no items" % (var, params[var]))
174 if "task.foreach" in jobp:
175 if arvados.current_task()['sequence'] == 0:
176 # This is the first task to start the other tasks and exit
177 fvars = jobp["task.foreach"]
178 if isinstance(fvars, basestring):
180 if not isinstance(fvars, list) or len(fvars) == 0:
181 logger.error("value of task.foreach must be a string or non-empty list")
183 recursive_foreach(jobp, jobp["task.foreach"])
184 if "task.vwd" in jobp:
185 # Set output of the first task to the base vwd collection so it
186 # will be merged with output fragments from the other tasks by
188 arvados.current_task().set_output(subst.do_substitution(jobp, jobp["task.vwd"]))
190 arvados.current_task().set_output(None)
193 # This is the only task so taskp/jobp are the same
196 if "task.vwd" in taskp:
197 # Populate output directory with symlinks to files in collection
198 vwd.checkout(subst.do_substitution(taskp, taskp["task.vwd"]), outdir)
200 if "task.cwd" in taskp:
201 os.chdir(subst.do_substitution(taskp, taskp["task.cwd"]))
203 cmd = expand_list(taskp, taskp["command"])
205 if "task.stdin" in taskp:
206 stdinname = subst.do_substitution(taskp, taskp["task.stdin"])
207 stdinfile = open(stdinname, "rb")
209 if "task.stdout" in taskp:
210 stdoutname = subst.do_substitution(taskp, taskp["task.stdout"])
211 stdoutfile = open(stdoutname, "wb")
213 logger.info("{}{}{}".format(' '.join(cmd), (" < " + stdinname) if stdinname is not None else "", (" > " + stdoutname) if stdoutname is not None else ""))
214 except subst.SubstitutionError as e:
216 logger.error("task parameters were:")
217 logger.error(pprint.pformat(taskp))
219 except Exception as e:
220 logger.exception("caught exception")
221 logger.error("task parameters were:")
222 logger.error(pprint.pformat(taskp))
226 sp = subprocess.Popen(cmd, shell=False, stdin=stdinfile, stdout=stdoutfile)
229 # forward signals to the process.
230 signal.signal(signal.SIGINT, lambda signum, frame: sig.send_signal(sp, signum))
231 signal.signal(signal.SIGTERM, lambda signum, frame: sig.send_signal(sp, signum))
232 signal.signal(signal.SIGQUIT, lambda signum, frame: sig.send_signal(sp, signum))
234 # wait for process to complete.
237 if sig.sig is not None:
238 logger.critical("terminating on signal %s" % sig.sig)
241 logger.info("completed with exit code %i (%s)" % (rcode, "success" if rcode == 0 else "failed"))
243 except Exception as e:
244 logger.exception("caught exception")
246 # restore default signal handlers.
247 signal.signal(signal.SIGINT, signal.SIG_DFL)
248 signal.signal(signal.SIGTERM, signal.SIG_DFL)
249 signal.signal(signal.SIGQUIT, signal.SIG_DFL)
254 logger.info("the following output files will be saved to keep:")
256 subprocess.call(["find", ".", "-type", "f", "-printf", "run-command: %12.12s %h/%f\\n"], stdout=sys.stderr)
258 logger.info("start writing output to keep")
260 if "task.vwd" in taskp:
261 if "task.foreach" in jobp:
262 # This is a subtask, so don't merge with the original collection, that will happen at the end
263 outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=False).manifest_text()
265 # Just a single task, so do merge with the original collection
266 outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=True).manifest_text()
268 outcollection = robust_put.upload(outdir, logger)
270 api.job_tasks().update(uuid=arvados.current_task()['uuid'],
272 'output': outcollection,
273 'success': (rcode == 0),