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 if "list" in c and "index" in c:
96 items = get_items(p, p[var])
97 return items[int(c["index"])]
98 elif isinstance(c, list):
99 return expand_list(p, c)
100 elif isinstance(c, str) or isinstance(c, unicode):
101 return [subst.do_substitution(p, c)]
105 def expand_list(p, l):
106 return [exp for arg in l for exp in expand_item(p, arg)]
108 def add_to_group(gr, match):
109 m = ('^_^').join(match.groups())
112 gr[m].append(match.group(0))
114 def get_items(p, value):
115 if isinstance(value, dict):
116 if "filter" in value and "regex" in value:
117 pattern = re.compile(value["regex"])
118 items = get_items(p, value["group"])
119 return [i for i in items if pattern.match(i)]
121 if "group" in value and "regex" in value:
122 pattern = re.compile(value["regex"])
123 items = get_items(p, value["group"])
128 add_to_group(groups, p)
129 return [groups[k] for k in groups]
131 if "extract" in value and "regex" in value:
132 pattern = re.compile(value["regex"])
133 items = get_items(p, value["group"])
134 return [p.groups() for i in items if p = pattern.match(i)]
136 if isinstance(value, list):
137 return expand_list(p, value)
139 fn = subst.do_substitution(p, value)
140 mode = os.stat(fn).st_mode
141 prefix = fn[len(os.environ['TASK_KEEPMOUNT'])+1:]
143 if stat.S_ISDIR(mode):
144 items = ["$(dir %s/%s/)" % (prefix, l) for l in os.listdir(fn)]
145 elif stat.S_ISREG(mode):
147 items = [line.rstrip("\r\n") for line in f]
158 def recursive_foreach(params, fvars):
161 items = get_items(params, params[var])
162 logger.info("parallelizing on %s with items %s" % (var, items))
163 if items is not None:
165 params = copy.copy(params)
168 recursive_foreach(params, fvars)
170 arvados.api().job_tasks().create(body={
171 'job_uuid': arvados.current_job()['uuid'],
172 'created_by_job_task_uuid': arvados.current_task()['uuid'],
178 logger.error("parameter %s with value %s in task.foreach yielded no items" % (var, params[var]))
182 if "task.foreach" in jobp:
183 if arvados.current_task()['sequence'] == 0:
184 # This is the first task to start the other tasks and exit
185 fvars = jobp["task.foreach"]
186 if isinstance(fvars, basestring):
188 if not isinstance(fvars, list) or len(fvars) == 0:
189 logger.error("value of task.foreach must be a string or non-empty list")
191 recursive_foreach(jobp, jobp["task.foreach"])
192 if "task.vwd" in jobp:
193 # Set output of the first task to the base vwd collection so it
194 # will be merged with output fragments from the other tasks by
196 arvados.current_task().set_output(subst.do_substitution(jobp, jobp["task.vwd"]))
198 arvados.current_task().set_output(None)
201 # This is the only task so taskp/jobp are the same
204 if "task.vwd" in taskp:
205 # Populate output directory with symlinks to files in collection
206 vwd.checkout(subst.do_substitution(taskp, taskp["task.vwd"]), outdir)
208 if "task.cwd" in taskp:
209 os.chdir(subst.do_substitution(taskp, taskp["task.cwd"]))
211 cmd = expand_list(taskp, taskp["command"])
213 if "task.stdin" in taskp:
214 stdinname = subst.do_substitution(taskp, taskp["task.stdin"])
215 stdinfile = open(stdinname, "rb")
217 if "task.stdout" in taskp:
218 stdoutname = subst.do_substitution(taskp, taskp["task.stdout"])
219 stdoutfile = open(stdoutname, "wb")
221 logger.info("{}{}{}".format(' '.join(cmd), (" < " + stdinname) if stdinname is not None else "", (" > " + stdoutname) if stdoutname is not None else ""))
222 except subst.SubstitutionError as e:
224 logger.error("task parameters were:")
225 logger.error(pprint.pformat(taskp))
227 except Exception as e:
228 logger.exception("caught exception")
229 logger.error("task parameters were:")
230 logger.error(pprint.pformat(taskp))
234 sp = subprocess.Popen(cmd, shell=False, stdin=stdinfile, stdout=stdoutfile)
237 # forward signals to the process.
238 signal.signal(signal.SIGINT, lambda signum, frame: sig.send_signal(sp, signum))
239 signal.signal(signal.SIGTERM, lambda signum, frame: sig.send_signal(sp, signum))
240 signal.signal(signal.SIGQUIT, lambda signum, frame: sig.send_signal(sp, signum))
242 # wait for process to complete.
245 if sig.sig is not None:
246 logger.critical("terminating on signal %s" % sig.sig)
249 logger.info("completed with exit code %i (%s)" % (rcode, "success" if rcode == 0 else "failed"))
251 except Exception as e:
252 logger.exception("caught exception")
254 # restore default signal handlers.
255 signal.signal(signal.SIGINT, signal.SIG_DFL)
256 signal.signal(signal.SIGTERM, signal.SIG_DFL)
257 signal.signal(signal.SIGQUIT, signal.SIG_DFL)
262 logger.info("the following output files will be saved to keep:")
264 subprocess.call(["find", ".", "-type", "f", "-printf", "run-command: %12.12s %h/%f\\n"], stdout=sys.stderr)
266 logger.info("start writing output to keep")
268 if "task.vwd" in taskp:
269 if "task.foreach" in jobp:
270 # This is a subtask, so don't merge with the original collection, that will happen at the end
271 outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=False).manifest_text()
273 # Just a single task, so do merge with the original collection
274 outcollection = vwd.checkin(subst.do_substitution(taskp, taskp["task.vwd"]), outdir, merge=True).manifest_text()
276 outcollection = robust_put.upload(outdir, logger)
278 api.job_tasks().update(uuid=arvados.current_task()['uuid'],
280 'output': outcollection,
281 'success': (rcode == 0),