def word_to_phone_lattice_torque(target, source, env):
args = source[-1].read()
data_list, lattice_list, wordpron, dic = source[0:4]
args["DICTIONARY"] = dic.abspath
args["DATA_FILE"] = data_list.abspath
args["FSMGZ_FORMAT"] = "true"
args["CONFUSION_NETWORK"] = ""
args["FSM_DIR"] = "temp"
args["WORDPRONSYMTABLE"] = wordpron.abspath
argstr = "-d %(DICTIONARY)s -D %(DATA_FILE)s -t %(FSMGZ_FORMAT)s -s %(WORDPRONSYMTABLE)s -S %(EPSILON_SYMBOLS)s %(CONFUSION_NETWORK)s -P %(PRUNE_THRESHOLD)d %(FSM_DIR)s" % args
if not os.path.exists(os.path.dirname(target[0].rstr())):
os.makedirs(os.path.dirname(target[0].rstr()))
command = env.subst("cat ${SOURCES[1].abspath}|${WRD2PHLATTICE} %s" % (argstr), source=source)
interval = args.get("interval", 10)
args["path"] = args.get("path", target[0].get_dir())
stdout = env.Dir(args.get("stdout", args["path"])).Dir("stdout").rstr()
stderr = env.Dir(args.get("stderr", args["path"])).Dir("stderr").rstr()
if not os.path.exists(stdout):
os.makedirs(stdout)
if not os.path.exists(stderr):
os.makedirs(stderr)
job = torque.Job(args.get("name", "scons-wrd2phlattice"),
commands=["mkdir -p temp", command],
path=args.get("path", target[0].get_dir()).rstr(),
stdout_path=stdout,
stderr_path=stderr,
other=args.get("other", ["#PBS -W group_list=yeticcls"]),
)
job.submit(commit=True)
while job.job_id in [x[0] for x in torque.get_jobs(True)]:
logging.info("sleeping...")
time.sleep(interval)
with meta_open(target[0].rstr(), "w") as ofd:
ofd.write(time.asctime() + "\n")
return None
def torque_executor(env, commands):
"""Runs a list of commands as separate Torque jobs using environment variables."""
args = {}
running = []
resources = {
"mem" : env["TORQUE_MEMORY"],
"cput" : env["TORQUE_TIME"],
"walltime" : env["TORQUE_TIME"],
}
for cmd in commands:
stdout = env.subst("${TORQUE_LOG}/")
stderr = env.subst("${TORQUE_LOG}/")
job = torque.Job(args.get("name", "scons"),
commands=[env.subst("source ${BASHRC_TXT}"), env.subst(cmd)],
resources=resources,
path=args.get("path", os.getcwd()),
stdout_path=stdout,
stderr_path=stderr,
array=args.get("array", 0),
other=args.get("other", ["#PBS -W group_list=yeticcls"]))
job.submit(commit=True)
running.append(job.job_id)
while len(running) > 0:
logging.info("monitoring %d torque jobs..." % (len(running)))
time.sleep(int(env["TORQUE_INTERVAL"]))
active = [x[0] for x in torque.get_jobs(True)]
running = [x for x in running if x in active]
return None
def run_asr_experiment_torque(target, source, env):
args = source[-1].read()
construct_command = env.subst("${ATTILA_INTERPRETER} ${SOURCES[1].abspath}", source=source)
out, err, success = run_command(construct_command)
if not success:
return out + err
stdout = env.Dir(args.get("stdout", args["path"])).Dir("stdout").rstr()
stderr = env.Dir(args.get("stderr", args["path"])).Dir("stderr").rstr()
if not os.path.exists(stdout):
os.makedirs(stdout)
if not os.path.exists(stderr):
os.makedirs(stderr)
command = env.subst("${ATTILA_INTERPRETER} ${SOURCES[2].abspath} -n ${TORQUE_JOBS_PER_SCONS_INSTANCE} -j $${PBS_ARRAYID} -w ${ACOUSTIC_WEIGHT} -l 1", source=source)
interval = args.get("interval", 10)
job = torque.Job(args.get("name", "scons"),
commands=[command],
path=args["path"],
stdout_path=stdout,
stderr_path=stderr,
array=args.get("array", 0),
other=args.get("other", ["#PBS -W group_list=yeticcls"]),
)
if env["HAS_TORQUE"]:
job.submit(commit=True)
while job.job_id in [x[0] for x in torque.get_jobs(True)]:
logging.debug("sleeping...")
time.sleep(interval)
else:
logging.info("no Torque server, but I would submit:\n%s" % (job))
with meta_open(target[0].rstr(), "w") as ofd:
ofd.write(time.asctime() + "\n")
return None
def standard_search_torque(target, source, env):
"""
Usage: /mnt/calculon-minor/lorelei_svn/KWS/bin64/stdsearch [-opts] [result_file] [query_file]
Options:
-d file data file [data.list] with lines in the following format :
utt_name start_time fst_path (default: data.list)
-f filt filter fst (default: none)
-i fst index fst (default: index.fst)
-n N return N-best results (default: return all)
-p fst pad fst (default: fspad.fst)
-s symbols arc symbols (default: word.list)
-t threshold min score needed to decide YES,
(if specified it overrides term-spec-threshold (default))
-T true/false true (default)=queries are in text format, false=fst format
txtformat: query list is a list of queries,
fstformat: query list is a list of full-paths to query fsts
-J int job-batch (for parallel run)
-N int total number of jobs (for parallel run)
-a string title on results list (default: stdbn.tlist.xml)
-b string prefix on termid (default: TERM-0)
-m string termid numerical formatting string (default: -1524500936)
-O if specified, don't optimize (default : optimize = true)
-v (verbose) if specified, print all debug outputs to stderr
-? info/options
"""
data_list, isym, idx, pad, queryph = source[0:5]
args = source[-1].read()
interval = args.get("interval", 30)
if source[-2].stat().st_size == 0:
with meta_open(target[0].rstr(), "w") as ofd:
ofd.write("""<stdlist termlist_filename="std.xml" indexing_time="68.51" language="english" index_size="" system_id="" />\n""")
return None
command = env.subst("${STDSEARCH} -F ${TARGET} -i ${SOURCES[2]} -b KW%(LANGUAGE_ID)s- -s ${SOURCES[1]} -p ${SOURCES[3]} -d ${SOURCES[0]} -a %(TITLE)s -m %(PRECISION)s ${SOURCES[4]}" % args, target=target, source=source)
job = torque.Job(args.get("name", "scons-stdsearch"),
commands=[command],
path=args.get("path", target[0].get_dir()).rstr(),
stdout_path=pjoin(target[0].get_dir().rstr(), "stdout"),
stderr_path=pjoin(target[0].get_dir().rstr(), "stderr"),
other=args.get("other", ["#PBS -W group_list=yeticcls"]),
)
if env["HAS_TORQUE"]:
print str(job)
job.submit(commit=True)
while job.job_id in [x[0] for x in torque.get_jobs(True)]:
logging.info("sleeping...")
time.sleep(interval)
else:
logging.info("no Torque server, but I would submit:\n%s" % (job))
with meta_open(target[0].rstr(), "w") as ofd:
ofd.write(time.asctime() + "\n")
return None
def submit_job(target, source, env):
args = source[-1].read()
stdout = args.get("stdout", os.path.join(args["path"], "stdout"))
stderr = args.get("stderr", os.path.join(args["path"], "stderr"))
if not os.path.exists(stdout):
os.makedirs(stdout)
if not os.path.exists(stderr):
os.makedirs(stderr)
interval = args.get("interval", 10)
job = torque.Job(args.get("name", "scons"),
commands=[env.subst(x) for x in args["commands"]],
path=args["path"],
stdout_path=stdout,
stderr_path=stderr,
array=args.get("array", 0),
other=args.get("other", []))
job.submit(commit=True)
while job.job_id in [x[0] for x in torque.get_jobs(True)]:
logging.info("sleeping...")
time.sleep(interval)
meta_open(target[0].rstr(), "w")
return None