def main():
# merge command-line and file-specified arguments
config = conf.config(__doc__, sys.argv[1:])
logger = loghelper.create(LOGGER,
log_level=config.get('log.level'),
log_fmt=config.get('log.format'),
log_file=config.get('log.file'))
dry_run = config.get('dry-run')
after_job = config.get('after-job')
max_dom = config['max_dom']
# either the start time is exactly specified, or else we calculate it from base time, delay and cycles
if config.get('start'):
init_time = config['start']
else:
init_time = shared.get_time(base_time=config.get('base-time'),
delay=config.get('delay'),
round=config.get('cycles'))
if config.get('end'):
end_init = config['end']
init_interval = config['init_interval']
init_times = list(
rrule.rrule(freq=rrule.HOURLY,
interval=init_interval,
dtstart=init_time,
until=end_init))
else:
init_times = [init_time]
jobs = config['jobs']
run_jobs = [j for j in jobs if true_like(j['run'])]
parallel = int(config.get('parallel'))
logger.debug("allow %d parallel simulations" % parallel)
after_job = None
previous_sim = None
working_dir = config.get('working-dir')
for n, init_time in enumerate(init_times):
# one-argument function to do initial-time substitution in strings
expand = lambda s: substitute.sub_date(str(s), init_time=init_time)
wdir = expand(working_dir)
if not os.path.exists(wdir):
logger.error("could not find %s, skipping" % wdir)
continue
# allow individual simulations to run in parallel, but up to a limit of N
# every time N reaches parallel limit, enforce a dependency
if parallel == 0 or (n % parallel) == 0:
after_job = previous_sim
logger.debug("dependency: %s" % after_job)
previous_sim = submit(run_jobs,
expand,
after_job=after_job,
array_job=max_dom,
dry_run=dry_run)
def update_namelist_wps(template, target, max_dom, init_time, fcst_hours, bdy_interval, geo_em_dir, met_em_dir, geogrid_tbl, metgrid_tbl, prefix, constants_name=None):
""" Updates the namelist.wps to reflect updated settings in config
Arguments:
expand -- function for exanding any strings
template -- base namelist_wps file to alter
target -- name of modified file to write to
"""
# consider refactoring this method, perhaps relacing it with a more general namelist updating function,
# and pushing some of the logic into the configuration file?
logger = shared.get_logger()
logger.debug('*** UPDATING namelist.wps ***')
# function to expand any date/time placeholders
expand = lambda s : substitute.sub_date(s, init_time=init_time)
#wps_run_dir= expand(target)
bdy_times = shared.get_bdy_times(init_time, fcst_hours, bdy_interval)
met_em_dir = expand(met_em_dir)
geo_em_dir = expand(geo_em_dir)
interval_seconds = bdy_interval * 60 * 60
logger.debug('reading namelist.wps <--------- %s' % template)
namelist = shared.read_namelist(template)
#
# Update some options based on the forecast config file
#
namelist.update('max_dom', max_dom)
namelist.update('opt_output_from_geogrid_path', geo_em_dir, section='share')
namelist.update('opt_output_from_metgrid_path', met_em_dir, section='metgrid')
namelist.update('opt_geogrid_tbl_path', expand(geogrid_tbl), section='geogrid')
namelist.update('opt_metgrid_tbl_path', expand(metgrid_tbl), section='metgrid')
namelist.update('interval_seconds', [interval_seconds])
#
# Generate formatted strings for inclusion in the namelist.wps file
#
start_str = bdy_times[0].strftime("%Y-%m-%d_%H:%M:%S")
end_str = bdy_times[-1].strftime("%Y-%m-%d_%H:%M:%S")
logger.debug("Updating namelist.wps start and end")
namelist.update('start_date', [start_str]*max_dom)
namelist.update('end_date', [end_str]*max_dom)
if constants_name:
namelist.update('constants_name', constants_name, section='metgrid')
namelist.to_file(target)
logger.debug('*** FINISHED UPDATING namelist.wps ***')
def main():
# merge command-line and file-specified arguments
config = conf.config(__doc__, sys.argv[1:])
logger = loghelper.create(LOGGER,
log_level=config.get('log.level'),
log_fmt=config.get('log.format'))
if config.get('log.file'):
log_file = config['log.file']
logger.addHandler(
loghelper.file_handler(log_file, config['log.level'],
config['log.format']))
dry_run = config.get('dry-run')
# either the start time is exactly specified, or else we calculate it from base time, delay and cycles
if config.get('start'):
init_time = config['start']
else:
init_time = shared.get_time(base_time=config.get('base-time'),
delay=config.get('delay'),
round=config.get('cycles'))
if config.get('end'):
end_init = config['end']
init_interval = config['init_interval']
init_times = list(
rrule.rrule(freq=rrule.HOURLY,
interval=init_interval,
dtstart=init_time,
until=end_init))
else:
init_times = [init_time]
for init_time in init_times:
# one-argument function to do initial-time substitution in strings
expand = lambda s: substitute.sub_date(str(s), init_time=init_time)
replacements = substitute.date_replacements(init_time=init_time)
jobs = config['jobs']
# get an ordered list of all the ones which we will run
run_jobs = [
jobs[j] for j in sorted(jobs.keys()) if jobs[j]['run'] == True
]
#for key, entry in config['jobs'].items():
for entry in run_jobs:
template = expand(entry['template'])
target = expand(entry['target'])
tm.fill_template(template, target, replacements)
def main():
# merge command-line and file-specified arguments
config = conf.config(__doc__, sys.argv[1:])
logger = loghelper.create(LOGGER, log_level=config.get('log.level'), log_fmt=config.get('log.format'), log_file=config.get('log.file'))
dry_run = config.get('dry-run')
after_job = config.get('after-job')
max_dom = config['max_dom']
# either the start time is exactly specified, or else we calculate it from base time, delay and cycles
if config.get('start'):
init_time = config['start']
else:
init_time = shared.get_time(base_time=config.get('base-time'), delay=config.get('delay'), round=config.get('cycles'))
if config.get('end'):
end_init = config['end']
init_interval = config['init_interval']
init_times = list(rrule.rrule(freq=rrule.HOURLY, interval=init_interval, dtstart=init_time, until=end_init))
else:
init_times = [init_time]
jobs = config['jobs']
run_jobs = [ j for j in jobs if true_like(j['run'])]
parallel = int(config.get('parallel'))
logger.debug("allow %d parallel simulations" % parallel)
after_job=None
previous_sim=None
working_dir = config.get('working-dir')
for n,init_time in enumerate(init_times):
# one-argument function to do initial-time substitution in strings
expand = lambda s : substitute.sub_date(str(s), init_time=init_time)
wdir = expand(working_dir)
if not os.path.exists(wdir):
logger.error("could not find %s, skipping" % wdir)
continue
# allow individual simulations to run in parallel, but up to a limit of N
# every time N reaches parallel limit, enforce a dependency
if parallel==0 or (n%parallel)==0:
after_job=previous_sim
logger.debug("dependency: %s" % after_job)
previous_sim = submit(run_jobs, expand, after_job=after_job, array_job=max_dom, dry_run=dry_run)
def main():
# merge command-line and file-specified arguments
config = conf.config(__doc__, sys.argv[1:])
logger = loghelper.create(LOGGER, log_level=config.get("log.level"), log_fmt=config.get("log.format"))
if config.get("log.file"):
log_file = config["log.file"]
logger.addHandler(loghelper.file_handler(log_file, config["log.level"], config["log.format"]))
dry_run = config.get("dry-run")
# either the start time is exactly specified, or else we calculate it from base time, delay and cycles
if config.get("start"):
init_time = config["start"]
else:
init_time = shared.get_time(
base_time=config.get("base-time"), delay=config.get("delay"), round=config.get("cycles")
)
if config.get("end"):
end_init = config["end"]
init_interval = config["init_interval"]
init_times = list(rrule.rrule(freq=rrule.HOURLY, interval=init_interval, dtstart=init_time, until=end_init))
else:
init_times = [init_time]
for init_time in init_times:
# one-argument function to do initial-time substitution in strings
expand = lambda s: substitute.sub_date(str(s), init_time=init_time)
replacements = substitute.date_replacements(init_time=init_time)
jobs = config["jobs"]
# get an ordered list of all the ones which we will run
run_jobs = [jobs[j] for j in sorted(jobs.keys()) if jobs[j]["run"] == True]
# for key, entry in config['jobs'].items():
for entry in run_jobs:
template = expand(entry["template"])
target = expand(entry["target"])
tm.fill_template(template, target, replacements)
def main():
# merge command-line and file-specified arguments
config = conf.config(__doc__, sys.argv[1:])
logger = loghelper.create(LOGGER, log_level=config.get('log.level'),
log_fmt=config.get('log.format'),
log_file=config.get('log.file'))
if not os.path.exists(config['namelist_wps']):
logger.error("No namelist.wps found, %s was specifed as template, but does not exist" % config['namelist_wps'])
sys.exit()
if not os.path.exists(config['namelist_input']):
logger.error("No namelist.input found, %s was specifed as template, but does not exist" % config['namlist_input'])
sys.exit()
dry_run = config.get('dry-run')
rmtree = config.get('rmtree')
max_dom = config['max_dom']
bdy_interval = config['bdy_interval']
fcst_hours = config['fcst_hours']
logger.debug(fcst_hours)
history_interval = config['history_interval']
link_boundaries = config.get('link-boundaries')
# either the start time is exactly specified, or else we calculate it
if config.get('start'):
init_time = config['start']
else:
init_time = shared.get_time(base_time=config.get('base-time'), delay=config.get('delay'), round=config.get('cycles'))
if config.get('end'):
end_init = config['end']
logger.debug(end_init)
init_interval = config['init_interval']
init_times = list(rrule.rrule(freq=rrule.HOURLY, interval=init_interval, dtstart=init_time, until=end_init))
else:
init_times = [init_time]
for init_time in init_times:
try:
logger.info("**** Running simulation for %s *****" % init_time)
# one-argument function to do initial-time substitution in strings
expand = lambda s : substitute.sub_date(s, init_time=init_time) if type(s)==type("") else s
date_replacements = substitute.date_replacements(init_time=init_time)
working_dir = expand(config['working_dir'])
logger.info("working dir: %s " % working_dir)
if rmtree:
safe_remove(working_dir, dry_run)
create_directory_structure(expand, remove=config.get('prepare.remove'), create=config.get('prepare.create'), copy=config.get('prepare.copy'), link=config.get('prepare.link'), dry_run=dry_run)
if config.get('prepare.template'):
for entry in config['prepare.template']:
tokens = expand(entry).split()
source = tokens[0]
target = tokens[1] if len(tokens)>1 else tokens[0]
templater.fill_template(source, target, date_replacements)
bdy_times = shared.get_bdy_times(init_time, fcst_hours, bdy_interval)
working_namelist = working_dir+"/namelist.wps"
# this can be made cleaner
prefix=''
update_namelist_wps(config['namelist_wps'], working_namelist, config['max_dom'],
init_time, fcst_hours, config['bdy_interval'],
config['geo_em_dir'], config['met_em_dir'], config['geogrid_run_dir'], config['metgrid_run_dir'], prefix,
config.get('constants_name'))
working_namelist = working_dir+"/namelist.input"
logger.debug(fcst_hours)
update_namelist_input(config['namelist_input'], working_namelist, max_dom, init_time, fcst_hours, history_interval, bdy_interval*60*60, metadata=config.get('metadata'))
logger.debug(fcst_hours)
# apply any additional specified namelist updates (consider getting rid of this section)
namelist_updates = config.get('namelist_updates')
if namelist_updates:
for key in sorted(namelist_updates.keys()):
entry = namelist_updates[key]
logger.debug('processing namelist update entry %s' % key)
template = expand(entry['template'])
target = expand(entry['target'])
logger.debug('%s\t---->\t%s' %(template.ljust(20), target.ljust(20)))
namelist = shared.read_namelist(template)
if entry.get('update'):
for old,new in entry['update'].items():
logger.debug('\t%s\t:\t%s' %(old.ljust(20), expand(new).ljust(20)))
namelist.update(old,expand(new))
namelist.to_file(target)
# link in input files for all ungrib jobs
# update namelist.wps to modify start and end time
if config.get('ungrib'):
for key,entry in config['ungrib'].items():
# apply any delay and rounding to the init_time to get correct time for dataset
# note that sometimes it is necessary to use a different time e.g. for SST field is delayed by one day
run_dir = expand(entry['run_dir'])
base_time = shared.get_time(init_time, delay=entry.get('delay'), round=entry.get('cycles'))
ungrib_len = int(entry['ungrib_len'])
bdy_times = shared.get_bdy_times(base_time, ungrib_len, bdy_interval)
namelist = shared.read_namelist(run_dir+"/namelist.wps")
start_str = base_time.strftime("%Y-%m-%d_%H:%M:%S")
end_str = bdy_times[-1].strftime("%Y-%m-%d_%H:%M:%S")
namelist.update('start_date', [start_str]*max_dom)
namelist.update('end_date', [end_str]*max_dom)
namelist.to_file(run_dir+"/namelist.wps")
# link in vtable
vtable = entry['vtable']
cmd = "%s %s/Vtable" % (vtable, run_dir)
shared.link(cmd, dry_run=dry_run)
if link_boundaries:
file_pattern = entry['files']
# create an ordered set to ensure filenames only appear once
filenames = shared.ordered_set([substitute.sub_date(file_pattern, init_time=base_time, valid_time=t) for t in bdy_times])
missing_files = []
for f in filenames:
if not os.path.exists(f):
missing_files.append(f)
logger.error("%s \t missing" % f)
if missing_files!=[]:
if rmtree:
safe_remove(working_dir, dry_run)
raise IOError("some files could not be found")
args = ' '.join(filenames)
cmd = '%s/link_grib.csh %s' %(run_dir,args)
shared.run_cmd(cmd, dry_run=dry_run, cwd=run_dir, log=False)
except IOError as e:
logger.error(e)
