def split_samples(sh, num_events=150000):
"""
Split a sample handler's samples into smaller samples of size num_events.
This function does not split individual files.
"""
from ROOT import SH
splitSH = SH.SampleHandler()
SH.scanNEvents(sh)
for sample in sh:
splitSH.add(SH.splitSample(sample, num_events))
return splitSH
def load_samples(args):
"""
Build a SampleHandler.
Not all these options are compatible with each other.
It might thus be better to split this up somehow.
"""
from ROOT import SH
if args.sampleHandler:
sh = SH.SampleHandler()
sh.load(args.sampleHandler)
else:
sh = scan_samples(args.scanDir, args.samplePattern)
# Split samples by file if requested
if args.splitSamples:
sh = split_samples(sh, 1)
# Choose samples according to task ID
if args.task:
task, numTasks = map(int, args.task.split(':'))
sh = select_by_task(sh, task, numTasks)
if args.eventsPerWorker:
from ROOT import EL
SH.scanNEvents(sh)
sh.setMetaDouble(EL.Job.optEventsPerWorker, args.eventsPerWorker)
return sh
def _split_samples_worker(sample, num_events):
"""Worker process function for split_samples_mp"""
from ROOT import SH
SH.scanNEvents(sample)
return SH.splitSample(sample, num_events)
