modify_metadata_with_upmu_sample_type.py

Go to the documentation of this file.

#!/usr/bin/env python

import samweb_client as swc
import datetime
import ephem
import sys

def getSampleType(t_start):
    if t_start ==0:
        return "NULL"
    deg = ephem.degrees
    sun = ephem.Sun()
    nova = ephem.Observer()
    nova.lat = '48:22.715780'
    nova.lon = '92:49.876930'
    nova.date = datetime.datetime.fromtimestamp(
        int(t_start)).strftime('%Y/%m/%d %H:%M')
    sun.compute(nova)
    angle = sun.alt
    if angle >deg('10'):
        stype = "Day"
    elif angle <=deg('10') and angle >deg('-10'):
        stype="Twilight"
    elif angle <=deg('-10') and angle > deg('-20'):
        stype="SignalExtended"
    else:
        stype="SignalPrimary"
    #print "{}\t{}\t{}".format(nova.date, angle, stype)
    return stype

def getSubRunType(sam, filename):
    metadata = sam.getMetadata(filename)
    key_start = "Analysis.UpMuSampleType"
    key_end = "Analysis.UpMuSampleTypeEnd"
    if key_start not in metadata or key_end not in metadata:
        isCurrent = False
    else:
        isCurrent = True
    # Get subrun start/end time, fix them if it's 0.
    subrun_start = metadata["Online.SubRunStartTime"]
    subrun_end = metadata["Online.SubRunEndTime"]
    if subrun_start == 0:
        if metadata["Online.Subrun"]==0:
            subrun_start = metadata["Online.RunStartTime"]
    if subrun_end == 0:
        try:
            fsplit = filename.split("_")
            adjsub="%s_%s_s%d_%s" %(fsplit[0], fsplit[1],
                                    int(md["Online.Subrun"])+1 ,fsplit[3])
            md2=sam.getMetadata(adjsub)
            subrun_end=md2["Online.SubRunStartTime"]
        except:
            print "Error: Retrieving adjacent subrun for run"\
                    " {}, using 3 min duration".format(f)
            subrun_end = subrun_start+180
    # Get subrun type (start_type and end_type) based on subrun start/end time.
    start_type = getSampleType(subrun_start)
    end_type = getSampleType(subrun_end)
    # Set the flag if modifying metadata is necessary.
    if isCurrent and (start_type != metadata[key_start]
                      or end_type != metadata[key_end]):
        isCurrent = False
    return (start_type, end_type, isCurrent)


if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser(
            prog='modify_metadata_with_upmu_sample_type.py',
            description="Add UpMu sample type to metadata.",
            epilog="Questions and Comments to dingpf@fnal.gov")
    parser.add_argument('-w', '--wet-run',action="store_true",
                        help="Run the script and modify metadata;")
    parser.add_argument('-s', '--stream_type',default='DDSun',
                        choices = ['DDMoon', 'DDSun', 'DDContained', 'DDupmu','t02'],
                        help="Data stream type;")
    parser.add_argument('-t','--data_tier', default='artdaq',
                        choices = ['artdaq','pidpart'],
                        help="Data tier type;")
    parser.add_argument('-rb', '--run-begin',default=21000, type=int,
                        help="Run range lower limit;")
    parser.add_argument('-re', '--run-end',default=21350, type=int,
                        help="Run range upper limit.")
    args = parser.parse_args()

    print 80*"/"
    rep = ''
    if not args.wet_run:
        print "Running without '-w' option."
        print "File metadata will NOT BE MODIFIED."
        print "Use option -w to modify file metadata."
        rep = raw_input( 'Do you want to proceed? y/[n] ' )
        if 1 < len(rep):
            rep = rep[0]
        if rep not in [ 'y', 'Y']:
            sys.exit(1)

    if args.wet_run:
        print "Running with '-w' option."
        print "File metadata may be MODIFIED."
        print "Remove option -w to only show the changes to file metadata."
        rep = raw_input( 'Do you want to proceed? y/[n] ' )
        if 1 < len(rep):
            rep = rep[0]
        if rep not in [ 'y', 'Y']:
            sys.exit(1)
    print 80*"*"

    sam = swc.SAMWebClient("nova")

    dim_run_range='(run_number >= {} and run_number <={})'.format(
        args.run_begin, args.run_end)


    dim_file_name='(file_name like "%{}%")'.format(args.stream_type)
    #cp
    dim_data_tier='(data_tier {})'.format(args.data_tier) 
    theDIM = dim_run_range + ' and ' + dim_file_name + ' and ' + dim_data_tier

    flist=sam.listFiles(theDIM)

    nfile = len(flist)
    print "%d files matching dimension." % (len(flist))

    i = 0
    for f in flist:
        #print f
        print "-"*80
        print "{}/{} Checking file: {}".format(i+1, nfile, f)
        start_type, end_type, isCurrent = getSubRunType(sam, f)
        mfmetadata = {"Analysis.UpMuSampleType": start_type,
                      "Analysis.UpMuSampleTypeEnd": end_type}
        if not isCurrent:
            if args.wet_run:
                sam.modifyFileMetadata(f,mfmetadata)
                print "MODIFY:       %s -- %s" %(f, mfmetadata)
            else:
                print "TO-BE-MODIFY: %s -- %s" %(f, mfmetadata)
        else:
            print "UP-TO-DATE:   %s -- %s" %(f, mfmetadata)
        i += 1
    print " -- Done"
    print 80*"*"