hist_file_dump Namespace Reference

Functions
def	fixprecision (x, precision=15)
def	jsonfixup (instr, fuzzyarray=False)
def	fuzzytreehash (tkey)
def	dumpdir (d)
def	key (x)

Variables
	ordering = args.rankorder
dictionary	accounting = {}
	f = ROOT.TFile.Open(args.filename)
	d = f.Get(args.path.rstrip('/'))
	sortedl = sorted(accounting.items(), key=key, reverse=True)

Function Documentation

dumpdir()

def hist_file_dump.dumpdir

(

d

)

Definition at line 101 of file hist_file_dump.py.

def dumpdir(d):
    thispath = d.GetPath()
    if ':' in thispath:
        thispath = thispath.split(':', 1)[1]
    #print thispath
    subdirs = []
    for k in d.GetListOfKeys():
        if not args.metadata and k.GetName() == 'metadata' and k.GetClassName() == 'TTree':
            continue
        if k.GetClassName().startswith('TDirectory'):
            subdirs.append(k)
        else:
            if args.tree_entries and k.GetClassName() == 'TTree':
                lhash = fuzzytreehash(k)
            elif args.hash:
                if k.GetClassName() != 'TEfficiency':
                    fixedjson = jsonfixup(ROOT.getjson(k), args.fuzzy_histbins)
                else:
                    j0 = json.loads(ROOT.getjson(k).Data())
                    for subh in ('fPassedHistogram', 'fTotalHistogram'):
                        j0[subh] = json.loads(jsonfixup(json.dumps(j0[subh])))
                    fixedjson = json.dumps(j0, sort_keys=True)
                lhash = zlib.adler32(fixedjson.encode())
                if lhash < 0:
                    lhash += 2**32
            else:
                lhash = 0
            idxname = os.path.join(thispath, k.GetName())
            accounting[idxname] = (k.GetObjlen(), k.GetNbytes()-k.GetKeylen())
            hashes[idxname] = lhash
            types[idxname] = k.GetClassName()
    for k in subdirs:
        dumpdir(k.ReadObj())
 

fixprecision()

def hist_file_dump.fixprecision	(		x,
			precision = 15
	)

Definition at line 13 of file hist_file_dump.py.

def fixprecision(x, precision=15):
    import math
    if not isinstance(x, float):
        return x
    else:
        mantissa, exponent = math.frexp(x)
        sm = '%.12g' % mantissa
        return _formatFloat (float(sm[:precision]) * 2**exponent)
 

fuzzytreehash()

def hist_file_dump.fuzzytreehash

(

tkey

)

Definition at line 92 of file hist_file_dump.py.

def fuzzytreehash(tkey):
    t = tkey.ReadObj()
    rv = zlib.adler32(((' '.join(_.GetName() for _ in t.GetListOfBranches()))
                        + (' '.join(_.GetName() + _.GetTypeName() for _ in t.GetListOfLeaves()))
                        + ' ' + str(t.GetEntries())).encode()
                        )
    del t
    return rv
 

jsonfixup()

def hist_file_dump.jsonfixup	(		instr,
			fuzzyarray = False
	)

Definition at line 22 of file hist_file_dump.py.

def jsonfixup(instr, fuzzyarray=False):
    if not isinstance(instr, str):
        instr = instr.Data()
    j=json.loads(instr)
    # the following are very subject to floating point numeric effects
    # are doubles, keep 15 decimal digits of precision
    for badkey in ('fTsumw', 'fTsumwx', 'fTsumw2', 'fTsumwx2', 'fTsumwy', 'fTsumwy2', 'fTsumwxy',
                   'fTsumwz', 'fTsumwz2', 'fTsumwxz', 'fTsumwyz' ):
        if badkey in j:
            j[badkey] = fixprecision(float(j[badkey]))
            #print(type(j["fTsumwx"]))
    # member, digits of precision
    arrkeys: List[Tuple[str,int,Callable]] = [('fSumw2', 15, float)]
    if fuzzyarray:
        arrkeys += [('fBinEntries', 15, float), ('fBinSumw2', 15, float)]
        # apply different precision for fArray depending on object type
        if '_typename' in j:
            if j['_typename'] in ('TH1F', 'TH2F', 'TH3F'):
                arrkeys.append(('fArray', 6, float))
            elif j['_typename'] in ('TH1C', 'TH1I', 'TH1S', 'TH2C', 'TH2I', 'TH2S',
                                    'TH3C', 'TH3I', 'TH3S'):
                # precision here isn't relevant, *should* be an integer
                arrkeys.append(('fArray', 15, lambda x: x))
            else:
                arrkeys.append(('fArray', 15, float))
        else:
            arrkeys.append(('fArray', 15, float))
    for badkey, precision, func in arrkeys:
        if badkey in j:
            j[badkey] = [fixprecision(func(_), precision) for _ in j[badkey]]
    # the following ignores small layout fluctuations in TTrees
    if 'fBranches' in j:
        for branch in j['fBranches']['arr']:
            branch['fBasketSeek'] = []
    return json.dumps(j, sort_keys=True)
 
parser=argparse.ArgumentParser()
parser.add_argument('filename',
                    help='Input HIST file name')
parser.add_argument('-r', '--rankorder', default='onfile',
                    choices=['onfile', 'uncompressed', 'name'],
                    help='rankorder is "onfile" (default), "uncompressed" or "name"')
parser.add_argument('-p', '--path',
                    help='Only look under this directory')
parser.add_argument('--hash', action='store_true',
                    help='Print hashes of objects')
parser.add_argument('--metadata', action='store_true',
                    help='Include metadata trees')
parser.add_argument('--no_onfile', action='store_true',
                    help="Don't show on file size")
parser.add_argument('--no_inmem', action='store_true',
                    help="Don't show in memory size")
parser.add_argument('--tree_entries', action='store_true',
                    help="Use more robust hash of TTree branches + entries")
parser.add_argument('--fuzzy_histbins', action='store_true',
                    help="Allow small variations in histogram bin content")
args=parser.parse_args()
 

key()

def hist_file_dump.key

(

x

)

Definition at line 146 of file hist_file_dump.py.

    def key(x): return (x[1][1], x[1][0], x[0])

Variable Documentation

accounting

dictionary hist_file_dump.accounting = {}

Definition at line 82 of file hist_file_dump.py.

d

hist_file_dump.d = f.Get(args.path.rstrip('/'))

Definition at line 137 of file hist_file_dump.py.

f

hist_file_dump.f = ROOT.TFile.Open(args.filename)

Definition at line 135 of file hist_file_dump.py.

ordering

hist_file_dump.ordering = args.rankorder

Definition at line 80 of file hist_file_dump.py.

sortedl

hist_file_dump.sortedl = sorted(accounting.items(), key=key, reverse=True)

Definition at line 151 of file hist_file_dump.py.