This file is indexed.

/usr/bin/rivet-rmgaps is in rivet 1.8.3-1.3.

This file is owned by root:root, with mode 0o755.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
#! /usr/bin/env python

"""
%prog mcfile.aida [outputfile.aida]
%prog --no-rivet-refs datafile.aida mcfile.aida [outputfile.aida]

Remove bins in Rivet-generated MC AIDA files which are actually binning gaps in
the reference histogram. Rivet's histogramming system currently has no way to
leave gaps between bins, hence this clean-up script.

If the output file is not specified, the input MC file will be overwritten.
"""

import sys
if sys.version_info[:3] < (2,4,0):
    print "rivet scripts require Python version >= 2.4.0... exiting"
    sys.exit(1)


import os, tempfile

class Inputdata:
    def __init__(self, filenames):
        self.histos = {}
        self.description = {}
        self.description['DrawOnly'] = []
        if not hasattr(filenames, "__iter__"):
            filenames = [filenames]
        for filename in filenames:
            f = open(filename+'.dat', 'r')
            for line in f:
                if (line.count('#',0,1)):
                    if (line.count('BEGIN HISTOGRAM')):
                        title = line.split('BEGIN HISTOGRAM', 1)[1].strip()
                        self.description['DrawOnly'].append(title)
                        self.histos[title] = Histogram(f)
            f.close()


class Histogram:
    def __init__(self, f):
        self.read_input(f)

    def read_input(self, f):
        self.description = {}
        self.data = []
        for line in f:
            if (line.count('#',0,1)):
                if (line.count('END HISTOGRAM')):
                    break
            else:
                line = line.rstrip()
                if line.count('='):
                    linearray = line.split('=', 1)
                    self.description[linearray[0]] = linearray[1]
                else:
                    linearray = line.split('\t')
                    if len(linearray) == 4:
                        self.data.append({'LowEdge': float(linearray[0]),
                                          'UpEdge':  float(linearray[1]),
                                          'Content': float(linearray[2]),
                                          'Error':   [float(linearray[3]),float(linearray[3])]})
                    elif len(linearray) == 5:
                        self.data.append({'LowEdge': float(linearray[0]),
                                          'UpEdge':  float(linearray[1]),
                                          'Content': float(linearray[2]),
                                          'Error':   [float(linearray[3]),float(linearray[4])]})

    def write_datapoint(self, f, xval, xerr, yval, yerr):
        f.write('    <dataPoint>\n')
        f.write('      <measurement errorPlus="%e" value="%e" errorMinus="%e"/>\n' %(xerr, xval, xerr))
        f.write('      <measurement errorPlus="%e" value="%e" errorMinus="%e"/>\n' %(yerr[1], yval, yerr[0]))
        f.write('    </dataPoint>\n')

    def write_datapointset_header(self, f):
        title = self.description.setdefault('Title', '')
        xlabel = self.description.setdefault('XLabel', '')
        ylabel = self.description.setdefault('YLabel', '')
        path = self.description.setdefault('AidaPath', None)
        if path is not None:
            path = path.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;')
        f.write('  <dataPointSet name="%s" dimension="2"\n' % path.split('/')[-1])
        f.write('    path="%s" title="%s">\n' % (os.path.abspath(path.replace(path.split('/')[-1], '')),
                                                 title.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;')))
        f.write('    <annotation>\n')
        f.write('      <item key="Title" value="%s" sticky="true"/>\n' % title.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;'))
        f.write('      <item key="XLabel" value="%s" sticky="true"/>\n' % xlabel.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;'))
        f.write('      <item key="YLabel" value="%s" sticky="true"/>\n' % ylabel.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;'))
        f.write('      <item key="AidaPath" value="%s" sticky="true"/>\n' % path)
        f.write('      <item key="FullPath" value="/%s.aida%s" sticky="true"/>\n' % (filename.split('/')[-1], path))
        f.write('    </annotation>\n')
        f.write('    <dimension dim="0" title="%s" />\n' % xlabel.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;'))
        f.write('    <dimension dim="1" title="%s" />\n' % ylabel.replace('>', '&gt;').replace('<', '&lt;').replace('"', '&quot;'))

    def write_datapointset_footer(self, f):
        f.write('  </dataPointSet>\n')

    def write_datapointset(self, f):
        self.write_datapointset_header(f)
        for bin, bindata in enumerate(self.data):
            xval = 0.5*(bindata['UpEdge'] + bindata['LowEdge'])
            #if bindata['UpEdge'] == bindata['LowEdge']:
            #    xerr = 0.5
            #else:
            #    xerr = 0.5*(bindata['UpEdge'] - bindata['LowEdge'])
            xerr = 0.5*(bindata['UpEdge'] - bindata['LowEdge'])
            yval = bindata['Content']
            yerr = bindata['Error']
            self.write_datapoint(f, xval, xerr, yval, yerr)
        self.write_datapointset_footer(f)

    def remove_gaps(self):
        ## Only look at histograms which are present in the reference file:
        try:
            refhist = refdata.histos['/REF%s' % self.description['AidaPath']]
        except:
            return

        ## Check for differences in the binning and remove superfluous MC bins:
        if len(refhist.data) != len(self.data):
            numrm = abs(len(self.data) - len(refhist.data))
            if numrm != 1:
                plural = "s"
            else:
                plural = ""
            logging.info("Stripping %d bin%s from %s" % (numrm, plural, self.description['AidaPath']))
            newdata = []
            for i in xrange(len(self.data)):
                if (len(refhist.data) == i):  ## MC has additional bins above the upper edge of the refdata
                    break
                if self.data[i]['LowEdge'] == refhist.data[i]['LowEdge'] and \
                   self.data[i]['UpEdge'] == refhist.data[i]['UpEdge']:
                    newdata.append(self.data[i])
                else:
                    logging.debug('Deleted bin %d' % i)
                    refhist.data.insert(i, self.data[i])
            self.data = newdata


## Command line parsing
import logging
from optparse import OptionParser
parser = OptionParser(usage=__doc__)
parser.add_option("-R", "--rivet-refs", dest="USE_RIVETREFS", action="store_true", default=True,
                  help="use the Rivet reference data files for comparison (default)")
parser.add_option("--no-rivet-refs", dest="USE_RIVETREFS", action="store_false", default=True,
                  help="don't use the Rivet reference data files for comparison")
parser.add_option("-v", "--verbose", action="store_const", const=logging.DEBUG, dest="LOGLEVEL",
                  default=logging.INFO, help="print debug (very verbose) messages")
parser.add_option("-q", "--quiet", action="store_const", const=logging.WARNING, dest="LOGLEVEL",
                  default=logging.INFO, help="be very quiet")
opts, args = parser.parse_args()

## Configure logging
logging.basicConfig(level=opts.LOGLEVEL, format="%(message)s")


if opts.USE_RIVETREFS:
    try:
        import rivet
        rivet.getAnalysisRefPaths()
    except:
        logging.error("Could not find the Rivet ref paths because the 'rivet' Python module could not be loaded")
        sys.exit(1)


if opts.USE_RIVETREFS:
    if len(args) < 1:
        logging.error("Must specify at least the MC input file")
        sys.exit(1)
    if len(args) >= 1:
        REFFILES = []
        INFILE = args[0]
        OUTFILE = args[0]
    if len(args) == 2:
        OUTFILE = args[1]
    if len(args) > 2:
        logging.error("Maximum of two arguments with the -R argument flag!")
        sys.exit(1)
else:
    if len(args) < 2:
        logging.error("Must specify at least the reference file and the MC input file (or use the -R option)")
        sys.exit(1)
    if len(args) >= 2:
        REFFILES = [args[0]]
        INFILE = args[1]
        OUTFILE = args[1]
    if len(args) == 3:
        OUTFILE = args[2]
    if len(args) > 3:
        logging.error("Maximum of three arguments with the -R argument flag!")
        sys.exit(1)


## Convert the aida input files to flat files we can parse:
tempdir = tempfile.mkdtemp('.gap_removal')

filename = INFILE.replace(".aida", "")
os.system("%s/aida2flat %s.aida > %s/%s.dat" % \
              (os.path.dirname(os.path.realpath(sys.argv[0])),
               filename, tempdir, os.path.basename(filename)))
mcdata = Inputdata(os.path.join(tempdir, os.path.basename(filename)))


## Build list of Rivet ref files
if opts.USE_RIVETREFS:
    anas = set()
    for hpath in mcdata.description['DrawOnly']:
        ana = hpath[1:].split("/")[0]
        anas.add(ana)
    #print anas
    for a in anas:
        apath = rivet.findAnalysisRefFile(a+".aida")
        if apath:
            REFFILES.append(apath)
            break
    #print REFFILES


## Run over ref files
refdatfiles = []
for rf in REFFILES:
    filename = rf.replace(".aida", "")
    os.system("%s/aida2flat %s.aida > %s/%s.dat" % \
                  (os.path.dirname(os.path.realpath(sys.argv[0])),
                   filename, tempdir, os.path.basename(filename)))
    refdatfiles.append( os.path.join(tempdir, os.path.basename(filename)) )
refdata = Inputdata(refdatfiles)


## Clean up
for i in os.listdir(tempdir):
    os.unlink('%s/%s' %(tempdir, i))
os.rmdir(tempdir)


## Remove gap bins
for i in mcdata.description['DrawOnly']:
    mcdata.histos[i].remove_gaps()


## Write the new aida file with removed gap bins:
f = open(OUTFILE, 'w')
f.write('<?xml version="1.0" encoding="ISO-8859-1" ?>\n')
f.write('<!DOCTYPE aida SYSTEM "http://aida.freehep.org/schemas/3.3/aida.dtd">\n')
f.write('<aida version="3.3">\n')
f.write('  <implementation version="1.1" package="FreeHEP"/>\n')
for i in mcdata.description['DrawOnly']:
    mcdata.histos[i].write_datapointset(f)
f.write('</aida>\n')
f.close