No OneTemporary
Actions

Size

20 KB

Subscribers

None

View Options

	diff --git a/bin/rivet b/bin/rivet
	--- a/bin/rivet
	+++ b/bin/rivet
	@@ -1,528 +1,530 @@
	#! /usr/bin/env python

	"""\
	Run Rivet analyses on inputted events from file or Unix pipe

	Examples:
	%prog [options] <hepmcfile> [<hepmcfile2> ...]
	my_generator -o myfifo & \ %prog [options] myfifo
	agile-runmc <genname> -n 100k -o- \| %prog [options]

	ENVIRONMENT:
	* RIVET_ANALYSIS_PATH: list of paths to be searched for plugin
	analysis libraries at runtime
	* RIVET_REF_PATH: list of paths to be searched for reference
	data files
	* RIVET_INFO_PATH: list of paths to be searched for analysis
	metadata files
	"""

	import os, sys

	## Load the rivet module
	try:
	import rivet
	except:
	## If rivet loading failed, try to bootstrap the Python path!
	try:
	# TODO: Is this a good idea? Maybe just notify the user that their PYTHONPATH is wrong?
	import commands
	modname = sys.modules[__name__].__file__
	binpath = os.path.dirname(modname)
	rivetconfigpath = os.path.join(binpath, "rivet-config")
	rivetpypath = commands.getoutput(rivetconfigpath + " --pythonpath")
	sys.path.append(rivetpypath)
	import rivet
	except:
	sys.stderr.write("The rivet Python module could not be loaded: is your PYTHONPATH set correctly?\n")
	sys.exit(1)

	rivet.util.check_python_version()
	rivet.util.set_process_name("rivet")

	import time, datetime, logging, signal

	## Parse command line options
	from optparse import OptionParser, OptionGroup
	parser = OptionParser(usage=__doc__, version="rivet v%s" % rivet.version())

	anagroup = OptionGroup(parser, "Analysis handling")
	anagroup.add_option("-a", "--analysis", "--analyses", dest="ANALYSES", action="append",
	default=[], metavar="ANA",
	help="add an analysis (or comma-separated list of analyses) to the processing list.")
	anagroup.add_option("--list-analyses", "--list", dest="LIST_ANALYSES", action="store_true",
	default=False, help="show the list of available analyses' names. With -v, it shows the descriptions, too")
	anagroup.add_option("--list-used-analyses", action="store_true", dest="LIST_USED_ANALYSES",
	default=False, help="list the analyses used by this command (after subtraction of inappropriate ones)")
	anagroup.add_option("--show-analysis", "--show-analyses", "--show", dest="SHOW_ANALYSES", action="append",
	default=[], help="show the details of an analysis")
	anagroup.add_option("--analysis-path", dest="ANALYSIS_PATH", metavar="PATH", default=None,
	help="specify the analysis search path (cf. $RIVET_ANALYSIS_PATH).")
	# TODO: remove/deprecate the append?
	anagroup.add_option("--analysis-path-append", dest="ANALYSIS_PATH_APPEND", metavar="PATH", default=None,
	help="append to the analysis search path (cf. $RIVET_ANALYSIS_PATH).")
	anagroup.add_option("--pwd", dest="ANALYSIS_PATH_PWD", action="store_true", default=False,
	help="append the current directory (pwd) to the analysis search path (cf. $RIVET_ANALYSIS_PATH).")
	# TODO: add control for more paths?
	parser.add_option_group(anagroup)


	extragroup = OptionGroup(parser, "Extra run settings")
	extragroup.add_option("-H", "--histo-file", dest="HISTOFILE",
	default="Rivet.yoda", help="specify the output histo file path (default = %default)")
	extragroup.add_option("-x", "--cross-section", dest="CROSS_SECTION",
	default=None, metavar="XS",
	help="specify the signal process cross-section in pb")
	extragroup.add_option("-n", "--nevts", dest="MAXEVTNUM", type="int",
	default=None, metavar="NUM",
	help="restrict the max number of events to read.")
	extragroup.add_option("--runname", dest="RUN_NAME", default=None, metavar="NAME",
	help="give an optional run name, to be prepended as a 'top level directory' in histo paths")
	extragroup.add_option("--ignore-beams", dest="IGNORE_BEAMS", action="store_true", default=False,
	help="Ignore input event beams when checking analysis compatibility. "
	"Warning: some analyses may not work with incorrect beams")
	parser.add_option_group(extragroup)


	timinggroup = OptionGroup(parser, "Timeouts and periodic operations")
	timinggroup.add_option("--event-timeout", dest="EVENT_TIMEOUT", type="int",
	default=21600, metavar="NSECS",
	help="max time in whole seconds to wait for an event to be generated from the specified source (default = %default)")
	timinggroup.add_option("--run-timeout", dest="RUN_TIMEOUT", type="int",
	default=None, metavar="NSECS",
	help="max time in whole seconds to wait for the run to finish. This can be useful on batch systems such "
	"as the LCG Grid where tokens expire on a fixed wall-clock and can render long Rivet runs unable to write "
	"out the final histogram file (default = unlimited)")
	timinggroup.add_option("--histo-interval", dest="HISTO_WRITE_INTERVAL", type=int,
	default=10000, help="[experimental!] specify the number of events between histogram file updates. "
	"Set to 0 to only write out at the end of the run. Note that intermediate histograms will be those "
	"from the analyze step only: analysis finalizing is currently not executed until the end of the run.")
	parser.add_option_group(timinggroup)


	verbgroup = OptionGroup(parser, "Verbosity control")
	parser.add_option("-l", dest="NATIVE_LOG_STRS", action="append",
	default=[], help="set a log level in the Rivet library")
	verbgroup.add_option("-v", "--verbose", action="store_const", const=logging.DEBUG, dest="LOGLEVEL",
	default=logging.INFO, help="print debug (very verbose) messages")
	verbgroup.add_option("-q", "--quiet", action="store_const", const=logging.WARNING, dest="LOGLEVEL",
	default=logging.INFO, help="be very quiet")
	parser.add_option_group(verbgroup)
	opts, args = parser.parse_args()


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

	## Control native Rivet library logger
	for l in opts.NATIVE_LOG_STRS:
	name, level = None, None
	try:
	name, level = l.split("=")
	except:
	name = "Rivet"
	level = l
	## Fix name
	if name != "Rivet" and not name.startswith("Rivet."):
	name = "Rivet." + name
	try:
	## Get right error type
	level = rivet.LEVELS.get(level.upper(), None)
	logging.debug("Setting log level: %s %d" % (name, level))
	rivet.setLogLevel(name, level)
	except:
	logging.warning("Couldn't process logging string '%s'" % l)


	## We allow comma-separated lists of analysis names -- normalise the list here
	newanas = []
	for a in opts.ANALYSES:
	if "," in a:
	newanas += a.split(",")
	else:
	newanas.append(a)
	opts.ANALYSES = newanas


	## Parse supplied cross-section
	if opts.CROSS_SECTION is not None:
	xsstr = opts.CROSS_SECTION
	try:
	opts.CROSS_SECTION = float(xsstr)
	except:
	import re
	suffmatch = re.search(r"[^\d.]", xsstr)
	if not suffmatch:
	raise ValueError("Bad cross-section string: %s" % xsstr)
	factor = base = None
	suffstart = suffmatch.start()
	if suffstart != -1:
	base = xsstr[:suffstart]
	suffix = xsstr[suffstart:].lower()
	if suffix == "mb":
	factor = 1e+9
	elif suffix == "mub":
	factor = 1e+6
	elif suffix == "nb":
	factor = 1e+3
	elif suffix == "pb":
	factor = 1
	elif suffix == "fb":
	factor = 1e-3
	elif suffix == "ab":
	factor = 1e-6
	if factor is None or base is None:
	raise ValueError("Bad cross-section string: %s" % xsstr)
	xs = float(base) * factor
	opts.CROSS_SECTION = xs


	## Print the available CLI options!
	#if opts.LIST_OPTIONS:
	# for o in parser.option_list:
	# print o.get_opt_string()
	# sys.exit(0)


	## Set up signal handling
	RECVD_KILL_SIGNAL = None
	def handleKillSignal(signum, frame):
	"Declare us as having been signalled, and return to default handling behaviour"
	global RECVD_KILL_SIGNAL
	logging.critical("Signal handler called with signal " + str(signum))
	RECVD_KILL_SIGNAL = signum
	signal.signal(signum, signal.SIG_DFL)
	## Signals to handle
	signal.signal(signal.SIGTERM, handleKillSignal);
	signal.signal(signal.SIGHUP, handleKillSignal);
	signal.signal(signal.SIGINT, handleKillSignal);
	signal.signal(signal.SIGUSR1, handleKillSignal);
	signal.signal(signal.SIGUSR2, handleKillSignal);
	try:
	signal.signal(signal.SIGXCPU, handleKillSignal);
	except:
	pass


	## Override/modify analysis search path
	if opts.ANALYSIS_PATH:
	rivet.setAnalysisLibPaths(opts.ANALYSIS_PATH.split(":"))
	if opts.ANALYSIS_PATH_APPEND:
	for ap in opts.ANALYSIS_PATH_APPEND.split(":"):
	rivet.addAnalysisLibPath(ap)
	if opts.ANALYSIS_PATH_PWD:
	rivet.addAnalysisLibPath(".")


	## List of analyses
	all_analyses = rivet.AnalysisLoader.analysisNames()
	if opts.LIST_ANALYSES:
	## Treat args as case-insensitive regexes if present
	regexes = None
	if args:
	import re
	regexes = [re.compile(arg, re.I) for arg in args]
	for aname in all_analyses:
	if not regexes:
	toshow = True
	else:
	toshow = False
	for regex in regexes:
	if regex.search(aname):
	toshow = True
	break
	if toshow:
	msg = aname
	if opts.LOGLEVEL <= logging.INFO:
	a = rivet.AnalysisLoader.getAnalysis(aname)
	st = "" if a.status() == "VALIDATED" else ("[" + a.status() + "] ")
	msg = "%-25s %s" % (aname, st + rivet.util.detex(a.summary()))
	print msg
	sys.exit(0)


	## Show analyses' details
	if len(opts.SHOW_ANALYSES) > 0:
	toshow = []
	for i, a in enumerate(opts.SHOW_ANALYSES):
	a_up = a.upper()
	if a_up in all_analyses and a_up not in toshow:
	toshow.append(a_up)
	else:
	## Treat as a case-insensitive regex
	import re
	regex = re.compile(a, re.I)
	for ana in all_analyses:
	if regex.search(ana) and a_up not in toshow:
	toshow.append(ana)

	## Show the matching analyses' details
	import textwrap
	for i, name in enumerate(sorted(toshow)):
	ana = rivet.AnalysisLoader.getAnalysis(name)
	print ""
	print name
	print len(name) * "="
	print ""
	print rivet.util.detex(ana.summary())
	print ""
	print "Status: %s" % ana.status()
	print ""

	if ana.inspireId():
	print "Inspire ID: %s" % ana.inspireId()
	print "Inspire URL: http://inspire-hep.net/record/%s" % ana.inspireId()
	print "HepData URL: http://hepdata.cedar.ac.uk/view/ins%s" % ana.inspireId()
	elif ana.spiresId():
	print "Spires ID: %s" % ana.spiresId()
	print "Inspire URL: http://inspire-hep.net/search?p=find+key+%s" % ana.spiresId()
	print "HepData URL: http://hepdata.cedar.ac.uk/view/irn%s" % ana.spiresId()

	if ana.experiment():
	print "Experiment: %s" % ana.experiment(),
	if ana.collider():
	print "(%s)" % ana.collider()

	if ana.year():
	print "Year of publication: %s" % ana.year()

	print "Authors:"
	for a in ana.authors():
	print " " + a

	print ""

	print "Description:"
	twrap = textwrap.TextWrapper(width=75, initial_indent=2" ", subsequent_indent=2" ")
	print twrap.fill(rivet.util.detex(ana.description()))

	print ""

	if ana.requiredBeams():
	def pid_to_str(pid):
	if pid == 11:
	return "e-"
	elif pid == -11:
	return "e+"
	elif pid == 2212:
	return "p+"
	elif pid == -2212:
	return "p-"
	elif pid == 10000:
	return "*"
	else:
	return str(pid)
	beamstrs = []
	for bp in ana.requiredBeams():
	beamstrs.append(pid_to_str(bp[0]) + " " + pid_to_str(bp[1]))
	print "Beams:", ", ".join(beamstrs)

	if ana.requiredEnergies():
	print "Beam energies:", "; ".join(["(%0.1f, %0.1f)" % (epair[0], epair[1]) for epair in ana.requiredEnergies()]), "GeV"
	else:
	print "Beam energies: ANY"

	if ana.runInfo():
	print "Run details:"
	twrap = textwrap.TextWrapper(width=75, initial_indent=2" ", subsequent_indent=4" ")
	for l in ana.runInfo().split("\n"):
	print twrap.fill(l)

	if ana.references():
	print ""
	print "References:"
	for r in ana.references():
	url = None
	if r.startswith("arXiv:"):
	code = r.split()[0].replace("arXiv:", "")
	url = "http://arxiv.org/abs/" + code
	elif r.startswith("doi:"):
	code = r.replace("doi:", "")
	url = "http://dx.doi.org/" + code
	if url is not None:
	r += " - " + url
	print " %s" % r

	if i+1 < len(toshow):
	print "\n"
	sys.exit(0)


	## Identify HepMC files/streams
	## TODO: check readability, deal with stdin
	if len(args) > 0:
	HEPMCFILES = args
	else:
	HEPMCFILES = ["-"]


	## Event number logging
	def logNEvt(n, starttime, maxevtnum):
	if n % 10000 == 0:
	nevtloglevel = logging.CRITICAL
	elif n % 1000 == 0:
	nevtloglevel = logging.WARNING
	elif n % 100 == 0:
	nevtloglevel = logging.INFO
	else:
	nevtloglevel = logging.DEBUG
	currenttime = datetime.datetime.now().replace(microsecond=0)
	elapsedtime = currenttime - starttime
	logging.log(nevtloglevel, "Event %d (%s elapsed)" % (n, str(elapsedtime)))
	# if maxevtnum is None:
	# logging.log(nevtloglevel, "Event %d (%s elapsed)" % (n, str(elapsedtime)))
	# else:
	# remainingtime = (maxevtnum-n) * elapsedtime.total_seconds() / float(n)
	# eta = time.strftime("%a %b %d %H:%M", datetime.localtime(currenttime + remainingtime))
	# logging.log(nevtloglevel, "Event %d (%d s elapsed / %d s left) -> ETA: %s" %
	# (n, elapsedtime, remainingtime, eta))


	## Do some checks on output histo file, before we stat the event loop
	histo_parentdir = os.path.dirname(os.path.abspath(opts.HISTOFILE))

	if not os.path.exists(histo_parentdir):
	logging.error('Parent path of output histogram file does not exist: %s\nExiting.' % histo_parentdir)
	sys.exit(1)

	if not os.access(histo_parentdir,os.W_OK):
	logging.error('Insufficient permissions to write output histogram file to directory %s\nExiting.' % histo_parentdir)
	sys.exit(1)


	## Set up analysis handler
	RUNNAME = opts.RUN_NAME or ""
	ah = rivet.AnalysisHandler(RUNNAME)
	ah.setIgnoreBeams(opts.IGNORE_BEAMS)
	for a in opts.ANALYSES:
	#a_up = a.upper()
	## Print warning message and exit if not a valid analysis name
	if not a in all_analyses:
	logging.warning("'%s' is not a known Rivet analysis! Do you need to set RIVET_ANALYSIS_PATH or use the --pwd switch?\n" % a)
	# TODO: lay out more neatly, or even try for a "did you mean XXXX?" heuristic?
	logging.warning("There are %d currently available analyses:\n" % len(all_analyses) + ", ".join(all_analyses))
	sys.exit(1)
	logging.debug("Adding analysis '%s'" % a)
	ah.addAnalysis(a)

	## Read and process events
	run = rivet.Run(ah)
	if opts.CROSS_SECTION is not None:
	logging.info("User-supplied cross-section = %e pb" % opts.CROSS_SECTION)
	run.setCrossSection(opts.CROSS_SECTION)
	if opts.LIST_USED_ANALYSES is not None:
	run.setListAnalyses(opts.LIST_USED_ANALYSES)

	## Print platform type
	import platform
	starttime = datetime.datetime.now().replace(microsecond=0)
	logging.info("Rivet %s running on machine %s (%s) at %s" % \
	(rivet.version(), platform.node(), platform.machine(), str(starttime)))


	def min_nonnull(a, b):
	"A version of min which considers None to always be greater than a real number"
	rtn = min(a, b)
	if rtn is not None:
	return rtn
	if a is not None:
	return a
	return b


	## Set up an event timeout handler
	class TimeoutException(Exception):
	pass
	if opts.EVENT_TIMEOUT or opts.RUN_TIMEOUT:
	def evttimeouthandler(signum, frame):
	logging.warn("It has taken more than %d secs to get an event! Is the input event stream working?" %
	min_nonnull(opts.EVENT_TIMEOUT, opts.RUN_TIMEOUT))
	raise TimeoutException("Event timeout")
	signal.signal(signal.SIGALRM, evttimeouthandler)


	## Init run based on one event
	hepmcfile = HEPMCFILES[0]
	## Apply a file-level weight derived from the filename
	hepmcfileweight = 1.0
	if ":" in hepmcfile:
	hepmcfile, hepmcfileweight = hepmcfile.rsplit(":", 1)
	hepmcfileweight = float(hepmcfileweight)
	try:
	if opts.EVENT_TIMEOUT or opts.RUN_TIMEOUT:
	signal.alarm(min_nonnull(opts.EVENT_TIMEOUT, opts.RUN_TIMEOUT))
	init_ok = run.init(hepmcfile, hepmcfileweight)
	signal.alarm(0)
	if not init_ok:
	logging.error("Failed to initialise using event file '%s'... exiting" % hepmcfile)
	sys.exit(2)
	except TimeoutException, te:
	logging.error("Timeout in initialisation from event file '%s'... exiting" % hepmcfile)
	sys.exit(3)


	## Event loop
	evtnum = 0
	for fileidx, hepmcfile in enumerate(HEPMCFILES):
	## Apply a file-level weight derived from the filename
	hepmcfileweight = 1.0
	if ":" in hepmcfile:
	hepmcfile, hepmcfileweight = hepmcfile.rsplit(":", 1)
	hepmcfileweight = float(hepmcfileweight)
	## Open next HepMC file (NB. this doesn't apply to the first file: it was already used for the run init)
	if fileidx > 0:
	run.openFile(hepmcfile, hepmcfileweight)
	if not run.readEvent():
	logging.warning("Could not read events from '%s'" % hepmcfile)
	continue
	msg = "Reading events from '%s'" % hepmcfile
	if hepmcfileweight != 1.0:
	msg += " (file weight = %e)" % hepmcfileweight
	logging.info(msg)
	while opts.MAXEVTNUM is None or evtnum < opts.MAXEVTNUM:
	evtnum += 1
	logNEvt(evtnum, starttime, opts.MAXEVTNUM)

	## Process this event
	processed_ok = run.processEvent()
	if not processed_ok:
	logging.warn("Event processing failed for evt #%i!" % evtnum)
	break

	## Set flag to exit event loop if run timeout exceeded
	if opts.RUN_TIMEOUT and (time.time() - starttime) > opts.RUN_TIMEOUT:
	logging.warning("Run timeout of %d secs exceeded... exiting gracefully" % opts.RUN_TIMEOUT)
	RECVD_KILL_SIGNAL = True

	## Exit the loop if signalled
	if RECVD_KILL_SIGNAL is not None:
	break

	## Read next event (with timeout handling if requested)
	try:
	if opts.EVENT_TIMEOUT:
	signal.alarm(opts.EVENT_TIMEOUT)
	read_ok = run.readEvent()
	signal.alarm(0)
	if not read_ok:
	break
	except TimeoutException, te:
	logging.error("Timeout in reading event from '%s'... exiting" % hepmcfile)
	sys.exit(3)

	## Write a histo file snapshot if appropriate
	if opts.HISTO_WRITE_INTERVAL is not None and opts.HISTO_WRITE_INTERVAL > 0:
	if evtnum % opts.HISTO_WRITE_INTERVAL == 0:
	ah.writeData(opts.HISTOFILE)

	loopendtime = datetime.datetime.now().replace(microsecond=0)
	logging.info("Finished event loop at %s" % str(loopendtime))
	logging.info("Cross-section = %e pb" % ah.crossSection())
	print

	## Finalize and write out data file
	run.finalize()
	ah.writeData(opts.HISTOFILE)
	print
	endtime = datetime.datetime.now().replace(microsecond=0)
	logging.info("Rivet run completed at %s, time elapsed = %s" % (str(endtime), str(endtime-starttime)))
	+print
	+logging.info("Histograms written to", os.path.abspath(opts.HISTOFILE)

File Metadata

Mime Type: text/x-diff
Expires: Tue, Nov 19, 6:25 PM (1 d, 20 h)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 3805582
Default Alt Text: (20 KB)

No OneTemporaryActions

View Options

File Metadata

Event Timeline

No OneTemporary
Actions