#! /usr/bin/env python
# ______________________________________________________________________
#
# Filename: memstats
#
# Created: 20-JAN-2001,   Harry Melanson
#
# ______________________________________________________________________

"""Summarize D0 framework job memory statistics.

USAGE:

  memstats [-hv] framework.log

OPTIONS:

  -h, --help     --  Print help text and exit.
  -v             --  Verbose print (list all increases).

Description:

 memstats analyzes the log file of a framework job, and looks for
 increases in memory usage.  The framework executable must have been
 run with the -mem option.  By default, only the top ten memory
 increases are listed.  Selecting the -v option results in a complete
 list.
 
Limitations:

 The current version seems to only work with the output of d0reco.
 The framework appears to change the format of the output of its
 monitoring of memory usage, depending on the use of the 'Controller
 pattern'.  memstats was written to support the pattern used by
 d0reco.

"""

# ______________________________________________________________________

import sys, os
import glob
import string

# ______________________________________________________________________

__author__ = 'Harry Melanson, melanson@fnal.gov'
__version__ = '01.00.00'
__filename__ = 'memstats'
__doc__ = __doc__ % vars()

# ______________________________________________________________________


def process_command():
    """Process the command."""

    from scriptutil import getOptions
    optlist, args = getOptions('v', [], __doc__)

    #  Process command line arguments
    
    if len(args) == 0:
	print "\nERROR: No RECO log file specified."
        print "\nNAME: ", (os.path.basename(sys.argv[0]))
        print  __doc__
	sys.exit(-1)

    outfile = args[0]
    if len(glob.glob(outfile)) == 0:
	print "\nERROR: RECO output file", outfile, "does not exist.\n"
	sys.exit(-1)
    elif len(glob.glob(outfile)) != 1:
	print "\nERROR: More than one file found?\n"
	sys.exit(-1)

    VERBOSE = 0
    for opt in optlist:
	if opt.flag == '-v':
	    VERBOSE = 1
	
    # ________________________________________
    #
    
    result = extractMemUsage(outfile)
    printSummary(result, VERBOSE)
    
    
# ______________________________________________________________________
#
def extractMemUsage(outfile):
    """Return list of memory usage from a framework log file."""

    #  Process the file
    
    file = open(outfile,'r')

    results = []
    found = 0
    for line in file.readlines():

        
	#  Look for lines from framework with package/algorithm
	
	if string.find(line, '%ERLOG-i mem:') != -1:

	    items = string.split(line)
	    mem = string.atof(items[-2])
	    found = 1

	#  If previous line was package/algo, this line has
	#  the cpu mean time
	
	elif found == 1:
	    
	    found = 0
	    if (string.find(line, 'process') == -1 and
		string.find(line, 'generator') == -1 and
		string.find(line, 'decide') == -1 and
		string.find(line, 'analyze') == -1 and
		string.find(line, 'builder') == -1 and
		string.find(line, 'rcfgChanged') == -1 and
		string.find(line, 'output') == -1 and
		string.find(line, 'jobSummary') == -1 and
		string.find(line, 'runInit') == -1 and
		string.find(line, 'fileOpen') == -1 ):
		pass
	    else:
		items = string.split(line)
		package = items[-1]
		type = items[0]
		results.append( MemUse(package, type, mem) )

    file.close()

    return results

# ______________________________________________________________________
#

class MemUse:

    def __init__(self, package, type, mem):
	self.package = package
	self.type = type
	self.mem = mem
	self.count = 1
	self.leak = 0

def sortMemUse(a, b):
    if a.leak > b.leak:
	return 1
    elif b.leak > a.leak:
	return -1
    else:
	return 0

# ______________________________________________________________________
#

def printSummary(results, level):
    """Print summary of results at specified level of detail."""

    requests = {}

    #  Go through each memory request, and determine per package.

    for item in results:

	if not requests.has_key(item.package):
	    requests[item.package] = item
	else:
	    requests[item.package].count = requests[item.package].count + 1

    #  Try to associate memory increases to each package

    oldmem = results[0].mem
    for item in results[1:]:
	requests[item.package].leak = requests[item.package].leak + (item.mem - oldmem)
	oldmem = item.mem
    
    #  Make a sorted list based on number of memory requests
    
    list = []
    for key in requests.keys():
	list.append(requests[key])

    list.sort(sortMemUse)
    list.reverse()

    print "\nFinal memory usage: ", results[-1].mem, "MB\n"
    
    format1 = '%-29s %10s %10s'
    format2 = '%-35s %3i       %5.1f'

    print format1 % ('Package', 'Count', 'Increase (MB)')
    print format1 % ('-------', '-----', '-------------')

    if (level == 0):
	for item in list[:10]:
	    print format2 % (item.package, item.count, item.leak)
	    
    elif (level == 1):
	for item in list:
	    print format2 % (item.package, item.count, item.leak)

# ______________________________________________________________________
#
#  Execute command

if __name__ == '__main__':
    process_command()
    sys.exit(0)