plot-usage

来自「EM算法的改进」· 代码 · 共 293 行

#!/bin/csh
#
# $Id: plot-usage 1339 2006-09-21 19:46:28Z tbailey $
# $Log$
# Revision 1.2  2005/10/05 06:18:35  nadya
# use full path for "rm". Asssume everybody has /bin/rm.
#
# Revision 1.1.1.1  2005/07/29 00:01:33  nadya
# Importing from meme-3.0.14, and adding configure/make
#
#

set months = "JanFebMarAprMayJunJulAugSepOctNovDec"
set format = "meme" 
set t = 0
set ts = "runs"
set pgm = $0; set pgm = $pgm:t
set args = ($*)
if ($#argv < 0) then
  usage:
  more << USAGE
  USAGE:
        $pgm [<logfile>] [-meme] [-mast] [-metameme] [-r] [-u] [-m] [-c] [-s]
 
	[<logfile>]	MEME server log file; default: read standard input
	[-meme]		log file is in MEME format; (default)
	[-mast]		log file is in MAST format; default: MEME format
	[-metameme]	log file is in MetaMEME format; default: MEME format
	[-r]		plot number of runs (default)
	[-u]            plot number of users; default: number of runs
	[-m]		plot maximum runs by any user; default: number of runs
	[-c]		plot cpu time for all runs; default: number of runs
			note: meme only; cpu time not recorded in mast-log
	[-s]            plot to screen
 
        Plot the number of runs or users of MEME, MAST or MetaMEME
	month-by-month.

        Number of users is based on the number of distinct email addresses
        given for MEME and MAST.
        Number of users is based on the number of distinct IP addresses
        for MetaMEME.

	Creates files:
		$pgm.<type>[.<logfile>].[ps tex]
	where <type> is "runs", "max", "users" or "cpu".
USAGE
  exit 1
endif

unlimit cputime
onintr cleanup

# get input arguments
set logfile = ""
set name = ""
while ("$1" != "")
  switch ($1)
  case -h:
    goto usage
    break
  case -meme:
    breaksw
  case -mast:
    set format = "mast";
    breaksw
  case -metameme:
    set format = "metameme";
    breaksw
  case -r:
    breaksw
  case -u:
    set t = 1
    set ts = "users"
    breaksw
  case -m:
    set t = 2
    set ts = "max"
    breaksw
  case -c: 
    set t = 3
    set ts = "cpu"
    breaksw
  case -s: 
    set s = 1
    breaksw
  default:
    set logfile = $1
    set name = ".$1"
  endsw
  shift
end

# set name to meme/mast/metameme if input is standard input
if ($logfile == "") then
  set name = ".$format"
endif

if ($t == 3 && $format == "mast") then
  echo "CPU time is not recorded in mast-log so -c not available"
  exit 1
endif

set gawk = $pgm.$$.gawk.tmp
cat << "END" > $gawk
BEGIN {
  big = 100000; 
  little = -1;
  min_year = min_month = big; 
  max_year = max_month = -1;
  for (i=0; i<12; i++) smonth[i+1] = substr(months, 3*i+1, 1);
}
# read next record in file
NF >= 4 {
  ok = 1;
  if (format == "meme" || format == "metameme") {	# meme format
    # check for missing host and repair it
    if ($2 == "submit:") {
      for (i=NF+1; i>=2; i--) {
        $i = $(i-1);			# shift everything 1 right
      }
      $1 = "host";
    }
    # check for a bad MEME record where the submit date is missing
    # and try to repair it by using start date/time for submit date/time
    if ($4 == "start:") {
      # insert the start time in place of the submit time
      for (i=NF+2; i>=6; i--) {
        $i = $(i-2);			# shift everything 2 right
      }
      $4 = $7;				# copy start time & date
      $5 = $8; 
    }
    # check for a bad Meta-MEME record where the start date is missing
    # and try to repair it by using submit date/time for start date/time
    if ($6 == "end:") {
      # insert the submit time in place of the start time
      for (i=NF+3; i>=9; i--) {
        $i = $(i-3);			# shift everything 3 right
      }
      $6 = "start:";
      $7 = $10;				# copy start time & date
      $7 = $11; 
    }
    # check for a bad Meta-MEME record where the "start" is missing
    if (0 && $8 == "end:") {
      # insert the "start:" 
      for (i=NF+2; i>=9; i--) {
        $i = $(i-1);			# shift everything 1 right
      }
      $6 = $9;				# copy start time & date
      $7 = $10; 
    }
    # check for "end" instead of "end:"
    if ($9 == "end") {$9 = "end:";}
    # check for bad record 
    if ($3 != "submit:" || $6 != "start:" || $9 != "end:") {
      print "bad record number", NR, "\n" > "/dev/stderr"
      print $0, "\n" > "/dev/stderr"
      ok = 0;
    } 
    date = $4;
    if (format == "meme") {
      day = substr(date, 1, 2) + 0;
      imonth = substr(date, 4, 2) + 0;
    } else {
      day = substr(date, 4, 2) + 0;
      imonth = substr(date, 1, 2) + 0;
    }
    year = substr(date, 7, 2) + 0;
    if (year < 96) {year = 2000 + year;} else {year = 1900 + year;}
    if (t == 3) {				# get cpu time
      start = $8;				# start time hh:mm:ss
      sex1 = substr(start,7,2)+60*(substr(start,4,2)+(60*substr(start,1,2)));
      end = $11;				# end time hh:mm:ss
      sex2 = substr(end,7,2)+60*(substr(end,4,2)+(60*substr(end,1,2)));
      cpu_hrs = sex2-sex1;			# elapsed sconds
      if (cpu_hrs < 0) cpu_hrs += 24*60*60;	# day change
      cpu_hrs /= 3600;			# convert to hours
      #print $2, cpu_hrs;
    }
  } else {					# mast format
    month = $4;
    year = $8;
    imonth = int(index(months, month)/3)+1;     # get index of month
  }
  if (ok) {
    user = $NF;
    runs[year,imonth]++;			# increment runs
    if (t==3) cpu[year,imonth] += cpu_hrs;	# increase cpu time
    if (!(users[year,imonth,user])) {
      nusers[year,imonth]++;
      users[year,imonth,user] = 1;
    }
    nruns[year,imonth,user]++;
    if (nruns[year,imonth,user] > maxruns[year,imonth]) {
      maxruns[year,imonth] = nruns[year,imonth,user];
    }
    # save minimum date
    if (year < min_year) {
      min_year = year; 
      min_month = imonth;
    }
    if (year == min_year && imonth < min_month) min_month = imonth;
    # save maximum date
    if (year > max_year) {
      #print "new year: ", year, " new month:, imonth, NR > "/dev/stderr";
      max_year = year;
      max_month = imonth;
    }
    if (year == max_year && imonth > max_month) max_month = imonth;
  } # ok
}
END {
  # print number of runs/users/cpu_time for each month
  # print "min", min_year, min_month, "max", max_year, max_month > plotdata
  x = 1;
  ymax = 0;
  printf("set xtics (") > com;
  #print "max_year", max_year, "max_month", max_month;
  for (year=min_year; year<=max_year; year++) {
    if (year==min_year) { min = min_month; } else { min = 1; }
    if (year==max_year) { max = max_month; } else { max = 12; }
    for (imonth=min; imonth<=max; imonth++) {
      printf("\'%s\' %d", smonth[imonth], x) > com;
      if (! (year==max_year && imonth==max) ) printf(", ") > com;
      if (t == 1) {
        y = nusers[year,imonth];
      } else if (t == 2) {  
        y = maxruns[year,imonth];
      } else if (t == 3) {  
        y = cpu[year,imonth];
      } else {
        y = runs[year,imonth];
      }
      if (y > ymax) ymax = y; 
      print x, y+.001 > plotdata;
      x++;
    }
  }
  printf(")\n") > com;
  print "set xlabel '", min_year, "-", max_year, "' 0,-1 font '1,40'" > com
  if (t == 1) {
    print "set ylabel 'USERS PER MONTH' font '1,40'" > com
  } else if (t == 2) {  
    print "set ylabel 'MAX RUNS PER MONTH' font '1,40'" > com
  } else if (t == 3) {  
    print "set ylabel 'CPU HOURS PER MONTH' font '1,40'" > com
  } else {
    print "set ylabel 'RUNS PER MONTH' font '1,40'" > com
  }
  print "set xrange [0:", x, "]" > com
  print "set yrange [0:", ymax, "]" > com
  print "set data style boxes" > com
  print "plot '" plotdata "' notitle" > com
  # print pointsize
  point = int(784/(x-1));
  if (point > 30) {point = 30;}
  print point;
}
"END"

# create file to hold points
set plotdata = $pgm.$$.plotdata.tmp
set com = $pgm.$$.com.tmp

# execute script
set point = `gawk -v format=$format -v plotdata=$plotdata -v com=$com -v months=$months -v t=$t -f $gawk $logfile`

#echo point = $point
#cat $com
#echo $plotdata
#cat $plotdata

# set up for screen or postscript
if ($?s) then
  set ps = "-ps _screen_" 
else
  set ps = "-ps $pgm.$ts$name"
endif

plotgen -f -nocat -color \
  -point $point \
  -file $com \
  $ps

if (!($?s)) then
  /bin/rm $pgm.$ts$name.tex
endif

cleanup:
/bin/rm $pgm.$$.*.tmp