histogram-diff.sh
#!/bin/sh
#
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
# histogram-diff.sh [-c <count>] <base> <incr>
#
# Compute incremental memory growth from histogram in file <base> to
# histogram in file <incr>, displaying at most <count> rows.
# How many rows are we gonna show?
COUNT=20
# Read arguments
while [ $# -gt 0 ]; do
case "$1" in
-c) COUNT=$2
shift 2
;;
*) break
;;
esac
done
BASE=$1
INCR=$2
# Sort the base and incremental files so that we can `join' them on
# the type name
sort $BASE > /tmp/$$.left
sort $INCR > /tmp/$$.right
# Do the join. The `awk' script computes the difference between
# the base and the incremental files.
join /tmp/$$.left /tmp/$$.right \
| awk '{ print $1, $2, $3, $4, $5, $4 - $2, $5 - $3; }' \
> /tmp/$$.joined
rm -f /tmp/$$.left /tmp/$$.right
# Now compute a `TOTAL' row.
awk '{ tobj1 += $2; tbytes1 += $3; tobj2 += $4; tbytes2 += $5; tdobj += $6; tdbytes += $7; } END { print "TOTAL", tobj1, tbytes1, tobj2, tbytes2, tdobj, tdbytes; }' /tmp/$$.joined \
> /tmp/$$.sorted
# Then, we sort by the largest delta in bytes.
sort -nr +6 /tmp/$$.joined >> /tmp/$$.sorted
rm -f /tmp/$$.joined
# Pretty-print, including percentages
cat <<EOF > /tmp/$$.awk
BEGIN {
print " ---- Base ---- ---- Incr ---- ----- Difference ----";
print "Type Count Bytes Count Bytes Count Bytes %Total";
}
\$1 == "TOTAL" {
tbytes = \$7;
}
NR <= $COUNT {
printf "%-22s %6d %8d %6d %8d %6d %8d %6.2lf\n", \$1, \$2, \$3, \$4, \$5, \$6, \$7, 100.0 * \$7 / tbytes;
}
NR > $COUNT {
oobjs1 += \$2; obytes1 += \$3;
oobjs2 += \$4; obytes2 += \$5;
odobjs += \$6; odbytes += \$7;
}
END {
printf "%-22s %6d %8d %6d %8d %6d %8d %6.2lf\n", "OTHER", oobjs1, obytes1, oobjs2, obytes2, odobjs, odbytes, odbytes * 100.0 / tbytes;
}
EOF
# Now pretty print the file, and spit it out on stdout.
awk -f /tmp/$$.awk /tmp/$$.sorted
rm -f /tmp/$$.awk /tmp/$$.sorted