#!/usr/bin/env perl -w
########################################################################
# sum-io - Given one or more files containing dtruss output and a
#          pattern matching the file name(s) you are interested in,
#          report the number of bytes written to and read from the
#          matched file(s) according to dtruss.
#
# To collect suitable output, say something like this:
#
#    $ sudo dtruss -p $(pgrep TargetProgram) 2>&1 | tee results
#
# Hit Ctrl-C when you've collected sufficient results.  (The tee(1) lets
# you monitor TargetProgram to see when it stops issuing interesting
# syscalls.)  Then run this script on the results file:
#
#    $ sum-io file_re results
#
# The file_re argument can be any Perl regex.
#
# Created 2016.09.03 by Warren Young
#
# LICENSE: https://creativecommons.org/licenses/by-sa/3.0/
########################################################################

use File::Basename;
use List::MoreUtils qw(uniq);
use Number::Format qw(format_bytes);

my ($file_re, @dtruss_results) = @ARGV;
die "usage: $0 <filename-regex> <dtruss-results> [more-results...]\n\n"
		unless length($file_re) && @dtruss_results;

my (%read, %written, $target_fd, $target_fn) = (0, 0);
for my $results (@dtruss_results) {
	open my $rf, '<', $results
			or die "Cannot read dtruss result file $results: $!\n";

	while (<$rf>) {
		my ($sc, $path, $ret) = /^(\w+)\((".*\\0")?.*= (\d+)/;
		next unless length($sc) && length($ret);

		if ($sc eq 'open' && /$file_re/) {
			# Save the return value of open(2) as the target FD.  The
			# base conversion is necessary because dtruss prints syscall
			# results as dec but syscall integer *args* as hex.  Grrr.
			$target_fd = sprintf("0x%X", int($ret));
			$target_fn = basename(substr $path, 1, -3);
		}
		elsif ($target_fd) {
			# Does this look like a syscall involving the target file?
			my ($call_fd) = /^\w+\((0x[0-9A-F]+)/;
			if ($call_fd && $call_fd eq $target_fd) {
				if ($sc eq 'close') {
					# The target file is closed now.  Forget it.
					undef $target_fd;
					undef $target_fn;
				}
				elsif ($sc =~ /read/) {
					$read{$target_fn} += $ret;
				}
				elsif ($sc =~ /write/) {
					$written{$target_fn} += $ret;
				}
				# else, syscall is uninteresting
			}
		}
		# else, haven't found the open(2) call for the target file yet
	}

	close $rf;
}

for my $f (uniq(sort(keys %read, keys %written))) {
	my ($r, $w) = ($read{$f}, $written{$f});
	if ($r or $w) {
		print "Data I/O for file $f:\n";
		print "    ", format_bytes($read{$f}), " read\n" if $r;
		print "    ", format_bytes($written{$f}), " written\n" if $w;
	}
}