mirror of
https://github.com/qpdf/qpdf.git
synced 2024-11-05 21:07:50 +00:00
3630a8c597
It is common to just read a few objects. Checking extraction of the first page exercises this to make sure we don't accidentally introduce a change that makes that case worse, such as adding an unnecessary traversal of the file, prematurely resolving objects we don't need, etc.
370 lines
8.6 KiB
Perl
Executable File
370 lines
8.6 KiB
Perl
Executable File
#!/usr/bin/env perl
|
|
require 5.008;
|
|
use warnings;
|
|
use strict;
|
|
use File::Basename;
|
|
use Time::HiRes qw(gettimeofday tv_interval);
|
|
use File::Path qw(make_path);
|
|
use IPC::Open3;
|
|
use IO::Pipe;
|
|
|
|
my $whoami = basename($0);
|
|
$| = 1;
|
|
|
|
# [ name, [ args ] ]
|
|
# If <IN> appears, it is replaced with the input file name. Otherwise,
|
|
# the input file name is added to the end of the arguments.
|
|
my @tests = (
|
|
['no arguments', []],
|
|
['generate object streams', ['--object-streams=generate']],
|
|
['disable object streams', ['--object-streams=disable']],
|
|
['split pages', ['--split-pages', '--remove-unreferenced-resources=no']],
|
|
['shared resource check', ['--split-pages', '--remove-unreferenced-resources=auto']],
|
|
['linearize', ['--linearize']],
|
|
['encrypt', ['--encrypt', 'u', 'o', '256', '--']],
|
|
['extract first page', ['--empty', '--pages', '<IN>', '1', '--']],
|
|
['json-output', ['--json-output']],
|
|
['json-input', ['--json-input']],
|
|
);
|
|
|
|
# If arg is not found in help output, look here. If not here, skip test.
|
|
# { new => old } -- if new is not found, replace with old; if old is
|
|
# if old is empty, remove argument
|
|
my %arg_compat = (
|
|
'--remove-unreferenced-resources=no' => '--preserve-unreferenced-resources',
|
|
'--remove-unreferenced-resources=yes' => '',
|
|
'--remove-unreferenced-resources=auto' => undef,
|
|
'--report-memory-usage' => '',
|
|
);
|
|
|
|
my $executable = undef;
|
|
my $test_dir = undef;
|
|
my $test_file = undef;
|
|
my $workdir = undef;
|
|
my $maxtime = undef;
|
|
my $iterations = undef;
|
|
|
|
my $default_executable = 'build/qpdf/qpdf';
|
|
my $default_test_dir = '../performance-test-files';
|
|
my $default_test_file = undef;
|
|
my $default_workdir = 'build/perf';
|
|
my $default_maxtime = 20;
|
|
my $default_iterations = 20;
|
|
|
|
sub usage
|
|
{
|
|
warn "
|
|
Usage: $whoami [ args ]
|
|
--dir dir test on all files in dir (default: $default_test_dir)
|
|
--file file test only on the named file
|
|
--executable qpdf use the specified qpdf (default: $default_executable)
|
|
--workdir where to write output pdfs (default: $default_workdir)
|
|
--maxtime maximum time for a test; 0 means unlimited (default: $default_maxtime)
|
|
--iterations number of iterations (default: $default_iterations)
|
|
--test regexp run only tests that match specified pattern
|
|
|
|
Populate $test_dir with files you want to use for performance
|
|
benchmarking. PDF files and qpdf JSON files are allowed. The qpdf
|
|
release process uses a clone of
|
|
https://github.com/qpdf/performance-test-files for this purpose.
|
|
|
|
Tests:
|
|
";
|
|
foreach my $t (@tests)
|
|
{
|
|
warn " $t->[0]\n";
|
|
}
|
|
exit 2;
|
|
}
|
|
|
|
my $test_re = undef;
|
|
while (@ARGV)
|
|
{
|
|
my $arg = shift(@ARGV);
|
|
if ('--dir' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$test_dir = shift(@ARGV);
|
|
$test_file = undef;
|
|
}
|
|
elsif ('--file' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$test_file = shift(@ARGV);
|
|
$test_dir = undef;
|
|
}
|
|
elsif ('--executable' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$executable = shift(@ARGV);
|
|
}
|
|
elsif ('--workdir' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$workdir = shift(@ARGV);
|
|
}
|
|
elsif ('--maxtime' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$maxtime = shift(@ARGV);
|
|
}
|
|
elsif ('--iterations' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$iterations = shift(@ARGV);
|
|
}
|
|
elsif ('--test' eq $arg)
|
|
{
|
|
usage() unless @ARGV;
|
|
$test_re = shift(@ARGV);
|
|
}
|
|
else
|
|
{
|
|
usage();
|
|
}
|
|
}
|
|
|
|
if ((! defined $test_dir) && (! defined $test_file))
|
|
{
|
|
$test_dir = $default_test_dir;
|
|
}
|
|
if (! defined $executable)
|
|
{
|
|
$executable = $default_executable;
|
|
}
|
|
if (! defined $workdir)
|
|
{
|
|
$workdir = $default_workdir;
|
|
}
|
|
if (! defined $maxtime)
|
|
{
|
|
$maxtime = $default_maxtime;
|
|
}
|
|
if (! defined $iterations)
|
|
{
|
|
$iterations = $default_iterations;
|
|
}
|
|
|
|
my @test_files = ();
|
|
my @json_test_files = ();
|
|
{ # private scope
|
|
my @tmp = ();
|
|
if (defined $test_file)
|
|
{
|
|
push(@tmp, $test_file);
|
|
}
|
|
else
|
|
{
|
|
opendir(D, $test_dir) or die "
|
|
$whoami: can't open test directory: $!
|
|
|
|
Configured test directory: $test_dir
|
|
|
|
Populate $test_dir with a clone of the
|
|
qpdf/performance-test-files github repository.
|
|
Run $whoami --help for details.
|
|
|
|
Repository URL: https://github.com/qpdf/performance-test-files
|
|
|
|
";
|
|
my @entries = readdir(D);
|
|
closedir(D);
|
|
for (sort @entries)
|
|
{
|
|
my $file = "$test_dir/$_";
|
|
if (-f $file && $file =~ m/.(pdf|json)$/i)
|
|
{
|
|
push(@tmp, $file);
|
|
}
|
|
}
|
|
}
|
|
foreach my $i (@tmp)
|
|
{
|
|
if ($i =~ m/.json$/)
|
|
{
|
|
push(@json_test_files, $i);
|
|
}
|
|
else
|
|
{
|
|
push(@test_files, $i);
|
|
}
|
|
}
|
|
}
|
|
|
|
my $report_mem = filter_args(["--report-memory-usage"]);
|
|
{
|
|
my ($r, $mem) = run_cmd($executable, @$report_mem,
|
|
"--empty", File::Spec->devnull());
|
|
if ($r != 0)
|
|
{
|
|
die "$whoami: $executable doesn't seem to work\n";
|
|
}
|
|
if ($mem == 0)
|
|
{
|
|
print "** Note: memory information is not available **\n";
|
|
}
|
|
}
|
|
|
|
run_tests();
|
|
print "\n";
|
|
|
|
sub filter_args
|
|
{
|
|
my $args = shift;
|
|
my $help = `$executable --help=all`;
|
|
my $new_args = [];
|
|
foreach my $arg (@$args)
|
|
{
|
|
my $to_check = $arg;
|
|
$to_check =~ s/=.*$//;
|
|
if (($to_check =~ m/^-/) && (index($help, $to_check) == -1))
|
|
{
|
|
my $new_arg = $arg_compat{$arg};
|
|
if (! defined $new_arg)
|
|
{
|
|
return undef;
|
|
}
|
|
if ($new_arg ne '')
|
|
{
|
|
print " replacing $arg with $new_arg\n";
|
|
push(@$new_args, $new_arg);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
push(@$new_args, $arg);
|
|
}
|
|
}
|
|
$new_args;
|
|
}
|
|
|
|
sub run_tests
|
|
{
|
|
my $args = shift;
|
|
|
|
chomp(my $commit = `git describe @`);
|
|
print "commit: $commit\n";
|
|
print "Format: time-in-seconds RAM-in-MiB filename\n";
|
|
make_path($workdir);
|
|
foreach my $test (@tests)
|
|
{
|
|
my ($name, $args) = @$test;
|
|
if ((defined $test_re) && $name !~ m/$test_re/)
|
|
{
|
|
print " skipping test $name\n";
|
|
next;
|
|
}
|
|
print " test: $name\n";
|
|
$args = filter_args($args);
|
|
if (! defined $args)
|
|
{
|
|
print " skipping (unknown arguments)\n";
|
|
next;
|
|
}
|
|
my $test_files = \@test_files;
|
|
foreach my $arg (@$args)
|
|
{
|
|
if ($arg eq '--json-input')
|
|
{
|
|
$test_files = \@json_test_files;
|
|
last;
|
|
}
|
|
}
|
|
foreach my $file (@$test_files)
|
|
{
|
|
my $time = run_test($file, $args);
|
|
if (defined $time)
|
|
{
|
|
print " $time " . basename($file) ."\n";
|
|
}
|
|
else
|
|
{
|
|
print " $file skipped\n";
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
sub run_test
|
|
{
|
|
my ($file, $args) = @_;
|
|
|
|
my $outfile = "out.pdf";
|
|
my $found_in = 0;
|
|
my @cmd = ($executable, @$report_mem);
|
|
for (@$args)
|
|
{
|
|
my $arg = $_;
|
|
if ($arg eq '--json-output')
|
|
{
|
|
$outfile = "out.json";
|
|
}
|
|
elsif ($arg eq '<IN>')
|
|
{
|
|
$found_in = 1;
|
|
$arg = $file;
|
|
}
|
|
push(@cmd, $arg);
|
|
}
|
|
if (! $found_in)
|
|
{
|
|
push(@cmd, $file);
|
|
}
|
|
push(@cmd, "$workdir/$outfile");
|
|
# Run once and discard to update caches
|
|
system("sync");
|
|
run_cmd(@cmd);
|
|
my $i = 0;
|
|
my $total = 0;
|
|
my $max_mem = 0;
|
|
while ($i < $iterations)
|
|
{
|
|
my $start = [gettimeofday];
|
|
my ($r, $mem) = run_cmd(@cmd);
|
|
if ($r == 2)
|
|
{
|
|
# interrupt
|
|
exit(2);
|
|
}
|
|
my $end = [gettimeofday];
|
|
if ($r != 0)
|
|
{
|
|
print " command failed; ignoring results\n";
|
|
return undef;
|
|
}
|
|
$max_mem = $mem > $max_mem ? $mem : $max_mem;
|
|
my $elapsed = tv_interval($start, $end);
|
|
$total += $elapsed;
|
|
++$i;
|
|
if (($maxtime > 0) && ($total >= $maxtime) && ($i >= 3))
|
|
{
|
|
# This is taking too long, so take what we have
|
|
last;
|
|
}
|
|
}
|
|
return sprintf("%8.4f %8.4f", $total / $i, $max_mem / 1048576);
|
|
}
|
|
|
|
sub run_cmd
|
|
{
|
|
my @cmd = @_;
|
|
my $pipe = IO::Pipe->new();
|
|
my $pid = open3(my $child_in, '>&STDOUT', $pipe->writer(), @cmd);
|
|
$child_in->close();
|
|
waitpid($pid, 0);
|
|
my $r = $?;
|
|
my $mem = 0;
|
|
while (<$pipe>)
|
|
{
|
|
if (m/qpdf-max-memory-usage (\d+)/)
|
|
{
|
|
$mem = $1;
|
|
}
|
|
else
|
|
{
|
|
warn $_;
|
|
}
|
|
}
|
|
($r, $mem);
|
|
}
|