writing more secure perl cgi scripts
1.0.0
如何编写更安全的Perl CGI脚本。指南有一些技巧。
2020年3月更新：将本教程从2008年迁移到Github，以进行档案目的。 （如今，有更好的编程语言和网络框架）。
本指南包含一些示例CGI脚本，这些脚本演示了如何改善CGI脚本中的安全性。示例count7.cgi具有最佳的安全性。
如果您在Perl中编写CGI脚本，则应该
使用-T标志。使用-w标志也是一个好主意。因此，脚本将从
#!/usr/bin/perl -Tw
避免尽可能多地使用临时文件。如果要调用外部程序，请尝试使用open2()调用。这是在假设外部程序可以从标准输入中读取输入数据并将结果写入标准输出的假设。例如，这样
use IPC::Open2;
my $childpid = open2(*READ, *WRITE, $config::blastall_path , "-m", "7", "-d" , $db, "-p", "blastp", "-b", $max_hits, "-v", $max_hits, "-M", "BLOSUM62" )
or die "Could not open pipe: $!";
print WRITE $blastinput;
close(WRITE);
my $blastoutput;
while(<READ>) {
$blastoutput .= $_;
}
close READ;
waitpid($childpid, 0);
永远不要使用这样的东西
system("myprogram -f $cgiparam");
而是使用逗号分隔的语法
system("myprogram","-f","$cgiparam");
Count.py是一个Python脚本，所有示例CGI脚本都使用。
#!/usr/bin/python
import sys
if len(sys.argv) == 2:
sys.exit()
elif len(sys.argv) == 3:
f=open(sys.argv[2], "r")
else:
sys.exit()
#f.seek(0)
lines=0
bytes=0
for line in f:
lines += 1
bytes += len( line )
if (sys.argv[1] == "bytes"):
print str(bytes)
if (sys.argv[1] == "lines"):
print str(lines)
配置Apache HTTPD允许服务器端包括（SSI）是危险的。该选项称为Includes 。
黑客可以使用SSI Exec语句上传文件。然后，这些命令将在Web浏览器试图访问上传文件后立即执行。通常，SSI仅用于以“ .shtml ”结尾的文件名激活。在此CGI脚本中，网络冲浪者选择文件名，并可以选择一个以“ .shtml ”结尾的文件名。
#!/usr/bin/perl
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
"filename",
$cgi->textfield(-name=>'fname'),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $fname = $cgi->param('fname');
my $content = $cgi->param('content');
my $type = $cgi->param('type');
my $filename="tmp/" . $fname;
my $outfh;
open($outfh, ">", $filename)
or die "Couldn't open $tmpfile for writing: $!";
print $outfh $content;
close $outfh;
my $result=`/home/esjolund/public_html/cgi-bin/count.py $type $filename`;
print "The file <b>$fname</b> has $result $type";
print $cgi->end_html;
}
检查use File::Temp qw(tempfile); 。如果您无法避免使用临时文件，请使用此PERL模块来创建它们。尽管在这里不可能，但您应该最好将UNLINK => 1用作tempfile()函数的参数。
#!/usr/bin/perl
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use File::Temp qw(tempfile);
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $content = $cgi->param('content');
my $type = $cgi->param('type');
File::Temp->safe_level( File::Temp::HIGH );
my ( $outfh, $filename ) =
tempfile( "tmp/myfiles.XXXXXX", UNLINK => 0 );
if ( !defined $outfh ) {
die "error: no temporary filen";
}
print $outfh $content;
close $outfh;
my $result=`/home/esjolund/public_html/cgi-bin/count.py $type $filename`;
print "Content has $result $type";
print $cgi->end_html;
}
通常可以使用管道避免临时文件。在perl中，这是用open2()完成的
#!/usr/bin/perl
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use IPC::Open2;
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $content = $cgi->param('content');
my $type = $cgi->param('type');
my($child_out, $child_in);
$pid = open2($child_out, $child_in, "/home/esjolund/public_html/cgi-bin/count.py", $type,"/dev/stdin");
print $child_in $content;
close($child_in);
my $result=<$child_out>;
waitpid($pid,0);
print "Content has $result $type";
print $cgi->end_html;
}
上一个示例count3.cgi是不安全的，因为它将$type变量直接传递到命令行（shell）。相反，我们应该使用逗号分隔的open2()的语法，如下所示：
#!/usr/bin/perl
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use IPC::Open2;
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $content = $cgi->param('content');
my $type = $cgi->param('type');
my($child_out, $child_in);
$pid = open2($child_out, $child_in, "/home/esjolund/public_html/cgi-bin/count.py",$type);
print $child_in $content;
close($child_in);
my $result=<$child_out>;
waitpid($pid,0);
print "Content has $result $type";
print $cgi->end_html;
}
检查污点模式， -T标志
#!/usr/bin/perl -T
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use IPC::Open2;
use Scalar::Util qw(tainted);
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
if (tainted($cgi->param('content')))
{ print 'variable content tainted!'; }
else
{ print 'variable content not tainted!'; }
print `echo $cgi->param('content')`;
my $content = $cgi->param('content');
my $type = $cgi->param('type');
my($child_out, $child_in);
$pid = open2($child_out, $child_in, "/home/esjolund/public_html/cgi-bin/count.py",$type);
print $child_in $content;
close($child_in);
my $result=<$child_out>;
waitpid($pid,0);
print "Content has $result $type";
print $cgi->end_html;
}
如果我们尝试运行上一个示例Count5.cgi ，Perl将抱怨该PATH变量是不安全的。如果我们使用-T标志，则需要设置PATH环境变量。
#!/usr/bin/perl -T
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use IPC::Open2;
use Scalar::Util qw(tainted);
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
$ENV{'PATH'} = '/bin:/usr/bin';
delete @ENV{'IFS', 'CDPATH', 'ENV', 'BASH_ENV'};
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $content = $cgi->param('content');
my $type = $cgi->param('type');
if ( tainted($type) )
{ print 'tainted!'; }
else
{ print 'not tainted!'; }
print $cgi->hr;
print `echo $type`;
my($child_out, $child_in);
$pid = open2($child_out, $child_in, "/home/esjolund/public_html/cgi-bin/count.py",$type);
print $child_in $content;
close($child_in);
my $result=<$child_out>;
waitpid($pid,0);
print "Content has $result $type";
print $cgi->end_html;
}
如果我们尝试运行上一个示例count6.cgi perl会抱怨我们正在以不安全的方式使用未污染的变量。要使用CGI输入变量，我们需要通过使用正则表达式处理它们来取消它们
#!/usr/bin/perl -T
use CGI qw(:standard);
use CGI::Carp qw(fatalsToBrowser);
use IPC::Open2;
use Scalar::Util qw(tainted);
my $cgi = new CGI;
if ($cgi->cgi_error()) { die() };
$ENV{'PATH'} = '/bin:/usr/bin';
delete @ENV{'IFS', 'CDPATH', 'ENV', 'BASH_ENV'};
print
$cgi->header(),
$cgi->start_html( '' ),
$cgi->h1('Count'),
$cgi->start_form(-method=>'GET'),
"type:",
$cgi->popup_menu(-name=>'type',-values=>['lines','bytes']),
$cgi->p,
$cgi->textarea(-name=>'content', -rows=>10, -columns=>50),
$cgi->p,
$cgi->submit(),
$cgi->end_form,
$cgi->hr;
if ($cgi->param) {
my $type;
if ( $cgi->param('type')=~/^(bytes|lines)$/ ) {
$type = $1;
} else {
print "variable type needs to be bytes or lines";
exit(0);
}
my $content;
if ( $cgi->param('content')=~/^(.*)$/s ) {
$content = $1;
} else {
exit(0);
}
if ( tainted($type) )
{ print 'tainted!'; }
else
{ print 'not tainted!'; }
print $cgi->hr;
print `echo $type`;
my($child_out, $child_in);
$pid = open2($child_out, $child_in, "/home/esjolund/public_html/cgi-bin/count.py",$type);
print $child_in $content;
close($child_in);
my $result=<$child_out>;
waitpid($pid,0);
print "Content has $result $type";
print $cgi->end_html;
}
