383 lines
11 KiB
Perl
383 lines
11 KiB
Perl
package Urupam::Validation;
|
|
|
|
use Mojo::Base -base;
|
|
use Mojo::URL;
|
|
use Mojo::UserAgent;
|
|
use Mojo::Promise;
|
|
use Urupam::Utils qw(sanitize_url);
|
|
use Socket
|
|
qw(getaddrinfo getnameinfo NI_NUMERICHOST NI_NUMERICSERV AF_INET AF_INET6 SOCK_STREAM);
|
|
|
|
my $MAX_URL_LENGTH = 2048;
|
|
my $CONNECT_TIMEOUT = 10;
|
|
my $REQUEST_TIMEOUT = 10;
|
|
my $MAX_REDIRECTS = 3;
|
|
|
|
my $DNS_ERROR_PATTERN =
|
|
qr/Name or service not known|getaddrinfo failed|Could not resolve|DNS|hostname|Name resolution|nodename nor servname provided/i;
|
|
my $SSL_ERROR_PATTERN = qr/SSL|certificate|TLS|verification failed/i;
|
|
my $CONNECTION_ERROR_PATTERN = qr/Connection refused|Can't connect|timeout/i;
|
|
|
|
my $PRIVATE_IP_PATTERN =
|
|
qr/^(127\.|192\.168\.|10\.|172\.(1[6-9]|2[0-9]|3[01])\.)/;
|
|
|
|
my @BLOCKED_DOMAINS = qw(
|
|
localhost 127.0.0.1 0.0.0.0 ::1
|
|
);
|
|
|
|
has ua => sub {
|
|
my $self = shift;
|
|
Mojo::UserAgent->new(
|
|
connect_timeout => $self->connect_timeout,
|
|
request_timeout => $self->request_timeout,
|
|
max_redirects => $self->max_redirects,
|
|
insecure => 0
|
|
);
|
|
};
|
|
|
|
has connect_timeout => sub { $CONNECT_TIMEOUT };
|
|
has request_timeout => sub { $REQUEST_TIMEOUT };
|
|
has max_redirects => sub { $MAX_REDIRECTS };
|
|
|
|
sub _parse_url {
|
|
my ( $self, $url ) = @_;
|
|
return undef unless defined $url;
|
|
my $parsed;
|
|
eval { $parsed = Mojo::URL->new($url); 1 } or return undef;
|
|
return $parsed;
|
|
}
|
|
|
|
sub _classify_error {
|
|
my ( $self, $err_str ) = @_;
|
|
|
|
return 'ssl' if $err_str =~ $SSL_ERROR_PATTERN;
|
|
return 'dns' if $err_str =~ $DNS_ERROR_PATTERN;
|
|
return 'connection' if $err_str =~ $CONNECTION_ERROR_PATTERN;
|
|
return 'unknown';
|
|
}
|
|
|
|
sub _format_error_message {
|
|
my ( $self, $error_type, $err_str ) = @_;
|
|
|
|
return "SSL certificate error: $err_str" if $error_type eq 'ssl';
|
|
return "DNS resolution failed: $err_str" if $error_type eq 'dns';
|
|
return "Cannot reach URL: $err_str" if $error_type eq 'connection';
|
|
return "URL validation failed: $err_str";
|
|
}
|
|
|
|
sub is_valid_url_length {
|
|
my ( $self, $url ) = @_;
|
|
return 0 unless defined $url && length($url) > 0;
|
|
return length($url) <= $MAX_URL_LENGTH;
|
|
}
|
|
|
|
sub _is_valid_ipv4 {
|
|
my ( $self, $ip ) = @_;
|
|
return 0 unless defined $ip;
|
|
return 0 unless $ip =~ /^(\d+)\.(\d+)\.(\d+)\.(\d+)$/;
|
|
my ( $a, $b, $c, $d ) = ( $1, $2, $3, $4 );
|
|
return
|
|
$a >= 0
|
|
&& $a <= 255
|
|
&& $b >= 0
|
|
&& $b <= 255
|
|
&& $c >= 0
|
|
&& $c <= 255
|
|
&& $d >= 0
|
|
&& $d <= 255;
|
|
}
|
|
|
|
sub _is_private_ipv4 {
|
|
my ( $self, $ip ) = @_;
|
|
return 0 unless $self->_is_valid_ipv4($ip);
|
|
return $ip =~ $PRIVATE_IP_PATTERN;
|
|
}
|
|
|
|
sub _is_private_ipv6 {
|
|
my ( $self, $ip ) = @_;
|
|
return 0 unless defined $ip;
|
|
|
|
$ip = lc($ip);
|
|
$ip =~ s/^\[|\]$//g;
|
|
|
|
return 1 if $ip eq '::1';
|
|
return 1 if $ip eq '::';
|
|
|
|
if ( $ip =~ /^::ffff:(.+)$/ ) {
|
|
return $self->_is_private_ipv4($1) ? 1 : 0;
|
|
}
|
|
|
|
return 0 unless $ip =~ /^([0-9a-f]{0,4}:)+[0-9a-f]{0,4}$/ || $ip =~ /^::/;
|
|
|
|
my @parts = split /:/, $ip;
|
|
return 0 unless @parts > 0;
|
|
|
|
my $first_part = $parts[0] || '';
|
|
return 0 unless length($first_part) > 0;
|
|
|
|
my $first = hex($first_part);
|
|
|
|
if ( $first >= 0xfc && $first <= 0xfd ) {
|
|
return 1;
|
|
}
|
|
|
|
if ( $first == 0xfe && @parts > 1 ) {
|
|
my $second_part = $parts[1] || '';
|
|
if ( length($second_part) > 0 ) {
|
|
my $second = hex($second_part);
|
|
if ( $second >= 0x80 && $second <= 0xbf ) {
|
|
return 1;
|
|
}
|
|
}
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
sub _resolve_host {
|
|
my ( $self, $host ) = @_;
|
|
return Mojo::Promise->resolve( [] )
|
|
unless defined $host && length($host) > 0;
|
|
|
|
if ( $self->_is_valid_ipv4($host) ) {
|
|
return Mojo::Promise->resolve( [ { type => 'ipv4', ip => $host } ] );
|
|
}
|
|
|
|
my $ipv6_host = $host;
|
|
$ipv6_host =~ s/^\[|\]$//g;
|
|
if ( $self->_is_private_ipv6($ipv6_host) ) {
|
|
return Mojo::Promise->resolve(
|
|
[ { type => 'ipv6', ip => $ipv6_host } ] );
|
|
}
|
|
|
|
if ( $ipv6_host =~ /^([0-9a-f]{0,4}:){2,}[0-9a-f]{0,4}$/i
|
|
|| $ipv6_host =~ /^::/ )
|
|
{
|
|
return Mojo::Promise->resolve(
|
|
[ { type => 'ipv6', ip => $ipv6_host } ] );
|
|
}
|
|
|
|
my ( $err, @results ) =
|
|
getaddrinfo( $host, undef, { socktype => SOCK_STREAM } );
|
|
return Mojo::Promise->resolve( [] ) if $err;
|
|
|
|
my @addresses;
|
|
for my $res (@results) {
|
|
my ( $hostnum, undef ) =
|
|
getnameinfo( $res->{addr}, NI_NUMERICHOST | NI_NUMERICSERV );
|
|
next unless defined $hostnum && length $hostnum;
|
|
|
|
if ( $res->{family} == AF_INET ) {
|
|
push @addresses, { type => 'ipv4', ip => $hostnum };
|
|
}
|
|
elsif ( $res->{family} == AF_INET6 ) {
|
|
push @addresses, { type => 'ipv6', ip => $hostnum };
|
|
}
|
|
}
|
|
|
|
return Mojo::Promise->resolve( \@addresses );
|
|
}
|
|
|
|
sub is_blocked_url {
|
|
my ( $self, $url ) = @_;
|
|
return Mojo::Promise->resolve(0) unless defined $url;
|
|
|
|
my $parsed = $self->_parse_url($url);
|
|
return Mojo::Promise->resolve(0) unless $parsed;
|
|
|
|
my $host = lc( $parsed->host // '' );
|
|
return Mojo::Promise->resolve(0) unless length($host) > 0;
|
|
|
|
for my $blocked (@BLOCKED_DOMAINS) {
|
|
return Mojo::Promise->resolve(1) if $host eq $blocked;
|
|
}
|
|
|
|
if ( $self->_is_private_ipv4($host) ) {
|
|
return Mojo::Promise->resolve(1);
|
|
}
|
|
|
|
if ( $self->_is_private_ipv6($host) ) {
|
|
return Mojo::Promise->resolve(1);
|
|
}
|
|
|
|
return $self->_resolve_host($host)->then(
|
|
sub {
|
|
my $addresses = shift;
|
|
for my $addr (@$addresses) {
|
|
if ( $addr->{type} eq 'ipv4'
|
|
&& $self->_is_private_ipv4( $addr->{ip} ) )
|
|
{
|
|
return 1;
|
|
}
|
|
if ( $addr->{type} eq 'ipv6'
|
|
&& $self->_is_private_ipv6( $addr->{ip} ) )
|
|
{
|
|
return 1;
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
);
|
|
}
|
|
|
|
sub _create_ssrf_safe_ua {
|
|
my $self = shift;
|
|
return Mojo::UserAgent->new(
|
|
connect_timeout => $self->connect_timeout,
|
|
request_timeout => $self->request_timeout,
|
|
max_redirects => 0,
|
|
insecure => 0
|
|
);
|
|
}
|
|
|
|
sub _follow_redirect_with_validation {
|
|
my ( $self, $ua, $url, $redirect_count ) = @_;
|
|
$redirect_count //= 0;
|
|
|
|
return Mojo::Promise->reject('Too many redirects')
|
|
if $redirect_count > $self->max_redirects;
|
|
|
|
return $ua->head_p($url)->then(
|
|
sub {
|
|
my $tx = shift;
|
|
my $code = $tx->result->code;
|
|
|
|
if ( $code >= 300 && $code < 400 ) {
|
|
my $location = $tx->result->headers->location;
|
|
return Mojo::Promise->reject('Redirect without Location header')
|
|
unless defined $location;
|
|
|
|
my $redirect_url = Mojo::URL->new($location)->to_abs($url);
|
|
return $self->is_blocked_url($redirect_url)->then(
|
|
sub {
|
|
my $blocked = shift;
|
|
if ($blocked) {
|
|
return Mojo::Promise->reject(
|
|
'Redirect to blocked domain or local address');
|
|
}
|
|
return $self->_follow_redirect_with_validation( $ua,
|
|
$redirect_url, $redirect_count + 1 );
|
|
}
|
|
);
|
|
}
|
|
|
|
return 1 if $code >= 200 && $code < 400;
|
|
if ( $code == 403 || $code == 404 || $code == 405 ) {
|
|
return $ua->get_p($url)->then(
|
|
sub {
|
|
my $get_tx = shift;
|
|
my $get_code = $get_tx->result->code;
|
|
return 1 if $get_code >= 200 && $get_code < 400;
|
|
return 1
|
|
if $get_code == 403
|
|
|| $get_code == 404
|
|
|| $get_code == 405;
|
|
return Mojo::Promise->reject(
|
|
"URL returned $get_code error")
|
|
if $get_code >= 400;
|
|
return Mojo::Promise->reject(
|
|
"URL returned unexpected status: $get_code");
|
|
}
|
|
);
|
|
}
|
|
return Mojo::Promise->reject("URL returned $code error")
|
|
if $code >= 400;
|
|
return Mojo::Promise->reject(
|
|
"URL returned unexpected status: $code");
|
|
}
|
|
)->catch(
|
|
sub {
|
|
my $err = shift;
|
|
my $err_str = "$err";
|
|
my $error_type = $self->_classify_error($err_str);
|
|
return Mojo::Promise->reject(
|
|
$self->_format_error_message( $error_type, $err_str ) );
|
|
}
|
|
);
|
|
}
|
|
|
|
sub check_url_reachable {
|
|
my ( $self, $url ) = @_;
|
|
|
|
return Mojo::Promise->reject('URL is required')
|
|
unless defined $url && length($url) > 0;
|
|
|
|
my $ssrf_ua = $self->_create_ssrf_safe_ua;
|
|
return $self->_follow_redirect_with_validation( $ssrf_ua, $url );
|
|
}
|
|
|
|
sub check_ssl_certificate {
|
|
my ( $self, $url ) = @_;
|
|
|
|
return Mojo::Promise->reject('URL is required')
|
|
unless defined $url && length($url) > 0;
|
|
|
|
my $parsed = $self->_parse_url($url);
|
|
return Mojo::Promise->resolve(1)
|
|
unless $parsed && $parsed->scheme && $parsed->scheme eq 'https';
|
|
|
|
return $self->ua->head_p($url)->then( sub { return 1; } )->catch(
|
|
sub {
|
|
my $err = shift;
|
|
my $err_str = "$err";
|
|
my $error_type = $self->_classify_error($err_str);
|
|
|
|
if ( $error_type eq 'ssl' ) {
|
|
return Mojo::Promise->reject(
|
|
"Invalid SSL certificate: $err_str");
|
|
}
|
|
|
|
return Mojo::Promise->reject(
|
|
$self->_format_error_message( $error_type, $err_str ) );
|
|
}
|
|
);
|
|
}
|
|
|
|
sub validate_short_code {
|
|
my ( $self, $code ) = @_;
|
|
return defined $code && length($code) == 12 && $code =~ /^[0-9a-zA-Z\-_]+$/;
|
|
}
|
|
|
|
sub validate_url_with_checks {
|
|
my ( $self, $url ) = @_;
|
|
|
|
return Mojo::Promise->reject('URL is required')
|
|
unless defined $url && length($url) > 0;
|
|
|
|
my $sanitized = sanitize_url($url);
|
|
return Mojo::Promise->reject('Invalid URL format')
|
|
unless defined $sanitized;
|
|
|
|
my $parsed = $self->_parse_url($sanitized);
|
|
return Mojo::Promise->reject('Invalid URL format') unless $parsed;
|
|
return Mojo::Promise->reject('Invalid URL format')
|
|
unless $parsed->scheme && $parsed->scheme =~ /^https?$/i;
|
|
return Mojo::Promise->reject('Invalid URL format') unless $parsed->host;
|
|
|
|
my $normalized = $parsed->to_string;
|
|
|
|
return Mojo::Promise->reject(
|
|
"URL exceeds maximum length of $MAX_URL_LENGTH characters")
|
|
unless $self->is_valid_url_length($normalized);
|
|
|
|
return $self->is_blocked_url($normalized)->then(
|
|
sub {
|
|
my $blocked = shift;
|
|
return Mojo::Promise->reject(
|
|
'This URL cannot be shortened (blocked domain or local address)'
|
|
) if $blocked;
|
|
|
|
my $ssl_check =
|
|
$parsed->scheme eq 'https'
|
|
? $self->check_ssl_certificate($normalized)
|
|
: Mojo::Promise->resolve(1);
|
|
|
|
return $ssl_check->then(
|
|
sub { return $self->check_url_reachable($normalized); } )
|
|
->then( sub { return $normalized; } );
|
|
}
|
|
);
|
|
}
|
|
|
|
1;
|