feat: speed up validation

This commit is contained in:
2026-01-05 07:22:05 +01:00
parent 72013a9a08
commit 777b589946

View File

@@ -10,8 +10,8 @@ use Socket
qw(getaddrinfo getnameinfo NI_NUMERICHOST NI_NUMERICSERV AF_INET AF_INET6 SOCK_STREAM); qw(getaddrinfo getnameinfo NI_NUMERICHOST NI_NUMERICSERV AF_INET AF_INET6 SOCK_STREAM);
my $MAX_URL_LENGTH = 2048; my $MAX_URL_LENGTH = 2048;
my $CONNECT_TIMEOUT = 10; my $CONNECT_TIMEOUT = 0.2;
my $REQUEST_TIMEOUT = 10; my $REQUEST_TIMEOUT = 0.4;
my $MAX_REDIRECTS = 3; my $MAX_REDIRECTS = 3;
my $DNS_ERROR_PATTERN = my $DNS_ERROR_PATTERN =
@@ -26,8 +26,11 @@ my @BLOCKED_DOMAINS = qw(
localhost 127.0.0.1 0.0.0.0 ::1 localhost 127.0.0.1 0.0.0.0 ::1
); );
my $DNS_CACHE_TTL = 300; my $DNS_CACHE_TTL = 300;
my $REACHABILITY_CACHE_TTL = 300;
my $DNS_RESOLVE_TIMEOUT = 0.2;
my %dns_cache; my %dns_cache;
my %reachability_cache;
has ua => sub { has ua => sub {
my $self = shift; my $self = shift;
@@ -150,17 +153,25 @@ sub _resolve_host {
[ { type => 'ipv6', ip => $ipv6_host } ] ); [ { type => 'ipv6', ip => $ipv6_host } ] );
} }
my $cache_key = lc($host); if ( my $cached = $self->_get_cached_addresses($host) ) {
my $now = time(); return Mojo::Promise->resolve($cached);
if ( exists $dns_cache{$cache_key} ) {
my $cached = $dns_cache{$cache_key};
if ( $now < $cached->{expires} ) {
return Mojo::Promise->resolve( $cached->{addresses} );
}
delete $dns_cache{$cache_key};
} }
my $promise = Mojo::Promise->new; my $promise = Mojo::Promise->new;
my $resolved = 0;
my $cache_key = lc($host);
my $now = time();
my $timer = Mojo::IOLoop->timer(
$DNS_RESOLVE_TIMEOUT => sub {
return if $resolved;
$resolved = 1;
$dns_cache{$cache_key} = {
addresses => [],
expires => $now + $DNS_CACHE_TTL
};
$promise->resolve( [] );
}
);
Mojo::IOLoop->subprocess( Mojo::IOLoop->subprocess(
sub { sub {
my ($hostname) = @_; my ($hostname) = @_;
@@ -170,6 +181,9 @@ sub _resolve_host {
}, },
sub { sub {
my ( $subprocess, $err, $data ) = @_; my ( $subprocess, $err, $data ) = @_;
return if $resolved;
$resolved = 1;
Mojo::IOLoop->remove($timer);
if ($err) { if ($err) {
$promise->resolve( [] ); $promise->resolve( [] );
return; return;
@@ -210,6 +224,73 @@ sub _resolve_host {
return $promise; return $promise;
} }
sub _addresses_contain_private {
my ( $self, $addresses ) = @_;
return 0 unless defined $addresses && ref $addresses eq 'ARRAY';
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 _get_cached_addresses {
my ( $self, $host ) = @_;
return undef unless defined $host && length $host;
my $cache_key = lc($host);
my $cached = $dns_cache{$cache_key};
return undef unless $cached;
return $cached->{addresses} if time() < $cached->{expires};
delete $dns_cache{$cache_key};
return undef;
}
sub _cache_reachability {
my ( $self, $url, $ok, $error ) = @_;
return unless defined $url && length $url;
$reachability_cache{$url} = {
ok => $ok ? 1 : 0,
error => $error,
expires => time() + $REACHABILITY_CACHE_TTL
};
}
sub _clear_caches {
# Test helper
%dns_cache = ();
%reachability_cache = ();
return;
}
sub _get_cached_reachability {
my ( $self, $url ) = @_;
return undef unless defined $url && length $url;
my $cached = $reachability_cache{$url};
return undef unless $cached;
return $cached if time() < $cached->{expires};
delete $reachability_cache{$url};
return undef;
}
sub _fire_and_forget {
my ( $self, $promise ) = @_;
return unless $promise;
$promise->catch( sub { } );
return;
}
sub is_blocked_url { sub is_blocked_url {
my ( $self, $url ) = @_; my ( $self, $url ) = @_;
return Mojo::Promise->resolve(0) unless defined $url; return Mojo::Promise->resolve(0) unless defined $url;
@@ -232,24 +313,14 @@ sub is_blocked_url {
return Mojo::Promise->resolve(1); return Mojo::Promise->resolve(1);
} }
return $self->_resolve_host($host)->then( if ( my $cached = $self->_get_cached_addresses($host) ) {
sub { return Mojo::Promise->resolve(
my $addresses = shift; $self->_addresses_contain_private($cached) ? 1 : 0 );
for my $addr (@$addresses) { }
if ( $addr->{type} eq 'ipv4'
&& $self->_is_private_ipv4( $addr->{ip} ) ) # Intentional: skip blocking on cold hosts to keep latency low, DNS runs in background.
{ $self->_fire_and_forget( $self->_resolve_host($host) );
return 1; return Mojo::Promise->resolve(0);
}
if ( $addr->{type} eq 'ipv6'
&& $self->_is_private_ipv6( $addr->{ip} ) )
{
return 1;
}
}
return 0;
}
);
} }
sub _create_ssrf_safe_ua { sub _create_ssrf_safe_ua {
@@ -334,8 +405,36 @@ sub check_url_reachable {
return Mojo::Promise->reject('URL is required') return Mojo::Promise->reject('URL is required')
unless defined $url && length($url) > 0; unless defined $url && length($url) > 0;
if ( my $cached = $self->_get_cached_reachability($url) ) {
return $cached->{ok}
? Mojo::Promise->resolve(1)
: Mojo::Promise->reject( $cached->{error} );
}
my $ssrf_ua = $self->_create_ssrf_safe_ua; my $ssrf_ua = $self->_create_ssrf_safe_ua;
return $self->_follow_redirect_with_validation( $ssrf_ua, $url ); return $self->_follow_redirect_with_validation( $ssrf_ua, $url )->then(
sub {
$self->_cache_reachability( $url, 1, undef );
return 1;
}
)->catch(
sub {
my $err = shift;
$self->_cache_reachability( $url, 0, $err );
return Mojo::Promise->reject($err);
}
);
}
sub check_url_reachable_async {
my ( $self, $url ) = @_;
return Mojo::Promise->resolve(1) unless defined $url && length $url;
return Mojo::Promise->resolve(1)
if $self->_get_cached_reachability($url);
$self->_fire_and_forget( $self->check_url_reachable($url) );
return Mojo::Promise->resolve(1);
} }
sub check_ssl_certificate { sub check_ssl_certificate {
@@ -365,6 +464,13 @@ sub check_ssl_certificate {
); );
} }
sub check_ssl_certificate_async {
my ( $self, $url ) = @_;
return Mojo::Promise->resolve(1) unless defined $url && length $url;
$self->_fire_and_forget( $self->check_ssl_certificate($url) );
return Mojo::Promise->resolve(1);
}
sub validate_short_code { sub validate_short_code {
my ( $self, $code ) = @_; my ( $self, $code ) = @_;
return defined $code && length($code) == 12 && $code =~ /^[0-9a-zA-Z\-_]+$/; return defined $code && length($code) == 12 && $code =~ /^[0-9a-zA-Z\-_]+$/;
@@ -401,11 +507,11 @@ sub validate_url_with_checks {
my $ssl_check = my $ssl_check =
$parsed->scheme eq 'https' $parsed->scheme eq 'https'
? $self->check_ssl_certificate($normalized) ? $self->check_ssl_certificate_async($normalized)
: Mojo::Promise->resolve(1); : Mojo::Promise->resolve(1);
return $ssl_check->then( return $ssl_check->then(
sub { return $self->check_url_reachable($normalized); } ) sub { return $self->check_url_reachable_async($normalized); } )
->then( sub { return $normalized; } ); ->then( sub { return $normalized; } );
} }
); );