File Coverage

blib/lib/WWW/Crawl4AI/Strategy/CloakBrowser.pm

Criterion	Covered	Total	%
statement	24	27	88.8
branch	4	4	100.0
condition	1	2	50.0
subroutine	7	9	77.7
pod	3	4	75.0
total	39	46	84.7

package WWW::Crawl4AI::Strategy::CloakBrowser;
# ABSTRACT: Crawl4AI strategy attaching to CloakBrowser over CDP
use Moo;
use URI ();
use WWW::Crawl4AI::Request ();
with 'WWW::Crawl4AI::Strategy';

our $VERSION = '0.001';


sub name       { 'crawl4ai_cloakbrowser' }


sub cost_class { 'stealth' }


sub applicable {
  my ( $self, $crawler ) = @_;
  return $crawler->cloakbrowser_url ? 1 : 0;
}


sub build_request {
  my ( $self, $crawler, $url ) = @_;
  return $self->_request(
    $url,
    browser => {
      browser_mode            => 'custom',
      cdp_url                 => $self->_cdp_url( $crawler, $url ),
      cache_cdp_connection    => WWW::Crawl4AI::Request::JSON_true(),
      create_isolated_context => WWW::Crawl4AI::Request::JSON_true(),
    },
    crawler => { wait_until => 'networkidle' },
  );
}

# Per-domain stable fingerprint: if the configured CDP URL has no query string,
# append ?fingerprint= so each domain gets a consistent CloakBrowser
# identity. CloakBrowser requires the seed to be a NON-NEGATIVE INTEGER — a
# non-numeric value (e.g. a raw host string) is rejected with HTTP 400. So the
# host is folded into a deterministic 32-bit hash: same domain → same seed,
# different domains → different seeds. A URL that already carries query params
# is used verbatim.
sub _cdp_url {
  my ( $self, $crawler, $url ) = @_;
  my $cdp = $crawler->cloakbrowser_url;
  return $cdp if $cdp =~ /\?/;
  my $host = eval { URI->new($url)->host } || 'default';
  ( my $base = $cdp ) =~ s{/+$}{};
  return "$base?fingerprint=" . $self->_fingerprint_seed($host);
}

# Deterministic 32-bit FNV-1a hash of the host. Zero-dependency, stable across
# runs and processes, yields a non-negative integer CloakBrowser accepts.
sub _fingerprint_seed {
  my ( $self, $host ) = @_;
  my $hash = 2166136261;                       # FNV-1a 32-bit offset basis
  for my $byte ( unpack 'C*', $host ) {
    $hash ^= $byte;
    $hash = ( $hash * 16777619 ) & 0xFFFFFFFF;  # FNV prime, wrap to 32 bits
  }
  return $hash;
}

1;

__END__