File Coverage

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

Criterion	Covered	Total	%
statement	17	17	100.0
branch	5	6	83.3
condition	2	6	33.3
subroutine	6	6	100.0
pod	4	4	100.0
total	34	39	87.1

package WWW::Crawl4AI::Strategy::Callback;
# ABSTRACT: last-resort Crawl4AI strategy delegating to a user coderef
use Moo;
with 'WWW::Crawl4AI::Strategy';
use WWW::Crawl4AI::Markdown qw( resolve_markdown_chain );

our $VERSION = '0.001';


sub name       { 'external_callback' }


sub cost_class { 'paid' }


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


# Does not go through Crawl4AI at all — hand the URL to the user's coderef and
# normalize whatever page-shaped hashref it returns. We accept markdown either
# as a plain string or as Crawl4AI's structured object, so Detect and Result
# see the same shape they get from every (Crawl4AI-backed) strategy.
sub crawl {
  my ( $self, $crawler, $url, %opts ) = @_;
  my $page = $crawler->callback->( $url, %opts );
  return undef unless ref $page eq 'HASH';
  $page->{markdown}  = resolve_markdown_chain( $page->{markdown} ) if ref $page->{markdown} eq 'HASH';
  $page->{url}       //= $url;
  $page->{final_url} //= $page->{url};
  return $page;
}


1;

__END__