pandoc-iri



#!/usr/bin/perl

# canonicalize URIs links, beautify their text representations
# set NETWORK_TESTS=1 to also validate and report failures to STDERR

use warnings;
use strict;

use Pandoc::Filter 0.05;
use Pandoc::Elements;

use URI;
use URI::Escape;
use URI::Find;
use HTTP::Tiny 0.014;

my ($ua, %links);

pandoc_filter(
    \&link_normalize,
);

sub link_normalize {
    my $self = shift;
    return [ Link(
        $self->content,
        [ pp_uri($self->target->[0]) => '' ]
    )]
        if ( $self->name eq 'Link' );
    return [ Str pp_string($self->content) ]
        if ( $self->name eq 'Str' );
    return;
}

sub normalize_uri {
    my $uri = URI->new(uri_unescape(shift))->canonical;
    return $uri unless $ENV{'NETWORK_TESTS'};
    return $uri if ( defined $links{$uri} );
    $ua = HTTP::Tiny->new()
        unless ( defined $ua );
    $links{$uri} = $ua->head($uri);
    printf STDERR "Failed fetching <%s>: %u %s\n",
        $uri, $links{$uri}->{status}, $links{$uri}->{reason}
        unless ($links{$uri}->{success});
    return $uri;
}

sub pp_uri {
    return normalize_uri(shift)->as_string;
}

sub pp_iri {
    return normalize_uri(shift)->as_iri;
}

sub pp_string {
    my $string = shift;
    my $finder = URI::Find->new( \&pp_iri );
    $finder->find(\$string);
    return $string;
}