summaryrefslogtreecommitdiff
path: root/pandoc-iri
diff options
context:
space:
mode:
Diffstat (limited to 'pandoc-iri')
-rwxr-xr-xpandoc-iri23
1 files changed, 21 insertions, 2 deletions
diff --git a/pandoc-iri b/pandoc-iri
index 0f3fb0c..9774532 100755
--- a/pandoc-iri
+++ b/pandoc-iri
@@ -1,5 +1,8 @@
#!/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;
@@ -9,6 +12,9 @@ use Pandoc::Elements;
use URI;
use URI::Escape;
use URI::Find;
+use HTTP::Tiny 0.014;
+
+my ($ua, %links);
pandoc_filter(
\&link_normalize,
@@ -26,12 +32,25 @@ sub link_normalize {
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 URI->new(uri_unescape(shift))->canonical->as_string;
+ return normalize_uri(shift)->as_string;
}
sub pp_iri {
- return URI->new(uri_unescape(shift))->canonical->as_iri;
+ return normalize_uri(shift)->as_iri;
}
sub pp_string {