Optionally validate URLs, enabled in new flightcheck mode.

2 files changed, 23 insertions, 2 deletions

diff --git a/Makefile b/Makefile
index 505254c..25079fe 100644
--- a/Makefile
+++ b/Makefile
@@ -1,3 +1,4 @@
+# expensive checks enabled: make -B flightcheck=1
 # Final document: make -B final=1
 
 stem = eut
@@ -17,6 +18,7 @@ localfilters = $(filter ./%,$(filters))
 title = Ensuring utmost transparency
 subtitle = Free Software and Open Standards under the Rules of Procedure of the European Parliament
 env_filter += $(citeproc_file:%=PANDOC_CITEPROC_FILE=%)
+env_filter += $(if $(flightcheck),NETWORK_TESTS=1)
 args_filter += $(filters:%=--filter %)
 args_meta += $(citeproc_file:%=-M bibliography=%)
 args_meta += -V lang=english -V langoption=variant=british
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 {