From 6f98b4dfb2b80e3eb3072dd3f140351581746db2 Mon Sep 17 00:00:00 2001 From: Jonas Smedegaard Date: Tue, 30 Dec 2014 01:50:08 +0100 Subject: Move mediawiki tweaks to separate scripts. --- mediawiki-uri-escape | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) create mode 100755 mediawiki-uri-escape (limited to 'mediawiki-uri-escape') diff --git a/mediawiki-uri-escape b/mediawiki-uri-escape new file mode 100755 index 0000000..40dd10c --- /dev/null +++ b/mediawiki-uri-escape @@ -0,0 +1,24 @@ +#!/usr/bin/perl + +# work around modest URL parsing + +use warnings; +use strict; + +use Path::Tiny; +use Regexp::Common qw(net); +use URI::Escape; + +my $infile = shift; +my $outfile = shift || $infile; + +$_ = path($infile)->slurp_utf8; + +my $uri_path = qr!https?://$RE{net}{domain}{-nospace}(?::\d+)?\K(/\S*?)!; +my $uri_end = qr!(?=[.,;]?[\s<\"\|}])!; + +s|\b$uri_path$uri_end|uri_escape(uri_unescape($1), "^A-Za-z0-9/\.")|eg; + +path($outfile)->spew_utf8($_); + +1; -- cgit v1.2.3