commit d1580a62edf2a5b18b3775ff6580b563025d9831
author Mike Ryan <mikeryan@lacklustre.net> Thu Aug 11 13:19:45 2016 -0700
committer Marcel Holtmann <marcel@holtmann.org> Thu Sep 15 22:05:15 2016 +0200
tree 4b5b113c36503f8913a31e9d988ddcbd2a4801c6
parent 3aa056ca200dddc725d9e8ca5666ba229025d5aa

tools: add more entities and better error handling to parse_companies

Add the remaining lowercase acute accented vowel HTML entities to
parse_companies.pl. On unknown entity, print an error to STDERR so the
maintainer can more clearly understand the failure.

tools/parse_companies.pl

diff --git a/tools/parse_companies.pl b/tools/parse_companies.pl
index 99d3f31..d5b2815 100755
--- a/tools/parse_companies.pl
+++ b/tools/parse_companies.pl
@@ -8,7 +8,11 @@
 
 my %known_entities = (
     'nbsp' => ' ',
+    'aacute' => 'á',
     'eacute' => 'é',
+    'iacute' => 'í',
+    'oacute' => 'ó',
+    'uacute' => 'ú',
     'auml' => 'ä',
     'uuml' => 'ü',
     'Uuml' => 'Ü',
@@ -23,8 +27,7 @@
     }
     foreach my $entity (map { lc $_ } $name =~ /&([^;]+);/g) {
         if ($entity ne 'amp') {
-            print "Unable to convert &$entity;, giving up\n";
-            exit 1;
+            die "\nparse_companies.pl: Unable to convert &$entity; giving up\n";
         }
     }
     $name =~ s/&/&/ig;