From 8aac36c6e2d16c7122eb1b18febba4b5b9374323 Mon Sep 17 00:00:00 2001
From: Guilhem Moulin <guilhem.moulin@ens-lyon.org>
Date: Fri, 2 Mar 2012 09:23:22 +0100
Subject: strings from DBI are lacking the UTF-8 flag

---
 videoadd.pl | 18 ++++++++++++++----
 1 file changed, 14 insertions(+), 4 deletions(-)

diff --git a/videoadd.pl b/videoadd.pl
index 765ed34..1c8b0db 100755
--- a/videoadd.pl
+++ b/videoadd.pl
@@ -641,7 +641,7 @@ sub mkcast {
 
 # Try to find a matching entry in the database.
 sub getfile {
-    my $crit = $_[0];
+    my $crit = &iconv($_[0]);
     return $crit if defined $ignoredb_flag;
     
     my $SELECT = "SELECT id, imdbid, filename, title, year FROM $config{prefix}videodata
@@ -672,7 +672,7 @@ sub getfile {
             my @options;
             foreach my $id (keys $res) {
                 printf "%04d - ", $res->{$id}->{id} or die "Can't printf: $!";
-                print &iconv($res->{$id}->{title}. " (" .$res->{$id}->{year}. ")\n")
+                print &iconv($res->{$id}->{title}. " (" .$res->{$id}->{year}. ")\n", 'utf8')
                     or die "Can't print: $!";
                 push @options, sprintf ("%04d", $res->{$id}->{id});
             }
@@ -798,12 +798,22 @@ sub YNquestion {
 }
 
 # Convert to UTF-8
+# (Strings from the database are in UTF-8, but are lacking the UTF-8 flag)
 sub iconv {
-    my $string = $_[0];
+    my ($string, $enc) = @_;
     return unless defined $string;
 
     unless (Encode::is_utf8($string)) {
-        $string = Encode::decode( 'latin1', my $copy = $string, 1 );
+        my @encodings = qw/latin1 utf8/;
+        @encodings = $enc if defined $enc;
+
+        my $decoded;
+        foreach my $enc (@encodings) {
+            $decoded = Encode::decode( $enc, $string);
+            next if $@ or not $decoded;
+            last if utf8::valid($decoded);
+        }
+        $string = $decoded;
     }
 
     # The UTF-8 flag should be on now
-- 
cgit v1.2.3