From 2f41bca4e8ef591f5fa5c9ee620c7aacc9409f71 Mon Sep 17 00:00:00 2001
From: Guilhem Moulin <guilhem.moulin@ens-lyon.org>
Date: Fri, 2 Mar 2012 11:53:49 +0100
Subject: cleaner way to set restore the UTF-8 flag for the output of SELECT
 queries

---
 videoadd.pl | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)

(limited to 'videoadd.pl')

diff --git a/videoadd.pl b/videoadd.pl
index 1c8b0db..0d39eae 100755
--- a/videoadd.pl
+++ b/videoadd.pl
@@ -267,6 +267,7 @@ unless ( defined $ignoredb_flag ) {
     }
     $dbh = DBI->connect($dsn, $config{user}, $config{password})
              or die "Error: Can't connect do database.\n";
+    $dbh->do( "SET NAMES UTF8" ) or die "Error: Can't set names to UTF-8.\n";
 }
 
 my $id;
@@ -658,6 +659,15 @@ sub getfile {
 
     my $res = $dbh->selectall_hashref( $SELECT, 'id' )
         or die "Can't select: $!\n";
+
+    # The database is using UTF-8, but the UTF-8 flag may be missing;
+    # restore it.
+    foreach my $id (keys $res) {
+        foreach my $k (keys $res->{$id}) {
+            Encode::_utf8_on($res->{$id}->{$k}); 
+        }
+    }
+
     # We have a new file to add to the database
     if (scalar (keys $res) > 0) {
         my $answer;
@@ -672,7 +682,7 @@ 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", 'utf8')
+                print $res->{$id}->{title}. " (" .$res->{$id}->{year}. ")\n"
                     or die "Can't print: $!";
                 push @options, sprintf ("%04d", $res->{$id}->{id});
             }
@@ -798,22 +808,13 @@ sub YNquestion {
 }
 
 # Convert to UTF-8
-# (Strings from the database are in UTF-8, but are lacking the UTF-8 flag)
+# The input string should be in latin1 unless the UTF-8 flag is on.
 sub iconv {
-    my ($string, $enc) = @_;
+    my $string = $_[0];
     return unless defined $string;
 
     unless (Encode::is_utf8($string)) {
-        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;
+        $string = Encode::decode( 'latin1', $string, 1 );
     }
 
     # The UTF-8 flag should be on now
-- 
cgit v1.2.3