From 7944c09286c112f5d2880c4c4d3fdf9a4a86aa1a Mon Sep 17 00:00:00 2001
From: Guilhem Moulin <guilhem.moulin@ens-lyon.org>
Date: Wed, 7 Mar 2012 01:10:19 +0100
Subject: unicode, re

---
 videorm.pl | 28 ++++++++++++++++++++++++----
 1 file changed, 24 insertions(+), 4 deletions(-)

(limited to 'videorm.pl')

diff --git a/videorm.pl b/videorm.pl
index bbdc5c2..3f58d4d 100755
--- a/videorm.pl
+++ b/videorm.pl
@@ -1,4 +1,4 @@
-#!/usr/bin/perl -CAL
+#!/usr/bin/perl -CADS
 
 # This program is free software. It comes without any warranty, to the
 # extent permitted by applicable law. You can redistribute it and/or
@@ -11,6 +11,8 @@ $VERSION = "0.1, 09 January 2012";
 
 use warnings;
 use strict;
+use utf8;
+use feature "unicode_strings";
 
 use Getopt::Long qw/:config posix_default no_ignore_case gnu_compat
                             bundling auto_version auto_help/;
@@ -19,7 +21,6 @@ use DBI;
 use File::Basename;
 use File::Spec::Functions qw /catfile catdir splitdir updir/;
 use File::Copy;
-use Cwd qw /realpath/;
 use Env qw /HOME/;
 
 ################################################################################
@@ -53,7 +54,8 @@ B<videorm.pl> [B<-q>] [B<-f>] I<path/to/file>
 
 =head1 DISCLAIMER
 
-Your collection is assumed to have the following structure: two
+Your collection is assumed to be encoded in UTF-8, and to have the following
+structure: two
 folders, I<DIRECTORS> and I<MOVIES>, that have the same parent.
 I<DIRECTORS> contains one subdirectory for each director, and each movie
 lies (B<as a regular file>) in the subdirectory of its director.
@@ -155,9 +157,11 @@ my $file = basename ($ARGV[0]);
 my $file_s = catfile ($symlinks, $file);
 my $file_d;
 $file_d = catfile ($symlinks, readlink $file_s) if -l $file_s;
+Encode::_utf8_on($file_d); 
 my $director;
 if (-l $file_s) {
     my @director = File::Spec->splitdir(readlink $file_s);
+    map {Encode::_utf8_on($_)} @director;
     $director = $director[2] if $#director >= 2;
 }
 
@@ -176,7 +180,7 @@ $dbh->do( "SET NAMES UTF8" ) or die "Error: Can't set names to UTF-8.\n";
 
 # Lookup for the file in the videodb database
 my $res = $dbh->selectall_arrayref ( "SELECT id FROM $config{prefix}videodata WHERE filename = ?",
-                                     undef, $file )
+                                     undef, &iconv($file) )
     or die "Can't select: " .$dbh->errstr. "\n";
 
 die "Error: Your collection is not sane! (and `--force' is not set).\n"
@@ -229,3 +233,19 @@ sub ack {
         $r = 1;
     }
 }
+
+# Convert to UTF-8
+# The input string should be in latin1 unless the UTF-8 flag is on.
+sub iconv {
+    my $string = $_[0];
+    return unless defined $string;
+
+    unless (Encode::is_utf8($string)) {
+        $string = Encode::decode( 'latin1', $string, 1 );
+    }
+
+    # The UTF-8 flag should be on now
+    warn "Warning: Not a valid Unicode string: \"$string\".\n"
+        unless utf8::valid($string);
+    return $string;
+}
-- 
cgit v1.2.3