From b1fb698c49b9591ad0a6ca3db5ce6a25e2309944 Mon Sep 17 00:00:00 2001
From: alecpl <alec@alec.pl>
Date: Wed, 03 Jun 2009 14:45:39 -0400
Subject: [PATCH] - added more charset aliases
---
program/include/main.inc | 14 ++++++++++----
1 files changed, 10 insertions(+), 4 deletions(-)
diff --git a/program/include/main.inc b/program/include/main.inc
index ddecd6e..b9408cf 100644
--- a/program/include/main.inc
+++ b/program/include/main.inc
@@ -243,7 +243,7 @@
$str = utf8_encode($str);
}
else if ($from != 'UTF-8' && $conv) {
- $from = preg_replace(array('/^WINDOWS-*/', '/^CP-/'), array('CP', 'CP'), $from);
+ $from = preg_replace(array('/^WINDOWS-*125([0-8])$/', '/^CP-/'), array('CP125\\1', 'CP'), $from);
$conv->loadCharset($from);
$str = $conv->strToUtf8($str);
}
@@ -264,7 +264,7 @@
return utf8_decode($str);
}
else if ($to != 'UTF-8' && $conv) {
- $to = preg_replace(array('/^WINDOWS-*/', '/^CP-/'), array('CP', 'CP'), $to);
+ $to = preg_replace(array('/^WINDOWS-*125([0-8])$/', '/^CP-/'), array('CP125\\1', 'CP'), $to);
$conv->loadCharset($to);
return $conv->utf8ToStr($str);
}
@@ -305,6 +305,7 @@
# RFC1642
$charset = str_replace('UNICODE-1-1-', '', $charset);
+ # Aliases: some of them from HTML5 spec.
$aliases = array(
'USASCII' => 'WINDOWS-1252',
'ANSIX31101983' => 'WINDOWS-1252',
@@ -313,9 +314,14 @@
'XUNKNOWN' => 'ISO-8859-15',
'XUSERDEFINED' => 'ISO-8859-15',
'KSC56011987' => 'EUC-KR',
- 'GB2312' => 'GB18030',
+ 'GB2312' => 'GBK',
+ 'GB231280' => 'GBK',
'UNICODE' => 'UTF-8',
- 'UTF7IMAP' => 'UTF7-IMAP'
+ 'UTF7IMAP' => 'UTF7-IMAP',
+ 'XXBIG5' => 'BIG5',
+ 'TIS620' => 'WINDOWS-874',
+ 'ISO88599' => 'WINDOWS-1254',
+ 'ISO885911' => 'WINDOWS-874',
);
$str = preg_replace('/[^a-z0-9]/i', '', $charset);
--
Gitblit v1.9.1