From 56a0ec9638a1bb212ee05a114f5debf8c62c8b2a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Michal=20=C4=8Ciha=C5=99?= <michal@cihar.com>
Date: Tue, 21 Feb 2006 20:28:27 +0000
Subject: [PATCH] Skip byte order marks (bug #1423362).

---
 ChangeLog                |  1 +
 libraries/import.lib.php | 13 +++++++++++++
 2 files changed, 14 insertions(+)

diff --git a/ChangeLog b/ChangeLog
index d68259139..a80175a1e 100755
--- a/ChangeLog
+++ b/ChangeLog
@@ -16,6 +16,7 @@ $Source$
     * libraries/iconv_wrapper.lib.php, libraries/charset_conversion.lib.php,
       libraries/database_interface.lib.php: Compatibility with iconv charset
       names on AIX (patch #1420704, thanks to Björn Wiberg - bwiberg).
+    * libraries/import.lib.php: Skip byte order marks (bug #1423362).
 
 2006-02-21 Sebastian Mendel <cybot_tm@users.sourceforge.net>
     * libraries/common.lib.php PMA_getUvaCondition():
diff --git a/libraries/import.lib.php b/libraries/import.lib.php
index 0562575d5..a99606aa0 100644
--- a/libraries/import.lib.php
+++ b/libraries/import.lib.php
@@ -251,6 +251,19 @@ function PMA_importGetNextChunk($size = 32768)
     if ($charset_conversion) {
         return PMA_convert_string($charset_of_file, $charset, $result);
     } else {
+        // Skip possible byte order marks (I do not think we need more
+        // charsets, but feel free to add more, you can use wikipedia for
+        // reference: <http://en.wikipedia.org/wiki/Byte_Order_Mark>)
+        // @TODO: BOM could be used for charset autodetection
+        if ($offset == $size) {
+            // UTF-8
+            if (strncmp($result, "\xEF\xBB\xBF", 3) == 0) {
+                $result = substr($result, 3);
+            // UTF-16 BE, LE
+            } elseif (strncmp($result, "\xFE\xFF", 2) == 0 || strncmp($result, "\xFF\xFE", 2) == 0) {
+                $result = substr($result, 2);
+            }
+        }
         return $result; 
     }
 }