From: Adam Dickmeiss <adam@indexdata.dk>
Date: Wed, 9 Jan 2008 14:53:26 +0000 (+0000)
Subject: Fixed bug #2002: Zebra crashes during merge using ICU indexing.
X-Git-Tag: ZEBRA.2.0.24~18
X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=commitdiff_plain;h=6b75f88ead401c95ebf4e6d6761447264789c01a

Fixed bug #2002: Zebra crashes during merge using ICU indexing.

The ICU stuff may produce long strings which results in buffer overflow.
This patch fixes that. The old charmap system produced srings no
longer than IT_MAX_WORD which is considerably less than INP_NAME_MAX
in kinput.c. This fixes key_file_read so that strings longer than
IT_MAX_WORD are truncated.
---

diff --git a/index/kinput.c b/index/kinput.c
index 9101fa7..e0181bc 100644
--- a/index/kinput.c
+++ b/index/kinput.c
@@ -1,4 +1,4 @@
-/* $Id: kinput.c,v 1.86 2007-10-31 16:56:14 adam Exp $
+/* $Id: kinput.c,v 1.87 2008-01-09 14:53:26 adam Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -188,8 +188,12 @@ int key_file_read(struct key_file *f, char *key)
     {
         i = 0;
         key[i++] = c;
-        while ((key[i++] = key_file_getc(f)))
-            ;
+        while ((c = key_file_getc(f)))
+        {
+            if (i < IT_MAX_WORD)
+                key[i++] = c;
+        }
+        key[i++] = '\0';
         strcpy(f->prev_name, key);
 	iscz1_reset(f->decode_handle);
     }