From 216a697f438942ac99f6fbd01822f376378efe2d Mon Sep 17 00:00:00 2001
From: Tomasz Kojm <tkojm@clamav.net>
Date: Tue, 24 Aug 2004 00:50:29 +0000
Subject: [PATCH] only enable signature file type recognition for text files

git-svn: trunk@788
---
 clamav-devel/ChangeLog             |  4 ++++
 clamav-devel/libclamav/filetypes.c | 16 ++++++++++++----
 clamav-devel/libclamav/filetypes.h |  3 ++-
 clamav-devel/libclamav/scanners.c  |  9 ++++++---
 4 files changed, 24 insertions(+), 8 deletions(-)

diff --git a/clamav-devel/ChangeLog b/clamav-devel/ChangeLog
index 92ff0b9ca..b288fa2d1 100644
--- a/clamav-devel/ChangeLog
+++ b/clamav-devel/ChangeLog
@@ -1,3 +1,7 @@
+Tue Aug 24 02:30:28 CEST 2004 (tk)
+----------------------------------
+  * libclamav: only enable signature file type recognition for text files
+
 Mon Aug 23 22:32:02 CEST 2004 (tk)
 ----------------------------------
   * include database info in version string (requested by Jason Haar
diff --git a/clamav-devel/libclamav/filetypes.c b/clamav-devel/libclamav/filetypes.c
index 9bfbed72b..cb077095a 100644
--- a/clamav-devel/libclamav/filetypes.c
+++ b/clamav-devel/libclamav/filetypes.c
@@ -24,6 +24,7 @@
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
+#include <ctype.h>
 
 #include "clamav.h"
 #include "filetypes.h"
@@ -110,7 +111,7 @@ static const struct cli_magic_s cli_magic[] = {
     {0,  "\060\046\262\165\216\146\317", 7, "WMA/WMV/ASF",	  CL_DATAFILE},
     {0,  ".RMF" ,			 4, "Real Media File",	  CL_DATAFILE},
 
-    {-1, NULL,				 0, NULL,              CL_UNKNOWN_TYPE}
+    {-1, NULL,				 0, NULL,              CL_UNKNOWN_DATA_TYPE}
 };
 
 static const struct cli_smagic_s cli_smagic[] = {
@@ -136,12 +137,13 @@ static const struct cli_smagic_s cli_smagic[] = {
     {"3c536372697074", "HTML data", CL_HTMLFILE},	/* <Script */
     {"3c534352495054", "HTML data", CL_HTMLFILE},	/* <SCRIPT */
 
-    {NULL,  NULL,   CL_UNKNOWN_TYPE}
+    {NULL,  NULL,   CL_UNKNOWN_DATA_TYPE}
 };
 
 cli_file_t cli_filetype(const char *buf, size_t buflen)
 {
-	int i;
+	int i, ascii = 1;
+
 
     for(i = 0; cli_magic[i].magic; i++) {
 	if(buflen >= cli_magic[i].offset+cli_magic[i].length) {
@@ -152,7 +154,13 @@ cli_file_t cli_filetype(const char *buf, size_t buflen)
 	}
     }
 
-    return CL_UNKNOWN_TYPE;
+    for(i = 0; i < buflen; i++)
+	if(isprint(buf[i])) { /* FIXME: do we need to handle intern. chars? */
+	    ascii = 0;
+	    break;
+	}
+
+    return ascii ? CL_UNKNOWN_TEXT_TYPE : CL_UNKNOWN_DATA_TYPE;
 }
 
 int cli_addtypesigs(struct cl_node *root)
diff --git a/clamav-devel/libclamav/filetypes.h b/clamav-devel/libclamav/filetypes.h
index 9299bd3ed..409999ede 100644
--- a/clamav-devel/libclamav/filetypes.h
+++ b/clamav-devel/libclamav/filetypes.h
@@ -24,7 +24,8 @@
 #define CL_TYPENO 500
 
 typedef enum {
-    CL_UNKNOWN_TYPE = CL_TYPENO,
+    CL_UNKNOWN_TEXT_TYPE = CL_TYPENO,
+    CL_UNKNOWN_DATA_TYPE,
     CL_DOSEXE,
     CL_DATAFILE,
     CL_GZFILE,
diff --git a/clamav-devel/libclamav/scanners.c b/clamav-devel/libclamav/scanners.c
index 975fbfab3..ae293a4fb 100644
--- a/clamav-devel/libclamav/scanners.c
+++ b/clamav-devel/libclamav/scanners.c
@@ -1125,12 +1125,15 @@ int cli_magic_scandesc(int desc, const char **virname, long int *scanned, const
 	    {
 		struct stat s;
 		if(fstat(desc, &s) == 0 && S_ISREG(s.st_mode) && s.st_size < 65536)
-		type = CL_UNKNOWN_TYPE;
+		type = CL_UNKNOWN_DATA_TYPE;
 	    }
 
-	default:
+	case CL_UNKNOWN_DATA_TYPE:
 	    ret = cli_scan_mydoom_log(desc, virname, scanned, root, limits, options, arec, mrec);
 	    break;
+
+	default:
+	    break;
     }
 
     type == CL_MAILFILE ? (*mrec)-- : (*arec)--;
@@ -1138,7 +1141,7 @@ int cli_magic_scandesc(int desc, const char **virname, long int *scanned, const
     if(type != CL_DATAFILE && ret != CL_VIRUS) { /* scan the raw file */
 	    int typerec;
 
-	type == CL_UNKNOWN_TYPE ? (typerec = 1) : (typerec = 0);
+	type == CL_UNKNOWN_TEXT_TYPE ? (typerec = 1) : (typerec = 0);
 	lseek(desc, 0, SEEK_SET);
 
 	if((nret = cli_scandesc(desc, virname, scanned, root, typerec)) == CL_VIRUS) {