@@ -1,3 +1,7 @@

+Sun Feb  3 10:38:08 EET 2008 (edwin)

+------------------------------------

+  * libclamav/entconv.c: don't leave return value uninitialized. (bb #808).

+

 Sat Feb  2 18:51:55 EET 2008 (edwin)

 -----------------------------------

   * libclamav/entconv.h: enum encodings was missing (bb #807)

@@ -366,7 +366,7 @@ static int iconv(iconv_t iconv_struct,char **inbuf, size_t *inbytesleft,

 static inline const char* detect_encoding(const unsigned char* bom, uint8_t* bom_found, uint8_t* enc_width)

 {

-	const char* encoding;

+	const char* encoding = NULL;

 	int has_bom = 0;

 	uint8_t enc_bytes = 1; /* default is UTF8, which has a minimum of 1 bytes */

 	/* undecided 32-bit encodings are treated as ucs4, and

@@ -473,11 +473,16 @@ static inline const char* detect_encoding(const unsigned char* bom, uint8_t* bom

 /* detects UTF-16(LE/BE), UCS-4(all 4 variants).

  * UTF-8 and simple ASCII are ignored, because we can process those as text */

-const char* encoding_detect_bom(const unsigned char* bom)

+const char* encoding_detect_bom(const unsigned char* bom, const size_t length)

 {

 	uint8_t has_bom;

 	uint8_t enc_width;

-	const char* encoding = detect_encoding(bom, &has_bom, &enc_width);

+	const char* encoding;

+

+	if(length < 4) {

+		return NULL;

+	}

+	encoding = detect_encoding(bom, &has_bom, &enc_width);

 	return enc_width > 1 ? encoding : NULL;

 }

@@ -51,7 +51,7 @@ enum encodings {E_UCS4,E_UTF16,E_UCS4_1234,E_UCS4_4321,E_UCS4_2143,E_UCS4_3412,E

 unsigned char* u16_normalize_tobuffer(uint16_t u16, unsigned char* dst, size_t dst_size);

 const char* entity_norm(struct entity_conv* conv,const unsigned char* entity);

-const char* encoding_detect_bom(const unsigned char* bom);

+const char* encoding_detect_bom(const unsigned char* bom, const size_t length);

 int encoding_normalize_toascii(const m_area_t* in_m_area, const char* initial_encoding, m_area_t* out_m_area);

 #endif

@@ -187,7 +187,7 @@ cli_file_t cli_filetype2(int desc, const struct cl_engine *engine)

 		    /* check if we can autodetect this encoding.

 		     * If we can't don't try to detect HTML sig, since

 		     * we just tried that above, and failed */

-		    if((encoding = encoding_detect_bom(smallbuff))) {

+		    if((encoding = encoding_detect_bom(smallbuff, bread))) {

 			    unsigned char decodedbuff[sizeof(smallbuff)*2];

 			    m_area_t in_area, out_area;