Pull get_unicode_or_ascii_string() into packet-smb.c.

Nobody else needs its extra stuff, as they're not pulling data out of an
SMB byte buffer, so they don't need the extra length check, and nobody
else uses it - or should use it, as we now have support for string
encodings in the tvbuff and protocol-tree code.

Change-Id: I8d249ea2c89a744eef12a05ad755811c07ca463a
Reviewed-on: https://code.wireshark.org/review/33581
Petri-Dish: Guy Harris <guy@alum.mit.edu>
Tested-by: Petri Dish Buildbot
Reviewed-by: Guy Harris <guy@alum.mit.edu>

1 files changed, 157 insertions, 0 deletions

diff --git a/epan/dissectors/packet-smb.c b/epan/dissectors/packet-smb.c
index b2a9c98c88..6d0fc69f63 100644
--- a/epan/dissectors/packet-smb.c
+++ b/epan/dissectors/packet-smb.c
@@ -1718,7 +1718,164 @@ static GSList *conv_tables = NULL;
    End of request/response matching functions
    XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX */
 
+/* Max string length for displaying Unicode strings.  */
+#define	MAX_UNICODE_STR_LEN	256
+
+/* Turn a little-endian Unicode '\0'-terminated string into a string we
+   can display.
+   XXX - for now, we just handle the ISO 8859-1 characters.
+   If exactlen==TRUE then us_lenp contains the exact len of the string in
+   bytes. It might not be null terminated !
+   bc specifies the number of bytes in the byte parameters; Windows 2000,
+   at least, appears, in some cases, to put only 1 byte of 0 at the end
+   of a Unicode string if the byte count
+*/
+static gchar *
+unicode_to_str(tvbuff_t *tvb, int offset, int *us_lenp, gboolean exactlen,
+	       guint16 bc)
+{
+	gchar    *cur;
+	gchar    *p;
+	guint16   uchar;
+	int       len;
+	int       us_len;
+	gboolean  overflow = FALSE;
+
+	cur=(gchar *)wmem_alloc(wmem_packet_scope(), MAX_UNICODE_STR_LEN+3+1);
+	p = cur;
+	len = MAX_UNICODE_STR_LEN;
+	us_len = 0;
+	for (;;) {
+		if (bc == 0)
+			break;
+
+		if (bc == 1) {
+			/* XXX - explain this */
+			if (!exactlen)
+				us_len += 1;	/* this is a one-byte null terminator */
+			break;
+		}
+
+		uchar = tvb_get_letohs(tvb, offset);
+		if (uchar == 0) {
+			us_len += 2;	/* this is a two-byte null terminator */
+			break;
+		}
 
+		if (len > 0) {
+			if ((uchar & 0xFF00) == 0)
+				*p++ = (gchar) uchar;	/* ISO 8859-1 */
+			else
+				*p++ = '?';	/* not 8859-1 */
+			len--;
+		} else
+			overflow = TRUE;
+
+		offset += 2;
+		bc -= 2;
+		us_len += 2;
+
+		if(exactlen){
+			if(us_len>= *us_lenp){
+				break;
+			}
+		}
+	}
+	if (overflow) {
+		/* Note that we're not showing the full string.  */
+		*p++ = '.';
+		*p++ = '.';
+		*p++ = '.';
+	}
+
+	*p = '\0';
+	*us_lenp = us_len;
+
+	return cur;
+}
+
+/* nopad == TRUE : Do not add any padding before this string
+ * exactlen == TRUE : len contains the exact len of the string in bytes.
+ * bc: pointer to variable with amount of data left in the byte parameters
+ *   region
+ */
+const gchar *
+get_unicode_or_ascii_string(tvbuff_t *tvb, int *offsetp,
+			    gboolean useunicode, int *len, gboolean nopad, gboolean exactlen,
+			    guint16 *bcp)
+{
+	gchar       *cur;
+	const gchar *string;
+	int          string_len = 0;
+	int          copylen;
+	gboolean     overflow   = FALSE;
+
+	if (*bcp == 0) {
+		/* Not enough data in buffer */
+		return NULL;
+	}
+
+	if (useunicode) {
+		if ((!nopad) && (*offsetp % 2)) {
+			(*offsetp)++;   /* Looks like a pad byte there sometimes */
+			(*bcp)--;
+
+			if (*bcp == 0) {
+				/* Not enough data in buffer */
+				return NULL;
+			}
+		}
+
+		if(exactlen){
+			string_len = *len;
+			if (string_len < 0) {
+				/* This probably means it's a very large unsigned number; just set
+				   it to the largest signed number, so that we throw the appropriate
+				   exception. */
+				string_len = INT_MAX;
+			}
+		}
+
+		string = unicode_to_str(tvb, *offsetp, &string_len, exactlen, *bcp);
+
+	} else {
+		if(exactlen){
+			/*
+			 * The string we return must be null-terminated.
+			 */
+			cur=(gchar *)wmem_alloc(wmem_packet_scope(), MAX_UNICODE_STR_LEN+3+1);
+			copylen = *len;
+
+			if (copylen < 0) {
+				/* This probably means it's a very large unsigned number; just set
+				   it to the largest signed number, so that we throw the appropriate
+				   exception. */
+				copylen = INT_MAX;
+			}
+
+			tvb_ensure_bytes_exist(tvb, *offsetp, copylen);
+
+			if (copylen > MAX_UNICODE_STR_LEN) {
+				copylen = MAX_UNICODE_STR_LEN;
+				overflow = TRUE;
+			}
+
+			tvb_memcpy(tvb, (guint8 *)cur, *offsetp, copylen);
+			cur[copylen] = '\0';
+
+			if (overflow)
+				g_strlcat(cur, "...",MAX_UNICODE_STR_LEN+3+1);
+
+			string_len = *len;
+			string = cur;
+		} else {
+			string = tvb_get_const_stringz(tvb, *offsetp, &string_len);
+		}
+	}
+
+	*len = string_len;
+	return string;
+}
 
 typedef struct _smb_uid_t {
 	char *domain;