dfilter: Make string slices a return an FT_STRING

Allow string slices (indexing) to work with internationalized
strings. The old behavior of indexing on byte boundaries can
be obtained using raw slices.

7 files changed, 82 insertions, 45 deletions

diff --git a/epan/ftypes/ftype-bytes.c b/epan/ftypes/ftype-bytes.c
index 5973a85c4c..61e9c0569d 100644
--- a/epan/ftypes/ftype-bytes.c
+++ b/epan/ftypes/ftype-bytes.c
@@ -501,10 +501,7 @@ len(fvalue_t *fv)
 static void
 slice(fvalue_t *fv, GByteArray *bytes, guint offset, guint length)
 {
-	guint8* data;
-
-	data = (guint8 *)g_bytes_get_data(fv->value.bytes, NULL) + offset;
-
+	const guint8 *data = (const guint8 *)g_bytes_get_data(fv->value.bytes, NULL) + offset;
 	g_byte_array_append(bytes, data, length);
 }
 
@@ -627,7 +624,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -664,7 +661,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -701,7 +698,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -738,7 +735,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -775,7 +772,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -812,7 +809,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -849,7 +846,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -886,7 +883,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -923,7 +920,7 @@ ftype_register_bytes(void)
 		bytes_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bytes_bitwise_and,		/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
diff --git a/epan/ftypes/ftype-ipv4.c b/epan/ftypes/ftype-ipv4.c
index dc22944a0f..e38c056c00 100644
--- a/epan/ftypes/ftype-ipv4.c
+++ b/epan/ftypes/ftype-ipv4.c
@@ -203,7 +203,7 @@ ftype_register_ipv4(void)
 		is_zero,
 		NULL,
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bitwise_and,
 		NULL,				/* unary_minus */
 		NULL,				/* add */
diff --git a/epan/ftypes/ftype-ipv6.c b/epan/ftypes/ftype-ipv6.c
index 3be91b9e30..32c949b021 100644
--- a/epan/ftypes/ftype-ipv6.c
+++ b/epan/ftypes/ftype-ipv6.c
@@ -232,7 +232,7 @@ ftype_register_ipv6(void)
 		is_zero,
 		NULL,
 		len,
-		slice,
+		(FvalueSlice)slice,
 		bitwise_and,
 		NULL,				/* unary_minus */
 		NULL,				/* add */
diff --git a/epan/ftypes/ftype-protocol.c b/epan/ftypes/ftype-protocol.c
index 8e5f6f4ba6..8aa6ccbe93 100644
--- a/epan/ftypes/ftype-protocol.c
+++ b/epan/ftypes/ftype-protocol.c
@@ -396,7 +396,7 @@ ftype_register_tvbuff(void)
 		is_zero,
 		NULL,
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,
 		NULL,				/* unary_minus */
 		NULL,				/* add */
diff --git a/epan/ftypes/ftype-string.c b/epan/ftypes/ftype-string.c
index 195267f7aa..6f9eb1d970 100644
--- a/epan/ftypes/ftype-string.c
+++ b/epan/ftypes/ftype-string.c
@@ -127,17 +127,28 @@ string_is_zero(const fvalue_t *fv)
 static guint
 len(fvalue_t *fv)
 {
-	return (guint)fv->value.strbuf->len;
+	/* g_utf8_strlen returns glong for no apparent reason*/
+	glong len = g_utf8_strlen(fv->value.strbuf->str, -1);
+	if (len < 0)
+		return 0;
+	return (guint)len;
 }
 
 static void
-slice(fvalue_t *fv, GByteArray *bytes, guint offset, guint length)
+slice(fvalue_t *fv, wmem_strbuf_t *buf, guint offset, guint length)
 {
-	guint8* data;
-
-	data = (guint8*)fv->value.strbuf->str + offset;
-
-	g_byte_array_append(bytes, data, length);
+	const char *str = fv->value.strbuf->str;
+
+	/* Go to the starting offset */
+	const char *p = g_utf8_offset_to_pointer(str, (glong)offset);
+	const char *n;
+	/* Copy 'length' codepoints to dst. Skip the terminating NULL */
+	while (*p != '\0' && length-- > 0) {
+		n = g_utf8_next_char(p);
+		/* Append n - p bytes (one codepoint)*/
+		wmem_strbuf_append_len(buf, p, n - p);
+		p = n;
+	}
 }
 
 static enum ft_result
@@ -213,7 +224,7 @@ ftype_register_string(void)
 		string_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,				/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -249,7 +260,7 @@ ftype_register_string(void)
 		string_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,				/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -285,7 +296,7 @@ ftype_register_string(void)
 		string_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,				/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -321,7 +332,7 @@ ftype_register_string(void)
 		string_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,				/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
@@ -357,7 +368,7 @@ ftype_register_string(void)
 		string_is_zero,			/* is_zero */
 		NULL,				/* is_negative */
 		len,
-		slice,
+		(FvalueSlice)slice,
 		NULL,				/* bitwise_and */
 		NULL,				/* unary_minus */
 		NULL,				/* add */
diff --git a/epan/ftypes/ftypes-int.h b/epan/ftypes/ftypes-int.h
index af7c7a9ae6..4f15fb26dd 100644
--- a/epan/ftypes/ftypes-int.h
+++ b/epan/ftypes/ftypes-int.h
@@ -86,7 +86,7 @@ typedef enum ft_result (*FvalueMatches)(const fvalue_t*, const ws_regex_t*, gboo
 typedef gboolean (*FvalueIs)(const fvalue_t*);
 typedef guint (*FvalueLen)(fvalue_t*);
 typedef guint (*FvalueHashFunc)(const fvalue_t *);
-typedef void (*FvalueSlice)(fvalue_t*, GByteArray *, guint offset, guint length);
+typedef void (*FvalueSlice)(fvalue_t*, void *, guint offset, guint length);
 typedef enum ft_result (*FvalueUnaryOp)(fvalue_t *, const fvalue_t*, gchar **);
 typedef enum ft_result (*FvalueBinaryOp)(fvalue_t *, const fvalue_t*, const fvalue_t*, gchar **);
 
diff --git a/epan/ftypes/ftypes.c b/epan/ftypes/ftypes.c
index 35dafb2674..e48264272c 100644
--- a/epan/ftypes/ftypes.c
+++ b/epan/ftypes/ftypes.c
@@ -577,16 +577,16 @@ fvalue_to_sinteger64(const fvalue_t *fv, gint64 *repr)
 
 typedef struct {
 	fvalue_t	*fv;
-	GByteArray	*bytes;
+	void		*ptr;
 	gboolean	slice_failure;
 } slice_data_t;
 
 static gboolean
-compute_drnode(guint field_length, drange_node *drnode, guint *offset_ptr, guint *length_ptr)
+compute_drnode(gsize field_length, drange_node *drnode, gsize *offset_ptr, gsize *length_ptr)
 {
-	gint		start_offset;
-	gint		length = 0;
-	gint		end_offset = 0;
+	gssize		start_offset;
+	gssize		length = 0;
+	gssize		end_offset = 0;
 	drange_node_end_t	ending;
 
 	start_offset = drange_node_get_start_offset(drnode);
@@ -640,8 +640,8 @@ slice_func(gpointer data, gpointer user_data)
 {
 	drange_node	*drnode = (drange_node	*)data;
 	slice_data_t	*slice_data = (slice_data_t *)user_data;
-	gint		start_offset;
-	gint		length = 0;
+	gsize		start_offset;
+	gsize		length = 0;
 	fvalue_t	*fv;
 
 	if (slice_data->slice_failure) {
@@ -654,19 +654,40 @@ slice_func(gpointer data, gpointer user_data)
 		return;
 	}
 
-	ws_assert(start_offset >=0 && length > 0);
-	fv->ftype->slice(fv, slice_data->bytes, start_offset, length);
+	ws_assert(length > 0);
+	fv->ftype->slice(fv, slice_data->ptr, (guint)start_offset, (guint)length);
 }
 
-/* Returns a new FT_BYTES fvalue_t* if possible, otherwise NULL */
-fvalue_t*
-fvalue_slice(fvalue_t *fv, drange_t *d_range)
+static fvalue_t *
+slice_string(fvalue_t *fv, drange_t *d_range)
+{
+	slice_data_t	slice_data;
+	fvalue_t	*new_fv;
+
+	slice_data.fv = fv;
+	slice_data.ptr = wmem_strbuf_create(NULL);
+	slice_data.slice_failure = FALSE;
+
+	/* XXX - We could make some optimizations here based on
+	 * drange_has_total_length() and
+	 * drange_get_max_offset().
+	 */
+
+	drange_foreach_drange_node(d_range, slice_func, &slice_data);
+
+	new_fv = fvalue_new(FT_STRING);
+	fvalue_set_strbuf(new_fv, slice_data.ptr);
+	return new_fv;
+}
+
+static fvalue_t *
+slice_bytes(fvalue_t *fv, drange_t *d_range)
 {
 	slice_data_t	slice_data;
 	fvalue_t	*new_fv;
 
 	slice_data.fv = fv;
-	slice_data.bytes = g_byte_array_new();
+	slice_data.ptr = g_byte_array_new();
 	slice_data.slice_failure = FALSE;
 
 	/* XXX - We could make some optimizations here based on
@@ -677,12 +698,20 @@ fvalue_slice(fvalue_t *fv, drange_t *d_range)
 	drange_foreach_drange_node(d_range, slice_func, &slice_data);
 
 	new_fv = fvalue_new(FT_BYTES);
-	GBytes *bytes = g_byte_array_free_to_bytes(slice_data.bytes);
-	fvalue_set_bytes(new_fv, bytes);
-	g_bytes_unref(bytes);
+	fvalue_set_byte_array(new_fv, slice_data.ptr);
 	return new_fv;
 }
 
+/* Returns a new slice fvalue_t* if possible, otherwise NULL */
+fvalue_t*
+fvalue_slice(fvalue_t *fv, drange_t *d_range)
+{
+	if (FT_IS_STRING(fvalue_type_ftenum(fv))) {
+		return slice_string(fv, d_range);
+	}
+	return slice_bytes(fv, d_range);
+}
+
 void
 fvalue_set_bytes(fvalue_t *fv, GBytes *value)
 {