#include <dirent.h>
#endif
-#include <ctype.h>
-#include <stdlib.h>
#include <string.h>
-#include <stdarg.h>
#include <errno.h>
-#include <stdio.h>
-
#include <glib.h>
#include <wsutil/str_util.h>
static range_t *global_xml_tcp_range = NULL;
static range_t *xml_tcp_range = NULL;
+static gboolean pref_heuristic_unicode = FALSE;
+
#define XML_CDATA -1000
#define XML_SCOPED_NAME -1001
-GArray* hf_arr;
-GArray* ett_arr;
+static GArray* hf_arr;
+static GArray* ett_arr;
static const gchar* default_media_types[] = {
"text/xml",
"application/rls-services+xml",
"application/smil",
"application/simple-filter+xml",
+ "application/simservs+xml",
"application/soap+xml",
+ "application/vnd.etsi.aoc+xml",
+ "application/vnd.etsi.cug+xml",
+ "application/vnd.etsi.iptvcommand+xml",
+ "application/vnd.etsi.iptvdiscovery+xml",
+ "application/vnd.etsi.iptvprofile+xml",
+ "application/vnd.etsi.iptvsad-bc+xml",
+ "application/vnd.etsi.iptvsad-cod+xml",
+ "application/vnd.etsi.iptvsad-npvr+xml",
+ "application/vnd.etsi.iptvueprofile+xml",
+ "application/vnd.etsi.mcid+xml",
+ "application/vnd.etsi.sci+xml",
+ "application/vnd.etsi.simservs+xml",
+ "application/vnd.3gpp.cw+xml",
"application/vnd.wv.csp+xml",
"application/vnd.wv.csp.xml",
"application/watcherinfo+xml",
"application/x-wms-logconnectstats",
"application/x-wms-logplaystats",
"application/x-wms-sendevent",
- "application/rss+xml",
+ "application/rss+xml",
"image/svg+xml",
};
static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child) {
- new_child->firts_child = NULL;
+ new_child->first_child = NULL;
new_child->last_child = NULL;
new_child->parent = parent;
new_child->prev_sibling = NULL;
if (parent == NULL) return; /* root */
- if (parent->firts_child == NULL) { /* the 1st child */
- parent->firts_child = new_child;
+ if (parent->first_child == NULL) { /* the 1st child */
+ parent->first_child = new_child;
} else { /* following children */
parent->last_child->next_sibling = new_child;
new_child->prev_sibling = parent->last_child;
dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree)
{
tvbparse_t* tt;
- tvbparse_elem_t* tok = NULL;
static GPtrArray* stack = NULL;
xml_frame_t* current_frame;
char* colinfo_str;
-
- if(!tree) return;
if (stack != NULL)
g_ptr_array_free(stack,TRUE);
current_frame = ep_alloc(sizeof(xml_frame_t));
current_frame->type = XML_FRAME_ROOT;
current_frame->name = NULL;
+ current_frame->name_orig_case = NULL;
+ current_frame->value = NULL;
insert_xml_frame(NULL, current_frame);
g_ptr_array_add(stack,current_frame);
ascii_strup_inplace(colinfo_str);
}
- if (check_col(pinfo->cinfo, COL_PROTOCOL))
- col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
+ col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
current_frame->ns = root_ns;
current_frame->tree = proto_item_add_subtree(current_frame->item,current_frame->ns->ett);
current_frame->last_item = current_frame->item;
- while(( tok = tvbparse_get(tt, want) )) ;
+ while(tvbparse_get(tt, want)) ;
pinfo->private_data = current_frame; /* pass XML structure to the dissector calling XML */
}
static gboolean dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree) {
- if ( (pref_heuristic_media || pref_heuristic_tcp || pref_heuristic_udp)
- && tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
- dissect_xml(tvb, pinfo, tree);
- return TRUE;
- } else {
- return FALSE;
+ if (pref_heuristic_media || pref_heuristic_tcp || pref_heuristic_udp) {
+ if (tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
+ dissect_xml(tvb, pinfo, tree);
+ return TRUE;
+ } else if (pref_heuristic_unicode) {
+ const guint8 *data = tvb_get_ephemeral_unicode_string(tvb, 0, tvb_length(tvb)/2, ENC_LITTLE_ENDIAN);
+ tvbuff_t *unicode_tvb = tvb_new_child_real_data(tvb, data, tvb_length(tvb)/2, tvb_length(tvb)/2);
+ if (tvbparse_peek(tvbparse_init(unicode_tvb,0,-1,NULL,want_ignore), want_heur)) {
+ add_new_data_source(pinfo, unicode_tvb, "UTF8");
+ dissect_xml(unicode_tvb, pinfo, tree);
+ return TRUE;
+ }
+ }
+ }
+ return FALSE;
+}
+
+xml_frame_t *xml_get_tag(xml_frame_t *frame, const gchar *name) {
+ xml_frame_t *tag = NULL;
+
+ xml_frame_t *xml_item = frame->first_child;
+ while (xml_item) {
+ if (xml_item->type == XML_FRAME_TAG) {
+ if (!name) { /* get the 1st tag */
+ tag = xml_item;
+ break;
+ } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
+ tag = xml_item;
+ break;
+ }
+ }
+ xml_item = xml_item->next_sibling;
+ }
+
+ return tag;
+}
+
+xml_frame_t *xml_get_attrib(xml_frame_t *frame, const gchar *name) {
+ xml_frame_t *attr = NULL;
+
+ xml_frame_t *xml_item = frame->first_child;
+ while (xml_item) {
+ if ((xml_item->type == XML_FRAME_ATTRIB) &&
+ xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
+ attr = xml_item;
+ break;
+ }
+ xml_item = xml_item->next_sibling;
}
+
+ return attr;
+}
+
+xml_frame_t *xml_get_cdata(xml_frame_t *frame) {
+ xml_frame_t *cdata = NULL;
+
+ xml_frame_t *xml_item = frame->first_child;
+ while (xml_item) {
+ if (xml_item->type == XML_FRAME_CDATA) {
+ cdata = xml_item;
+ break;
+ }
+ xml_item = xml_item->next_sibling;
+ }
+
+ return cdata;
}
static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
GPtrArray* stack = tvbparse_data;
xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
int hfid;
+ gboolean is_cdata = FALSE;
proto_item* pi;
+ xml_frame_t* new_frame;
if (tok->id == XML_CDATA) {
hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
+ is_cdata = TRUE;
} else if ( tok->id > 0) {
hfid = tok->id;
} else {
proto_item_set_text(pi, "%s",
tvb_format_text(tok->tvb,tok->offset,tok->len));
+
+ if (is_cdata) {
+ new_frame = ep_alloc(sizeof(xml_frame_t));
+ new_frame->type = XML_FRAME_CDATA;
+ new_frame->name = NULL;
+ new_frame->name_orig_case = NULL;
+ new_frame->value = tvb_new_subset(tok->tvb, tok->offset, tok->len, tok->len);
+ insert_xml_frame(current_frame, new_frame);
+ new_frame->item = pi;
+ new_frame->last_item = pi;
+ new_frame->tree = NULL;
+ new_frame->start_offset = tok->offset;
+ new_frame->ns = NULL;
+ }
}
static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
new_frame = ep_alloc(sizeof(xml_frame_t));
new_frame->type = XML_FRAME_XMPLI;
new_frame->name = name;
+ new_frame->name_orig_case = name;
+ new_frame->value = NULL;
insert_xml_frame(current_frame, new_frame);
new_frame->item = pi;
new_frame->last_item = pi;
xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
proto_tree_add_text(current_frame->tree,
- tok->tvb, tok->offset, tok->len,
+ tok->tvb, tok->offset, tok->len, "%s",
tvb_format_text(tok->tvb,tok->offset,tok->len));
if (stack->len > 1) {
xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
tvbparse_elem_t* name_tok = tok->sub->next;
gchar* root_name;
- gchar* name = NULL;
+ gchar *name = NULL, *name_orig_case = NULL;
xml_ns_t* ns;
xml_frame_t* new_frame;
proto_item* pi;
root_name = (gchar*)tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
name = (gchar*)tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
+ name_orig_case = name;
nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
} else {
name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
+ name_orig_case = ep_strdup(name);
ascii_strdown_inplace(name);
if(current_frame->ns) {
new_frame = ep_alloc(sizeof(xml_frame_t));
new_frame->type = XML_FRAME_TAG;
new_frame->name = name;
+ new_frame->name_orig_case = name_orig_case;
+ new_frame->value = NULL;
insert_xml_frame(current_frame, new_frame);
new_frame->item = pi;
new_frame->last_item = pi;
xml_frame_t* new_frame;
tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
- name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
+ name_tok->tvb, name_tok->offset, name_tok->len, ENC_ASCII|ENC_NA);
proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
new_frame = ep_alloc(sizeof(xml_frame_t));
new_frame->type = XML_FRAME_DTD_DOCTYPE;
new_frame->name = (gchar*)tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
+ new_frame->name_orig_case = new_frame->name;
+ new_frame->value = NULL;
insert_xml_frame(current_frame, new_frame);
new_frame->item = dtd_item;
new_frame->last_item = dtd_item;
static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
GPtrArray* stack = tvbparse_data;
xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
- gchar* name = (gchar*)tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
- tvbparse_elem_t* value = tok->sub->next->next->data;
+ gchar *name = NULL, *name_orig_case = NULL;
+ tvbparse_elem_t* value;
+ tvbparse_elem_t* value_part = tok->sub->next->next->data;
int* hfidp;
int hfid;
+ proto_item* pi;
+ xml_frame_t* new_frame;
+ name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
+ name_orig_case = ep_strdup(name);
ascii_strdown_inplace(name);
+
if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
hfid = *hfidp;
+ value = value_part;
} else {
hfid = hf_unknowwn_attrib;
value = tok;
}
- current_frame->last_item = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
- proto_item_set_text(current_frame->last_item, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
+ pi = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
+ proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
+
+ current_frame->last_item = pi;
+
+ new_frame = ep_alloc(sizeof(xml_frame_t));
+ new_frame->type = XML_FRAME_ATTRIB;
+ new_frame->name = name;
+ new_frame->name_orig_case = name_orig_case;
+ new_frame->value = tvb_new_subset(value_part->tvb, value_part->offset, value_part->len, value_part->len);
+ insert_xml_frame(current_frame, new_frame);
+ new_frame->item = pi;
+ new_frame->last_item = pi;
+ new_frame->tree = NULL;
+ new_frame->start_offset = tok->offset;
+ new_frame->ns = NULL;
}
static void init_xml_parser(void) {
- tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
- tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
+ tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
+ tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
want_name,
hfri.hfinfo.display = BASE_NONE;
hfri.hfinfo.strings = NULL;
hfri.hfinfo.bitmask = 0x0;
- hfri.hfinfo.blurb = "";
+ hfri.hfinfo.blurb = NULL;
hfri.hfinfo.id = 0;
hfri.hfinfo.parent = 0;
- hfri.hfinfo.ref_count = 0;
+ hfri.hfinfo.ref_type = HF_REF_TYPE_NONE;
hfri.hfinfo.bitshift = 0;
hfri.hfinfo.same_name_next = NULL;
hfri.hfinfo.same_name_prev = NULL;
}
static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
-
- if(dtd_data->proto_name) g_free(dtd_data->proto_name);
- if(dtd_data->media_type) g_free(dtd_data->media_type);
- if(dtd_data->description) g_free(dtd_data->description);
- if(dtd_data->proto_root) g_free(dtd_data->proto_root);
+ g_free(dtd_data->proto_name);
+ g_free(dtd_data->media_type);
+ g_free(dtd_data->description);
+ g_free(dtd_data->proto_root);
g_string_free(dtd_data->error,TRUE);
-
while(dtd_data->elements->len) {
dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
g_ptr_array_free(nl->list,TRUE);
g_ptr_array_free(dtd_data->attributes,TRUE);
g_free(dtd_data);
-
}
-
static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
gchar* key = g_strdup(k);
int* value = g_malloc(sizeof(int));
GString* s = g_string_new(proto_name);
gchar* str;
g_string_append(s,".");
-
+
for (i = 1; i < hier->len; i++) {
g_string_append_printf(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
}
dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
- if (!element) {
- g_string_append_printf(errors,"element %s is not defined\n", nl->name);
+ if (element) {
+ while(nl->list->len) {
+ gchar* name = g_ptr_array_remove_index(nl->list,0);
+ int* id_p = g_malloc(sizeof(int));
- goto next_attribute;
+ *id_p = -1;
+ g_hash_table_insert(element->attributes,name,id_p);
+ }
}
-
- while(nl->list->len) {
- gchar* name = g_ptr_array_remove_index(nl->list,0);
- int* id_p = g_malloc(sizeof(int));
-
- *id_p = -1;
- g_hash_table_insert(element->attributes,name,id_p);
+ else {
+ g_string_append_printf(errors,"element %s is not defined\n", nl->name);
}
-next_attribute:
g_free(nl->name);
g_ptr_array_free(nl->list,TRUE);
g_free(nl);
/* if a proto_root is defined in the dtd we'll use that as root */
if( dtd_data->proto_root ) {
- if(root_name)
- g_free(root_name);
+ g_free(root_name);
root_name = g_strdup(dtd_data->proto_root);
}
hfs = hf_arr;
etts = ett_arr;
g_ptr_array_add(hier,g_strdup("xml"));
- root_element = &xml_ns;
} else {
/*
* if we were given a proto_name the namespace will be registered
- * as an indipendent protocol with its own hf and ett arrays.
+ * as an independent protocol with its own hf and ett arrays.
*/
hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
etts = g_array_new(FALSE,FALSE,sizeof(gint*));
/*
* if we were given a proto_name the namespace will be registered
- * as an indipendent protocol.
+ * as an independent protocol.
*/
if( dtd_data->proto_name ) {
gint* ett_p;
g_hash_table_destroy(elements);
destroy_dtd_data(dtd_data);
-
- if (root_name)
- g_free(root_name);
+ g_free(root_name);
}
# define DIRECTORY_T GDir
guint namelen;
filename = GETFNAME_OP(file);
- namelen = strlen(filename);
+ namelen = (int)strlen(filename);
if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4),".dtd") == 0 ) ) {
GString* errors = g_string_new("");
GString* preparsed = dtd_preparse(dirname, filename, errors);
}
static void range_delete_xml_tcp_callback(guint32 port) {
- dissector_delete("tcp.port", port, xml_handle);
+ dissector_delete_uint("tcp.port", port, xml_handle);
}
static void range_add_xml_tcp_callback(guint32 port) {
- dissector_add("tcp.port", port, xml_handle);
+ dissector_add_uint("tcp.port", port, xml_handle);
}
static void apply_prefs(void) {
pref_heuristic_media_save = FALSE;
}
}
-
+
if (pref_heuristic_tcp_save != pref_heuristic_tcp ) {
if (pref_heuristic_tcp) {
heur_dissector_add("tcp", dissect_xml_heur, xml_ns.hf_tag);
pref_heuristic_tcp_save = FALSE;
}
}
-
+
if (pref_heuristic_udp_save != pref_heuristic_udp ) {
if (pref_heuristic_udp) {
heur_dissector_add("udp", dissect_xml_heur, xml_ns.hf_tag);
range_foreach(xml_tcp_range, range_delete_xml_tcp_callback);
g_free(xml_tcp_range);
xml_tcp_range = range_copy(global_xml_tcp_range);
- range_foreach(xml_tcp_range, range_add_xml_tcp_callback);
+ range_foreach(xml_tcp_range, range_add_xml_tcp_callback);
}
void
};
static hf_register_info hf_base[] = {
- { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
- { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }}
+ { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
+ { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }}
};
module_t* xml_module;
prefs_register_bool_preference(xml_module, "heuristic", "Use Heuristics for media types",
"Try to recognize XML for unknown media types",
&pref_heuristic_media);
- prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for tcp",
+ prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for TCP",
"Try to recognize XML for unknown TCP ports",
&pref_heuristic_tcp);
prefs_register_range_preference(xml_module, "tcp.port", "TCP Ports",
prefs_register_bool_preference(xml_module, "heuristic_udp", "Use Heuristics for UDP",
"Try to recognize XML for unknown UDP ports",
&pref_heuristic_udp);
-
+ prefs_register_bool_preference(xml_module, "heuristic_unicode", "Use Unicode in heuristics",
+ "Try to recognize XML encoded in Unicode (UCS-2)",
+ &pref_heuristic_unicode);
+
g_array_free(hf_arr,FALSE);
g_array_free(ett_arr,TRUE);