2 * wireshark's xml dissector .
4 * (C) 2005, Luis E. Garcia Ontanon.
8 * Refer to the AUTHORS file or the AUTHORS section in the man page
9 * for contacting the author(s) of this file.
11 * Wireshark - Network traffic analyzer
12 * By Gerald Combs <gerald@wireshark.org>
13 * Copyright 1998 Gerald Combs
15 * This program is free software; you can redistribute it and/or
16 * modify it under the terms of the GNU General Public License
17 * as published by the Free Software Foundation; either version 2
18 * of the License, or (at your option) any later version.
20 * This program is distributed in the hope that it will be useful,
21 * but WITHOUT ANY WARRANTY; without even the implied warranty of
22 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23 * GNU General Public License for more details.
25 * You should have received a copy of the GNU General Public License
26 * along with this program; if not, write to the Free Software
27 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
48 #include <wsutil/str_util.h>
50 #include <epan/emem.h>
51 #include <epan/packet.h>
52 #include <epan/tvbparse.h>
54 #include <epan/report_err.h>
55 #include <epan/filesystem.h>
56 #include <epan/prefs.h>
57 #include <epan/garrayfix.h>
59 #include "packet-xml.h"
61 struct _attr_reg_data {
67 static gint ett_dtd = -1;
68 static gint ett_xmpli = -1;
70 static int hf_unknowwn_attrib = -1;
71 static int hf_comment = -1;
72 static int hf_xmlpi = -1;
73 static int hf_dtd_tag = -1;
74 static int hf_doctype = -1;
76 /* dissector handles */
77 static dissector_handle_t xml_handle;
79 /* parser definitions */
80 static tvbparse_wanted_t* want;
81 static tvbparse_wanted_t* want_ignore;
82 static tvbparse_wanted_t* want_heur;
84 static GHashTable* xmpli_names;
85 static GHashTable* media_types;
87 static xml_ns_t xml_ns = {"xml","/",-1,-1,-1,NULL,NULL,NULL};
88 static xml_ns_t unknown_ns = {"unknown","?",-1,-1,-1,NULL,NULL,NULL};
89 static xml_ns_t* root_ns;
91 static gboolean pref_heuristic_media = FALSE;
92 static gboolean pref_heuristic_tcp = FALSE;
93 static gboolean pref_heuristic_udp = FALSE;
94 static gboolean pref_heuristic_media_save = FALSE;
95 static gboolean pref_heuristic_tcp_save = FALSE;
96 static gboolean pref_heuristic_udp_save = FALSE;
97 static range_t *global_xml_tcp_range = NULL;
98 static range_t *xml_tcp_range = NULL;
100 #define XML_CDATA -1000
101 #define XML_SCOPED_NAME -1001
107 static const gchar* default_media_types[] = {
114 "application/auth-policy+xml",
115 "application/cpim-pidf+xml",
116 "application/cpl+xml",
117 "application/mathml+xml",
118 "application/media_control+xml",
119 "application/note+xml",
120 "application/pidf+xml",
121 "application/poc-settings+xml",
122 "application/rdf+xml",
123 "application/reginfo+xml",
124 "application/resource-lists+xml",
125 "application/rlmi+xml",
126 "application/rls-services+xml",
128 "application/simple-filter+xml",
129 "application/soap+xml",
130 "application/vnd.wv.csp+xml",
131 "application/vnd.wv.csp.xml",
132 "application/watcherinfo+xml",
133 "application/xcap-att+xml",
134 "application/xcap-caps+xml",
135 "application/xcap-diff+xml",
136 "application/xcap-el+xml",
137 "application/xcap-error+xml",
138 "application/xcap-ns+xml",
140 "application/xml-dtd",
141 "application/xpidf+xml",
142 "application/xslt+xml",
143 "application/x-wms-logconnectstats",
144 "application/x-wms-logplaystats",
145 "application/x-wms-sendevent",
146 "application/rss+xml",
150 static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child) {
151 new_child->firts_child = NULL;
152 new_child->last_child = NULL;
154 new_child->parent = parent;
155 new_child->next_sibling = NULL;
156 new_child->prev_sibling = NULL;
157 if (parent == NULL) return; /* root */
159 if (parent->firts_child == NULL) { /* the 1st child */
160 parent->firts_child = new_child;
161 } else { /* following children */
162 parent->last_child->next_sibling = new_child;
163 new_child->prev_sibling = parent->last_child;
165 parent->last_child = new_child;
169 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree)
172 tvbparse_elem_t* tok = NULL;
173 static GPtrArray* stack = NULL;
174 xml_frame_t* current_frame;
178 g_ptr_array_free(stack,TRUE);
180 stack = g_ptr_array_new();
181 current_frame = ep_alloc(sizeof(xml_frame_t));
182 current_frame->type = XML_FRAME_ROOT;
183 current_frame->name = NULL;
184 current_frame->name_orig_case = NULL;
185 current_frame->value = NULL;
186 insert_xml_frame(NULL, current_frame);
187 g_ptr_array_add(stack,current_frame);
189 tt = tvbparse_init(tvb,0,-1,stack,want_ignore);
190 current_frame->start_offset = 0;
194 if (pinfo->match_string)
195 root_ns = g_hash_table_lookup(media_types,pinfo->match_string);
199 colinfo_str = "/XML";
201 colinfo_str = ep_strdup_printf("/%s",root_ns->name);
202 ascii_strup_inplace(colinfo_str);
205 if (check_col(pinfo->cinfo, COL_PROTOCOL))
206 col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
208 current_frame->ns = root_ns;
210 current_frame->item = proto_tree_add_item(tree,current_frame->ns->hf_tag,tvb,0,-1,FALSE);
211 current_frame->tree = proto_item_add_subtree(current_frame->item,current_frame->ns->ett);
212 current_frame->last_item = current_frame->item;
214 while(( tok = tvbparse_get(tt, want) )) ;
216 pinfo->private_data = current_frame; /* pass XML structure to the dissector calling XML */
219 static gboolean dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree) {
220 if ( (pref_heuristic_media || pref_heuristic_tcp || pref_heuristic_udp)
221 && tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
222 dissect_xml(tvb, pinfo, tree);
229 xml_frame_t *xml_get_tag(xml_frame_t *frame, const gchar *name) {
230 xml_frame_t *tag = NULL;
232 xml_frame_t *xml_item = frame->firts_child;
234 if ((xml_item->type == XML_FRAME_TAG)) {
235 if (!name) { /* get the 1st tag */
238 } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
243 xml_item = xml_item->next_sibling;
249 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const gchar *name) {
250 xml_frame_t *attr = NULL;
252 xml_frame_t *xml_item = frame->firts_child;
254 if ((xml_item->type == XML_FRAME_ATTRIB) &&
255 xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
259 xml_item = xml_item->next_sibling;
265 xml_frame_t *xml_get_cdata(xml_frame_t *frame) {
266 xml_frame_t *cdata = NULL;
268 xml_frame_t *xml_item = frame->firts_child;
270 if ((xml_item->type == XML_FRAME_CDATA)) {
274 xml_item = xml_item->next_sibling;
280 static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
281 GPtrArray* stack = tvbparse_data;
282 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
284 gboolean is_cdata = FALSE;
286 xml_frame_t* new_frame;
288 if (tok->id == XML_CDATA) {
289 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
291 } else if ( tok->id > 0) {
294 hfid = xml_ns.hf_cdata;
297 pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, FALSE);
299 proto_item_set_text(pi, "%s",
300 tvb_format_text(tok->tvb,tok->offset,tok->len));
303 new_frame = ep_alloc(sizeof(xml_frame_t));
304 new_frame->type = XML_FRAME_CDATA;
305 new_frame->name = NULL;
306 new_frame->name_orig_case = NULL;
307 new_frame->value = tvb_new_subset(tok->tvb, tok->offset, tok->len, tok->len);
308 insert_xml_frame(current_frame, new_frame);
309 new_frame->item = pi;
310 new_frame->last_item = pi;
311 new_frame->tree = NULL;
312 new_frame->start_offset = tok->offset;
313 new_frame->ns = NULL;
317 static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
318 GPtrArray* stack = tvbparse_data;
319 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
322 tvbparse_elem_t* name_tok = tok->sub->next;
323 gchar* name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
324 xml_ns_t* ns = g_hash_table_lookup(xmpli_names,name);
325 xml_frame_t* new_frame;
330 ascii_strdown_inplace(name);
339 pi = proto_tree_add_item(current_frame->tree,hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
341 proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
343 pt = proto_item_add_subtree(pi,ett);
345 new_frame = ep_alloc(sizeof(xml_frame_t));
346 new_frame->type = XML_FRAME_XMPLI;
347 new_frame->name = name;
348 new_frame->name_orig_case = name;
349 new_frame->value = NULL;
350 insert_xml_frame(current_frame, new_frame);
351 new_frame->item = pi;
352 new_frame->last_item = pi;
353 new_frame->tree = pt;
354 new_frame->start_offset = tok->offset;
357 g_ptr_array_add(stack,new_frame);
361 static void after_xmlpi(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
362 GPtrArray* stack = tvbparse_data;
363 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
365 proto_tree_add_text(current_frame->tree,
366 tok->tvb, tok->offset, tok->len, "%s",
367 tvb_format_text(tok->tvb,tok->offset,tok->len));
369 if (stack->len > 1) {
370 g_ptr_array_remove_index_fast(stack,stack->len - 1);
372 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened xmpli tag ]");
376 static void before_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
377 GPtrArray* stack = tvbparse_data;
378 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
379 tvbparse_elem_t* name_tok = tok->sub->next;
381 gchar *name = NULL, *name_orig_case = NULL;
383 xml_frame_t* new_frame;
387 if (name_tok->sub->id == XML_SCOPED_NAME) {
388 tvbparse_elem_t* root_tok = name_tok->sub->sub;
389 tvbparse_elem_t* leaf_tok = name_tok->sub->sub->next->next;
390 xml_ns_t* nameroot_ns;
392 root_name = (gchar*)tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
393 name = (gchar*)tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
394 name_orig_case = name;
396 nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
399 ns = g_hash_table_lookup(nameroot_ns->elements,name);
408 name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
409 name_orig_case = ep_strdup(name);
410 ascii_strdown_inplace(name);
412 if(current_frame->ns) {
413 ns = g_hash_table_lookup(current_frame->ns->elements,name);
416 if (! ( ns = g_hash_table_lookup(root_ns->elements,name) ) ) {
425 pi = proto_tree_add_item(current_frame->tree,ns->hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
426 proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
428 pt = proto_item_add_subtree(pi,ns->ett);
430 new_frame = ep_alloc(sizeof(xml_frame_t));
431 new_frame->type = XML_FRAME_TAG;
432 new_frame->name = name;
433 new_frame->name_orig_case = name_orig_case;
434 new_frame->value = NULL;
435 insert_xml_frame(current_frame, new_frame);
436 new_frame->item = pi;
437 new_frame->last_item = pi;
438 new_frame->tree = pt;
439 new_frame->start_offset = tok->offset;
442 g_ptr_array_add(stack,new_frame);
446 static void after_open_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
447 GPtrArray* stack = tvbparse_data;
448 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
450 proto_item_append_text(current_frame->last_item,">");
453 static void after_closed_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
454 GPtrArray* stack = tvbparse_data;
455 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
457 proto_item_append_text(current_frame->last_item,"/>");
459 if (stack->len > 1) {
460 g_ptr_array_remove_index_fast(stack,stack->len - 1);
462 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
466 static void after_untag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
467 GPtrArray* stack = tvbparse_data;
468 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
470 proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
472 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
473 tvb_format_text(tok->tvb,tok->offset,tok->len));
475 if (stack->len > 1) {
476 g_ptr_array_remove_index_fast(stack,stack->len - 1);
478 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
479 "[ ERROR: Closing an unopened tag ]");
483 static void before_dtd_doctype(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
484 GPtrArray* stack = tvbparse_data;
485 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
486 xml_frame_t* new_frame;
487 tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
488 proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
489 name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
491 proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
493 new_frame = ep_alloc(sizeof(xml_frame_t));
494 new_frame->type = XML_FRAME_DTD_DOCTYPE;
495 new_frame->name = (gchar*)tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
496 new_frame->name_orig_case = new_frame->name;
497 new_frame->value = NULL;
498 insert_xml_frame(current_frame, new_frame);
499 new_frame->item = dtd_item;
500 new_frame->last_item = dtd_item;
501 new_frame->tree = proto_item_add_subtree(dtd_item,ett_dtd);
502 new_frame->start_offset = tok->offset;
503 new_frame->ns = NULL;
505 g_ptr_array_add(stack,new_frame);
508 static void pop_stack(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
509 GPtrArray* stack = tvbparse_data;
510 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
512 if (stack->len > 1) {
513 g_ptr_array_remove_index_fast(stack,stack->len - 1);
515 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
516 "[ ERROR: Closing an unopened tag ]");
520 static void after_dtd_close(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
521 GPtrArray* stack = tvbparse_data;
522 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
524 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
525 tvb_format_text(tok->tvb,tok->offset,tok->len));
526 if (stack->len > 1) {
527 g_ptr_array_remove_index_fast(stack,stack->len - 1);
529 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
533 static void get_attrib_value(void* tvbparse_data _U_, const void* wanted_data _U_, tvbparse_elem_t* tok) {
534 tok->data = tok->sub;
537 static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
538 GPtrArray* stack = tvbparse_data;
539 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
540 gchar *name = NULL, *name_orig_case = NULL;
541 tvbparse_elem_t* value;
542 tvbparse_elem_t* value_part = tok->sub->next->next->data;
546 xml_frame_t* new_frame;
548 name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
549 name_orig_case = ep_strdup(name);
550 ascii_strdown_inplace(name);
552 if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
556 hfid = hf_unknowwn_attrib;
560 pi = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
561 proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
563 new_frame = ep_alloc(sizeof(xml_frame_t));
564 new_frame->type = XML_FRAME_ATTRIB;
565 new_frame->name = name;
566 new_frame->name_orig_case = name_orig_case;
567 new_frame->value = tvb_new_subset(value_part->tvb, value_part->offset, value_part->len, value_part->len);
568 insert_xml_frame(current_frame, new_frame);
569 new_frame->item = pi;
570 new_frame->last_item = pi;
571 new_frame->tree = NULL;
572 new_frame->start_offset = tok->offset;
573 new_frame->ns = NULL;
577 static void unrecognized_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_){
578 GPtrArray* stack = tvbparse_data;
579 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
581 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Unrecognized text ]");
587 static void init_xml_parser(void) {
588 tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
589 tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
591 tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
593 tvbparse_char(-1,":",NULL,NULL,NULL),
597 tvbparse_wanted_t* want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
602 tvbparse_wanted_t* want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
603 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\"','\\'),
604 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\'','\\'),
605 tvbparse_chars(-1,1,0,"0123456789",NULL,NULL,NULL),
609 tvbparse_wanted_t* want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
610 tvbparse_set_seq(-1, NULL, NULL, after_attrib,
612 tvbparse_char(-1,"=",NULL,NULL,NULL),
616 tvbparse_wanted_t* want_stoptag = tvbparse_set_oneof(-1,NULL,NULL,NULL,
617 tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
618 tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
621 tvbparse_wanted_t* want_stopxmlpi = tvbparse_string(-1,"?>",NULL,NULL,after_xmlpi);
623 tvbparse_wanted_t* want_comment = tvbparse_set_seq(hf_comment,NULL,NULL,after_token,
624 tvbparse_string(-1,"<!--",NULL,NULL,NULL),
625 tvbparse_until(-1,NULL,NULL,NULL,
626 tvbparse_string(-1,"-->",NULL,NULL,NULL),
630 tvbparse_wanted_t* want_xmlpi = tvbparse_set_seq(hf_xmlpi,NULL,before_xmpli,NULL,
631 tvbparse_string(-1,"<?",NULL,NULL,NULL),
633 tvbparse_set_oneof(-1,NULL,NULL,NULL,
635 tvbparse_set_seq(-1,NULL,NULL,NULL,
642 tvbparse_wanted_t* want_closing_tag = tvbparse_set_seq(0,NULL,NULL,after_untag,
643 tvbparse_char(-1, "<", NULL, NULL, NULL),
644 tvbparse_char(-1, "/", NULL, NULL, NULL),
646 tvbparse_char(-1, ">", NULL, NULL, NULL),
649 tvbparse_wanted_t* want_doctype_start = tvbparse_set_seq(-1,NULL,before_dtd_doctype,NULL,
650 tvbparse_char(-1,"<",NULL,NULL,NULL),
651 tvbparse_char(-1,"!",NULL,NULL,NULL),
652 tvbparse_casestring(-1,"DOCTYPE",NULL,NULL,NULL),
653 tvbparse_set_oneof(-1,NULL,NULL,NULL,
654 tvbparse_set_seq(-1,NULL,NULL,NULL,
656 tvbparse_char(-1,"[",NULL,NULL,NULL),
658 tvbparse_set_seq(-1,NULL,NULL,pop_stack,
660 tvbparse_set_oneof(-1,NULL,NULL,NULL,
661 tvbparse_casestring(-1,"PUBLIC",NULL,NULL,NULL),
662 tvbparse_casestring(-1,"SYSTEM",NULL,NULL,NULL),
664 tvbparse_until(-1,NULL,NULL,NULL,
665 tvbparse_char(-1,">",NULL,NULL,NULL),
671 tvbparse_wanted_t* want_dtd_tag = tvbparse_set_seq(hf_dtd_tag,NULL,NULL,after_token,
672 tvbparse_char(-1,"<",NULL,NULL,NULL),
673 tvbparse_char(-1,"!",NULL,NULL,NULL),
674 tvbparse_until(-1,NULL,NULL,NULL,
675 tvbparse_char(-1, ">", NULL, NULL, NULL),
679 tvbparse_wanted_t* want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
680 tvbparse_char(-1,"<",NULL,NULL,NULL),
682 tvbparse_set_oneof(-1,NULL,NULL,NULL,
683 tvbparse_set_seq(-1,NULL,NULL,NULL,
691 tvbparse_wanted_t* want_dtd_close = tvbparse_set_seq(-1,NULL,NULL,after_dtd_close,
692 tvbparse_char(-1,"]",NULL,NULL,NULL),
693 tvbparse_char(-1,">",NULL,NULL,NULL),
696 want_ignore = tvbparse_chars(-1,1,0," \t\r\n",NULL,NULL,NULL);
699 want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
707 tvbparse_not_chars(XML_CDATA,1,0,"<",NULL,NULL,after_token),
708 tvbparse_not_chars(-1,1,0," \t\r\n",NULL,NULL,unrecognized_token),
711 want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
722 static xml_ns_t* xml_new_namespace(GHashTable* hash, gchar* name, ...) {
723 xml_ns_t* ns = g_malloc(sizeof(xml_ns_t));
727 ns->name = g_strdup(name);
731 ns->attributes = g_hash_table_new(g_str_hash,g_str_equal);
732 ns->elements = g_hash_table_new(g_str_hash,g_str_equal);
736 while(( attr_name = va_arg(ap,gchar*) )) {
737 int* hfp = g_malloc(sizeof(int));
739 g_hash_table_insert(ns->attributes,g_strdup(attr_name),hfp);
744 g_hash_table_insert(hash,ns->name,ns);
750 static void add_xml_field(GArray* hfs, int* p_id, gchar* name, gchar* fqn) {
751 hf_register_info hfri;
754 hfri.hfinfo.name = name;
755 hfri.hfinfo.abbrev = fqn;
756 hfri.hfinfo.type = FT_STRING;
757 hfri.hfinfo.display = BASE_NONE;
758 hfri.hfinfo.strings = NULL;
759 hfri.hfinfo.bitmask = 0x0;
760 hfri.hfinfo.blurb = NULL;
762 hfri.hfinfo.parent = 0;
763 hfri.hfinfo.ref_count = HF_REF_TYPE_NONE;
764 hfri.hfinfo.bitshift = 0;
765 hfri.hfinfo.same_name_next = NULL;
766 hfri.hfinfo.same_name_prev = NULL;
768 g_array_append_val(hfs,hfri);
771 static void add_xml_attribute_names(gpointer k, gpointer v, gpointer p) {
772 struct _attr_reg_data* d = p;
773 gchar* basename = g_strdup_printf("%s.%s",d->basename,(gchar*)k);
774 add_xml_field(d->hf, (int*) v, (gchar*)k, basename);
778 static void add_xmlpi_namespace(gpointer k _U_, gpointer v, gpointer p) {
780 gchar* basename = g_strdup_printf("%s.%s",(gchar*)p,ns->name);
781 gint* ett_p = &(ns->ett);
782 struct _attr_reg_data d;
784 add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
786 g_array_append_val(ett_arr,ett_p);
788 d.basename = basename;
791 g_hash_table_foreach(ns->attributes,add_xml_attribute_names,&d);
795 static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
796 g_free(dtd_data->proto_name);
797 g_free(dtd_data->media_type);
798 g_free(dtd_data->description);
799 g_free(dtd_data->proto_root);
801 g_string_free(dtd_data->error,TRUE);
803 while(dtd_data->elements->len) {
804 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
805 g_ptr_array_free(nl->list,TRUE);
809 g_ptr_array_free(dtd_data->elements,TRUE);
811 while(dtd_data->attributes->len) {
812 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->attributes,0);
813 g_ptr_array_free(nl->list,TRUE);
817 g_ptr_array_free(dtd_data->attributes,TRUE);
822 static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
823 gchar* key = g_strdup(k);
824 int* value = g_malloc(sizeof(int));
828 g_hash_table_insert(dst,key,value);
832 static GHashTable* copy_attributes_hash(GHashTable* src) {
833 GHashTable* dst = g_hash_table_new(g_str_hash,g_str_equal);
835 g_hash_table_foreach(src,copy_attrib_item,dst);
840 static xml_ns_t* duplicate_element(xml_ns_t* orig) {
841 xml_ns_t* new_item = g_malloc(sizeof(xml_ns_t));
844 new_item->name = g_strdup(orig->name);
845 new_item->hf_tag = -1;
846 new_item->hf_cdata = -1;
848 new_item->attributes = copy_attributes_hash(orig->attributes);
849 new_item->elements = g_hash_table_new(g_str_hash,g_str_equal);
850 new_item->element_names = g_ptr_array_new();
852 for(i=0; i < orig->element_names->len; i++) {
853 g_ptr_array_add(new_item->element_names,
854 g_ptr_array_index(orig->element_names,i));
860 static gchar* fully_qualified_name(GPtrArray* hier, gchar* name, gchar* proto_name) {
862 GString* s = g_string_new(proto_name);
864 g_string_append(s,".");
866 for (i = 1; i < hier->len; i++) {
867 g_string_append_printf(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
870 g_string_append(s,name);
872 g_string_free(s,FALSE);
878 static xml_ns_t* make_xml_hier(gchar* elem_name,
880 GHashTable* elements,
890 struct _attr_reg_data d;
891 gboolean recurred = FALSE;
894 if ( g_str_equal(elem_name,root->name) ) {
898 if (! ( orig = g_hash_table_lookup(elements,elem_name) )) {
899 g_string_append_printf(error,"element '%s' is not defined\n", elem_name);
903 for (i = 0; i < hier->len; i++) {
904 if( strcmp(elem_name,(gchar*) g_ptr_array_index(hier,i) ) == 0 ) {
913 fqn = fully_qualified_name(hier,elem_name,proto_name);
915 new = duplicate_element(orig);
918 add_xml_field(hfs, &(new->hf_tag), g_strdup(elem_name), fqn);
919 add_xml_field(hfs, &(new->hf_cdata), g_strdup(elem_name), fqn);
922 g_array_append_val(etts,ett_p);
927 g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
929 while(new->element_names->len) {
930 gchar* child_name = g_ptr_array_remove_index(new->element_names,0);
931 xml_ns_t* child_element = NULL;
933 g_ptr_array_add(hier,elem_name);
934 child_element = make_xml_hier(child_name, root, elements, hier,error,hfs,etts,proto_name);
935 g_ptr_array_remove_index_fast(hier,hier->len - 1);
938 g_hash_table_insert(new->elements,child_element->name,child_element);
942 g_ptr_array_free(new->element_names,TRUE);
943 new->element_names = NULL;
947 static gboolean free_both(gpointer k, gpointer v, gpointer p _U_) {
953 static gboolean free_elements(gpointer k _U_, gpointer v, gpointer p _U_) {
956 g_hash_table_foreach_remove(e->attributes,free_both,NULL);
957 g_hash_table_destroy(e->attributes);
958 g_hash_table_destroy(e->elements);
960 while (e->element_names->len) {
961 g_free(g_ptr_array_remove_index(e->element_names,0));
964 g_ptr_array_free(e->element_names,TRUE);
970 static void register_dtd(dtd_build_data_t* dtd_data, GString* errors) {
971 GHashTable* elements = g_hash_table_new(g_str_hash,g_str_equal);
972 gchar* root_name = NULL;
973 xml_ns_t* root_element = NULL;
978 GPtrArray* element_names = g_ptr_array_new();
980 /* we first populate elements with the those coming from the parser */
981 while(dtd_data->elements->len) {
982 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->elements,0);
983 xml_ns_t* element = g_malloc(sizeof(xml_ns_t));
985 /* we will use the first element found as root in case no other one was given. */
986 if (root_name == NULL)
987 root_name = g_strdup(nl->name);
989 element->name = nl->name;
990 element->element_names = nl->list;
991 element->hf_tag = -1;
992 element->hf_cdata = -1;
994 element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
995 element->elements = g_hash_table_new(g_str_hash,g_str_equal);
997 if( g_hash_table_lookup(elements,element->name) ) {
998 g_string_append_printf(errors,"element %s defined more than once\n", element->name);
999 free_elements(NULL,element,NULL);
1001 g_hash_table_insert(elements,element->name,element);
1002 g_ptr_array_add(element_names,g_strdup(element->name));
1008 /* then we add the attributes to its relative elements */
1009 while(dtd_data->attributes->len) {
1010 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
1011 xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
1014 g_string_append_printf(errors,"element %s is not defined\n", nl->name);
1016 goto next_attribute;
1019 while(nl->list->len) {
1020 gchar* name = g_ptr_array_remove_index(nl->list,0);
1021 int* id_p = g_malloc(sizeof(int));
1024 g_hash_table_insert(element->attributes,name,id_p);
1029 g_ptr_array_free(nl->list,TRUE);
1033 /* if a proto_root is defined in the dtd we'll use that as root */
1034 if( dtd_data->proto_root ) {
1036 root_name = g_strdup(dtd_data->proto_root);
1039 /* we use a stack with the names to avoid recurring infinitelly */
1040 hier = g_ptr_array_new();
1043 * if a proto name was given in the dtd the dtd will be used as a protocol
1044 * or else the dtd will be loaded as a branch of the xml namespace
1046 if( ! dtd_data->proto_name ) {
1049 g_ptr_array_add(hier,g_strdup("xml"));
1050 root_element = &xml_ns;
1053 * if we were given a proto_name the namespace will be registered
1054 * as an independent protocol with its own hf and ett arrays.
1056 hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1057 etts = g_array_new(FALSE,FALSE,sizeof(gint*));
1060 /* the root element of the dtd's namespace */
1061 root_element = g_malloc(sizeof(xml_ns_t));
1062 root_element->name = g_strdup(root_name);
1063 root_element->fqn = dtd_data->proto_name ? g_strdup(dtd_data->proto_name) : root_element->name;
1064 root_element->hf_tag = -1;
1065 root_element->hf_cdata = -1;
1066 root_element->ett = -1;
1067 root_element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1068 root_element->element_names = element_names;
1071 * we can either create a namespace as a flat namespace
1072 * in which all the elements are at the root level
1073 * or we can create a recursive namespace
1075 if (dtd_data->recursion) {
1076 xml_ns_t* orig_root;
1078 make_xml_hier(root_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1080 g_hash_table_insert(root_element->elements,root_element->name,root_element);
1082 orig_root = g_hash_table_lookup(elements,root_name);
1084 /* if the root element was defined copy its attrlist to the child */
1086 struct _attr_reg_data d;
1088 d.basename = dtd_data->proto_name;
1091 root_element->attributes = copy_attributes_hash(orig_root->attributes);
1092 g_hash_table_foreach(root_element->attributes,add_xml_attribute_names,&d);
1094 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1097 /* we then create all the sub hierachies to catch the recurred cases */
1098 g_ptr_array_add(hier,root_name);
1100 while(root_element->element_names->len) {
1101 curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1103 if( ! g_hash_table_lookup(root_element->elements,curr_name) ) {
1104 xml_ns_t* new = make_xml_hier(curr_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1105 g_hash_table_insert(root_element->elements,new->name,new);
1112 /* a flat namespace */
1113 g_ptr_array_add(hier,root_name);
1115 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1117 while(root_element->element_names->len) {
1120 struct _attr_reg_data d;
1122 curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1123 new = duplicate_element(g_hash_table_lookup(elements,curr_name));
1124 new->fqn = fully_qualified_name(hier, curr_name, root_name);
1126 add_xml_field(hfs, &(new->hf_tag), curr_name, new->fqn);
1127 add_xml_field(hfs, &(new->hf_cdata), curr_name, new->fqn);
1129 d.basename = new->fqn;
1132 g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
1135 g_array_append_val(etts,ett_p);
1137 g_ptr_array_free(new->element_names,TRUE);
1139 g_hash_table_insert(root_element->elements,new->name,new);
1143 g_ptr_array_free(element_names,TRUE);
1145 g_ptr_array_free(hier,TRUE);
1148 * if we were given a proto_name the namespace will be registered
1149 * as an independent protocol.
1151 if( dtd_data->proto_name ) {
1154 if ( ! dtd_data->description) {
1155 dtd_data->description = g_strdup(root_name);
1158 ett_p = &root_element->ett;
1159 g_array_append_val(etts,ett_p);
1161 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1163 root_element->hf_tag = proto_register_protocol(dtd_data->description, dtd_data->proto_name, dtd_data->proto_name);
1164 proto_register_field_array(root_element->hf_tag, (hf_register_info*)g_array_data(hfs), hfs->len);
1165 proto_register_subtree_array((gint**)g_array_data(etts), etts->len);
1167 if (dtd_data->media_type) {
1168 g_hash_table_insert(media_types,dtd_data->media_type,root_element);
1169 dtd_data->media_type = NULL;
1172 dtd_data->description = NULL;
1173 dtd_data->proto_name = NULL;
1174 g_array_free(hfs,FALSE);
1175 g_array_free(etts,TRUE);
1178 g_hash_table_insert(xml_ns.elements,root_element->name,root_element);
1180 g_hash_table_foreach_remove(elements,free_elements,NULL);
1181 g_hash_table_destroy(elements);
1183 destroy_dtd_data(dtd_data);
1187 # define DIRECTORY_T GDir
1188 # define FILE_T gchar
1189 # define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1190 # define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1191 # define GETFNAME_OP(file) (file);
1192 # define CLOSEDIR_OP(dir) g_dir_close(dir)
1194 static void init_xml_names(void) {
1195 xml_ns_t* xmlpi_xml_ns;
1199 const gchar* filename;
1202 GError** dummy = g_malloc(sizeof(GError *));
1205 xmpli_names = g_hash_table_new(g_str_hash,g_str_equal);
1206 media_types = g_hash_table_new(g_str_hash,g_str_equal);
1208 unknown_ns.elements = xml_ns.elements = g_hash_table_new(g_str_hash,g_str_equal);
1209 unknown_ns.attributes = xml_ns.attributes = g_hash_table_new(g_str_hash,g_str_equal);
1211 xmlpi_xml_ns = xml_new_namespace(xmpli_names,"xml","version","encoding","standalone",NULL);
1213 g_hash_table_destroy(xmlpi_xml_ns->elements);
1214 xmlpi_xml_ns->elements = NULL;
1217 dirname = get_persconffile_path("dtds", FALSE, FALSE);
1219 if (test_for_directory(dirname) != EISDIR) {
1220 /* Although dir isn't a directory it may still use memory */
1222 dirname = get_datafile_path("dtds");
1225 if (test_for_directory(dirname) == EISDIR) {
1227 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1228 while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1230 filename = GETFNAME_OP(file);
1232 namelen = (int)strlen(filename);
1233 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4),".dtd") == 0 ) ) {
1234 GString* errors = g_string_new("");
1235 GString* preparsed = dtd_preparse(dirname, filename, errors);
1236 dtd_build_data_t* dtd_data;
1239 report_failure("Dtd Preparser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1243 dtd_data = dtd_parse(preparsed);
1245 g_string_free(preparsed,TRUE);
1247 if (dtd_data->error->len) {
1248 report_failure("Dtd Parser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,dtd_data->error->str);
1249 destroy_dtd_data(dtd_data);
1253 register_dtd(dtd_data,errors);
1256 report_failure("Dtd Registration in file: %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1257 g_string_free(errors,TRUE);
1269 for(i=0;i<array_length(default_media_types);i++) {
1270 if( ! g_hash_table_lookup(media_types,default_media_types[i]) ) {
1271 g_hash_table_insert(media_types,(gpointer)default_media_types[i],&xml_ns);
1275 g_hash_table_foreach(xmpli_names,add_xmlpi_namespace,"xml.xmlpi");
1280 static void range_delete_xml_tcp_callback(guint32 port) {
1281 dissector_delete("tcp.port", port, xml_handle);
1284 static void range_add_xml_tcp_callback(guint32 port) {
1285 dissector_add("tcp.port", port, xml_handle);
1288 static void apply_prefs(void) {
1289 if (pref_heuristic_media_save != pref_heuristic_media) {
1290 if (pref_heuristic_media) {
1291 heur_dissector_add("http", dissect_xml_heur, xml_ns.hf_tag);
1292 heur_dissector_add("sip", dissect_xml_heur, xml_ns.hf_tag);
1293 heur_dissector_add("media", dissect_xml_heur, xml_ns.hf_tag);
1294 pref_heuristic_media_save = TRUE;
1296 heur_dissector_delete("http", dissect_xml_heur, xml_ns.hf_tag);
1297 heur_dissector_delete("sip", dissect_xml_heur, xml_ns.hf_tag);
1298 heur_dissector_delete("media", dissect_xml_heur, xml_ns.hf_tag);
1299 pref_heuristic_media_save = FALSE;
1303 if (pref_heuristic_tcp_save != pref_heuristic_tcp ) {
1304 if (pref_heuristic_tcp) {
1305 heur_dissector_add("tcp", dissect_xml_heur, xml_ns.hf_tag);
1306 pref_heuristic_tcp_save = TRUE;
1308 heur_dissector_delete("tcp", dissect_xml_heur, xml_ns.hf_tag);
1309 pref_heuristic_tcp_save = FALSE;
1313 if (pref_heuristic_udp_save != pref_heuristic_udp ) {
1314 if (pref_heuristic_udp) {
1315 heur_dissector_add("udp", dissect_xml_heur, xml_ns.hf_tag);
1316 pref_heuristic_udp_save = TRUE;
1318 heur_dissector_delete("udp", dissect_xml_heur, xml_ns.hf_tag);
1319 pref_heuristic_udp_save = FALSE;
1323 range_foreach(xml_tcp_range, range_delete_xml_tcp_callback);
1324 g_free(xml_tcp_range);
1325 xml_tcp_range = range_copy(global_xml_tcp_range);
1326 range_foreach(xml_tcp_range, range_add_xml_tcp_callback);
1330 proto_register_xml(void) {
1331 static gint *ett_base[] = {
1338 static hf_register_info hf_base[] = {
1339 { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1340 { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1341 { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1342 { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1343 { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1344 { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1345 { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1346 { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }}
1348 module_t* xml_module;
1350 hf_arr = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1351 ett_arr = g_array_new(FALSE,FALSE,sizeof(gint*));
1353 g_array_append_vals(hf_arr,hf_base,array_length(hf_base));
1354 g_array_append_vals(ett_arr,ett_base,array_length(ett_base));
1358 xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1360 proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)g_array_data(hf_arr), hf_arr->len);
1361 proto_register_subtree_array((gint**)g_array_data(ett_arr), ett_arr->len);
1363 xml_module = prefs_register_protocol(xml_ns.hf_tag,apply_prefs);
1364 prefs_register_bool_preference(xml_module, "heuristic", "Use Heuristics for media types",
1365 "Try to recognize XML for unknown media types",
1366 &pref_heuristic_media);
1367 prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for tcp",
1368 "Try to recognize XML for unknown TCP ports",
1369 &pref_heuristic_tcp);
1370 prefs_register_range_preference(xml_module, "tcp.port", "TCP Ports",
1372 &global_xml_tcp_range, 65535);
1373 prefs_register_bool_preference(xml_module, "heuristic_udp", "Use Heuristics for UDP",
1374 "Try to recognize XML for unknown UDP ports",
1375 &pref_heuristic_udp);
1377 g_array_free(hf_arr,FALSE);
1378 g_array_free(ett_arr,TRUE);
1380 register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1384 xml_tcp_range = range_empty();
1389 static void add_dissector_media(gpointer k, gpointer v _U_, gpointer p _U_) {
1390 dissector_add_string("media_type", (gchar*)k, xml_handle);
1394 proto_reg_handoff_xml(void)
1397 xml_handle = find_dissector("xml");
1399 g_hash_table_foreach(media_types,add_dissector_media,NULL);