2 * ethereal'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 * Ethereal - Network traffic analyzer
12 * By Gerald Combs <gerald@ethereal.com>
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.
47 #include <epan/emem.h>
48 #include <epan/packet.h>
49 #include <epan/strutil.h>
50 #include <epan/tvbparse.h>
52 #include <epan/report_err.h>
53 #include <epan/filesystem.h>
55 typedef struct _xml_ns_t {
56 /* the name of this namespace */
59 /* its fully qualified name */
62 /* the contents of the whole element from <> to </> */
65 /* chunks of cdata from <> to </> excluding sub tags */
68 /* the subtree for its sub items */
71 GHashTable* attributes;
72 /* key: the attribute name
73 value: hf_id of what's between quotes */
75 /* the namespace's namespaces */
77 /* key: the element name
78 value: the child namespace */
80 GPtrArray* element_names;
81 /* imported directly from the parser and used while building the namespace */
88 proto_item* last_item;
93 struct _attr_reg_data {
99 static gint ett_dtd = -1;
100 static gint ett_xmpli = -1;
102 static int hf_unknowwn_attrib = -1;
103 static int hf_comment = -1;
104 static int hf_xmlpi = -1;
105 static int hf_dtd_tag = -1;
106 static int hf_doctype = -1;
108 /* dissector handles */
109 static dissector_handle_t xml_handle;
111 /* parser definitions */
112 static tvbparse_wanted_t* want;
113 static tvbparse_wanted_t* want_ignore;
115 static GHashTable* xmpli_names;
116 static GHashTable* media_types;
118 static xml_ns_t xml_ns = {"xml","/",-1,-1,-1,NULL,NULL,NULL};
119 static xml_ns_t unknown_ns = {"unknown","?",-1,-1,-1,NULL,NULL,NULL};
120 static xml_ns_t* root_ns;
122 #define XML_CDATA -1000
123 #define XML_SCOPED_NAME -1001
129 static const gchar* default_media_types[] = {
132 "application/soap+xml",
133 "application/xml-dtd",
139 "application/vnd.wv.csp+xml",
140 "application/xcap-el+xml",
141 "application/xcap-att+xml",
142 "application/xcap-error+xml",
143 "application/xcap-caps+xml",
144 "application/auth-policy+xml",
146 "application/cpim-pidf+xml",
147 "application/rdf+xml",
148 "application/xslt+xml",
149 "application/mathml+xml",
151 "application/vnd.wv.csp.xml",
156 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree)
159 tvbparse_elem_t* tok = NULL;
160 static GPtrArray* stack = NULL;
161 xml_frame_t* current_frame;
166 g_ptr_array_free(stack,TRUE);
168 stack = g_ptr_array_new();
169 current_frame = ep_alloc(sizeof(xml_frame_t));
170 g_ptr_array_add(stack,current_frame);
172 tt = tvbparse_init(tvb,0,-1,stack,want_ignore);
173 current_frame->start_offset = 0;
177 if (pinfo->match_string)
178 root_ns = g_hash_table_lookup(media_types,pinfo->match_string);
184 current_frame->ns = root_ns;
186 current_frame->item = proto_tree_add_item(tree,current_frame->ns->hf_tag,tvb,0,-1,FALSE);
187 current_frame->tree = proto_item_add_subtree(current_frame->item,current_frame->ns->ett);
188 current_frame->last_item = current_frame->item;
190 while(( tok = tvbparse_get(tt, want) )) ;
194 static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
195 GPtrArray* stack = tvbparse_data;
196 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
200 if (tok->id == XML_CDATA) {
201 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
202 } else if ( tok->id > 0) {
205 hfid = xml_ns.hf_cdata;
208 pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, FALSE);
210 proto_item_set_text(pi, "%s",
211 tvb_format_text(tok->tvb,tok->offset,tok->len));
214 static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
215 GPtrArray* stack = tvbparse_data;
216 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
219 tvbparse_elem_t* name_tok = tok->sub->next;
220 gchar* name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
221 xml_ns_t* ns = g_hash_table_lookup(xmpli_names,name);
235 pi = proto_tree_add_item(current_frame->tree,hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
237 proto_item_set_text(pi,tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
239 pt = proto_item_add_subtree(pi,ett);
241 current_frame = ep_alloc(sizeof(xml_frame_t));
242 current_frame->item = pi;
243 current_frame->last_item = pi;
244 current_frame->tree = pt;
245 current_frame->start_offset = tok->offset;
246 current_frame->ns = ns;
248 g_ptr_array_add(stack,current_frame);
252 static void after_xmlpi(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
253 GPtrArray* stack = tvbparse_data;
254 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
256 proto_tree_add_text(current_frame->tree,
257 tok->tvb, tok->offset, tok->len,
258 tvb_format_text(tok->tvb,tok->offset,tok->len));
260 if (stack->len > 1) {
261 g_ptr_array_remove_index_fast(stack,stack->len - 1);
263 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened xmpli tag ]");
267 static void before_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
268 GPtrArray* stack = tvbparse_data;
269 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
270 tvbparse_elem_t* name_tok = tok->sub->next;
274 xml_frame_t* new_frame;
278 if (name_tok->sub->id == XML_SCOPED_NAME) {
279 tvbparse_elem_t* root_tok = name_tok->sub->sub;
280 tvbparse_elem_t* leaf_tok = name_tok->sub->sub->next->next;
281 xml_ns_t* nameroot_ns;
283 root_name = tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
284 name = tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
286 nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
289 ns = g_hash_table_lookup(nameroot_ns->elements,name);
298 name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
301 if(current_frame->ns) {
302 ns = g_hash_table_lookup(current_frame->ns->elements,name);
305 if (! ( ns = g_hash_table_lookup(root_ns->elements,name) ) ) {
314 pi = proto_tree_add_item(current_frame->tree,ns->hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
315 proto_item_set_text(pi,tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
317 pt = proto_item_add_subtree(pi,ns->ett);
319 new_frame = ep_alloc(sizeof(xml_frame_t));
320 new_frame->item = pi;
321 new_frame->last_item = pi;
322 new_frame->tree = pt;
323 new_frame->start_offset = tok->offset;
326 g_ptr_array_add(stack,new_frame);
330 static void after_open_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
331 GPtrArray* stack = tvbparse_data;
332 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
334 proto_item_append_text(current_frame->last_item,">");
337 static void after_closed_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
338 GPtrArray* stack = tvbparse_data;
339 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
341 proto_item_append_text(current_frame->last_item,"/>");
343 if (stack->len > 1) {
344 g_ptr_array_remove_index_fast(stack,stack->len - 1);
346 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
350 void after_untag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
351 GPtrArray* stack = tvbparse_data;
352 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
354 proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
356 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
357 tvb_format_text(tok->tvb,tok->offset,tok->len));
359 if (stack->len > 1) {
360 g_ptr_array_remove_index_fast(stack,stack->len - 1);
362 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
363 "[ ERROR: Closing an unopened tag ]");
367 static void before_dtd_doctype(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
368 GPtrArray* stack = tvbparse_data;
369 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
370 tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
371 proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
372 name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
374 proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
376 current_frame = ep_alloc(sizeof(xml_frame_t));
377 current_frame->item = dtd_item;
378 current_frame->last_item = dtd_item;
379 current_frame->tree = proto_item_add_subtree(dtd_item,ett_dtd);
380 current_frame->start_offset = tok->offset;
381 current_frame->ns = NULL;
383 g_ptr_array_add(stack,current_frame);
386 static void pop_stack(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
387 GPtrArray* stack = tvbparse_data;
388 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
390 if (stack->len > 1) {
391 g_ptr_array_remove_index_fast(stack,stack->len - 1);
393 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
394 "[ ERROR: Closing an unopened tag ]");
398 static void after_dtd_close(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
399 GPtrArray* stack = tvbparse_data;
400 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
402 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
403 tvb_format_text(tok->tvb,tok->offset,tok->len));
404 if (stack->len > 1) {
405 g_ptr_array_remove_index_fast(stack,stack->len - 1);
407 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
411 static void get_attrib_value(void* tvbparse_data _U_, const void* wanted_data _U_, tvbparse_elem_t* tok) {
412 tok->data = tok->sub;
415 static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
416 GPtrArray* stack = tvbparse_data;
417 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
418 gchar* name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
419 tvbparse_elem_t* value = tok->sub->next->next->data;
424 if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
427 hfid = hf_unknowwn_attrib;
431 current_frame->last_item = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
432 proto_item_set_text(current_frame->last_item, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
436 static void unrecognized_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_){
437 GPtrArray* stack = tvbparse_data;
438 xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
440 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Unrecognized text ]");
446 void init_xml_parser(void) {
447 tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ",NULL,NULL,NULL);
448 tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz-_ABCDEFGHIJKLMNOPQRSTUVWXYZ:",NULL,NULL,NULL);
450 tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
452 tvbparse_char(-1,":",NULL,NULL,NULL),
456 tvbparse_wanted_t* want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
461 tvbparse_wanted_t* want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
462 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\"','\\'),
463 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\'','\\'),
464 tvbparse_chars(-1,1,0,"0123456789",NULL,NULL,NULL),
468 tvbparse_wanted_t* want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
469 tvbparse_set_seq(-1, NULL, NULL, after_attrib,
471 tvbparse_char(-1,"=",NULL,NULL,NULL),
475 tvbparse_wanted_t* want_stoptag = tvbparse_set_oneof(-1,NULL,NULL,NULL,
476 tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
477 tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
480 tvbparse_wanted_t* want_stopxmlpi = tvbparse_string(-1,"?>",NULL,NULL,after_xmlpi);
482 tvbparse_wanted_t* want_comment = tvbparse_set_seq(hf_comment,NULL,NULL,after_token,
483 tvbparse_string(-1,"<!--",NULL,NULL,NULL),
484 tvbparse_until(-1,NULL,NULL,NULL,
485 tvbparse_string(-1,"-->",NULL,NULL,NULL),
489 tvbparse_wanted_t* want_xmlpi = tvbparse_set_seq(hf_xmlpi,NULL,before_xmpli,NULL,
490 tvbparse_string(-1,"<?",NULL,NULL,NULL),
492 tvbparse_set_oneof(-1,NULL,NULL,NULL,
494 tvbparse_set_seq(-1,NULL,NULL,NULL,
501 tvbparse_wanted_t* want_closing_tag = tvbparse_set_seq(0,NULL,NULL,after_untag,
502 tvbparse_char(-1, "<", NULL, NULL, NULL),
503 tvbparse_char(-1, "/", NULL, NULL, NULL),
505 tvbparse_char(-1, ">", NULL, NULL, NULL),
508 tvbparse_wanted_t* want_doctype_start = tvbparse_set_seq(-1,NULL,before_dtd_doctype,NULL,
509 tvbparse_char(-1,"<",NULL,NULL,NULL),
510 tvbparse_char(-1,"!",NULL,NULL,NULL),
511 tvbparse_casestring(-1,"DOCTYPE",NULL,NULL,NULL),
512 tvbparse_set_oneof(-1,NULL,NULL,NULL,
513 tvbparse_set_seq(-1,NULL,NULL,NULL,
515 tvbparse_char(-1,"[",NULL,NULL,NULL),
517 tvbparse_set_seq(-1,NULL,NULL,pop_stack,
519 tvbparse_set_oneof(-1,NULL,NULL,NULL,
520 tvbparse_casestring(-1,"PUBLIC",NULL,NULL,NULL),
521 tvbparse_casestring(-1,"SYSTEM",NULL,NULL,NULL),
523 tvbparse_until(-1,NULL,NULL,NULL,
524 tvbparse_char(-1,">",NULL,NULL,NULL),
530 tvbparse_wanted_t* want_dtd_tag = tvbparse_set_seq(hf_dtd_tag,NULL,NULL,after_token,
531 tvbparse_char(-1,"<",NULL,NULL,NULL),
532 tvbparse_char(-1,"!",NULL,NULL,NULL),
533 tvbparse_until(-1,NULL,NULL,NULL,
534 tvbparse_char(-1, ">", NULL, NULL, NULL),
538 tvbparse_wanted_t* want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
539 tvbparse_char(-1,"<",NULL,NULL,NULL),
541 tvbparse_set_oneof(-1,NULL,NULL,NULL,
542 tvbparse_set_seq(-1,NULL,NULL,NULL,
550 tvbparse_wanted_t* want_dtd_close = tvbparse_set_seq(-1,NULL,NULL,after_dtd_close,
551 tvbparse_char(-1,"]",NULL,NULL,NULL),
552 tvbparse_char(-1,">",NULL,NULL,NULL),
555 want_ignore = tvbparse_chars(-1,1,0," \t\r\n",NULL,NULL,NULL);
558 want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
566 tvbparse_not_chars(XML_CDATA,1,0,"<",NULL,NULL,after_token),
567 tvbparse_not_chars(-1,1,0," \t\r\n",NULL,NULL,unrecognized_token),
574 xml_ns_t* xml_new_namespace(GHashTable* hash, gchar* name, ...) {
575 xml_ns_t* ns = g_malloc(sizeof(xml_ns_t));
579 ns->name = g_strdup(name);
583 ns->attributes = g_hash_table_new(g_str_hash,g_str_equal);
584 ns->elements = g_hash_table_new(g_str_hash,g_str_equal);
588 while(( attr_name = va_arg(ap,gchar*) )) {
589 int* hfp = g_malloc(sizeof(int));
591 g_hash_table_insert(ns->attributes,g_strdup(attr_name),hfp);
596 g_hash_table_insert(hash,ns->name,ns);
602 void add_xml_field(GArray* hfs, int* p_id, gchar* name, gchar* fqn) {
603 hf_register_info hfri;
606 hfri.hfinfo.name = name;
607 hfri.hfinfo.abbrev = fqn;
608 hfri.hfinfo.type = FT_STRING;
609 hfri.hfinfo.display = BASE_NONE;
610 hfri.hfinfo.strings = NULL;
611 hfri.hfinfo.bitmask = 0x0;
612 hfri.hfinfo.blurb = "";
614 hfri.hfinfo.parent = 0;
615 hfri.hfinfo.ref_count = 0;
616 hfri.hfinfo.bitshift = 0;
617 hfri.hfinfo.same_name_next = NULL;
618 hfri.hfinfo.same_name_prev = NULL;
620 g_array_append_val(hfs,hfri);
623 void add_xml_attribute_names(gpointer k, gpointer v, gpointer p) {
624 struct _attr_reg_data* d = p;
625 gchar* basename = g_strdup_printf("%s.%s",d->basename,(gchar*)k);
626 add_xml_field(d->hf, (int*) v, (gchar*)k, basename);
630 void add_xmlpi_namespace(gpointer k _U_, gpointer v, gpointer p) {
632 gchar* basename = g_strdup_printf("%s.%s",(gchar*)p,ns->name);
633 gint* ett_p = &(ns->ett);
634 struct _attr_reg_data d;
636 add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
638 g_array_append_val(ett_arr,ett_p);
640 d.basename = basename;
643 g_hash_table_foreach(ns->attributes,add_xml_attribute_names,&d);
647 static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
649 if(dtd_data->proto_name) g_free(dtd_data->proto_name);
650 if(dtd_data->media_type) g_free(dtd_data->media_type);
651 if(dtd_data->description) g_free(dtd_data->description);
652 if(dtd_data->proto_root) g_free(dtd_data->proto_root);
654 g_string_free(dtd_data->error,TRUE);
657 while(dtd_data->elements->len) {
658 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
659 g_ptr_array_free(nl->list,TRUE);
663 g_ptr_array_free(dtd_data->elements,TRUE);
665 while(dtd_data->attributes->len) {
666 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
667 g_ptr_array_free(nl->list,TRUE);
671 g_ptr_array_free(dtd_data->attributes,TRUE);
678 static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
679 gchar* key = g_strdup(k);
680 int* value = g_malloc(sizeof(int));
684 g_hash_table_insert(dst,key,value);
688 static GHashTable* copy_attributes_hash(GHashTable* src) {
689 GHashTable* dst = g_hash_table_new(g_str_hash,g_str_equal);
691 g_hash_table_foreach(src,copy_attrib_item,dst);
696 static xml_ns_t* duplicate_element(xml_ns_t* orig) {
697 xml_ns_t* new_item = g_malloc(sizeof(xml_ns_t));
700 new_item->name = g_strdup(orig->name);
701 new_item->hf_tag = -1;
702 new_item->hf_cdata = -1;
704 new_item->attributes = copy_attributes_hash(orig->attributes);
705 new_item->elements = g_hash_table_new(g_str_hash,g_str_equal);
706 new_item->element_names = g_ptr_array_new();
708 for(i=0; i < orig->element_names->len; i++) {
709 g_ptr_array_add(new_item->element_names,
710 g_ptr_array_index(orig->element_names,i));
716 static gchar* fully_qualified_name(GPtrArray* hier, gchar* name) {
718 GString* s = g_string_new("");
721 for (i = 0; i < hier->len; i++) {
722 g_string_sprintfa(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
725 g_string_append(s,name);
727 g_string_free(s,FALSE);
733 static xml_ns_t* make_xml_hier(gchar* elem_name,
735 GHashTable* elements,
744 struct _attr_reg_data d;
745 gboolean recurred = FALSE;
748 if ( g_str_equal(elem_name,root->name) ) {
752 if (! ( orig = g_hash_table_lookup(elements,elem_name) )) {
753 g_string_sprintfa(error,"element '%s' is not defined\n", elem_name);
757 for (i = 0; i < hier->len; i++) {
758 if( strcmp(elem_name,(gchar*) g_ptr_array_index(hier,i) ) == 0 ) {
767 fqn = fully_qualified_name(hier,elem_name);
769 new = duplicate_element(orig);
772 add_xml_field(hfs, &(new->hf_tag), g_strdup(elem_name), fqn);
773 add_xml_field(hfs, &(new->hf_cdata), g_strdup(elem_name), fqn);
776 g_array_append_val(etts,ett_p);
781 g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
783 while(new->element_names->len) {
784 gchar* child_name = g_ptr_array_remove_index(new->element_names,0);
785 xml_ns_t* child_element = NULL;
787 g_ptr_array_add(hier,elem_name);
788 child_element = make_xml_hier(child_name, root, elements, hier,error,hfs,etts);
789 g_ptr_array_remove_index_fast(hier,hier->len - 1);
792 g_hash_table_insert(new->elements,child_element->name,child_element);
796 g_ptr_array_free(new->element_names,TRUE);
797 new->element_names = NULL;
801 static gboolean free_both(gpointer k, gpointer v, gpointer p _U_) {
807 static gboolean free_elements(gpointer k _U_, gpointer v, gpointer p _U_) {
810 g_hash_table_foreach_remove(e->attributes,free_both,NULL);
811 g_hash_table_destroy(e->attributes);
812 g_hash_table_destroy(e->elements);
814 while (e->element_names->len) {
815 g_free(g_ptr_array_remove_index(e->element_names,0));
818 g_ptr_array_free(e->element_names,TRUE);
824 static void register_dtd(dtd_build_data_t* dtd_data, GString* errors) {
825 GHashTable* elements = g_hash_table_new(g_str_hash,g_str_equal);
826 gchar* root_name = NULL;
827 xml_ns_t* root_element = NULL;
832 GPtrArray* element_names = g_ptr_array_new();
834 /* we first populate elements with the those coming from the parser */
835 while(dtd_data->elements->len) {
836 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->elements,0);
837 xml_ns_t* element = g_malloc(sizeof(xml_ns_t));
839 /* we will use the first element found as root in case no other one was given. */
840 if (root_name == NULL)
841 root_name = g_strdup(nl->name);
843 element->name = nl->name;
844 element->element_names = nl->list;
845 element->hf_tag = -1;
846 element->hf_cdata = -1;
848 element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
849 element->elements = g_hash_table_new(g_str_hash,g_str_equal);
851 if( g_hash_table_lookup(elements,element->name) ) {
852 g_string_sprintfa(errors,"element %s defined more than once\n", element->name);
853 free_elements(NULL,element,NULL);
855 g_hash_table_insert(elements,element->name,element);
856 g_ptr_array_add(element_names,g_strdup(element->name));
862 /* then we add the attributes to its relative elements */
863 while(dtd_data->attributes->len) {
864 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
865 xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
868 g_string_sprintfa(errors,"element %s is not defined\n", nl->name);
873 while(nl->list->len) {
874 gchar* name = g_ptr_array_remove_index(nl->list,0);
875 int* id_p = g_malloc(sizeof(int));
878 g_hash_table_insert(element->attributes,name,id_p);
883 g_ptr_array_free(nl->list,TRUE);
887 /* if a proto_root is defined in the dtd we'll use that as root */
888 if( dtd_data->proto_root ) {
891 root_name = g_strdup(dtd_data->proto_root);
894 /* we use a stack with the names to avoid recurring infinitelly */
895 hier = g_ptr_array_new();
898 * if a proto name was given in the dtd the dtd will be used as a protocol
899 * or else the dtd will be loaded as a branch of the xml namespace
901 if( ! dtd_data->proto_name ) {
904 g_ptr_array_add(hier,g_strdup("xml"));
905 root_element = &xml_ns;
908 * if we were given a proto_name the namespace will be registered
909 * as an indipendent protocol with its own hf and ett arrays.
911 hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
912 etts = g_array_new(FALSE,FALSE,sizeof(gint*));
915 /* the root element of the dtd's namespace */
916 root_element = g_malloc(sizeof(xml_ns_t));
917 root_element->name = g_strdup(root_name);
918 root_element->fqn = root_element->name;
919 root_element->hf_tag = -1;
920 root_element->hf_cdata = -1;
921 root_element->ett = -1;
922 root_element->elements = g_hash_table_new(g_str_hash,g_str_equal);
923 root_element->element_names = element_names;
926 * we can either create a namespace as a flat namespace
927 * in which all the elements are at the root level
928 * or we can create a recursive namespace
930 if (dtd_data->recursion) {
933 make_xml_hier(root_name, root_element, elements,hier,errors,hfs,etts);
935 g_hash_table_insert(root_element->elements,root_element->name,root_element);
937 orig_root = g_hash_table_lookup(elements,root_name);
939 /* if the root element was defined copy its attrlist to the child */
941 struct _attr_reg_data d;
943 d.basename = root_name;
946 root_element->attributes = copy_attributes_hash(orig_root->attributes);
947 g_hash_table_foreach(root_element->attributes,add_xml_attribute_names,&d);
949 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
952 /* we then create all the sub hierachies to catch the recurred cases */
953 g_ptr_array_add(hier,root_name);
955 while(root_element->element_names->len) {
956 curr_name = g_ptr_array_remove_index(root_element->element_names,0);
958 if( ! g_hash_table_lookup(root_element->elements,curr_name) ) {
959 xml_ns_t* new = make_xml_hier(curr_name, root_element, elements,hier,errors,hfs,etts);
960 g_hash_table_insert(root_element->elements,new->name,new);
967 /* a flat namespace */
968 g_ptr_array_add(hier,root_name);
970 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
972 while(root_element->element_names->len) {
975 struct _attr_reg_data d;
977 curr_name = g_ptr_array_remove_index(root_element->element_names,0);
978 new = duplicate_element(g_hash_table_lookup(elements,curr_name));
979 new->fqn = fully_qualified_name(hier, curr_name);
981 add_xml_field(hfs, &(new->hf_tag), curr_name, new->fqn);
982 add_xml_field(hfs, &(new->hf_cdata), curr_name, new->fqn);
984 d.basename = new->fqn;
987 g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
990 g_array_append_val(etts,ett_p);
992 g_ptr_array_free(new->element_names,TRUE);
994 g_hash_table_insert(root_element->elements,new->name,new);
998 g_ptr_array_free(element_names,TRUE);
1000 g_ptr_array_free(hier,TRUE);
1003 * if we were given a proto_name the namespace will be registered
1004 * as an indipendent protocol.
1006 if( dtd_data->proto_name ) {
1009 if ( ! dtd_data->description) {
1010 dtd_data->description = g_strdup(root_name);
1013 ett_p = &root_element->ett;
1014 g_array_append_val(etts,ett_p);
1016 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1018 root_element->hf_tag = proto_register_protocol(dtd_data->description, dtd_data->proto_name, root_element->name);
1019 proto_register_field_array(root_element->hf_tag, (hf_register_info*)hfs->data, hfs->len);
1020 proto_register_subtree_array((gint**)etts->data, etts->len);
1022 if (dtd_data->media_type) {
1023 g_hash_table_insert(media_types,dtd_data->media_type,root_element);
1024 dtd_data->media_type = NULL;
1027 dtd_data->description = NULL;
1028 dtd_data->proto_name = NULL;
1029 g_array_free(hfs,FALSE);
1030 g_array_free(etts,TRUE);
1033 g_hash_table_insert(xml_ns.elements,root_element->name,root_element);
1035 g_hash_table_foreach_remove(elements,free_elements,NULL);
1036 g_hash_table_destroy(elements);
1038 destroy_dtd_data(dtd_data);
1044 #if GLIB_MAJOR_VERSION < 2
1045 # define DIRECTORY_T DIR
1046 # define FILE_T struct dirent
1047 # define OPENDIR_OP(name) opendir(name)
1048 # define DIRGETNEXT_OP(dir) readdir(dir)
1049 # define GETFNAME_OP(file) (gchar *)file->d_name
1050 # define CLOSEDIR_OP(dir) closedir(dir)
1052 # define DIRECTORY_T GDir
1053 # define FILE_T gchar
1054 # define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1055 # define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1056 # define GETFNAME_OP(file) (file);
1057 # define CLOSEDIR_OP(dir) g_dir_close(dir)
1060 void init_xml_names(void) {
1061 xml_ns_t* xmlpi_xml_ns;
1065 const gchar* filename;
1068 #if GLIB_MAJOR_VERSION >= 2
1069 GError** dummy = g_malloc(sizeof(GError *));
1073 xmpli_names = g_hash_table_new(g_str_hash,g_str_equal);
1074 media_types = g_hash_table_new(g_str_hash,g_str_equal);
1076 unknown_ns.elements = xml_ns.elements = g_hash_table_new(g_str_hash,g_str_equal);
1077 unknown_ns.attributes = xml_ns.attributes = g_hash_table_new(g_str_hash,g_str_equal);
1079 xmlpi_xml_ns = xml_new_namespace(xmpli_names,"xml","version","encoding","standalone",NULL);
1081 g_hash_table_destroy(xmlpi_xml_ns->elements);
1082 xmlpi_xml_ns->elements = NULL;
1085 dirname = get_persconffile_path("dtds", FALSE);
1087 if (test_for_directory(dirname) != EISDIR) {
1088 dirname = get_datafile_path("dtds");
1091 if (test_for_directory(dirname) == EISDIR) {
1093 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1094 while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1096 filename = GETFNAME_OP(file);
1098 namelen = strlen(filename);
1099 if ( namelen > 4 && ( g_strcasecmp(filename+(namelen-4),".dtd") == 0 ) ) {
1100 GString* errors = g_string_new("");
1101 GString* preparsed = dtd_preparse(dirname, filename, errors);
1102 dtd_build_data_t* dtd_data;
1105 report_failure("Dtd Preparser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1109 dtd_data = dtd_parse(preparsed);
1111 g_string_free(preparsed,TRUE);
1113 if (dtd_data->error->len) {
1114 report_failure("Dtd Parser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,dtd_data->error->str);
1115 destroy_dtd_data(dtd_data);
1119 register_dtd(dtd_data,errors);
1122 report_failure("Dtd Registration in file: %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1123 g_string_free(errors,TRUE);
1133 for(i=0;i<array_length(default_media_types);i++) {
1134 if( ! g_hash_table_lookup(media_types,default_media_types[i]) ) {
1135 g_hash_table_insert(media_types,(gpointer)default_media_types[i],&xml_ns);
1139 g_hash_table_foreach(xmpli_names,add_xmlpi_namespace,"xml.xmlpi");
1141 #if GLIB_MAJOR_VERSION >= 2
1147 proto_register_xml(void) {
1149 static gint *ett_base[] = {
1156 static hf_register_info hf_base[] = {
1157 { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1158 { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1159 { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1160 { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1161 { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1162 { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1163 { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }},
1164 { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, "", HFILL }}
1167 hf_arr = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1168 ett_arr = g_array_new(FALSE,FALSE,sizeof(gint*));
1170 g_array_append_vals(hf_arr,hf_base,array_length(hf_base));
1171 g_array_append_vals(ett_arr,ett_base,array_length(ett_base));
1175 xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1177 proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)hf_arr->data, hf_arr->len);
1178 proto_register_subtree_array((gint**)ett_arr->data, ett_arr->len);
1180 g_array_free(hf_arr,FALSE);
1181 g_array_free(ett_arr,TRUE);
1183 register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1189 void add_dissector_media(gpointer k, gpointer v _U_, gpointer p _U_) {
1190 dissector_add_string("media_type", (gchar*)k, xml_handle);
1194 proto_reg_handoff_xml(void)
1197 xml_handle = find_dissector("xml");
1199 g_hash_table_foreach(media_types,add_dissector_media,NULL);