Rename the routines that handle dissector tables with unsigned integer
[obnox/wireshark/wip.git] / epan / dissectors / packet-xml.c
1 /* packet-xml.c
2  * wireshark's xml dissector .
3  *
4  * (C) 2005, Luis E. Garcia Ontanon.
5  *
6  * $Id$
7  *
8  * Refer to the AUTHORS file or the AUTHORS section in the man page
9  * for contacting the author(s) of this file.
10  *
11  * Wireshark - Network traffic analyzer
12  * By Gerald Combs <gerald@wireshark.org>
13  * Copyright 1998 Gerald Combs
14  *
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.
19  *
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.
24  *
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.
28  */
29
30 #ifdef HAVE_CONFIG_H
31 #include "config.h"
32 #endif
33
34 #ifdef HAVE_DIRENT_H
35 #include <dirent.h>
36 #endif
37
38 #include <string.h>
39 #include <errno.h>
40
41 #include <glib.h>
42
43 #include <wsutil/str_util.h>
44
45 #include <epan/emem.h>
46 #include <epan/packet.h>
47 #include <epan/tvbparse.h>
48 #include <epan/dtd.h>
49 #include <epan/report_err.h>
50 #include <epan/filesystem.h>
51 #include <epan/prefs.h>
52 #include <epan/garrayfix.h>
53
54 #include "packet-xml.h"
55
56 struct _attr_reg_data {
57         GArray* hf;
58         gchar* basename;
59 };
60
61
62 static gint ett_dtd = -1;
63 static gint ett_xmpli = -1;
64
65 static int hf_unknowwn_attrib = -1;
66 static int hf_comment = -1;
67 static int hf_xmlpi = -1;
68 static int hf_dtd_tag = -1;
69 static int hf_doctype = -1;
70
71 /* dissector handles */
72 static dissector_handle_t xml_handle;
73
74 /* parser definitions */
75 static tvbparse_wanted_t* want;
76 static tvbparse_wanted_t* want_ignore;
77 static tvbparse_wanted_t* want_heur;
78
79 static GHashTable* xmpli_names;
80 static GHashTable* media_types;
81
82 static xml_ns_t xml_ns = {"xml","/",-1,-1,-1,NULL,NULL,NULL};
83 static xml_ns_t unknown_ns = {"unknown","?",-1,-1,-1,NULL,NULL,NULL};
84 static xml_ns_t* root_ns;
85
86 static gboolean pref_heuristic_media = FALSE;
87 static gboolean pref_heuristic_tcp = FALSE;
88 static gboolean pref_heuristic_udp = FALSE;
89 static gboolean pref_heuristic_media_save = FALSE;
90 static gboolean pref_heuristic_tcp_save = FALSE;
91 static gboolean pref_heuristic_udp_save = FALSE;
92 static range_t *global_xml_tcp_range = NULL;
93 static range_t *xml_tcp_range = NULL;
94
95 static gboolean pref_heuristic_unicode = FALSE;
96
97 #define XML_CDATA -1000
98 #define XML_SCOPED_NAME -1001
99
100
101 static GArray* hf_arr;
102 static GArray* ett_arr;
103
104 static const gchar* default_media_types[] = {
105         "text/xml",
106         "text/vnd.wap.wml",
107         "text/vnd.wap.si",
108         "text/vnd.wap.sl",
109         "text/vnd.wap.co",
110         "text/vnd.wap.emn",
111         "application/auth-policy+xml",
112         "application/cpim-pidf+xml",
113         "application/cpl+xml",
114         "application/mathml+xml",
115         "application/media_control+xml",
116         "application/note+xml",
117         "application/pidf+xml",
118         "application/poc-settings+xml",
119         "application/rdf+xml",
120         "application/reginfo+xml",
121         "application/resource-lists+xml",
122         "application/rlmi+xml",
123         "application/rls-services+xml",
124         "application/smil",
125         "application/simple-filter+xml",
126         "application/soap+xml",
127         "application/vnd.etsi.aoc+xml",
128         "application/vnd.etsi.cug+xml",
129         "application/vnd.etsi.iptvcommand+xml",
130         "application/vnd.etsi.iptvdiscovery+xml",
131         "application/vnd.etsi.iptvprofile+xml",
132         "application/vnd.etsi.iptvsad-bc+xml",
133         "application/vnd.etsi.iptvsad-cod+xml",
134         "application/vnd.etsi.iptvsad-npvr+xml",
135         "application/vnd.etsi.iptvueprofile+xml",
136         "application/vnd.etsi.mcid+xml",
137         "application/vnd.etsi.sci+xml",
138         "application/vnd.etsi.simservs+xml",
139         "application/vnd.wv.csp+xml",
140         "application/vnd.wv.csp.xml",
141         "application/watcherinfo+xml",
142         "application/xcap-att+xml",
143         "application/xcap-caps+xml",
144         "application/xcap-diff+xml",
145         "application/xcap-el+xml",
146         "application/xcap-error+xml",
147         "application/xcap-ns+xml",
148         "application/xml",
149         "application/xml-dtd",
150         "application/xpidf+xml",
151         "application/xslt+xml",
152         "application/x-wms-logconnectstats",
153         "application/x-wms-logplaystats",
154         "application/x-wms-sendevent",
155         "application/rss+xml",
156         "image/svg+xml",
157 };
158
159 static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child) {
160         new_child->first_child = NULL;
161         new_child->last_child = NULL;
162
163         new_child->parent = parent;
164         new_child->next_sibling = NULL;
165         new_child->prev_sibling = NULL;
166         if (parent == NULL) return;  /* root */
167
168         if (parent->first_child == NULL) {  /* the 1st child */
169                 parent->first_child = new_child;
170         } else {  /* following children */
171                 parent->last_child->next_sibling = new_child;
172                 new_child->prev_sibling = parent->last_child;
173         }
174         parent->last_child = new_child;
175 }
176
177 static void
178 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree)
179 {
180         tvbparse_t* tt;
181         tvbparse_elem_t* tok = NULL;
182         static GPtrArray* stack = NULL;
183         xml_frame_t* current_frame;
184         char* colinfo_str;
185
186         if (stack != NULL)
187                 g_ptr_array_free(stack,TRUE);
188
189         stack = g_ptr_array_new();
190         current_frame = ep_alloc(sizeof(xml_frame_t));
191         current_frame->type = XML_FRAME_ROOT;
192         current_frame->name = NULL;
193         current_frame->name_orig_case = NULL;
194         current_frame->value = NULL;
195         insert_xml_frame(NULL, current_frame);
196         g_ptr_array_add(stack,current_frame);
197
198         tt = tvbparse_init(tvb,0,-1,stack,want_ignore);
199         current_frame->start_offset = 0;
200
201         root_ns = NULL;
202
203         if (pinfo->match_string)
204                 root_ns = g_hash_table_lookup(media_types,pinfo->match_string);
205
206         if (! root_ns ) {
207                 root_ns = &xml_ns;
208                 colinfo_str = "/XML";
209         } else {
210                 colinfo_str = ep_strdup_printf("/%s",root_ns->name);
211                 ascii_strup_inplace(colinfo_str);
212         }
213
214         col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
215
216         current_frame->ns = root_ns;
217
218         current_frame->item = proto_tree_add_item(tree,current_frame->ns->hf_tag,tvb,0,-1,FALSE);
219         current_frame->tree = proto_item_add_subtree(current_frame->item,current_frame->ns->ett);
220         current_frame->last_item = current_frame->item;
221
222         while(( tok = tvbparse_get(tt, want) )) ;
223
224         pinfo->private_data = current_frame;  /* pass XML structure to the dissector calling XML */
225 }
226
227 static gboolean dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree) {
228         if (pref_heuristic_media || pref_heuristic_tcp || pref_heuristic_udp) {
229                 if (tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
230                         dissect_xml(tvb, pinfo, tree);
231                         return TRUE;
232                 } else if (pref_heuristic_unicode) {
233                         const guint8 *data = tvb_get_ephemeral_faked_unicode(tvb, 0, tvb_length(tvb)/2, TRUE);
234                         tvbuff_t *unicode_tvb = tvb_new_real_data(data, tvb_length(tvb)/2, tvb_length(tvb)/2);
235                         if (tvbparse_peek(tvbparse_init(unicode_tvb,0,-1,NULL,want_ignore), want_heur)) {
236                                 dissect_xml(unicode_tvb, pinfo, tree);
237                                 return TRUE;
238                         }
239                 }
240         }
241         return FALSE;
242 }
243
244 xml_frame_t *xml_get_tag(xml_frame_t *frame, const gchar *name) {
245         xml_frame_t *tag = NULL;
246
247         xml_frame_t *xml_item = frame->first_child;
248         while (xml_item) {
249                 if ((xml_item->type == XML_FRAME_TAG)) {
250                         if (!name) {  /* get the 1st tag */
251                         tag = xml_item;
252                                 break;
253                         } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
254                         tag = xml_item;
255                                 break;
256                         }
257                 }
258                 xml_item = xml_item->next_sibling;
259         }
260
261         return tag;
262 }
263
264 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const gchar *name) {
265         xml_frame_t *attr = NULL;
266
267         xml_frame_t *xml_item = frame->first_child;
268         while (xml_item) {
269                 if ((xml_item->type == XML_FRAME_ATTRIB) &&
270                         xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
271                 attr = xml_item;
272                         break;
273                 }
274                 xml_item = xml_item->next_sibling;
275         }
276
277         return attr;
278 }
279
280 xml_frame_t *xml_get_cdata(xml_frame_t *frame) {
281         xml_frame_t *cdata = NULL;
282
283         xml_frame_t *xml_item = frame->first_child;
284         while (xml_item) {
285                 if ((xml_item->type == XML_FRAME_CDATA)) {
286                 cdata = xml_item;
287                         break;
288                 }
289                 xml_item = xml_item->next_sibling;
290         }
291
292         return cdata;
293 }
294
295 static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
296         GPtrArray* stack = tvbparse_data;
297         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
298         int hfid;
299         gboolean is_cdata = FALSE;
300         proto_item* pi;
301         xml_frame_t* new_frame;
302
303         if (tok->id == XML_CDATA) {
304                 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
305                 is_cdata = TRUE;
306         } else if ( tok->id > 0) {
307                 hfid = tok->id;
308         } else {
309                 hfid = xml_ns.hf_cdata;
310         }
311
312         pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, FALSE);
313
314         proto_item_set_text(pi, "%s",
315                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
316
317         if (is_cdata) {
318                 new_frame = ep_alloc(sizeof(xml_frame_t));
319                 new_frame->type = XML_FRAME_CDATA;
320                 new_frame->name = NULL;
321                 new_frame->name_orig_case = NULL;
322                 new_frame->value = tvb_new_subset(tok->tvb, tok->offset, tok->len, tok->len);
323                 insert_xml_frame(current_frame, new_frame);
324                 new_frame->item = pi;
325                 new_frame->last_item = pi;
326                 new_frame->tree = NULL;
327                 new_frame->start_offset = tok->offset;
328                 new_frame->ns = NULL;
329         }
330 }
331
332 static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
333         GPtrArray* stack = tvbparse_data;
334         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
335         proto_item* pi;
336         proto_tree* pt;
337         tvbparse_elem_t* name_tok = tok->sub->next;
338         gchar* name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
339         xml_ns_t* ns = g_hash_table_lookup(xmpli_names,name);
340         xml_frame_t* new_frame;
341
342         int hf_tag;
343         gint ett;
344
345         ascii_strdown_inplace(name);
346         if (!ns) {
347                 hf_tag = hf_xmlpi;
348                 ett = ett_xmpli;
349         } else {
350                 hf_tag = ns->hf_tag;
351                 ett = ns->ett;
352         }
353
354         pi = proto_tree_add_item(current_frame->tree,hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
355
356         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
357
358         pt = proto_item_add_subtree(pi,ett);
359
360         new_frame = ep_alloc(sizeof(xml_frame_t));
361         new_frame->type = XML_FRAME_XMPLI;
362         new_frame->name = name;
363         new_frame->name_orig_case = name;
364         new_frame->value = NULL;
365         insert_xml_frame(current_frame, new_frame);
366         new_frame->item = pi;
367         new_frame->last_item = pi;
368         new_frame->tree = pt;
369         new_frame->start_offset = tok->offset;
370         new_frame->ns = ns;
371
372         g_ptr_array_add(stack,new_frame);
373
374 }
375
376 static void after_xmlpi(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
380         proto_tree_add_text(current_frame->tree,
381                                                    tok->tvb, tok->offset, tok->len, "%s",
382                                                    tvb_format_text(tok->tvb,tok->offset,tok->len));
383
384         if (stack->len > 1) {
385                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
386         } else {
387                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened xmpli tag ]");
388         }
389 }
390
391 static void before_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
392         GPtrArray* stack = tvbparse_data;
393         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
394         tvbparse_elem_t* name_tok = tok->sub->next;
395         gchar* root_name;
396         gchar *name = NULL, *name_orig_case = NULL;
397         xml_ns_t* ns;
398         xml_frame_t* new_frame;
399         proto_item* pi;
400         proto_tree* pt;
401
402         if (name_tok->sub->id == XML_SCOPED_NAME) {
403                 tvbparse_elem_t* root_tok = name_tok->sub->sub;
404                 tvbparse_elem_t* leaf_tok = name_tok->sub->sub->next->next;
405                 xml_ns_t* nameroot_ns;
406
407                 root_name = (gchar*)tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
408                 name = (gchar*)tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
409                 name_orig_case = name;
410
411                 nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
412
413                 if(nameroot_ns) {
414                         ns = g_hash_table_lookup(nameroot_ns->elements,name);
415                         if (!ns) {
416                                 ns = &unknown_ns;
417                         }
418                 } else {
419                         ns = &unknown_ns;
420                 }
421
422         } else {
423                 name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
424                 name_orig_case = ep_strdup(name);
425                 ascii_strdown_inplace(name);
426
427                 if(current_frame->ns) {
428                         ns = g_hash_table_lookup(current_frame->ns->elements,name);
429
430                         if (!ns) {
431                                 if (! ( ns = g_hash_table_lookup(root_ns->elements,name) ) ) {
432                                         ns = &unknown_ns;
433                                 }
434                         }
435                 } else {
436                         ns = &unknown_ns;
437                 }
438         }
439
440         pi = proto_tree_add_item(current_frame->tree,ns->hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
441         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
442
443         pt = proto_item_add_subtree(pi,ns->ett);
444
445         new_frame = ep_alloc(sizeof(xml_frame_t));
446         new_frame->type = XML_FRAME_TAG;
447         new_frame->name = name;
448         new_frame->name_orig_case = name_orig_case;
449         new_frame->value = NULL;
450         insert_xml_frame(current_frame, new_frame);
451         new_frame->item = pi;
452         new_frame->last_item = pi;
453         new_frame->tree = pt;
454         new_frame->start_offset = tok->offset;
455         new_frame->ns = ns;
456
457         g_ptr_array_add(stack,new_frame);
458
459 }
460
461 static void after_open_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
462         GPtrArray* stack = tvbparse_data;
463         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
464
465         proto_item_append_text(current_frame->last_item,">");
466 }
467
468 static void after_closed_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
469         GPtrArray* stack = tvbparse_data;
470         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
471
472         proto_item_append_text(current_frame->last_item,"/>");
473
474         if (stack->len > 1) {
475                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
476         } else {
477                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
478         }
479 }
480
481 static void after_untag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
482         GPtrArray* stack = tvbparse_data;
483         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
484
485         proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
486
487         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
488                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
489
490         if (stack->len > 1) {
491                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
492         } else {
493                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
494                                                         "[ ERROR: Closing an unopened tag ]");
495         }
496 }
497
498 static void before_dtd_doctype(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
499         GPtrArray* stack = tvbparse_data;
500         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
501         xml_frame_t* new_frame;
502         tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
503         proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
504                                                                                            name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
505
506         proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
507
508         new_frame = ep_alloc(sizeof(xml_frame_t));
509         new_frame->type = XML_FRAME_DTD_DOCTYPE;
510         new_frame->name = (gchar*)tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
511         new_frame->name_orig_case = new_frame->name;
512         new_frame->value = NULL;
513         insert_xml_frame(current_frame, new_frame);
514         new_frame->item = dtd_item;
515         new_frame->last_item = dtd_item;
516         new_frame->tree = proto_item_add_subtree(dtd_item,ett_dtd);
517         new_frame->start_offset = tok->offset;
518         new_frame->ns = NULL;
519
520         g_ptr_array_add(stack,new_frame);
521 }
522
523 static void pop_stack(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
524         GPtrArray* stack = tvbparse_data;
525         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
526
527         if (stack->len > 1) {
528                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
529         } else {
530                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
531                                                         "[ ERROR: Closing an unopened tag ]");
532         }
533 }
534
535 static void after_dtd_close(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
536         GPtrArray* stack = tvbparse_data;
537         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
538
539         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
540                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
541         if (stack->len > 1) {
542                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
543         } else {
544                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
545         }
546 }
547
548 static void get_attrib_value(void* tvbparse_data _U_, const void* wanted_data _U_, tvbparse_elem_t* tok) {
549         tok->data = tok->sub;
550 }
551
552 static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
553         GPtrArray* stack = tvbparse_data;
554         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
555         gchar *name = NULL, *name_orig_case = NULL;
556         tvbparse_elem_t* value;
557         tvbparse_elem_t* value_part = tok->sub->next->next->data;
558         int* hfidp;
559         int hfid;
560         proto_item* pi;
561         xml_frame_t* new_frame;
562
563         name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
564         name_orig_case = ep_strdup(name);
565         ascii_strdown_inplace(name);
566
567         if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
568                 hfid = *hfidp;
569                 value = value_part;
570         } else {
571                 hfid = hf_unknowwn_attrib;
572                 value = tok;
573         }
574
575         pi = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
576         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
577
578         current_frame->last_item = pi;
579
580         new_frame = ep_alloc(sizeof(xml_frame_t));
581         new_frame->type = XML_FRAME_ATTRIB;
582         new_frame->name = name;
583         new_frame->name_orig_case = name_orig_case;
584         new_frame->value = tvb_new_subset(value_part->tvb, value_part->offset, value_part->len, value_part->len);
585         insert_xml_frame(current_frame, new_frame);
586         new_frame->item = pi;
587         new_frame->last_item = pi;
588         new_frame->tree = NULL;
589         new_frame->start_offset = tok->offset;
590         new_frame->ns = NULL;
591
592 }
593
594 static void unrecognized_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_){
595         GPtrArray* stack = tvbparse_data;
596         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
597
598         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Unrecognized text ]");
599
600 }
601
602
603
604 static void init_xml_parser(void) {
605         tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
606         tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
607
608         tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
609                                                                want_name,
610                                                                tvbparse_char(-1,":",NULL,NULL,NULL),
611                                                                want_name,
612                                                                NULL);
613
614         tvbparse_wanted_t* want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
615                                                               want_scoped_name,
616                                                               want_name,
617                                                               NULL);
618
619         tvbparse_wanted_t* want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
620                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\"','\\'),
621                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\'','\\'),
622                                                                   tvbparse_chars(-1,1,0,"0123456789",NULL,NULL,NULL),
623                                                                   want_name,
624                                                                   NULL);
625
626         tvbparse_wanted_t* want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
627                                                                   tvbparse_set_seq(-1, NULL, NULL, after_attrib,
628                                                                                    want_attr_name,
629                                                                                    tvbparse_char(-1,"=",NULL,NULL,NULL),
630                                                                                    want_attrib_value,
631                                                                                    NULL));
632
633         tvbparse_wanted_t* want_stoptag = tvbparse_set_oneof(-1,NULL,NULL,NULL,
634                                                              tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
635                                                              tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
636                                                              NULL);
637
638         tvbparse_wanted_t* want_stopxmlpi = tvbparse_string(-1,"?>",NULL,NULL,after_xmlpi);
639
640         tvbparse_wanted_t* want_comment = tvbparse_set_seq(hf_comment,NULL,NULL,after_token,
641                                                            tvbparse_string(-1,"<!--",NULL,NULL,NULL),
642                                                            tvbparse_until(-1,NULL,NULL,NULL,
643                                                                           tvbparse_string(-1,"-->",NULL,NULL,NULL),
644                                                                           TP_UNTIL_INCLUDE),
645                                                            NULL);
646
647         tvbparse_wanted_t* want_xmlpi = tvbparse_set_seq(hf_xmlpi,NULL,before_xmpli,NULL,
648                                                          tvbparse_string(-1,"<?",NULL,NULL,NULL),
649                                                          want_name,
650                                                          tvbparse_set_oneof(-1,NULL,NULL,NULL,
651                                                                             want_stopxmlpi,
652                                                                             tvbparse_set_seq(-1,NULL,NULL,NULL,
653                                                                                              want_attributes,
654                                                                                              want_stopxmlpi,
655                                                                                              NULL),
656                                                                             NULL),
657                                                          NULL);
658
659         tvbparse_wanted_t* want_closing_tag = tvbparse_set_seq(0,NULL,NULL,after_untag,
660                                                                tvbparse_char(-1, "<", NULL, NULL, NULL),
661                                                                tvbparse_char(-1, "/", NULL, NULL, NULL),
662                                                                want_tag_name,
663                                                                tvbparse_char(-1, ">", NULL, NULL, NULL),
664                                                                NULL);
665
666         tvbparse_wanted_t* want_doctype_start = tvbparse_set_seq(-1,NULL,before_dtd_doctype,NULL,
667                                                                  tvbparse_char(-1,"<",NULL,NULL,NULL),
668                                                                  tvbparse_char(-1,"!",NULL,NULL,NULL),
669                                                                  tvbparse_casestring(-1,"DOCTYPE",NULL,NULL,NULL),
670                                                                  tvbparse_set_oneof(-1,NULL,NULL,NULL,
671                                                                                     tvbparse_set_seq(-1,NULL,NULL,NULL,
672                                                                                                      want_name,
673                                                                                                      tvbparse_char(-1,"[",NULL,NULL,NULL),
674                                                                                                      NULL),
675                                                                                     tvbparse_set_seq(-1,NULL,NULL,pop_stack,
676                                                                                                      want_name,
677                                                                                                      tvbparse_set_oneof(-1,NULL,NULL,NULL,
678                                                                                                                         tvbparse_casestring(-1,"PUBLIC",NULL,NULL,NULL),
679                                                                                                                         tvbparse_casestring(-1,"SYSTEM",NULL,NULL,NULL),
680                                                                                                                         NULL),
681                                                                                                      tvbparse_until(-1,NULL,NULL,NULL,
682                                                                                                                     tvbparse_char(-1,">",NULL,NULL,NULL),
683                                                                                                                     TP_UNTIL_INCLUDE),
684                                                                                                      NULL),
685                                                                                     NULL),
686                                                                  NULL);
687
688         tvbparse_wanted_t* want_dtd_tag = tvbparse_set_seq(hf_dtd_tag,NULL,NULL,after_token,
689                                                            tvbparse_char(-1,"<",NULL,NULL,NULL),
690                                                            tvbparse_char(-1,"!",NULL,NULL,NULL),
691                                                            tvbparse_until(-1,NULL,NULL,NULL,
692                                                                           tvbparse_char(-1, ">", NULL, NULL, NULL),
693                                                                           TP_UNTIL_INCLUDE),
694                                                            NULL);
695
696         tvbparse_wanted_t* want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
697                                                        tvbparse_char(-1,"<",NULL,NULL,NULL),
698                                                        want_tag_name,
699                                                        tvbparse_set_oneof(-1,NULL,NULL,NULL,
700                                                                           tvbparse_set_seq(-1,NULL,NULL,NULL,
701                                                                                            want_attributes,
702                                                                                            want_stoptag,
703                                                                                            NULL),
704                                                                           want_stoptag,
705                                                                           NULL),
706                                                        NULL);
707
708         tvbparse_wanted_t* want_dtd_close = tvbparse_set_seq(-1,NULL,NULL,after_dtd_close,
709                                                              tvbparse_char(-1,"]",NULL,NULL,NULL),
710                                                              tvbparse_char(-1,">",NULL,NULL,NULL),
711                                                              NULL);
712
713         want_ignore = tvbparse_chars(-1,1,0," \t\r\n",NULL,NULL,NULL);
714
715
716         want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
717                                   want_comment,
718                                   want_xmlpi,
719                                   want_closing_tag,
720                                   want_doctype_start,
721                                   want_dtd_close,
722                                   want_dtd_tag,
723                                   want_tag,
724                                   tvbparse_not_chars(XML_CDATA,1,0,"<",NULL,NULL,after_token),
725                                   tvbparse_not_chars(-1,1,0," \t\r\n",NULL,NULL,unrecognized_token),
726                                   NULL);
727
728         want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
729                                        want_comment,
730                                        want_xmlpi,
731                                        want_doctype_start,
732                                        want_dtd_tag,
733                                        want_tag,
734                                        NULL);
735
736 }
737
738
739 static xml_ns_t* xml_new_namespace(GHashTable* hash, gchar* name, ...) {
740         xml_ns_t* ns = g_malloc(sizeof(xml_ns_t));
741         va_list ap;
742         gchar* attr_name;
743
744         ns->name = g_strdup(name);
745         ns->hf_tag = -1;
746         ns->hf_cdata = -1;
747         ns->ett = -1;
748         ns->attributes = g_hash_table_new(g_str_hash,g_str_equal);
749         ns->elements = g_hash_table_new(g_str_hash,g_str_equal);
750
751         va_start(ap,name);
752
753         while(( attr_name = va_arg(ap,gchar*) )) {
754                 int* hfp = g_malloc(sizeof(int));
755                 *hfp = -1;
756                 g_hash_table_insert(ns->attributes,g_strdup(attr_name),hfp);
757         };
758
759         va_end(ap);
760
761         g_hash_table_insert(hash,ns->name,ns);
762
763         return ns;
764 }
765
766
767 static void add_xml_field(GArray* hfs, int* p_id, gchar* name, gchar* fqn) {
768         hf_register_info hfri;
769
770         hfri.p_id = p_id;
771         hfri.hfinfo.name = name;
772         hfri.hfinfo.abbrev = fqn;
773         hfri.hfinfo.type = FT_STRING;
774         hfri.hfinfo.display = BASE_NONE;
775         hfri.hfinfo.strings = NULL;
776         hfri.hfinfo.bitmask = 0x0;
777         hfri.hfinfo.blurb = NULL;
778         hfri.hfinfo.id = 0;
779         hfri.hfinfo.parent = 0;
780         hfri.hfinfo.ref_type = HF_REF_TYPE_NONE;
781         hfri.hfinfo.bitshift = 0;
782         hfri.hfinfo.same_name_next = NULL;
783         hfri.hfinfo.same_name_prev = NULL;
784
785         g_array_append_val(hfs,hfri);
786 }
787
788 static void add_xml_attribute_names(gpointer k, gpointer v, gpointer p) {
789         struct _attr_reg_data* d = p;
790         gchar* basename = g_strdup_printf("%s.%s",d->basename,(gchar*)k);
791         add_xml_field(d->hf, (int*) v, (gchar*)k, basename);
792 }
793
794
795 static void add_xmlpi_namespace(gpointer k _U_, gpointer v, gpointer p) {
796         xml_ns_t* ns = v;
797         gchar* basename = g_strdup_printf("%s.%s",(gchar*)p,ns->name);
798         gint* ett_p = &(ns->ett);
799         struct _attr_reg_data d;
800
801         add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
802
803         g_array_append_val(ett_arr,ett_p);
804
805         d.basename = basename;
806         d.hf = hf_arr;
807
808         g_hash_table_foreach(ns->attributes,add_xml_attribute_names,&d);
809
810 }
811
812 static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
813         g_free(dtd_data->proto_name);
814         g_free(dtd_data->media_type);
815         g_free(dtd_data->description);
816         g_free(dtd_data->proto_root);
817
818         g_string_free(dtd_data->error,TRUE);
819
820         while(dtd_data->elements->len) {
821                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
822                 g_ptr_array_free(nl->list,TRUE);
823                 g_free(nl);
824         }
825
826         g_ptr_array_free(dtd_data->elements,TRUE);
827
828         while(dtd_data->attributes->len) {
829                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->attributes,0);
830                 g_ptr_array_free(nl->list,TRUE);
831                 g_free(nl);
832         }
833
834         g_ptr_array_free(dtd_data->attributes,TRUE);
835
836         g_free(dtd_data);
837 }
838
839 static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
840         gchar* key = g_strdup(k);
841         int* value = g_malloc(sizeof(int));
842         GHashTable* dst = p;
843
844         *value = -1;
845         g_hash_table_insert(dst,key,value);
846
847 }
848
849 static GHashTable* copy_attributes_hash(GHashTable* src) {
850         GHashTable* dst = g_hash_table_new(g_str_hash,g_str_equal);
851
852         g_hash_table_foreach(src,copy_attrib_item,dst);
853
854         return dst;
855 }
856
857 static xml_ns_t* duplicate_element(xml_ns_t* orig) {
858         xml_ns_t* new_item  = g_malloc(sizeof(xml_ns_t));
859         guint i;
860
861         new_item->name = g_strdup(orig->name);
862         new_item->hf_tag = -1;
863         new_item->hf_cdata = -1;
864         new_item->ett = -1;
865         new_item->attributes = copy_attributes_hash(orig->attributes);
866         new_item->elements =  g_hash_table_new(g_str_hash,g_str_equal);
867         new_item->element_names = g_ptr_array_new();
868
869         for(i=0; i < orig->element_names->len; i++) {
870                 g_ptr_array_add(new_item->element_names,
871                                                    g_ptr_array_index(orig->element_names,i));
872         }
873
874         return new_item;
875 }
876
877 static gchar* fully_qualified_name(GPtrArray* hier, gchar* name, gchar* proto_name) {
878         guint i;
879         GString* s = g_string_new(proto_name);
880         gchar* str;
881         g_string_append(s,".");
882
883         for (i = 1; i < hier->len; i++) {
884                 g_string_append_printf(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
885         }
886
887         g_string_append(s,name);
888         str = s->str;
889         g_string_free(s,FALSE);
890
891         return str;
892 }
893
894
895 static xml_ns_t* make_xml_hier(gchar* elem_name,
896                                xml_ns_t* root,
897                                GHashTable* elements,
898                                GPtrArray* hier,
899                                GString* error,
900                                GArray* hfs,
901                                GArray* etts,
902                                char* proto_name) {
903         xml_ns_t* new;
904         xml_ns_t* orig;
905         gchar* fqn;
906         gint* ett_p;
907         struct _attr_reg_data d;
908         gboolean recurred = FALSE;
909         guint i;
910
911         if ( g_str_equal(elem_name,root->name) ) {
912                 return NULL;
913         }
914
915         if (! ( orig = g_hash_table_lookup(elements,elem_name) )) {
916                 g_string_append_printf(error,"element '%s' is not defined\n", elem_name);
917                 return NULL;
918         }
919
920         for (i = 0; i < hier->len; i++) {
921                 if( strcmp(elem_name,(gchar*) g_ptr_array_index(hier,i) ) == 0 ) {
922                         recurred = TRUE;
923                 }
924         }
925
926         if (recurred) {
927                 return NULL;
928         }
929
930         fqn = fully_qualified_name(hier,elem_name,proto_name);
931
932         new = duplicate_element(orig);
933         new->fqn = fqn;
934
935         add_xml_field(hfs, &(new->hf_tag), g_strdup(elem_name), fqn);
936         add_xml_field(hfs, &(new->hf_cdata), g_strdup(elem_name), fqn);
937
938         ett_p = &new->ett;
939         g_array_append_val(etts,ett_p);
940
941         d.basename = fqn;
942         d.hf = hfs;
943
944         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
945
946         while(new->element_names->len) {
947                 gchar* child_name = g_ptr_array_remove_index(new->element_names,0);
948                 xml_ns_t* child_element = NULL;
949
950                 g_ptr_array_add(hier,elem_name);
951                 child_element = make_xml_hier(child_name, root, elements, hier,error,hfs,etts,proto_name);
952                 g_ptr_array_remove_index_fast(hier,hier->len - 1);
953
954                 if (child_element) {
955                         g_hash_table_insert(new->elements,child_element->name,child_element);
956                 }
957         }
958
959         g_ptr_array_free(new->element_names,TRUE);
960         new->element_names = NULL;
961         return new;
962 }
963
964 static gboolean free_both(gpointer k, gpointer v, gpointer p _U_) {
965         g_free(k);
966         g_free(v);
967         return TRUE;
968 }
969
970 static gboolean free_elements(gpointer k _U_, gpointer v, gpointer p _U_) {
971         xml_ns_t* e = v;
972         g_free(e->name);
973         g_hash_table_foreach_remove(e->attributes,free_both,NULL);
974         g_hash_table_destroy(e->attributes);
975         g_hash_table_destroy(e->elements);
976
977         while (e->element_names->len) {
978                 g_free(g_ptr_array_remove_index(e->element_names,0));
979         }
980
981         g_ptr_array_free(e->element_names,TRUE);
982         g_free(e);
983
984         return TRUE;
985 }
986
987 static void register_dtd(dtd_build_data_t* dtd_data, GString* errors) {
988         GHashTable* elements = g_hash_table_new(g_str_hash,g_str_equal);
989         gchar* root_name = NULL;
990         xml_ns_t* root_element = NULL;
991         GArray* hfs;
992         GArray* etts;
993         GPtrArray* hier;
994         gchar* curr_name;
995         GPtrArray* element_names = g_ptr_array_new();
996
997         /* we first populate elements with the those coming from the parser */
998         while(dtd_data->elements->len) {
999                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->elements,0);
1000                 xml_ns_t* element = g_malloc(sizeof(xml_ns_t));
1001
1002                 /* we will use the first element found as root in case no other one was given. */
1003                 if (root_name == NULL)
1004                         root_name = g_strdup(nl->name);
1005
1006                 element->name = nl->name;
1007                 element->element_names = nl->list;
1008                 element->hf_tag = -1;
1009                 element->hf_cdata = -1;
1010                 element->ett = -1;
1011                 element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1012                 element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1013
1014                 if( g_hash_table_lookup(elements,element->name) ) {
1015                         g_string_append_printf(errors,"element %s defined more than once\n", element->name);
1016                         free_elements(NULL,element,NULL);
1017                 } else {
1018                         g_hash_table_insert(elements,element->name,element);
1019                         g_ptr_array_add(element_names,g_strdup(element->name));
1020                 }
1021
1022                 g_free(nl);
1023         }
1024
1025         /* then we add the attributes to its relative elements */
1026         while(dtd_data->attributes->len) {
1027                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
1028                 xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
1029
1030                 if (element) {
1031                 while(nl->list->len) {
1032                     gchar* name = g_ptr_array_remove_index(nl->list,0);
1033                     int* id_p = g_malloc(sizeof(int));
1034
1035                     *id_p = -1;
1036                     g_hash_table_insert(element->attributes,name,id_p);
1037                 }
1038                 }
1039                 else {
1040             g_string_append_printf(errors,"element %s is not defined\n", nl->name);
1041                 }
1042
1043                 g_free(nl->name);
1044                 g_ptr_array_free(nl->list,TRUE);
1045                 g_free(nl);
1046         }
1047
1048         /* if a proto_root is defined in the dtd we'll use that as root */
1049         if( dtd_data->proto_root ) {
1050                 g_free(root_name);
1051                 root_name = g_strdup(dtd_data->proto_root);
1052         }
1053
1054         /* we use a stack with the names to avoid recurring infinitelly */
1055         hier = g_ptr_array_new();
1056
1057         /*
1058          * if a proto name was given in the dtd the dtd will be used as a protocol
1059          * or else the dtd will be loaded as a branch of the xml namespace
1060          */
1061         if( ! dtd_data->proto_name ) {
1062                 hfs = hf_arr;
1063                 etts = ett_arr;
1064                 g_ptr_array_add(hier,g_strdup("xml"));
1065                 root_element = &xml_ns;
1066         } else {
1067                 /*
1068                  * if we were given a proto_name the namespace will be registered
1069                  * as an independent protocol with its own hf and ett arrays.
1070                  */
1071                 hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1072                 etts = g_array_new(FALSE,FALSE,sizeof(gint*));
1073         }
1074
1075         /* the root element of the dtd's namespace */
1076         root_element = g_malloc(sizeof(xml_ns_t));
1077         root_element->name = g_strdup(root_name);
1078         root_element->fqn = dtd_data->proto_name ? g_strdup(dtd_data->proto_name) : root_element->name;
1079         root_element->hf_tag = -1;
1080         root_element->hf_cdata = -1;
1081         root_element->ett = -1;
1082         root_element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1083         root_element->element_names = element_names;
1084
1085         /*
1086          * we can either create a namespace as a flat namespace
1087          * in which all the elements are at the root level
1088          * or we can create a recursive namespace
1089          */
1090         if (dtd_data->recursion) {
1091                 xml_ns_t* orig_root;
1092
1093                 make_xml_hier(root_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1094
1095                 g_hash_table_insert(root_element->elements,root_element->name,root_element);
1096
1097                 orig_root = g_hash_table_lookup(elements,root_name);
1098
1099                 /* if the root element was defined copy its attrlist to the child */
1100                 if(orig_root) {
1101                         struct _attr_reg_data d;
1102
1103                         d.basename = dtd_data->proto_name;
1104                         d.hf = hfs;
1105
1106                         root_element->attributes = copy_attributes_hash(orig_root->attributes);
1107                         g_hash_table_foreach(root_element->attributes,add_xml_attribute_names,&d);
1108                 } else {
1109                         root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1110                 }
1111
1112                 /* we then create all the sub hierachies to catch the recurred cases */
1113                 g_ptr_array_add(hier,root_name);
1114
1115                 while(root_element->element_names->len) {
1116                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1117
1118                         if( ! g_hash_table_lookup(root_element->elements,curr_name) ) {
1119                                 xml_ns_t* new = make_xml_hier(curr_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1120                                 g_hash_table_insert(root_element->elements,new->name,new);
1121                         }
1122
1123                         g_free(curr_name);
1124                 }
1125
1126         } else {
1127                 /* a flat namespace */
1128                 g_ptr_array_add(hier,root_name);
1129
1130                 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1131
1132                 while(root_element->element_names->len) {
1133                         xml_ns_t* new;
1134                         gint* ett_p;
1135                         struct _attr_reg_data d;
1136
1137                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1138                         new = duplicate_element(g_hash_table_lookup(elements,curr_name));
1139                         new->fqn = fully_qualified_name(hier, curr_name, root_name);
1140
1141                         add_xml_field(hfs, &(new->hf_tag), curr_name, new->fqn);
1142                         add_xml_field(hfs, &(new->hf_cdata), curr_name, new->fqn);
1143
1144                         d.basename = new->fqn;
1145                         d.hf = hfs;
1146
1147                         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
1148
1149                         ett_p = &new->ett;
1150                         g_array_append_val(etts,ett_p);
1151
1152                         g_ptr_array_free(new->element_names,TRUE);
1153
1154                         g_hash_table_insert(root_element->elements,new->name,new);
1155                 }
1156         }
1157
1158         g_ptr_array_free(element_names,TRUE);
1159
1160         g_ptr_array_free(hier,TRUE);
1161
1162         /*
1163          * if we were given a proto_name the namespace will be registered
1164          * as an independent protocol.
1165          */
1166         if( dtd_data->proto_name ) {
1167                 gint* ett_p;
1168
1169                 if ( ! dtd_data->description) {
1170                         dtd_data->description = g_strdup(root_name);
1171                 }
1172
1173                 ett_p = &root_element->ett;
1174                 g_array_append_val(etts,ett_p);
1175
1176                 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1177
1178                 root_element->hf_tag = proto_register_protocol(dtd_data->description, dtd_data->proto_name, dtd_data->proto_name);
1179                 proto_register_field_array(root_element->hf_tag, (hf_register_info*)g_array_data(hfs), hfs->len);
1180                 proto_register_subtree_array((gint**)g_array_data(etts), etts->len);
1181
1182                 if (dtd_data->media_type) {
1183                         g_hash_table_insert(media_types,dtd_data->media_type,root_element);
1184                         dtd_data->media_type = NULL;
1185                 }
1186
1187                 dtd_data->description = NULL;
1188                 dtd_data->proto_name = NULL;
1189                 g_array_free(hfs,FALSE);
1190                 g_array_free(etts,TRUE);
1191         }
1192
1193         g_hash_table_insert(xml_ns.elements,root_element->name,root_element);
1194
1195         g_hash_table_foreach_remove(elements,free_elements,NULL);
1196         g_hash_table_destroy(elements);
1197
1198         destroy_dtd_data(dtd_data);
1199         g_free(root_name);
1200 }
1201
1202 #  define DIRECTORY_T GDir
1203 #  define FILE_T gchar
1204 #  define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1205 #  define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1206 #  define GETFNAME_OP(file) (file);
1207 #  define CLOSEDIR_OP(dir) g_dir_close(dir)
1208
1209 static void init_xml_names(void) {
1210         xml_ns_t* xmlpi_xml_ns;
1211         guint i;
1212         DIRECTORY_T* dir;
1213         const FILE_T* file;
1214         const gchar* filename;
1215         gchar* dirname;
1216
1217         GError** dummy = g_malloc(sizeof(GError *));
1218         *dummy = NULL;
1219
1220         xmpli_names = g_hash_table_new(g_str_hash,g_str_equal);
1221         media_types = g_hash_table_new(g_str_hash,g_str_equal);
1222
1223         unknown_ns.elements = xml_ns.elements = g_hash_table_new(g_str_hash,g_str_equal);
1224         unknown_ns.attributes = xml_ns.attributes = g_hash_table_new(g_str_hash,g_str_equal);
1225
1226         xmlpi_xml_ns = xml_new_namespace(xmpli_names,"xml","version","encoding","standalone",NULL);
1227
1228         g_hash_table_destroy(xmlpi_xml_ns->elements);
1229         xmlpi_xml_ns->elements = NULL;
1230
1231
1232         dirname = get_persconffile_path("dtds", FALSE, FALSE);
1233
1234         if (test_for_directory(dirname) != EISDIR) {
1235                 /* Although dir isn't a directory it may still use memory */
1236                 g_free(dirname);
1237                 dirname = get_datafile_path("dtds");
1238         }
1239
1240         if (test_for_directory(dirname) == EISDIR) {
1241
1242                 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1243                         while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1244                                 guint namelen;
1245                                 filename = GETFNAME_OP(file);
1246
1247                                 namelen = (int)strlen(filename);
1248                                 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4),".dtd")  == 0 ) ) {
1249                                         GString* errors = g_string_new("");
1250                                         GString* preparsed = dtd_preparse(dirname, filename, errors);
1251                                         dtd_build_data_t* dtd_data;
1252
1253                                         if (errors->len) {
1254                                                 report_failure("Dtd Preparser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1255                                                 continue;
1256                                         }
1257
1258                                         dtd_data = dtd_parse(preparsed);
1259
1260                                         g_string_free(preparsed,TRUE);
1261
1262                                         if (dtd_data->error->len) {
1263                                                 report_failure("Dtd Parser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,dtd_data->error->str);
1264                                                 destroy_dtd_data(dtd_data);
1265                                                 continue;
1266                                         }
1267
1268                                         register_dtd(dtd_data,errors);
1269
1270                                         if (errors->len) {
1271                                                 report_failure("Dtd Registration in file: %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1272                                                 g_string_free(errors,TRUE);
1273                                                 continue;
1274                                         }
1275                                 }
1276                         }
1277
1278                         CLOSEDIR_OP(dir);
1279                 }
1280         }
1281
1282         g_free(dirname);
1283
1284         for(i=0;i<array_length(default_media_types);i++) {
1285                 if( ! g_hash_table_lookup(media_types,default_media_types[i]) ) {
1286                         g_hash_table_insert(media_types,(gpointer)default_media_types[i],&xml_ns);
1287                 }
1288         }
1289
1290         g_hash_table_foreach(xmpli_names,add_xmlpi_namespace,"xml.xmlpi");
1291
1292         g_free(dummy);
1293 }
1294
1295 static void range_delete_xml_tcp_callback(guint32 port) {
1296         dissector_delete_uint("tcp.port", port, xml_handle);
1297 }
1298
1299 static void range_add_xml_tcp_callback(guint32 port) {
1300         dissector_add_uint("tcp.port", port, xml_handle);
1301 }
1302
1303 static void apply_prefs(void) {
1304         if (pref_heuristic_media_save != pref_heuristic_media) {
1305                 if (pref_heuristic_media) {
1306                         heur_dissector_add("http", dissect_xml_heur, xml_ns.hf_tag);
1307                         heur_dissector_add("sip", dissect_xml_heur, xml_ns.hf_tag);
1308                         heur_dissector_add("media", dissect_xml_heur, xml_ns.hf_tag);
1309                         pref_heuristic_media_save = TRUE;
1310                 } else {
1311                         heur_dissector_delete("http", dissect_xml_heur, xml_ns.hf_tag);
1312                         heur_dissector_delete("sip", dissect_xml_heur, xml_ns.hf_tag);
1313                         heur_dissector_delete("media", dissect_xml_heur, xml_ns.hf_tag);
1314                         pref_heuristic_media_save = FALSE;
1315                 }
1316         }
1317
1318         if (pref_heuristic_tcp_save != pref_heuristic_tcp ) {
1319                 if (pref_heuristic_tcp) {
1320                         heur_dissector_add("tcp", dissect_xml_heur, xml_ns.hf_tag);
1321                         pref_heuristic_tcp_save = TRUE;
1322                 } else {
1323                         heur_dissector_delete("tcp", dissect_xml_heur, xml_ns.hf_tag);
1324                         pref_heuristic_tcp_save = FALSE;
1325                 }
1326         }
1327
1328         if (pref_heuristic_udp_save != pref_heuristic_udp ) {
1329                 if (pref_heuristic_udp) {
1330                         heur_dissector_add("udp", dissect_xml_heur, xml_ns.hf_tag);
1331                         pref_heuristic_udp_save = TRUE;
1332                 } else {
1333                         heur_dissector_delete("udp", dissect_xml_heur, xml_ns.hf_tag);
1334                         pref_heuristic_udp_save = FALSE;
1335                 }
1336         }
1337
1338         range_foreach(xml_tcp_range, range_delete_xml_tcp_callback);
1339         g_free(xml_tcp_range);
1340         xml_tcp_range = range_copy(global_xml_tcp_range);
1341         range_foreach(xml_tcp_range, range_add_xml_tcp_callback);
1342 }
1343
1344 void
1345 proto_register_xml(void) {
1346         static gint *ett_base[] = {
1347                 &unknown_ns.ett,
1348                 &xml_ns.ett,
1349                 &ett_dtd,
1350                 &ett_xmpli
1351         };
1352
1353         static hf_register_info hf_base[] = {
1354                 { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1355                 { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1356                 { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1357                 { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1358                 { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1359                 { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1360                 { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1361                 { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }}
1362     };
1363         module_t* xml_module;
1364
1365         hf_arr = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1366         ett_arr = g_array_new(FALSE,FALSE,sizeof(gint*));
1367
1368         g_array_append_vals(hf_arr,hf_base,array_length(hf_base));
1369         g_array_append_vals(ett_arr,ett_base,array_length(ett_base));
1370
1371         init_xml_names();
1372
1373         xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1374
1375         proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)g_array_data(hf_arr), hf_arr->len);
1376         proto_register_subtree_array((gint**)g_array_data(ett_arr), ett_arr->len);
1377
1378         xml_module = prefs_register_protocol(xml_ns.hf_tag,apply_prefs);
1379         prefs_register_bool_preference(xml_module, "heuristic", "Use Heuristics for media types",
1380                                    "Try to recognize XML for unknown media types",
1381                                    &pref_heuristic_media);
1382         prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for TCP",
1383                                    "Try to recognize XML for unknown TCP ports",
1384                                    &pref_heuristic_tcp);
1385         prefs_register_range_preference(xml_module, "tcp.port", "TCP Ports",
1386                                                                         "TCP Ports range",
1387                                                                         &global_xml_tcp_range, 65535);
1388         prefs_register_bool_preference(xml_module, "heuristic_udp", "Use Heuristics for UDP",
1389                                    "Try to recognize XML for unknown UDP ports",
1390                                    &pref_heuristic_udp);
1391         prefs_register_bool_preference(xml_module, "heuristic_unicode", "Use Unicode in heuristics",
1392                                    "Try to recognize XML encoded in Unicode (UCS-2)",
1393                                    &pref_heuristic_unicode);
1394
1395         g_array_free(hf_arr,FALSE);
1396         g_array_free(ett_arr,TRUE);
1397
1398         register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1399
1400         init_xml_parser();
1401
1402         xml_tcp_range = range_empty();
1403
1404
1405 }
1406
1407 static void add_dissector_media(gpointer k, gpointer v _U_, gpointer p _U_) {
1408         dissector_add_string("media_type", (gchar*)k, xml_handle);
1409 }
1410
1411 void
1412 proto_reg_handoff_xml(void)
1413 {
1414
1415         xml_handle = find_dissector("xml");
1416
1417         g_hash_table_foreach(media_types,add_dissector_media,NULL);
1418
1419 }