Add application/vnd.3gpp.cw+xml
[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.3gpp.cw+xml",
140         "application/vnd.wv.csp+xml",
141         "application/vnd.wv.csp.xml",
142         "application/watcherinfo+xml",
143         "application/xcap-att+xml",
144         "application/xcap-caps+xml",
145         "application/xcap-diff+xml",
146         "application/xcap-el+xml",
147         "application/xcap-error+xml",
148         "application/xcap-ns+xml",
149         "application/xml",
150         "application/xml-dtd",
151         "application/xpidf+xml",
152         "application/xslt+xml",
153         "application/x-wms-logconnectstats",
154         "application/x-wms-logplaystats",
155         "application/x-wms-sendevent",
156         "application/rss+xml",
157         "image/svg+xml",
158 };
159
160 static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child) {
161         new_child->first_child = NULL;
162         new_child->last_child = NULL;
163
164         new_child->parent = parent;
165         new_child->next_sibling = NULL;
166         new_child->prev_sibling = NULL;
167         if (parent == NULL) return;  /* root */
168
169         if (parent->first_child == NULL) {  /* the 1st child */
170                 parent->first_child = new_child;
171         } else {  /* following children */
172                 parent->last_child->next_sibling = new_child;
173                 new_child->prev_sibling = parent->last_child;
174         }
175         parent->last_child = new_child;
176 }
177
178 static void
179 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree)
180 {
181         tvbparse_t* tt;
182         tvbparse_elem_t* tok = NULL;
183         static GPtrArray* stack = NULL;
184         xml_frame_t* current_frame;
185         char* colinfo_str;
186
187         if (stack != NULL)
188                 g_ptr_array_free(stack,TRUE);
189
190         stack = g_ptr_array_new();
191         current_frame = ep_alloc(sizeof(xml_frame_t));
192         current_frame->type = XML_FRAME_ROOT;
193         current_frame->name = NULL;
194         current_frame->name_orig_case = NULL;
195         current_frame->value = NULL;
196         insert_xml_frame(NULL, current_frame);
197         g_ptr_array_add(stack,current_frame);
198
199         tt = tvbparse_init(tvb,0,-1,stack,want_ignore);
200         current_frame->start_offset = 0;
201
202         root_ns = NULL;
203
204         if (pinfo->match_string)
205                 root_ns = g_hash_table_lookup(media_types,pinfo->match_string);
206
207         if (! root_ns ) {
208                 root_ns = &xml_ns;
209                 colinfo_str = "/XML";
210         } else {
211                 colinfo_str = ep_strdup_printf("/%s",root_ns->name);
212                 ascii_strup_inplace(colinfo_str);
213         }
214
215         col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
216
217         current_frame->ns = root_ns;
218
219         current_frame->item = proto_tree_add_item(tree,current_frame->ns->hf_tag,tvb,0,-1,FALSE);
220         current_frame->tree = proto_item_add_subtree(current_frame->item,current_frame->ns->ett);
221         current_frame->last_item = current_frame->item;
222
223         while(( tok = tvbparse_get(tt, want) )) ;
224
225         pinfo->private_data = current_frame;  /* pass XML structure to the dissector calling XML */
226 }
227
228 static gboolean dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree) {
229         if (pref_heuristic_media || pref_heuristic_tcp || pref_heuristic_udp) {
230                 if (tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
231                         dissect_xml(tvb, pinfo, tree);
232                         return TRUE;
233                 } else if (pref_heuristic_unicode) {
234                         const guint8 *data = tvb_get_ephemeral_faked_unicode(tvb, 0, tvb_length(tvb)/2, TRUE);
235                         tvbuff_t *unicode_tvb = tvb_new_real_data(data, tvb_length(tvb)/2, tvb_length(tvb)/2);
236                         if (tvbparse_peek(tvbparse_init(unicode_tvb,0,-1,NULL,want_ignore), want_heur)) {
237                                 dissect_xml(unicode_tvb, pinfo, tree);
238                                 return TRUE;
239                         }
240                 }
241         }
242         return FALSE;
243 }
244
245 xml_frame_t *xml_get_tag(xml_frame_t *frame, const gchar *name) {
246         xml_frame_t *tag = NULL;
247
248         xml_frame_t *xml_item = frame->first_child;
249         while (xml_item) {
250                 if ((xml_item->type == XML_FRAME_TAG)) {
251                         if (!name) {  /* get the 1st tag */
252                         tag = xml_item;
253                                 break;
254                         } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
255                         tag = xml_item;
256                                 break;
257                         }
258                 }
259                 xml_item = xml_item->next_sibling;
260         }
261
262         return tag;
263 }
264
265 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const gchar *name) {
266         xml_frame_t *attr = NULL;
267
268         xml_frame_t *xml_item = frame->first_child;
269         while (xml_item) {
270                 if ((xml_item->type == XML_FRAME_ATTRIB) &&
271                         xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
272                 attr = xml_item;
273                         break;
274                 }
275                 xml_item = xml_item->next_sibling;
276         }
277
278         return attr;
279 }
280
281 xml_frame_t *xml_get_cdata(xml_frame_t *frame) {
282         xml_frame_t *cdata = NULL;
283
284         xml_frame_t *xml_item = frame->first_child;
285         while (xml_item) {
286                 if ((xml_item->type == XML_FRAME_CDATA)) {
287                 cdata = xml_item;
288                         break;
289                 }
290                 xml_item = xml_item->next_sibling;
291         }
292
293         return cdata;
294 }
295
296 static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
297         GPtrArray* stack = tvbparse_data;
298         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
299         int hfid;
300         gboolean is_cdata = FALSE;
301         proto_item* pi;
302         xml_frame_t* new_frame;
303
304         if (tok->id == XML_CDATA) {
305                 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
306                 is_cdata = TRUE;
307         } else if ( tok->id > 0) {
308                 hfid = tok->id;
309         } else {
310                 hfid = xml_ns.hf_cdata;
311         }
312
313         pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, FALSE);
314
315         proto_item_set_text(pi, "%s",
316                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
317
318         if (is_cdata) {
319                 new_frame = ep_alloc(sizeof(xml_frame_t));
320                 new_frame->type = XML_FRAME_CDATA;
321                 new_frame->name = NULL;
322                 new_frame->name_orig_case = NULL;
323                 new_frame->value = tvb_new_subset(tok->tvb, tok->offset, tok->len, tok->len);
324                 insert_xml_frame(current_frame, new_frame);
325                 new_frame->item = pi;
326                 new_frame->last_item = pi;
327                 new_frame->tree = NULL;
328                 new_frame->start_offset = tok->offset;
329                 new_frame->ns = NULL;
330         }
331 }
332
333 static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
334         GPtrArray* stack = tvbparse_data;
335         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
336         proto_item* pi;
337         proto_tree* pt;
338         tvbparse_elem_t* name_tok = tok->sub->next;
339         gchar* name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
340         xml_ns_t* ns = g_hash_table_lookup(xmpli_names,name);
341         xml_frame_t* new_frame;
342
343         int hf_tag;
344         gint ett;
345
346         ascii_strdown_inplace(name);
347         if (!ns) {
348                 hf_tag = hf_xmlpi;
349                 ett = ett_xmpli;
350         } else {
351                 hf_tag = ns->hf_tag;
352                 ett = ns->ett;
353         }
354
355         pi = proto_tree_add_item(current_frame->tree,hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
356
357         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
358
359         pt = proto_item_add_subtree(pi,ett);
360
361         new_frame = ep_alloc(sizeof(xml_frame_t));
362         new_frame->type = XML_FRAME_XMPLI;
363         new_frame->name = name;
364         new_frame->name_orig_case = name;
365         new_frame->value = NULL;
366         insert_xml_frame(current_frame, new_frame);
367         new_frame->item = pi;
368         new_frame->last_item = pi;
369         new_frame->tree = pt;
370         new_frame->start_offset = tok->offset;
371         new_frame->ns = ns;
372
373         g_ptr_array_add(stack,new_frame);
374
375 }
376
377 static void after_xmlpi(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
378         GPtrArray* stack = tvbparse_data;
379         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
380
381         proto_tree_add_text(current_frame->tree,
382                                                    tok->tvb, tok->offset, tok->len, "%s",
383                                                    tvb_format_text(tok->tvb,tok->offset,tok->len));
384
385         if (stack->len > 1) {
386                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
387         } else {
388                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened xmpli tag ]");
389         }
390 }
391
392 static void before_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
393         GPtrArray* stack = tvbparse_data;
394         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
395         tvbparse_elem_t* name_tok = tok->sub->next;
396         gchar* root_name;
397         gchar *name = NULL, *name_orig_case = NULL;
398         xml_ns_t* ns;
399         xml_frame_t* new_frame;
400         proto_item* pi;
401         proto_tree* pt;
402
403         if (name_tok->sub->id == XML_SCOPED_NAME) {
404                 tvbparse_elem_t* root_tok = name_tok->sub->sub;
405                 tvbparse_elem_t* leaf_tok = name_tok->sub->sub->next->next;
406                 xml_ns_t* nameroot_ns;
407
408                 root_name = (gchar*)tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
409                 name = (gchar*)tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
410                 name_orig_case = name;
411
412                 nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
413
414                 if(nameroot_ns) {
415                         ns = g_hash_table_lookup(nameroot_ns->elements,name);
416                         if (!ns) {
417                                 ns = &unknown_ns;
418                         }
419                 } else {
420                         ns = &unknown_ns;
421                 }
422
423         } else {
424                 name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
425                 name_orig_case = ep_strdup(name);
426                 ascii_strdown_inplace(name);
427
428                 if(current_frame->ns) {
429                         ns = g_hash_table_lookup(current_frame->ns->elements,name);
430
431                         if (!ns) {
432                                 if (! ( ns = g_hash_table_lookup(root_ns->elements,name) ) ) {
433                                         ns = &unknown_ns;
434                                 }
435                         }
436                 } else {
437                         ns = &unknown_ns;
438                 }
439         }
440
441         pi = proto_tree_add_item(current_frame->tree,ns->hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
442         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
443
444         pt = proto_item_add_subtree(pi,ns->ett);
445
446         new_frame = ep_alloc(sizeof(xml_frame_t));
447         new_frame->type = XML_FRAME_TAG;
448         new_frame->name = name;
449         new_frame->name_orig_case = name_orig_case;
450         new_frame->value = NULL;
451         insert_xml_frame(current_frame, new_frame);
452         new_frame->item = pi;
453         new_frame->last_item = pi;
454         new_frame->tree = pt;
455         new_frame->start_offset = tok->offset;
456         new_frame->ns = ns;
457
458         g_ptr_array_add(stack,new_frame);
459
460 }
461
462 static void after_open_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
463         GPtrArray* stack = tvbparse_data;
464         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
465
466         proto_item_append_text(current_frame->last_item,">");
467 }
468
469 static void after_closed_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
470         GPtrArray* stack = tvbparse_data;
471         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
472
473         proto_item_append_text(current_frame->last_item,"/>");
474
475         if (stack->len > 1) {
476                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
477         } else {
478                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
479         }
480 }
481
482 static void after_untag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
483         GPtrArray* stack = tvbparse_data;
484         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
485
486         proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
487
488         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
489                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
490
491         if (stack->len > 1) {
492                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
493         } else {
494                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
495                                                         "[ ERROR: Closing an unopened tag ]");
496         }
497 }
498
499 static void before_dtd_doctype(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
500         GPtrArray* stack = tvbparse_data;
501         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
502         xml_frame_t* new_frame;
503         tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
504         proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
505                                                                                            name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
506
507         proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
508
509         new_frame = ep_alloc(sizeof(xml_frame_t));
510         new_frame->type = XML_FRAME_DTD_DOCTYPE;
511         new_frame->name = (gchar*)tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
512         new_frame->name_orig_case = new_frame->name;
513         new_frame->value = NULL;
514         insert_xml_frame(current_frame, new_frame);
515         new_frame->item = dtd_item;
516         new_frame->last_item = dtd_item;
517         new_frame->tree = proto_item_add_subtree(dtd_item,ett_dtd);
518         new_frame->start_offset = tok->offset;
519         new_frame->ns = NULL;
520
521         g_ptr_array_add(stack,new_frame);
522 }
523
524 static void pop_stack(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
525         GPtrArray* stack = tvbparse_data;
526         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
527
528         if (stack->len > 1) {
529                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
530         } else {
531                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
532                                                         "[ ERROR: Closing an unopened tag ]");
533         }
534 }
535
536 static void after_dtd_close(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
537         GPtrArray* stack = tvbparse_data;
538         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
539
540         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
541                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
542         if (stack->len > 1) {
543                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
544         } else {
545                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
546         }
547 }
548
549 static void get_attrib_value(void* tvbparse_data _U_, const void* wanted_data _U_, tvbparse_elem_t* tok) {
550         tok->data = tok->sub;
551 }
552
553 static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
554         GPtrArray* stack = tvbparse_data;
555         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
556         gchar *name = NULL, *name_orig_case = NULL;
557         tvbparse_elem_t* value;
558         tvbparse_elem_t* value_part = tok->sub->next->next->data;
559         int* hfidp;
560         int hfid;
561         proto_item* pi;
562         xml_frame_t* new_frame;
563
564         name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
565         name_orig_case = ep_strdup(name);
566         ascii_strdown_inplace(name);
567
568         if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
569                 hfid = *hfidp;
570                 value = value_part;
571         } else {
572                 hfid = hf_unknowwn_attrib;
573                 value = tok;
574         }
575
576         pi = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
577         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
578
579         current_frame->last_item = pi;
580
581         new_frame = ep_alloc(sizeof(xml_frame_t));
582         new_frame->type = XML_FRAME_ATTRIB;
583         new_frame->name = name;
584         new_frame->name_orig_case = name_orig_case;
585         new_frame->value = tvb_new_subset(value_part->tvb, value_part->offset, value_part->len, value_part->len);
586         insert_xml_frame(current_frame, new_frame);
587         new_frame->item = pi;
588         new_frame->last_item = pi;
589         new_frame->tree = NULL;
590         new_frame->start_offset = tok->offset;
591         new_frame->ns = NULL;
592
593 }
594
595 static void unrecognized_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_){
596         GPtrArray* stack = tvbparse_data;
597         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
598
599         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Unrecognized text ]");
600
601 }
602
603
604
605 static void init_xml_parser(void) {
606         tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
607         tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
608
609         tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
610                                                                want_name,
611                                                                tvbparse_char(-1,":",NULL,NULL,NULL),
612                                                                want_name,
613                                                                NULL);
614
615         tvbparse_wanted_t* want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
616                                                               want_scoped_name,
617                                                               want_name,
618                                                               NULL);
619
620         tvbparse_wanted_t* want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
621                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\"','\\'),
622                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\'','\\'),
623                                                                   tvbparse_chars(-1,1,0,"0123456789",NULL,NULL,NULL),
624                                                                   want_name,
625                                                                   NULL);
626
627         tvbparse_wanted_t* want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
628                                                                   tvbparse_set_seq(-1, NULL, NULL, after_attrib,
629                                                                                    want_attr_name,
630                                                                                    tvbparse_char(-1,"=",NULL,NULL,NULL),
631                                                                                    want_attrib_value,
632                                                                                    NULL));
633
634         tvbparse_wanted_t* want_stoptag = tvbparse_set_oneof(-1,NULL,NULL,NULL,
635                                                              tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
636                                                              tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
637                                                              NULL);
638
639         tvbparse_wanted_t* want_stopxmlpi = tvbparse_string(-1,"?>",NULL,NULL,after_xmlpi);
640
641         tvbparse_wanted_t* want_comment = tvbparse_set_seq(hf_comment,NULL,NULL,after_token,
642                                                            tvbparse_string(-1,"<!--",NULL,NULL,NULL),
643                                                            tvbparse_until(-1,NULL,NULL,NULL,
644                                                                           tvbparse_string(-1,"-->",NULL,NULL,NULL),
645                                                                           TP_UNTIL_INCLUDE),
646                                                            NULL);
647
648         tvbparse_wanted_t* want_xmlpi = tvbparse_set_seq(hf_xmlpi,NULL,before_xmpli,NULL,
649                                                          tvbparse_string(-1,"<?",NULL,NULL,NULL),
650                                                          want_name,
651                                                          tvbparse_set_oneof(-1,NULL,NULL,NULL,
652                                                                             want_stopxmlpi,
653                                                                             tvbparse_set_seq(-1,NULL,NULL,NULL,
654                                                                                              want_attributes,
655                                                                                              want_stopxmlpi,
656                                                                                              NULL),
657                                                                             NULL),
658                                                          NULL);
659
660         tvbparse_wanted_t* want_closing_tag = tvbparse_set_seq(0,NULL,NULL,after_untag,
661                                                                tvbparse_char(-1, "<", NULL, NULL, NULL),
662                                                                tvbparse_char(-1, "/", NULL, NULL, NULL),
663                                                                want_tag_name,
664                                                                tvbparse_char(-1, ">", NULL, NULL, NULL),
665                                                                NULL);
666
667         tvbparse_wanted_t* want_doctype_start = tvbparse_set_seq(-1,NULL,before_dtd_doctype,NULL,
668                                                                  tvbparse_char(-1,"<",NULL,NULL,NULL),
669                                                                  tvbparse_char(-1,"!",NULL,NULL,NULL),
670                                                                  tvbparse_casestring(-1,"DOCTYPE",NULL,NULL,NULL),
671                                                                  tvbparse_set_oneof(-1,NULL,NULL,NULL,
672                                                                                     tvbparse_set_seq(-1,NULL,NULL,NULL,
673                                                                                                      want_name,
674                                                                                                      tvbparse_char(-1,"[",NULL,NULL,NULL),
675                                                                                                      NULL),
676                                                                                     tvbparse_set_seq(-1,NULL,NULL,pop_stack,
677                                                                                                      want_name,
678                                                                                                      tvbparse_set_oneof(-1,NULL,NULL,NULL,
679                                                                                                                         tvbparse_casestring(-1,"PUBLIC",NULL,NULL,NULL),
680                                                                                                                         tvbparse_casestring(-1,"SYSTEM",NULL,NULL,NULL),
681                                                                                                                         NULL),
682                                                                                                      tvbparse_until(-1,NULL,NULL,NULL,
683                                                                                                                     tvbparse_char(-1,">",NULL,NULL,NULL),
684                                                                                                                     TP_UNTIL_INCLUDE),
685                                                                                                      NULL),
686                                                                                     NULL),
687                                                                  NULL);
688
689         tvbparse_wanted_t* want_dtd_tag = tvbparse_set_seq(hf_dtd_tag,NULL,NULL,after_token,
690                                                            tvbparse_char(-1,"<",NULL,NULL,NULL),
691                                                            tvbparse_char(-1,"!",NULL,NULL,NULL),
692                                                            tvbparse_until(-1,NULL,NULL,NULL,
693                                                                           tvbparse_char(-1, ">", NULL, NULL, NULL),
694                                                                           TP_UNTIL_INCLUDE),
695                                                            NULL);
696
697         tvbparse_wanted_t* want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
698                                                        tvbparse_char(-1,"<",NULL,NULL,NULL),
699                                                        want_tag_name,
700                                                        tvbparse_set_oneof(-1,NULL,NULL,NULL,
701                                                                           tvbparse_set_seq(-1,NULL,NULL,NULL,
702                                                                                            want_attributes,
703                                                                                            want_stoptag,
704                                                                                            NULL),
705                                                                           want_stoptag,
706                                                                           NULL),
707                                                        NULL);
708
709         tvbparse_wanted_t* want_dtd_close = tvbparse_set_seq(-1,NULL,NULL,after_dtd_close,
710                                                              tvbparse_char(-1,"]",NULL,NULL,NULL),
711                                                              tvbparse_char(-1,">",NULL,NULL,NULL),
712                                                              NULL);
713
714         want_ignore = tvbparse_chars(-1,1,0," \t\r\n",NULL,NULL,NULL);
715
716
717         want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
718                                   want_comment,
719                                   want_xmlpi,
720                                   want_closing_tag,
721                                   want_doctype_start,
722                                   want_dtd_close,
723                                   want_dtd_tag,
724                                   want_tag,
725                                   tvbparse_not_chars(XML_CDATA,1,0,"<",NULL,NULL,after_token),
726                                   tvbparse_not_chars(-1,1,0," \t\r\n",NULL,NULL,unrecognized_token),
727                                   NULL);
728
729         want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
730                                        want_comment,
731                                        want_xmlpi,
732                                        want_doctype_start,
733                                        want_dtd_tag,
734                                        want_tag,
735                                        NULL);
736
737 }
738
739
740 static xml_ns_t* xml_new_namespace(GHashTable* hash, gchar* name, ...) {
741         xml_ns_t* ns = g_malloc(sizeof(xml_ns_t));
742         va_list ap;
743         gchar* attr_name;
744
745         ns->name = g_strdup(name);
746         ns->hf_tag = -1;
747         ns->hf_cdata = -1;
748         ns->ett = -1;
749         ns->attributes = g_hash_table_new(g_str_hash,g_str_equal);
750         ns->elements = g_hash_table_new(g_str_hash,g_str_equal);
751
752         va_start(ap,name);
753
754         while(( attr_name = va_arg(ap,gchar*) )) {
755                 int* hfp = g_malloc(sizeof(int));
756                 *hfp = -1;
757                 g_hash_table_insert(ns->attributes,g_strdup(attr_name),hfp);
758         };
759
760         va_end(ap);
761
762         g_hash_table_insert(hash,ns->name,ns);
763
764         return ns;
765 }
766
767
768 static void add_xml_field(GArray* hfs, int* p_id, gchar* name, gchar* fqn) {
769         hf_register_info hfri;
770
771         hfri.p_id = p_id;
772         hfri.hfinfo.name = name;
773         hfri.hfinfo.abbrev = fqn;
774         hfri.hfinfo.type = FT_STRING;
775         hfri.hfinfo.display = BASE_NONE;
776         hfri.hfinfo.strings = NULL;
777         hfri.hfinfo.bitmask = 0x0;
778         hfri.hfinfo.blurb = NULL;
779         hfri.hfinfo.id = 0;
780         hfri.hfinfo.parent = 0;
781         hfri.hfinfo.ref_type = HF_REF_TYPE_NONE;
782         hfri.hfinfo.bitshift = 0;
783         hfri.hfinfo.same_name_next = NULL;
784         hfri.hfinfo.same_name_prev = NULL;
785
786         g_array_append_val(hfs,hfri);
787 }
788
789 static void add_xml_attribute_names(gpointer k, gpointer v, gpointer p) {
790         struct _attr_reg_data* d = p;
791         gchar* basename = g_strdup_printf("%s.%s",d->basename,(gchar*)k);
792         add_xml_field(d->hf, (int*) v, (gchar*)k, basename);
793 }
794
795
796 static void add_xmlpi_namespace(gpointer k _U_, gpointer v, gpointer p) {
797         xml_ns_t* ns = v;
798         gchar* basename = g_strdup_printf("%s.%s",(gchar*)p,ns->name);
799         gint* ett_p = &(ns->ett);
800         struct _attr_reg_data d;
801
802         add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
803
804         g_array_append_val(ett_arr,ett_p);
805
806         d.basename = basename;
807         d.hf = hf_arr;
808
809         g_hash_table_foreach(ns->attributes,add_xml_attribute_names,&d);
810
811 }
812
813 static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
814         g_free(dtd_data->proto_name);
815         g_free(dtd_data->media_type);
816         g_free(dtd_data->description);
817         g_free(dtd_data->proto_root);
818
819         g_string_free(dtd_data->error,TRUE);
820
821         while(dtd_data->elements->len) {
822                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
823                 g_ptr_array_free(nl->list,TRUE);
824                 g_free(nl);
825         }
826
827         g_ptr_array_free(dtd_data->elements,TRUE);
828
829         while(dtd_data->attributes->len) {
830                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->attributes,0);
831                 g_ptr_array_free(nl->list,TRUE);
832                 g_free(nl);
833         }
834
835         g_ptr_array_free(dtd_data->attributes,TRUE);
836
837         g_free(dtd_data);
838 }
839
840 static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
841         gchar* key = g_strdup(k);
842         int* value = g_malloc(sizeof(int));
843         GHashTable* dst = p;
844
845         *value = -1;
846         g_hash_table_insert(dst,key,value);
847
848 }
849
850 static GHashTable* copy_attributes_hash(GHashTable* src) {
851         GHashTable* dst = g_hash_table_new(g_str_hash,g_str_equal);
852
853         g_hash_table_foreach(src,copy_attrib_item,dst);
854
855         return dst;
856 }
857
858 static xml_ns_t* duplicate_element(xml_ns_t* orig) {
859         xml_ns_t* new_item  = g_malloc(sizeof(xml_ns_t));
860         guint i;
861
862         new_item->name = g_strdup(orig->name);
863         new_item->hf_tag = -1;
864         new_item->hf_cdata = -1;
865         new_item->ett = -1;
866         new_item->attributes = copy_attributes_hash(orig->attributes);
867         new_item->elements =  g_hash_table_new(g_str_hash,g_str_equal);
868         new_item->element_names = g_ptr_array_new();
869
870         for(i=0; i < orig->element_names->len; i++) {
871                 g_ptr_array_add(new_item->element_names,
872                                                    g_ptr_array_index(orig->element_names,i));
873         }
874
875         return new_item;
876 }
877
878 static gchar* fully_qualified_name(GPtrArray* hier, gchar* name, gchar* proto_name) {
879         guint i;
880         GString* s = g_string_new(proto_name);
881         gchar* str;
882         g_string_append(s,".");
883
884         for (i = 1; i < hier->len; i++) {
885                 g_string_append_printf(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
886         }
887
888         g_string_append(s,name);
889         str = s->str;
890         g_string_free(s,FALSE);
891
892         return str;
893 }
894
895
896 static xml_ns_t* make_xml_hier(gchar* elem_name,
897                                xml_ns_t* root,
898                                GHashTable* elements,
899                                GPtrArray* hier,
900                                GString* error,
901                                GArray* hfs,
902                                GArray* etts,
903                                char* proto_name) {
904         xml_ns_t* new;
905         xml_ns_t* orig;
906         gchar* fqn;
907         gint* ett_p;
908         struct _attr_reg_data d;
909         gboolean recurred = FALSE;
910         guint i;
911
912         if ( g_str_equal(elem_name,root->name) ) {
913                 return NULL;
914         }
915
916         if (! ( orig = g_hash_table_lookup(elements,elem_name) )) {
917                 g_string_append_printf(error,"element '%s' is not defined\n", elem_name);
918                 return NULL;
919         }
920
921         for (i = 0; i < hier->len; i++) {
922                 if( strcmp(elem_name,(gchar*) g_ptr_array_index(hier,i) ) == 0 ) {
923                         recurred = TRUE;
924                 }
925         }
926
927         if (recurred) {
928                 return NULL;
929         }
930
931         fqn = fully_qualified_name(hier,elem_name,proto_name);
932
933         new = duplicate_element(orig);
934         new->fqn = fqn;
935
936         add_xml_field(hfs, &(new->hf_tag), g_strdup(elem_name), fqn);
937         add_xml_field(hfs, &(new->hf_cdata), g_strdup(elem_name), fqn);
938
939         ett_p = &new->ett;
940         g_array_append_val(etts,ett_p);
941
942         d.basename = fqn;
943         d.hf = hfs;
944
945         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
946
947         while(new->element_names->len) {
948                 gchar* child_name = g_ptr_array_remove_index(new->element_names,0);
949                 xml_ns_t* child_element = NULL;
950
951                 g_ptr_array_add(hier,elem_name);
952                 child_element = make_xml_hier(child_name, root, elements, hier,error,hfs,etts,proto_name);
953                 g_ptr_array_remove_index_fast(hier,hier->len - 1);
954
955                 if (child_element) {
956                         g_hash_table_insert(new->elements,child_element->name,child_element);
957                 }
958         }
959
960         g_ptr_array_free(new->element_names,TRUE);
961         new->element_names = NULL;
962         return new;
963 }
964
965 static gboolean free_both(gpointer k, gpointer v, gpointer p _U_) {
966         g_free(k);
967         g_free(v);
968         return TRUE;
969 }
970
971 static gboolean free_elements(gpointer k _U_, gpointer v, gpointer p _U_) {
972         xml_ns_t* e = v;
973         g_free(e->name);
974         g_hash_table_foreach_remove(e->attributes,free_both,NULL);
975         g_hash_table_destroy(e->attributes);
976         g_hash_table_destroy(e->elements);
977
978         while (e->element_names->len) {
979                 g_free(g_ptr_array_remove_index(e->element_names,0));
980         }
981
982         g_ptr_array_free(e->element_names,TRUE);
983         g_free(e);
984
985         return TRUE;
986 }
987
988 static void register_dtd(dtd_build_data_t* dtd_data, GString* errors) {
989         GHashTable* elements = g_hash_table_new(g_str_hash,g_str_equal);
990         gchar* root_name = NULL;
991         xml_ns_t* root_element = NULL;
992         GArray* hfs;
993         GArray* etts;
994         GPtrArray* hier;
995         gchar* curr_name;
996         GPtrArray* element_names = g_ptr_array_new();
997
998         /* we first populate elements with the those coming from the parser */
999         while(dtd_data->elements->len) {
1000                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->elements,0);
1001                 xml_ns_t* element = g_malloc(sizeof(xml_ns_t));
1002
1003                 /* we will use the first element found as root in case no other one was given. */
1004                 if (root_name == NULL)
1005                         root_name = g_strdup(nl->name);
1006
1007                 element->name = nl->name;
1008                 element->element_names = nl->list;
1009                 element->hf_tag = -1;
1010                 element->hf_cdata = -1;
1011                 element->ett = -1;
1012                 element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1013                 element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1014
1015                 if( g_hash_table_lookup(elements,element->name) ) {
1016                         g_string_append_printf(errors,"element %s defined more than once\n", element->name);
1017                         free_elements(NULL,element,NULL);
1018                 } else {
1019                         g_hash_table_insert(elements,element->name,element);
1020                         g_ptr_array_add(element_names,g_strdup(element->name));
1021                 }
1022
1023                 g_free(nl);
1024         }
1025
1026         /* then we add the attributes to its relative elements */
1027         while(dtd_data->attributes->len) {
1028                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
1029                 xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
1030
1031                 if (element) {
1032                 while(nl->list->len) {
1033                     gchar* name = g_ptr_array_remove_index(nl->list,0);
1034                     int* id_p = g_malloc(sizeof(int));
1035
1036                     *id_p = -1;
1037                     g_hash_table_insert(element->attributes,name,id_p);
1038                 }
1039                 }
1040                 else {
1041             g_string_append_printf(errors,"element %s is not defined\n", nl->name);
1042                 }
1043
1044                 g_free(nl->name);
1045                 g_ptr_array_free(nl->list,TRUE);
1046                 g_free(nl);
1047         }
1048
1049         /* if a proto_root is defined in the dtd we'll use that as root */
1050         if( dtd_data->proto_root ) {
1051                 g_free(root_name);
1052                 root_name = g_strdup(dtd_data->proto_root);
1053         }
1054
1055         /* we use a stack with the names to avoid recurring infinitelly */
1056         hier = g_ptr_array_new();
1057
1058         /*
1059          * if a proto name was given in the dtd the dtd will be used as a protocol
1060          * or else the dtd will be loaded as a branch of the xml namespace
1061          */
1062         if( ! dtd_data->proto_name ) {
1063                 hfs = hf_arr;
1064                 etts = ett_arr;
1065                 g_ptr_array_add(hier,g_strdup("xml"));
1066                 root_element = &xml_ns;
1067         } else {
1068                 /*
1069                  * if we were given a proto_name the namespace will be registered
1070                  * as an independent protocol with its own hf and ett arrays.
1071                  */
1072                 hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1073                 etts = g_array_new(FALSE,FALSE,sizeof(gint*));
1074         }
1075
1076         /* the root element of the dtd's namespace */
1077         root_element = g_malloc(sizeof(xml_ns_t));
1078         root_element->name = g_strdup(root_name);
1079         root_element->fqn = dtd_data->proto_name ? g_strdup(dtd_data->proto_name) : root_element->name;
1080         root_element->hf_tag = -1;
1081         root_element->hf_cdata = -1;
1082         root_element->ett = -1;
1083         root_element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1084         root_element->element_names = element_names;
1085
1086         /*
1087          * we can either create a namespace as a flat namespace
1088          * in which all the elements are at the root level
1089          * or we can create a recursive namespace
1090          */
1091         if (dtd_data->recursion) {
1092                 xml_ns_t* orig_root;
1093
1094                 make_xml_hier(root_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1095
1096                 g_hash_table_insert(root_element->elements,root_element->name,root_element);
1097
1098                 orig_root = g_hash_table_lookup(elements,root_name);
1099
1100                 /* if the root element was defined copy its attrlist to the child */
1101                 if(orig_root) {
1102                         struct _attr_reg_data d;
1103
1104                         d.basename = dtd_data->proto_name;
1105                         d.hf = hfs;
1106
1107                         root_element->attributes = copy_attributes_hash(orig_root->attributes);
1108                         g_hash_table_foreach(root_element->attributes,add_xml_attribute_names,&d);
1109                 } else {
1110                         root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1111                 }
1112
1113                 /* we then create all the sub hierachies to catch the recurred cases */
1114                 g_ptr_array_add(hier,root_name);
1115
1116                 while(root_element->element_names->len) {
1117                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1118
1119                         if( ! g_hash_table_lookup(root_element->elements,curr_name) ) {
1120                                 xml_ns_t* new = make_xml_hier(curr_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1121                                 g_hash_table_insert(root_element->elements,new->name,new);
1122                         }
1123
1124                         g_free(curr_name);
1125                 }
1126
1127         } else {
1128                 /* a flat namespace */
1129                 g_ptr_array_add(hier,root_name);
1130
1131                 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1132
1133                 while(root_element->element_names->len) {
1134                         xml_ns_t* new;
1135                         gint* ett_p;
1136                         struct _attr_reg_data d;
1137
1138                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1139                         new = duplicate_element(g_hash_table_lookup(elements,curr_name));
1140                         new->fqn = fully_qualified_name(hier, curr_name, root_name);
1141
1142                         add_xml_field(hfs, &(new->hf_tag), curr_name, new->fqn);
1143                         add_xml_field(hfs, &(new->hf_cdata), curr_name, new->fqn);
1144
1145                         d.basename = new->fqn;
1146                         d.hf = hfs;
1147
1148                         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
1149
1150                         ett_p = &new->ett;
1151                         g_array_append_val(etts,ett_p);
1152
1153                         g_ptr_array_free(new->element_names,TRUE);
1154
1155                         g_hash_table_insert(root_element->elements,new->name,new);
1156                 }
1157         }
1158
1159         g_ptr_array_free(element_names,TRUE);
1160
1161         g_ptr_array_free(hier,TRUE);
1162
1163         /*
1164          * if we were given a proto_name the namespace will be registered
1165          * as an independent protocol.
1166          */
1167         if( dtd_data->proto_name ) {
1168                 gint* ett_p;
1169
1170                 if ( ! dtd_data->description) {
1171                         dtd_data->description = g_strdup(root_name);
1172                 }
1173
1174                 ett_p = &root_element->ett;
1175                 g_array_append_val(etts,ett_p);
1176
1177                 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1178
1179                 root_element->hf_tag = proto_register_protocol(dtd_data->description, dtd_data->proto_name, dtd_data->proto_name);
1180                 proto_register_field_array(root_element->hf_tag, (hf_register_info*)g_array_data(hfs), hfs->len);
1181                 proto_register_subtree_array((gint**)g_array_data(etts), etts->len);
1182
1183                 if (dtd_data->media_type) {
1184                         g_hash_table_insert(media_types,dtd_data->media_type,root_element);
1185                         dtd_data->media_type = NULL;
1186                 }
1187
1188                 dtd_data->description = NULL;
1189                 dtd_data->proto_name = NULL;
1190                 g_array_free(hfs,FALSE);
1191                 g_array_free(etts,TRUE);
1192         }
1193
1194         g_hash_table_insert(xml_ns.elements,root_element->name,root_element);
1195
1196         g_hash_table_foreach_remove(elements,free_elements,NULL);
1197         g_hash_table_destroy(elements);
1198
1199         destroy_dtd_data(dtd_data);
1200         g_free(root_name);
1201 }
1202
1203 #  define DIRECTORY_T GDir
1204 #  define FILE_T gchar
1205 #  define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1206 #  define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1207 #  define GETFNAME_OP(file) (file);
1208 #  define CLOSEDIR_OP(dir) g_dir_close(dir)
1209
1210 static void init_xml_names(void) {
1211         xml_ns_t* xmlpi_xml_ns;
1212         guint i;
1213         DIRECTORY_T* dir;
1214         const FILE_T* file;
1215         const gchar* filename;
1216         gchar* dirname;
1217
1218         GError** dummy = g_malloc(sizeof(GError *));
1219         *dummy = NULL;
1220
1221         xmpli_names = g_hash_table_new(g_str_hash,g_str_equal);
1222         media_types = g_hash_table_new(g_str_hash,g_str_equal);
1223
1224         unknown_ns.elements = xml_ns.elements = g_hash_table_new(g_str_hash,g_str_equal);
1225         unknown_ns.attributes = xml_ns.attributes = g_hash_table_new(g_str_hash,g_str_equal);
1226
1227         xmlpi_xml_ns = xml_new_namespace(xmpli_names,"xml","version","encoding","standalone",NULL);
1228
1229         g_hash_table_destroy(xmlpi_xml_ns->elements);
1230         xmlpi_xml_ns->elements = NULL;
1231
1232
1233         dirname = get_persconffile_path("dtds", FALSE, FALSE);
1234
1235         if (test_for_directory(dirname) != EISDIR) {
1236                 /* Although dir isn't a directory it may still use memory */
1237                 g_free(dirname);
1238                 dirname = get_datafile_path("dtds");
1239         }
1240
1241         if (test_for_directory(dirname) == EISDIR) {
1242
1243                 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1244                         while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1245                                 guint namelen;
1246                                 filename = GETFNAME_OP(file);
1247
1248                                 namelen = (int)strlen(filename);
1249                                 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4),".dtd")  == 0 ) ) {
1250                                         GString* errors = g_string_new("");
1251                                         GString* preparsed = dtd_preparse(dirname, filename, errors);
1252                                         dtd_build_data_t* dtd_data;
1253
1254                                         if (errors->len) {
1255                                                 report_failure("Dtd Preparser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1256                                                 continue;
1257                                         }
1258
1259                                         dtd_data = dtd_parse(preparsed);
1260
1261                                         g_string_free(preparsed,TRUE);
1262
1263                                         if (dtd_data->error->len) {
1264                                                 report_failure("Dtd Parser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,dtd_data->error->str);
1265                                                 destroy_dtd_data(dtd_data);
1266                                                 continue;
1267                                         }
1268
1269                                         register_dtd(dtd_data,errors);
1270
1271                                         if (errors->len) {
1272                                                 report_failure("Dtd Registration in file: %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1273                                                 g_string_free(errors,TRUE);
1274                                                 continue;
1275                                         }
1276                                 }
1277                         }
1278
1279                         CLOSEDIR_OP(dir);
1280                 }
1281         }
1282
1283         g_free(dirname);
1284
1285         for(i=0;i<array_length(default_media_types);i++) {
1286                 if( ! g_hash_table_lookup(media_types,default_media_types[i]) ) {
1287                         g_hash_table_insert(media_types,(gpointer)default_media_types[i],&xml_ns);
1288                 }
1289         }
1290
1291         g_hash_table_foreach(xmpli_names,add_xmlpi_namespace,"xml.xmlpi");
1292
1293         g_free(dummy);
1294 }
1295
1296 static void range_delete_xml_tcp_callback(guint32 port) {
1297         dissector_delete_uint("tcp.port", port, xml_handle);
1298 }
1299
1300 static void range_add_xml_tcp_callback(guint32 port) {
1301         dissector_add_uint("tcp.port", port, xml_handle);
1302 }
1303
1304 static void apply_prefs(void) {
1305         if (pref_heuristic_media_save != pref_heuristic_media) {
1306                 if (pref_heuristic_media) {
1307                         heur_dissector_add("http", dissect_xml_heur, xml_ns.hf_tag);
1308                         heur_dissector_add("sip", dissect_xml_heur, xml_ns.hf_tag);
1309                         heur_dissector_add("media", dissect_xml_heur, xml_ns.hf_tag);
1310                         pref_heuristic_media_save = TRUE;
1311                 } else {
1312                         heur_dissector_delete("http", dissect_xml_heur, xml_ns.hf_tag);
1313                         heur_dissector_delete("sip", dissect_xml_heur, xml_ns.hf_tag);
1314                         heur_dissector_delete("media", dissect_xml_heur, xml_ns.hf_tag);
1315                         pref_heuristic_media_save = FALSE;
1316                 }
1317         }
1318
1319         if (pref_heuristic_tcp_save != pref_heuristic_tcp ) {
1320                 if (pref_heuristic_tcp) {
1321                         heur_dissector_add("tcp", dissect_xml_heur, xml_ns.hf_tag);
1322                         pref_heuristic_tcp_save = TRUE;
1323                 } else {
1324                         heur_dissector_delete("tcp", dissect_xml_heur, xml_ns.hf_tag);
1325                         pref_heuristic_tcp_save = FALSE;
1326                 }
1327         }
1328
1329         if (pref_heuristic_udp_save != pref_heuristic_udp ) {
1330                 if (pref_heuristic_udp) {
1331                         heur_dissector_add("udp", dissect_xml_heur, xml_ns.hf_tag);
1332                         pref_heuristic_udp_save = TRUE;
1333                 } else {
1334                         heur_dissector_delete("udp", dissect_xml_heur, xml_ns.hf_tag);
1335                         pref_heuristic_udp_save = FALSE;
1336                 }
1337         }
1338
1339         range_foreach(xml_tcp_range, range_delete_xml_tcp_callback);
1340         g_free(xml_tcp_range);
1341         xml_tcp_range = range_copy(global_xml_tcp_range);
1342         range_foreach(xml_tcp_range, range_add_xml_tcp_callback);
1343 }
1344
1345 void
1346 proto_register_xml(void) {
1347         static gint *ett_base[] = {
1348                 &unknown_ns.ett,
1349                 &xml_ns.ett,
1350                 &ett_dtd,
1351                 &ett_xmpli
1352         };
1353
1354         static hf_register_info hf_base[] = {
1355                 { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1356                 { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1357                 { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1358                 { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1359                 { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1360                 { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1361                 { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1362                 { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }}
1363     };
1364         module_t* xml_module;
1365
1366         hf_arr = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1367         ett_arr = g_array_new(FALSE,FALSE,sizeof(gint*));
1368
1369         g_array_append_vals(hf_arr,hf_base,array_length(hf_base));
1370         g_array_append_vals(ett_arr,ett_base,array_length(ett_base));
1371
1372         init_xml_names();
1373
1374         xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1375
1376         proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)g_array_data(hf_arr), hf_arr->len);
1377         proto_register_subtree_array((gint**)g_array_data(ett_arr), ett_arr->len);
1378
1379         xml_module = prefs_register_protocol(xml_ns.hf_tag,apply_prefs);
1380         prefs_register_bool_preference(xml_module, "heuristic", "Use Heuristics for media types",
1381                                    "Try to recognize XML for unknown media types",
1382                                    &pref_heuristic_media);
1383         prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for TCP",
1384                                    "Try to recognize XML for unknown TCP ports",
1385                                    &pref_heuristic_tcp);
1386         prefs_register_range_preference(xml_module, "tcp.port", "TCP Ports",
1387                                                                         "TCP Ports range",
1388                                                                         &global_xml_tcp_range, 65535);
1389         prefs_register_bool_preference(xml_module, "heuristic_udp", "Use Heuristics for UDP",
1390                                    "Try to recognize XML for unknown UDP ports",
1391                                    &pref_heuristic_udp);
1392         prefs_register_bool_preference(xml_module, "heuristic_unicode", "Use Unicode in heuristics",
1393                                    "Try to recognize XML encoded in Unicode (UCS-2)",
1394                                    &pref_heuristic_unicode);
1395
1396         g_array_free(hf_arr,FALSE);
1397         g_array_free(ett_arr,TRUE);
1398
1399         register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1400
1401         init_xml_parser();
1402
1403         xml_tcp_range = range_empty();
1404
1405
1406 }
1407
1408 static void add_dissector_media(gpointer k, gpointer v _U_, gpointer p _U_) {
1409         dissector_add_string("media_type", (gchar*)k, xml_handle);
1410 }
1411
1412 void
1413 proto_reg_handoff_xml(void)
1414 {
1415
1416         xml_handle = find_dissector("xml");
1417
1418         g_hash_table_foreach(media_types,add_dissector_media,NULL);
1419
1420 }