Remove unneeded #include <stdio.h>
[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 <ctype.h>
39 #include <stdlib.h>
40 #include <string.h>
41 #include <stdarg.h>
42 #include <errno.h>
43
44 #include <glib.h>
45
46 #include <wsutil/str_util.h>
47
48 #include <epan/emem.h>
49 #include <epan/packet.h>
50 #include <epan/tvbparse.h>
51 #include <epan/dtd.h>
52 #include <epan/report_err.h>
53 #include <epan/filesystem.h>
54 #include <epan/prefs.h>
55 #include <epan/garrayfix.h>
56
57 #include "packet-xml.h"
58
59 struct _attr_reg_data {
60         GArray* hf;
61         gchar* basename;
62 };
63
64
65 static gint ett_dtd = -1;
66 static gint ett_xmpli = -1;
67
68 static int hf_unknowwn_attrib = -1;
69 static int hf_comment = -1;
70 static int hf_xmlpi = -1;
71 static int hf_dtd_tag = -1;
72 static int hf_doctype = -1;
73
74 /* dissector handles */
75 static dissector_handle_t xml_handle;
76
77 /* parser definitions */
78 static tvbparse_wanted_t* want;
79 static tvbparse_wanted_t* want_ignore;
80 static tvbparse_wanted_t* want_heur;
81
82 static GHashTable* xmpli_names;
83 static GHashTable* media_types;
84
85 static xml_ns_t xml_ns = {"xml","/",-1,-1,-1,NULL,NULL,NULL};
86 static xml_ns_t unknown_ns = {"unknown","?",-1,-1,-1,NULL,NULL,NULL};
87 static xml_ns_t* root_ns;
88
89 static gboolean pref_heuristic_media = FALSE;
90 static gboolean pref_heuristic_tcp = FALSE;
91 static gboolean pref_heuristic_udp = FALSE;
92 static gboolean pref_heuristic_media_save = FALSE;
93 static gboolean pref_heuristic_tcp_save = FALSE;
94 static gboolean pref_heuristic_udp_save = FALSE;
95 static range_t *global_xml_tcp_range = NULL;
96 static range_t *xml_tcp_range = NULL;
97
98 #define XML_CDATA -1000
99 #define XML_SCOPED_NAME -1001
100
101
102 GArray* hf_arr;
103 GArray* ett_arr;
104
105 static const gchar* default_media_types[] = {
106         "text/xml",
107         "text/vnd.wap.wml",
108         "text/vnd.wap.si",
109         "text/vnd.wap.sl",
110         "text/vnd.wap.co",
111         "text/vnd.wap.emn",
112         "application/auth-policy+xml",
113         "application/cpim-pidf+xml",
114         "application/cpl+xml",
115         "application/mathml+xml",
116         "application/media_control+xml",
117         "application/note+xml",
118         "application/pidf+xml",
119         "application/poc-settings+xml",
120         "application/rdf+xml",
121         "application/reginfo+xml",
122         "application/resource-lists+xml",
123         "application/rlmi+xml",
124         "application/rls-services+xml",
125         "application/smil",
126         "application/simple-filter+xml",
127         "application/soap+xml",
128         "application/vnd.etsi.aoc+xml",
129         "application/vnd.etsi.cug+xml",
130         "application/vnd.etsi.iptvcommand+xml",
131         "application/vnd.etsi.iptvdiscovery+xml",
132         "application/vnd.etsi.iptvprofile+xml",
133         "application/vnd.etsi.iptvsad-bc+xml",
134         "application/vnd.etsi.iptvsad-cod+xml",
135         "application/vnd.etsi.iptvsad-npvr+xml",
136         "application/vnd.etsi.iptvueprofile+xml",
137         "application/vnd.etsi.mcid+xml",
138         "application/vnd.etsi.sci+xml",
139         "application/vnd.etsi.simservs+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->firts_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->firts_child == NULL) {  /* the 1st child */
170                 parent->firts_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              && tvbparse_peek(tvbparse_init(tvb,0,-1,NULL,want_ignore), want_heur)) {
231                 dissect_xml(tvb, pinfo, tree);
232                 return TRUE;
233         } else {
234                 return FALSE;
235         }
236 }
237
238 xml_frame_t *xml_get_tag(xml_frame_t *frame, const gchar *name) {
239         xml_frame_t *tag = NULL;
240
241         xml_frame_t *xml_item = frame->firts_child;
242         while (xml_item) {
243                 if ((xml_item->type == XML_FRAME_TAG)) {
244                         if (!name) {  /* get the 1st tag */
245                         tag = xml_item;
246                                 break;
247                         } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
248                         tag = xml_item;
249                                 break;
250                         }
251                 }
252                 xml_item = xml_item->next_sibling;
253         }
254
255         return tag;
256 }
257
258 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const gchar *name) {
259         xml_frame_t *attr = NULL;
260
261         xml_frame_t *xml_item = frame->firts_child;
262         while (xml_item) {
263                 if ((xml_item->type == XML_FRAME_ATTRIB) && 
264                         xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
265                 attr = xml_item;
266                         break;
267                 }
268                 xml_item = xml_item->next_sibling;
269         }
270
271         return attr;
272 }
273
274 xml_frame_t *xml_get_cdata(xml_frame_t *frame) {
275         xml_frame_t *cdata = NULL;
276
277         xml_frame_t *xml_item = frame->firts_child;
278         while (xml_item) {
279                 if ((xml_item->type == XML_FRAME_CDATA)) {
280                 cdata = xml_item;
281                         break;
282                 }
283                 xml_item = xml_item->next_sibling;
284         }
285
286         return cdata;
287 }
288
289 static void after_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
290         GPtrArray* stack = tvbparse_data;
291         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
292         int hfid;
293         gboolean is_cdata = FALSE;
294         proto_item* pi;
295         xml_frame_t* new_frame;
296
297         if (tok->id == XML_CDATA) {
298                 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
299                 is_cdata = TRUE;
300         } else if ( tok->id > 0) {
301                 hfid = tok->id;
302         } else {
303                 hfid = xml_ns.hf_cdata;
304         }
305
306         pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, FALSE);
307
308         proto_item_set_text(pi, "%s",
309                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
310
311         if (is_cdata) {
312                 new_frame = ep_alloc(sizeof(xml_frame_t));
313                 new_frame->type = XML_FRAME_CDATA;
314                 new_frame->name = NULL;
315                 new_frame->name_orig_case = NULL;
316                 new_frame->value = tvb_new_subset(tok->tvb, tok->offset, tok->len, tok->len);
317                 insert_xml_frame(current_frame, new_frame);
318                 new_frame->item = pi;
319                 new_frame->last_item = pi;
320                 new_frame->tree = NULL;
321                 new_frame->start_offset = tok->offset;
322                 new_frame->ns = NULL;
323         }
324 }
325
326 static void before_xmpli(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
327         GPtrArray* stack = tvbparse_data;
328         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
329         proto_item* pi;
330         proto_tree* pt;
331         tvbparse_elem_t* name_tok = tok->sub->next;
332         gchar* name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
333         xml_ns_t* ns = g_hash_table_lookup(xmpli_names,name);
334         xml_frame_t* new_frame;
335
336         int hf_tag;
337         gint ett;
338
339         ascii_strdown_inplace(name);
340         if (!ns) {
341                 hf_tag = hf_xmlpi;
342                 ett = ett_xmpli;
343         } else {
344                 hf_tag = ns->hf_tag;
345                 ett = ns->ett;
346         }
347
348         pi = proto_tree_add_item(current_frame->tree,hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
349
350         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
351
352         pt = proto_item_add_subtree(pi,ett);
353
354         new_frame = ep_alloc(sizeof(xml_frame_t));
355         new_frame->type = XML_FRAME_XMPLI;
356         new_frame->name = name;
357         new_frame->name_orig_case = name;
358         new_frame->value = NULL;
359         insert_xml_frame(current_frame, new_frame);
360         new_frame->item = pi;
361         new_frame->last_item = pi;
362         new_frame->tree = pt;
363         new_frame->start_offset = tok->offset;
364         new_frame->ns = ns;
365
366         g_ptr_array_add(stack,new_frame);
367
368 }
369
370 static void after_xmlpi(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
371         GPtrArray* stack = tvbparse_data;
372         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
373
374         proto_tree_add_text(current_frame->tree,
375                                                    tok->tvb, tok->offset, tok->len, "%s",
376                                                    tvb_format_text(tok->tvb,tok->offset,tok->len));
377
378         if (stack->len > 1) {
379                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
380         } else {
381                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened xmpli tag ]");
382         }
383 }
384
385 static void before_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
386         GPtrArray* stack = tvbparse_data;
387         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
388         tvbparse_elem_t* name_tok = tok->sub->next;
389         gchar* root_name;
390         gchar *name = NULL, *name_orig_case = NULL;
391         xml_ns_t* ns;
392         xml_frame_t* new_frame;
393         proto_item* pi;
394         proto_tree* pt;
395
396         if (name_tok->sub->id == XML_SCOPED_NAME) {
397                 tvbparse_elem_t* root_tok = name_tok->sub->sub;
398                 tvbparse_elem_t* leaf_tok = name_tok->sub->sub->next->next;
399                 xml_ns_t* nameroot_ns;
400
401                 root_name = (gchar*)tvb_get_ephemeral_string(root_tok->tvb,root_tok->offset,root_tok->len);
402                 name = (gchar*)tvb_get_ephemeral_string(leaf_tok->tvb,leaf_tok->offset,leaf_tok->len);
403                 name_orig_case = name;
404
405                 nameroot_ns = g_hash_table_lookup(xml_ns.elements,root_name);
406
407                 if(nameroot_ns) {
408                         ns = g_hash_table_lookup(nameroot_ns->elements,name);
409                         if (!ns) {
410                                 ns = &unknown_ns;
411                         }
412                 } else {
413                         ns = &unknown_ns;
414                 }
415
416         } else {
417                 name = tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
418                 name_orig_case = ep_strdup(name);
419                 ascii_strdown_inplace(name);
420
421                 if(current_frame->ns) {
422                         ns = g_hash_table_lookup(current_frame->ns->elements,name);
423
424                         if (!ns) {
425                                 if (! ( ns = g_hash_table_lookup(root_ns->elements,name) ) ) {
426                                         ns = &unknown_ns;
427                                 }
428                         }
429                 } else {
430                         ns = &unknown_ns;
431                 }
432         }
433
434         pi = proto_tree_add_item(current_frame->tree,ns->hf_tag,tok->tvb,tok->offset,tok->len,FALSE);
435         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,(name_tok->offset - tok->offset) + name_tok->len));
436
437         pt = proto_item_add_subtree(pi,ns->ett);
438
439         new_frame = ep_alloc(sizeof(xml_frame_t));
440         new_frame->type = XML_FRAME_TAG;
441         new_frame->name = name;
442         new_frame->name_orig_case = name_orig_case;
443         new_frame->value = NULL;
444         insert_xml_frame(current_frame, new_frame);
445         new_frame->item = pi;
446         new_frame->last_item = pi;
447         new_frame->tree = pt;
448         new_frame->start_offset = tok->offset;
449         new_frame->ns = ns;
450
451         g_ptr_array_add(stack,new_frame);
452
453 }
454
455 static void after_open_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
456         GPtrArray* stack = tvbparse_data;
457         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
458
459         proto_item_append_text(current_frame->last_item,">");
460 }
461
462 static void after_closed_tag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
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         if (stack->len > 1) {
469                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
470         } else {
471                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
472         }
473 }
474
475 static void after_untag(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
476         GPtrArray* stack = tvbparse_data;
477         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
478
479         proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
480
481         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
482                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
483
484         if (stack->len > 1) {
485                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
486         } else {
487                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
488                                                         "[ ERROR: Closing an unopened tag ]");
489         }
490 }
491
492 static void before_dtd_doctype(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
493         GPtrArray* stack = tvbparse_data;
494         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
495         xml_frame_t* new_frame;
496         tvbparse_elem_t* name_tok = tok->sub->next->next->next->sub->sub;
497         proto_tree* dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
498                                                                                            name_tok->tvb, name_tok->offset, name_tok->len, FALSE);
499
500         proto_item_set_text(dtd_item,"%s",tvb_format_text(tok->tvb,tok->offset,tok->len));
501
502         new_frame = ep_alloc(sizeof(xml_frame_t));
503         new_frame->type = XML_FRAME_DTD_DOCTYPE;
504         new_frame->name = (gchar*)tvb_get_ephemeral_string(name_tok->tvb,name_tok->offset,name_tok->len);
505         new_frame->name_orig_case = new_frame->name;
506         new_frame->value = NULL;
507         insert_xml_frame(current_frame, new_frame);
508         new_frame->item = dtd_item;
509         new_frame->last_item = dtd_item;
510         new_frame->tree = proto_item_add_subtree(dtd_item,ett_dtd);
511         new_frame->start_offset = tok->offset;
512         new_frame->ns = NULL;
513
514         g_ptr_array_add(stack,new_frame);
515 }
516
517 static void pop_stack(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_) {
518         GPtrArray* stack = tvbparse_data;
519         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
520
521         if (stack->len > 1) {
522                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
523         } else {
524                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,
525                                                         "[ ERROR: Closing an unopened tag ]");
526         }
527 }
528
529 static void after_dtd_close(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok){
530         GPtrArray* stack = tvbparse_data;
531         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
532
533         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"%s",
534                                                 tvb_format_text(tok->tvb,tok->offset,tok->len));
535         if (stack->len > 1) {
536                 g_ptr_array_remove_index_fast(stack,stack->len - 1);
537         } else {
538                 proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Closing an unopened tag ]");
539         }
540 }
541
542 static void get_attrib_value(void* tvbparse_data _U_, const void* wanted_data _U_, tvbparse_elem_t* tok) {
543         tok->data = tok->sub;
544 }
545
546 static void after_attrib(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok) {
547         GPtrArray* stack = tvbparse_data;
548         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
549         gchar *name = NULL, *name_orig_case = NULL;
550         tvbparse_elem_t* value;
551         tvbparse_elem_t* value_part = tok->sub->next->next->data;
552         int* hfidp;
553         int hfid;
554         proto_item* pi;
555         xml_frame_t* new_frame;
556
557         name = tvb_get_ephemeral_string(tok->sub->tvb,tok->sub->offset,tok->sub->len);
558         name_orig_case = ep_strdup(name);
559         ascii_strdown_inplace(name);
560
561         if(current_frame->ns && (hfidp = g_hash_table_lookup(current_frame->ns->attributes,name) )) {
562                 hfid = *hfidp;
563                 value = value_part;
564         } else {
565                 hfid = hf_unknowwn_attrib;
566                 value = tok;
567         }
568
569         pi = proto_tree_add_item(current_frame->tree,hfid,value->tvb,value->offset,value->len,FALSE);
570         proto_item_set_text(pi, "%s", tvb_format_text(tok->tvb,tok->offset,tok->len));
571
572         current_frame->last_item = pi;
573         
574         new_frame = ep_alloc(sizeof(xml_frame_t));
575         new_frame->type = XML_FRAME_ATTRIB;
576         new_frame->name = name;
577         new_frame->name_orig_case = name_orig_case;
578         new_frame->value = tvb_new_subset(value_part->tvb, value_part->offset, value_part->len, value_part->len);
579         insert_xml_frame(current_frame, new_frame);
580         new_frame->item = pi;
581         new_frame->last_item = pi;
582         new_frame->tree = NULL;
583         new_frame->start_offset = tok->offset;
584         new_frame->ns = NULL;
585
586 }
587
588 static void unrecognized_token(void* tvbparse_data, const void* wanted_data _U_, tvbparse_elem_t* tok _U_){
589         GPtrArray* stack = tvbparse_data;
590         xml_frame_t* current_frame = g_ptr_array_index(stack,stack->len - 1);
591
592         proto_tree_add_text(current_frame->tree,tok->tvb,tok->offset,tok->len,"[ ERROR: Unrecognized text ]");
593
594 }
595
596
597
598 static void init_xml_parser(void) {
599         tvbparse_wanted_t* want_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",NULL,NULL,NULL);
600         tvbparse_wanted_t* want_attr_name = tvbparse_chars(-1,1,0,"abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",NULL,NULL,NULL);
601
602         tvbparse_wanted_t* want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
603                                                                want_name,
604                                                                tvbparse_char(-1,":",NULL,NULL,NULL),
605                                                                want_name,
606                                                                NULL);
607
608         tvbparse_wanted_t* want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
609                                                               want_scoped_name,
610                                                               want_name,
611                                                               NULL);
612
613         tvbparse_wanted_t* want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
614                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\"','\\'),
615                                                                   tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb,'\'','\\'),
616                                                                   tvbparse_chars(-1,1,0,"0123456789",NULL,NULL,NULL),
617                                                                   want_name,
618                                                                   NULL);
619
620         tvbparse_wanted_t* want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
621                                                                   tvbparse_set_seq(-1, NULL, NULL, after_attrib,
622                                                                                    want_attr_name,
623                                                                                    tvbparse_char(-1,"=",NULL,NULL,NULL),
624                                                                                    want_attrib_value,
625                                                                                    NULL));
626
627         tvbparse_wanted_t* want_stoptag = tvbparse_set_oneof(-1,NULL,NULL,NULL,
628                                                              tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
629                                                              tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
630                                                              NULL);
631
632         tvbparse_wanted_t* want_stopxmlpi = tvbparse_string(-1,"?>",NULL,NULL,after_xmlpi);
633
634         tvbparse_wanted_t* want_comment = tvbparse_set_seq(hf_comment,NULL,NULL,after_token,
635                                                            tvbparse_string(-1,"<!--",NULL,NULL,NULL),
636                                                            tvbparse_until(-1,NULL,NULL,NULL,
637                                                                           tvbparse_string(-1,"-->",NULL,NULL,NULL),
638                                                                           TP_UNTIL_INCLUDE),
639                                                            NULL);
640
641         tvbparse_wanted_t* want_xmlpi = tvbparse_set_seq(hf_xmlpi,NULL,before_xmpli,NULL,
642                                                          tvbparse_string(-1,"<?",NULL,NULL,NULL),
643                                                          want_name,
644                                                          tvbparse_set_oneof(-1,NULL,NULL,NULL,
645                                                                             want_stopxmlpi,
646                                                                             tvbparse_set_seq(-1,NULL,NULL,NULL,
647                                                                                              want_attributes,
648                                                                                              want_stopxmlpi,
649                                                                                              NULL),
650                                                                             NULL),
651                                                          NULL);
652
653         tvbparse_wanted_t* want_closing_tag = tvbparse_set_seq(0,NULL,NULL,after_untag,
654                                                                tvbparse_char(-1, "<", NULL, NULL, NULL),
655                                                                tvbparse_char(-1, "/", NULL, NULL, NULL),
656                                                                want_tag_name,
657                                                                tvbparse_char(-1, ">", NULL, NULL, NULL),
658                                                                NULL);
659
660         tvbparse_wanted_t* want_doctype_start = tvbparse_set_seq(-1,NULL,before_dtd_doctype,NULL,
661                                                                  tvbparse_char(-1,"<",NULL,NULL,NULL),
662                                                                  tvbparse_char(-1,"!",NULL,NULL,NULL),
663                                                                  tvbparse_casestring(-1,"DOCTYPE",NULL,NULL,NULL),
664                                                                  tvbparse_set_oneof(-1,NULL,NULL,NULL,
665                                                                                     tvbparse_set_seq(-1,NULL,NULL,NULL,
666                                                                                                      want_name,
667                                                                                                      tvbparse_char(-1,"[",NULL,NULL,NULL),
668                                                                                                      NULL),
669                                                                                     tvbparse_set_seq(-1,NULL,NULL,pop_stack,
670                                                                                                      want_name,
671                                                                                                      tvbparse_set_oneof(-1,NULL,NULL,NULL,
672                                                                                                                         tvbparse_casestring(-1,"PUBLIC",NULL,NULL,NULL),
673                                                                                                                         tvbparse_casestring(-1,"SYSTEM",NULL,NULL,NULL),
674                                                                                                                         NULL),
675                                                                                                      tvbparse_until(-1,NULL,NULL,NULL,
676                                                                                                                     tvbparse_char(-1,">",NULL,NULL,NULL),
677                                                                                                                     TP_UNTIL_INCLUDE),
678                                                                                                      NULL),
679                                                                                     NULL),
680                                                                  NULL);
681
682         tvbparse_wanted_t* want_dtd_tag = tvbparse_set_seq(hf_dtd_tag,NULL,NULL,after_token,
683                                                            tvbparse_char(-1,"<",NULL,NULL,NULL),
684                                                            tvbparse_char(-1,"!",NULL,NULL,NULL),
685                                                            tvbparse_until(-1,NULL,NULL,NULL,
686                                                                           tvbparse_char(-1, ">", NULL, NULL, NULL),
687                                                                           TP_UNTIL_INCLUDE),
688                                                            NULL);
689
690         tvbparse_wanted_t* want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
691                                                        tvbparse_char(-1,"<",NULL,NULL,NULL),
692                                                        want_tag_name,
693                                                        tvbparse_set_oneof(-1,NULL,NULL,NULL,
694                                                                           tvbparse_set_seq(-1,NULL,NULL,NULL,
695                                                                                            want_attributes,
696                                                                                            want_stoptag,
697                                                                                            NULL),
698                                                                           want_stoptag,
699                                                                           NULL),
700                                                        NULL);
701
702         tvbparse_wanted_t* want_dtd_close = tvbparse_set_seq(-1,NULL,NULL,after_dtd_close,
703                                                              tvbparse_char(-1,"]",NULL,NULL,NULL),
704                                                              tvbparse_char(-1,">",NULL,NULL,NULL),
705                                                              NULL);
706
707         want_ignore = tvbparse_chars(-1,1,0," \t\r\n",NULL,NULL,NULL);
708
709
710         want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
711                                   want_comment,
712                                   want_xmlpi,
713                                   want_closing_tag,
714                                   want_doctype_start,
715                                   want_dtd_close,
716                                   want_dtd_tag,
717                                   want_tag,
718                                   tvbparse_not_chars(XML_CDATA,1,0,"<",NULL,NULL,after_token),
719                                   tvbparse_not_chars(-1,1,0," \t\r\n",NULL,NULL,unrecognized_token),
720                                   NULL);
721
722         want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
723                                        want_comment,
724                                        want_xmlpi,
725                                        want_doctype_start,
726                                        want_dtd_tag,
727                                        want_tag,
728                                        NULL);
729
730 }
731
732
733 static xml_ns_t* xml_new_namespace(GHashTable* hash, gchar* name, ...) {
734         xml_ns_t* ns = g_malloc(sizeof(xml_ns_t));
735         va_list ap;
736         gchar* attr_name;
737
738         ns->name = g_strdup(name);
739         ns->hf_tag = -1;
740         ns->hf_cdata = -1;
741         ns->ett = -1;
742         ns->attributes = g_hash_table_new(g_str_hash,g_str_equal);
743         ns->elements = g_hash_table_new(g_str_hash,g_str_equal);
744
745         va_start(ap,name);
746
747         while(( attr_name = va_arg(ap,gchar*) )) {
748                 int* hfp = g_malloc(sizeof(int));
749                 *hfp = -1;
750                 g_hash_table_insert(ns->attributes,g_strdup(attr_name),hfp);
751         };
752
753         va_end(ap);
754
755         g_hash_table_insert(hash,ns->name,ns);
756
757         return ns;
758 }
759
760
761 static void add_xml_field(GArray* hfs, int* p_id, gchar* name, gchar* fqn) {
762         hf_register_info hfri;
763
764         hfri.p_id = p_id;
765         hfri.hfinfo.name = name;
766         hfri.hfinfo.abbrev = fqn;
767         hfri.hfinfo.type = FT_STRING;
768         hfri.hfinfo.display = BASE_NONE;
769         hfri.hfinfo.strings = NULL;
770         hfri.hfinfo.bitmask = 0x0;
771         hfri.hfinfo.blurb = NULL;
772         hfri.hfinfo.id = 0;
773         hfri.hfinfo.parent = 0;
774         hfri.hfinfo.ref_type = HF_REF_TYPE_NONE;
775         hfri.hfinfo.bitshift = 0;
776         hfri.hfinfo.same_name_next = NULL;
777         hfri.hfinfo.same_name_prev = NULL;
778
779         g_array_append_val(hfs,hfri);
780 }
781
782 static void add_xml_attribute_names(gpointer k, gpointer v, gpointer p) {
783         struct _attr_reg_data* d = p;
784         gchar* basename = g_strdup_printf("%s.%s",d->basename,(gchar*)k);
785         add_xml_field(d->hf, (int*) v, (gchar*)k, basename);
786 }
787
788
789 static void add_xmlpi_namespace(gpointer k _U_, gpointer v, gpointer p) {
790         xml_ns_t* ns = v;
791         gchar* basename = g_strdup_printf("%s.%s",(gchar*)p,ns->name);
792         gint* ett_p = &(ns->ett);
793         struct _attr_reg_data d;
794
795         add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
796
797         g_array_append_val(ett_arr,ett_p);
798
799         d.basename = basename;
800         d.hf = hf_arr;
801
802         g_hash_table_foreach(ns->attributes,add_xml_attribute_names,&d);
803
804 }
805
806 static void destroy_dtd_data(dtd_build_data_t* dtd_data) {
807         g_free(dtd_data->proto_name);
808         g_free(dtd_data->media_type);
809         g_free(dtd_data->description);
810         g_free(dtd_data->proto_root);
811
812         g_string_free(dtd_data->error,TRUE);
813
814         while(dtd_data->elements->len) {
815                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->elements,0);
816                 g_ptr_array_free(nl->list,TRUE);
817                 g_free(nl);
818         }
819
820         g_ptr_array_free(dtd_data->elements,TRUE);
821
822         while(dtd_data->attributes->len) {
823                 dtd_named_list_t* nl = g_ptr_array_remove_index_fast(dtd_data->attributes,0);
824                 g_ptr_array_free(nl->list,TRUE);
825                 g_free(nl);
826         }
827
828         g_ptr_array_free(dtd_data->attributes,TRUE);
829
830         g_free(dtd_data);
831 }
832
833 static void copy_attrib_item(gpointer k, gpointer v _U_, gpointer p) {
834         gchar* key = g_strdup(k);
835         int* value = g_malloc(sizeof(int));
836         GHashTable* dst = p;
837
838         *value = -1;
839         g_hash_table_insert(dst,key,value);
840
841 }
842
843 static GHashTable* copy_attributes_hash(GHashTable* src) {
844         GHashTable* dst = g_hash_table_new(g_str_hash,g_str_equal);
845
846         g_hash_table_foreach(src,copy_attrib_item,dst);
847
848         return dst;
849 }
850
851 static xml_ns_t* duplicate_element(xml_ns_t* orig) {
852         xml_ns_t* new_item  = g_malloc(sizeof(xml_ns_t));
853         guint i;
854
855         new_item->name = g_strdup(orig->name);
856         new_item->hf_tag = -1;
857         new_item->hf_cdata = -1;
858         new_item->ett = -1;
859         new_item->attributes = copy_attributes_hash(orig->attributes);
860         new_item->elements =  g_hash_table_new(g_str_hash,g_str_equal);
861         new_item->element_names = g_ptr_array_new();
862
863         for(i=0; i < orig->element_names->len; i++) {
864                 g_ptr_array_add(new_item->element_names,
865                                                    g_ptr_array_index(orig->element_names,i));
866         }
867
868         return new_item;
869 }
870
871 static gchar* fully_qualified_name(GPtrArray* hier, gchar* name, gchar* proto_name) {
872         guint i;
873         GString* s = g_string_new(proto_name);
874         gchar* str;
875         g_string_append(s,".");
876         
877         for (i = 1; i < hier->len; i++) {
878                 g_string_append_printf(s, "%s.",(gchar*)g_ptr_array_index(hier,i));
879         }
880
881         g_string_append(s,name);
882         str = s->str;
883         g_string_free(s,FALSE);
884
885         return str;
886 }
887
888
889 static xml_ns_t* make_xml_hier(gchar* elem_name,
890                                xml_ns_t* root,
891                                GHashTable* elements,
892                                GPtrArray* hier,
893                                GString* error,
894                                GArray* hfs,
895                                GArray* etts,
896                                char* proto_name) {
897         xml_ns_t* new;
898         xml_ns_t* orig;
899         gchar* fqn;
900         gint* ett_p;
901         struct _attr_reg_data d;
902         gboolean recurred = FALSE;
903         guint i;
904
905         if ( g_str_equal(elem_name,root->name) ) {
906                 return NULL;
907         }
908
909         if (! ( orig = g_hash_table_lookup(elements,elem_name) )) {
910                 g_string_append_printf(error,"element '%s' is not defined\n", elem_name);
911                 return NULL;
912         }
913
914         for (i = 0; i < hier->len; i++) {
915                 if( strcmp(elem_name,(gchar*) g_ptr_array_index(hier,i) ) == 0 ) {
916                         recurred = TRUE;
917                 }
918         }
919
920         if (recurred) {
921                 return NULL;
922         }
923
924         fqn = fully_qualified_name(hier,elem_name,proto_name);
925
926         new = duplicate_element(orig);
927         new->fqn = fqn;
928
929         add_xml_field(hfs, &(new->hf_tag), g_strdup(elem_name), fqn);
930         add_xml_field(hfs, &(new->hf_cdata), g_strdup(elem_name), fqn);
931
932         ett_p = &new->ett;
933         g_array_append_val(etts,ett_p);
934
935         d.basename = fqn;
936         d.hf = hfs;
937
938         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
939
940         while(new->element_names->len) {
941                 gchar* child_name = g_ptr_array_remove_index(new->element_names,0);
942                 xml_ns_t* child_element = NULL;
943
944                 g_ptr_array_add(hier,elem_name);
945                 child_element = make_xml_hier(child_name, root, elements, hier,error,hfs,etts,proto_name);
946                 g_ptr_array_remove_index_fast(hier,hier->len - 1);
947
948                 if (child_element) {
949                         g_hash_table_insert(new->elements,child_element->name,child_element);
950                 }
951         }
952
953         g_ptr_array_free(new->element_names,TRUE);
954         new->element_names = NULL;
955         return new;
956 }
957
958 static gboolean free_both(gpointer k, gpointer v, gpointer p _U_) {
959         g_free(k);
960         g_free(v);
961         return TRUE;
962 }
963
964 static gboolean free_elements(gpointer k _U_, gpointer v, gpointer p _U_) {
965         xml_ns_t* e = v;
966         g_free(e->name);
967         g_hash_table_foreach_remove(e->attributes,free_both,NULL);
968         g_hash_table_destroy(e->attributes);
969         g_hash_table_destroy(e->elements);
970
971         while (e->element_names->len) {
972                 g_free(g_ptr_array_remove_index(e->element_names,0));
973         }
974
975         g_ptr_array_free(e->element_names,TRUE);
976         g_free(e);
977
978         return TRUE;
979 }
980
981 static void register_dtd(dtd_build_data_t* dtd_data, GString* errors) {
982         GHashTable* elements = g_hash_table_new(g_str_hash,g_str_equal);
983         gchar* root_name = NULL;
984         xml_ns_t* root_element = NULL;
985         GArray* hfs;
986         GArray* etts;
987         GPtrArray* hier;
988         gchar* curr_name;
989         GPtrArray* element_names = g_ptr_array_new();
990
991         /* we first populate elements with the those coming from the parser */
992         while(dtd_data->elements->len) {
993                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->elements,0);
994                 xml_ns_t* element = g_malloc(sizeof(xml_ns_t));
995
996                 /* we will use the first element found as root in case no other one was given. */
997                 if (root_name == NULL)
998                         root_name = g_strdup(nl->name);
999
1000                 element->name = nl->name;
1001                 element->element_names = nl->list;
1002                 element->hf_tag = -1;
1003                 element->hf_cdata = -1;
1004                 element->ett = -1;
1005                 element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1006                 element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1007
1008                 if( g_hash_table_lookup(elements,element->name) ) {
1009                         g_string_append_printf(errors,"element %s defined more than once\n", element->name);
1010                         free_elements(NULL,element,NULL);
1011                 } else {
1012                         g_hash_table_insert(elements,element->name,element);
1013                         g_ptr_array_add(element_names,g_strdup(element->name));
1014                 }
1015
1016                 g_free(nl);
1017         }
1018
1019         /* then we add the attributes to its relative elements */
1020         while(dtd_data->attributes->len) {
1021                 dtd_named_list_t* nl = g_ptr_array_remove_index(dtd_data->attributes,0);
1022                 xml_ns_t* element = g_hash_table_lookup(elements,nl->name);
1023
1024                 if (!element) {
1025                         g_string_append_printf(errors,"element %s is not defined\n", nl->name);
1026
1027                         goto next_attribute;
1028                 }
1029
1030                 while(nl->list->len) {
1031                         gchar* name = g_ptr_array_remove_index(nl->list,0);
1032                         int* id_p = g_malloc(sizeof(int));
1033
1034                         *id_p = -1;
1035                         g_hash_table_insert(element->attributes,name,id_p);
1036                 }
1037
1038 next_attribute:
1039                 g_free(nl->name);
1040                 g_ptr_array_free(nl->list,TRUE);
1041                 g_free(nl);
1042         }
1043
1044         /* if a proto_root is defined in the dtd we'll use that as root */
1045         if( dtd_data->proto_root ) {
1046                 g_free(root_name);
1047                 root_name = g_strdup(dtd_data->proto_root);
1048         }
1049
1050         /* we use a stack with the names to avoid recurring infinitelly */
1051         hier = g_ptr_array_new();
1052
1053         /*
1054          * if a proto name was given in the dtd the dtd will be used as a protocol
1055          * or else the dtd will be loaded as a branch of the xml namespace
1056          */
1057         if( ! dtd_data->proto_name ) {
1058                 hfs = hf_arr;
1059                 etts = ett_arr;
1060                 g_ptr_array_add(hier,g_strdup("xml"));
1061                 root_element = &xml_ns;
1062         } else {
1063                 /*
1064                  * if we were given a proto_name the namespace will be registered
1065                  * as an independent protocol with its own hf and ett arrays.
1066                  */
1067                 hfs = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1068                 etts = g_array_new(FALSE,FALSE,sizeof(gint*));
1069         }
1070
1071         /* the root element of the dtd's namespace */
1072         root_element = g_malloc(sizeof(xml_ns_t));
1073         root_element->name = g_strdup(root_name);
1074         root_element->fqn = dtd_data->proto_name ? g_strdup(dtd_data->proto_name) : root_element->name;
1075         root_element->hf_tag = -1;
1076         root_element->hf_cdata = -1;
1077         root_element->ett = -1;
1078         root_element->elements = g_hash_table_new(g_str_hash,g_str_equal);
1079         root_element->element_names = element_names;
1080
1081         /*
1082          * we can either create a namespace as a flat namespace
1083          * in which all the elements are at the root level
1084          * or we can create a recursive namespace
1085          */
1086         if (dtd_data->recursion) {
1087                 xml_ns_t* orig_root;
1088
1089                 make_xml_hier(root_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1090
1091                 g_hash_table_insert(root_element->elements,root_element->name,root_element);
1092
1093                 orig_root = g_hash_table_lookup(elements,root_name);
1094
1095                 /* if the root element was defined copy its attrlist to the child */
1096                 if(orig_root) {
1097                         struct _attr_reg_data d;
1098
1099                         d.basename = dtd_data->proto_name;
1100                         d.hf = hfs;
1101
1102                         root_element->attributes = copy_attributes_hash(orig_root->attributes);
1103                         g_hash_table_foreach(root_element->attributes,add_xml_attribute_names,&d);
1104                 } else {
1105                         root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1106                 }
1107
1108                 /* we then create all the sub hierachies to catch the recurred cases */
1109                 g_ptr_array_add(hier,root_name);
1110
1111                 while(root_element->element_names->len) {
1112                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1113
1114                         if( ! g_hash_table_lookup(root_element->elements,curr_name) ) {
1115                                 xml_ns_t* new = make_xml_hier(curr_name, root_element, elements,hier,errors,hfs,etts,dtd_data->proto_name);
1116                                 g_hash_table_insert(root_element->elements,new->name,new);
1117                         }
1118
1119                         g_free(curr_name);
1120                 }
1121
1122         } else {
1123                 /* a flat namespace */
1124                 g_ptr_array_add(hier,root_name);
1125
1126                 root_element->attributes = g_hash_table_new(g_str_hash,g_str_equal);
1127
1128                 while(root_element->element_names->len) {
1129                         xml_ns_t* new;
1130                         gint* ett_p;
1131                         struct _attr_reg_data d;
1132
1133                         curr_name = g_ptr_array_remove_index(root_element->element_names,0);
1134                         new = duplicate_element(g_hash_table_lookup(elements,curr_name));
1135                         new->fqn = fully_qualified_name(hier, curr_name, root_name);
1136
1137                         add_xml_field(hfs, &(new->hf_tag), curr_name, new->fqn);
1138                         add_xml_field(hfs, &(new->hf_cdata), curr_name, new->fqn);
1139
1140                         d.basename = new->fqn;
1141                         d.hf = hfs;
1142
1143                         g_hash_table_foreach(new->attributes,add_xml_attribute_names,&d);
1144
1145                         ett_p = &new->ett;
1146                         g_array_append_val(etts,ett_p);
1147
1148                         g_ptr_array_free(new->element_names,TRUE);
1149
1150                         g_hash_table_insert(root_element->elements,new->name,new);
1151                 }
1152         }
1153
1154         g_ptr_array_free(element_names,TRUE);
1155
1156         g_ptr_array_free(hier,TRUE);
1157
1158         /*
1159          * if we were given a proto_name the namespace will be registered
1160          * as an independent protocol.
1161          */
1162         if( dtd_data->proto_name ) {
1163                 gint* ett_p;
1164
1165                 if ( ! dtd_data->description) {
1166                         dtd_data->description = g_strdup(root_name);
1167                 }
1168
1169                 ett_p = &root_element->ett;
1170                 g_array_append_val(etts,ett_p);
1171
1172                 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1173
1174                 root_element->hf_tag = proto_register_protocol(dtd_data->description, dtd_data->proto_name, dtd_data->proto_name);
1175                 proto_register_field_array(root_element->hf_tag, (hf_register_info*)g_array_data(hfs), hfs->len);
1176                 proto_register_subtree_array((gint**)g_array_data(etts), etts->len);
1177
1178                 if (dtd_data->media_type) {
1179                         g_hash_table_insert(media_types,dtd_data->media_type,root_element);
1180                         dtd_data->media_type = NULL;
1181                 }
1182
1183                 dtd_data->description = NULL;
1184                 dtd_data->proto_name = NULL;
1185                 g_array_free(hfs,FALSE);
1186                 g_array_free(etts,TRUE);
1187         }
1188
1189         g_hash_table_insert(xml_ns.elements,root_element->name,root_element);
1190
1191         g_hash_table_foreach_remove(elements,free_elements,NULL);
1192         g_hash_table_destroy(elements);
1193
1194         destroy_dtd_data(dtd_data);
1195         g_free(root_name);
1196 }
1197
1198 #  define DIRECTORY_T GDir
1199 #  define FILE_T gchar
1200 #  define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1201 #  define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1202 #  define GETFNAME_OP(file) (file);
1203 #  define CLOSEDIR_OP(dir) g_dir_close(dir)
1204
1205 static void init_xml_names(void) {
1206         xml_ns_t* xmlpi_xml_ns;
1207         guint i;
1208         DIRECTORY_T* dir;
1209         const FILE_T* file;
1210         const gchar* filename;
1211         gchar* dirname;
1212
1213         GError** dummy = g_malloc(sizeof(GError *));
1214         *dummy = NULL;
1215
1216         xmpli_names = g_hash_table_new(g_str_hash,g_str_equal);
1217         media_types = g_hash_table_new(g_str_hash,g_str_equal);
1218
1219         unknown_ns.elements = xml_ns.elements = g_hash_table_new(g_str_hash,g_str_equal);
1220         unknown_ns.attributes = xml_ns.attributes = g_hash_table_new(g_str_hash,g_str_equal);
1221
1222         xmlpi_xml_ns = xml_new_namespace(xmpli_names,"xml","version","encoding","standalone",NULL);
1223
1224         g_hash_table_destroy(xmlpi_xml_ns->elements);
1225         xmlpi_xml_ns->elements = NULL;
1226
1227
1228         dirname = get_persconffile_path("dtds", FALSE, FALSE);
1229
1230         if (test_for_directory(dirname) != EISDIR) {
1231                 /* Although dir isn't a directory it may still use memory */
1232                 g_free(dirname);
1233                 dirname = get_datafile_path("dtds");
1234         }
1235
1236         if (test_for_directory(dirname) == EISDIR) {
1237
1238                 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1239                         while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1240                                 guint namelen;
1241                                 filename = GETFNAME_OP(file);
1242
1243                                 namelen = (int)strlen(filename);
1244                                 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4),".dtd")  == 0 ) ) {
1245                                         GString* errors = g_string_new("");
1246                                         GString* preparsed = dtd_preparse(dirname, filename, errors);
1247                                         dtd_build_data_t* dtd_data;
1248
1249                                         if (errors->len) {
1250                                                 report_failure("Dtd Preparser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1251                                                 continue;
1252                                         }
1253
1254                                         dtd_data = dtd_parse(preparsed);
1255
1256                                         g_string_free(preparsed,TRUE);
1257
1258                                         if (dtd_data->error->len) {
1259                                                 report_failure("Dtd Parser in file %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,dtd_data->error->str);
1260                                                 destroy_dtd_data(dtd_data);
1261                                                 continue;
1262                                         }
1263
1264                                         register_dtd(dtd_data,errors);
1265
1266                                         if (errors->len) {
1267                                                 report_failure("Dtd Registration in file: %s%c%s: %s",dirname,G_DIR_SEPARATOR,filename,errors->str);
1268                                                 g_string_free(errors,TRUE);
1269                                                 continue;
1270                                         }
1271                                 }
1272                         }
1273
1274                         CLOSEDIR_OP(dir);
1275                 }
1276         }
1277
1278         g_free(dirname);
1279
1280         for(i=0;i<array_length(default_media_types);i++) {
1281                 if( ! g_hash_table_lookup(media_types,default_media_types[i]) ) {
1282                         g_hash_table_insert(media_types,(gpointer)default_media_types[i],&xml_ns);
1283                 }
1284         }
1285
1286         g_hash_table_foreach(xmpli_names,add_xmlpi_namespace,"xml.xmlpi");
1287
1288         g_free(dummy);
1289 }
1290
1291 static void range_delete_xml_tcp_callback(guint32 port) {
1292         dissector_delete("tcp.port", port, xml_handle);
1293 }
1294
1295 static void range_add_xml_tcp_callback(guint32 port) {
1296         dissector_add("tcp.port", port, xml_handle);
1297 }
1298
1299 static void apply_prefs(void) {
1300         if (pref_heuristic_media_save != pref_heuristic_media) {
1301                 if (pref_heuristic_media) {
1302                         heur_dissector_add("http", dissect_xml_heur, xml_ns.hf_tag);
1303                         heur_dissector_add("sip", dissect_xml_heur, xml_ns.hf_tag);
1304                         heur_dissector_add("media", dissect_xml_heur, xml_ns.hf_tag);
1305                         pref_heuristic_media_save = TRUE;
1306                 } else {
1307                         heur_dissector_delete("http", dissect_xml_heur, xml_ns.hf_tag);
1308                         heur_dissector_delete("sip", dissect_xml_heur, xml_ns.hf_tag);
1309                         heur_dissector_delete("media", dissect_xml_heur, xml_ns.hf_tag);
1310                         pref_heuristic_media_save = FALSE;
1311                 }
1312         }
1313         
1314         if (pref_heuristic_tcp_save != pref_heuristic_tcp ) {
1315                 if (pref_heuristic_tcp) {
1316                         heur_dissector_add("tcp", dissect_xml_heur, xml_ns.hf_tag);
1317                         pref_heuristic_tcp_save = TRUE;
1318                 } else {
1319                         heur_dissector_delete("tcp", dissect_xml_heur, xml_ns.hf_tag);
1320                         pref_heuristic_tcp_save = FALSE;
1321                 }
1322         }
1323         
1324         if (pref_heuristic_udp_save != pref_heuristic_udp ) {
1325                 if (pref_heuristic_udp) {
1326                         heur_dissector_add("udp", dissect_xml_heur, xml_ns.hf_tag);
1327                         pref_heuristic_udp_save = TRUE;
1328                 } else {
1329                         heur_dissector_delete("udp", dissect_xml_heur, xml_ns.hf_tag);
1330                         pref_heuristic_udp_save = FALSE;
1331                 }
1332         }
1333
1334         range_foreach(xml_tcp_range, range_delete_xml_tcp_callback);
1335         g_free(xml_tcp_range);
1336         xml_tcp_range = range_copy(global_xml_tcp_range);
1337         range_foreach(xml_tcp_range, range_add_xml_tcp_callback);       
1338 }
1339
1340 void
1341 proto_register_xml(void) {
1342         static gint *ett_base[] = {
1343                 &unknown_ns.ett,
1344                 &xml_ns.ett,
1345                 &ett_dtd,
1346                 &ett_xmpli
1347         };
1348
1349         static hf_register_info hf_base[] = {
1350                 { &hf_xmlpi, {"XMLPI", "xml.xmlpi", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1351                 { &hf_comment, {"Comment", "xml.comment", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1352                 { &hf_unknowwn_attrib, {"Attribute", "xml.attribute", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1353                 { &hf_doctype, {"Doctype", "xml.doctype", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1354                 { &hf_dtd_tag, {"DTD Tag", "xml.dtdtag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1355                 { &unknown_ns.hf_cdata, {"CDATA", "xml.cdata", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1356                 { &unknown_ns.hf_tag, {"Tag", "xml.tag", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }},
1357                 { &xml_ns.hf_cdata, {"Unknown", "xml.unknown", FT_STRING, BASE_NONE, NULL, 0, NULL, HFILL }}
1358     };
1359         module_t* xml_module;
1360
1361         hf_arr = g_array_new(FALSE,FALSE,sizeof(hf_register_info));
1362         ett_arr = g_array_new(FALSE,FALSE,sizeof(gint*));
1363
1364         g_array_append_vals(hf_arr,hf_base,array_length(hf_base));
1365         g_array_append_vals(ett_arr,ett_base,array_length(ett_base));
1366
1367         init_xml_names();
1368
1369         xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1370
1371         proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)g_array_data(hf_arr), hf_arr->len);
1372         proto_register_subtree_array((gint**)g_array_data(ett_arr), ett_arr->len);
1373
1374         xml_module = prefs_register_protocol(xml_ns.hf_tag,apply_prefs);
1375         prefs_register_bool_preference(xml_module, "heuristic", "Use Heuristics for media types",
1376                                    "Try to recognize XML for unknown media types",
1377                                    &pref_heuristic_media);
1378         prefs_register_bool_preference(xml_module, "heuristic_tcp", "Use Heuristics for tcp",
1379                                    "Try to recognize XML for unknown TCP ports",
1380                                    &pref_heuristic_tcp);
1381         prefs_register_range_preference(xml_module, "tcp.port", "TCP Ports",
1382                                                                         "TCP Ports range",
1383                                                                         &global_xml_tcp_range, 65535);
1384         prefs_register_bool_preference(xml_module, "heuristic_udp", "Use Heuristics for UDP",
1385                                    "Try to recognize XML for unknown UDP ports",
1386                                    &pref_heuristic_udp);
1387         
1388         g_array_free(hf_arr,FALSE);
1389         g_array_free(ett_arr,TRUE);
1390
1391         register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1392
1393         init_xml_parser();
1394
1395         xml_tcp_range = range_empty();
1396
1397
1398 }
1399
1400 static void add_dissector_media(gpointer k, gpointer v _U_, gpointer p _U_) {
1401         dissector_add_string("media_type", (gchar*)k, xml_handle);
1402 }
1403
1404 void
1405 proto_reg_handoff_xml(void)
1406 {
1407
1408         xml_handle = find_dissector("xml");
1409
1410         g_hash_table_foreach(media_types,add_dissector_media,NULL);
1411
1412 }