?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Revert "TODO epan/dissectors/asn1/kerberos/packet-kerberos-template.c new GSS flags"
[wireshark-sm.git] / epan / dissectors / packet-xml.c
bloba50d2b1d9e686d0fec559b2b92e611a96e53ad4a
1 /* packet-xml.c
2 * wireshark's xml dissector .
3 *
4 * (C) 2005, Luis E. Garcia Ontanon.
5 *
6 * Refer to the AUTHORS file or the AUTHORS section in the man page
7 * for contacting the author(s) of this file.
8 *
9 * Wireshark - Network traffic analyzer
10 * By Gerald Combs <gerald@wireshark.org>
11 * Copyright 1998 Gerald Combs
12 *
13 * SPDX-License-Identifier: GPL-2.0-or-later
14 */
15
16 #include "config.h"
17
18 #include <string.h>
19 #include <errno.h>
20
21 #include <epan/packet.h>
22 #include <epan/tvbparse.h>
23 #include <epan/dtd.h>
24 #include <epan/proto_data.h>
25 #include <wsutil/filesystem.h>
26 #include <epan/prefs.h>
27 #include <epan/expert.h>
28 #include <epan/iana_charsets.h>
29 #include <epan/asn1.h>
30 #include <wsutil/str_util.h>
31 #include <wsutil/report_message.h>
32 #include <wsutil/wsgcrypt.h>
33 #include <wsutil/array.h>
34 #include "packet-kerberos.h"
35 #include "read_keytab_file.h"
36
37 #include "packet-xml.h"
38 #include "packet-acdr.h"
39
40 void proto_register_xml(void);
41 void proto_reg_handoff_xml(void);
42
43 struct _attr_reg_data {
44 wmem_array_t *hf;
45 const char *basename;
46 };
47
48
49 static int ett_dtd;
50 static int ett_xmpli;
51
52 static int hf_unknowwn_attrib;
53 static int hf_comment;
54 static int hf_xmlpi;
55 static int hf_dtd_tag;
56 static int hf_doctype;
57 static int hf_cdatasection;
58
59 static expert_field ei_xml_closing_unopened_tag;
60 static expert_field ei_xml_closing_unopened_xmpli_tag;
61 static expert_field ei_xml_unrecognized_text;
62
63 /* dissector handles */
64 static dissector_handle_t xml_handle;
65 static dissector_handle_t gssapi_handle;
66
67 /* Port 3702 is IANA-registered for Web Service Discovery, which uses
68 * SOAP-over-UDP to send XML */
69 #define XML_UDP_PORT_RANGE "3702"
70
71 /* parser definitions */
72 static tvbparse_wanted_t *want;
73 static tvbparse_wanted_t *want_ignore;
74 static tvbparse_wanted_t *want_heur;
75
76 static wmem_map_t *xmpli_names;
77 static wmem_map_t *media_types;
78
79 static xml_ns_t xml_ns = {"xml", "/", -1, -1, -1, NULL, NULL, NULL};
80 static xml_ns_t unknown_ns = {"unknown", "?", -1, -1, -1, NULL, NULL, NULL};
81 static xml_ns_t *root_ns;
82
83 static bool pref_heuristic_unicode;
84 static int pref_default_encoding = IANA_CS_UTF_8;
85
86
87 #define XML_CDATA -1000
88 #define XML_SCOPED_NAME -1001
89
90
91 static wmem_array_t *hf_arr;
92 static GArray *ett_arr;
93 static GRegex* encoding_pattern;
94
95 static const char *default_media_types[] = {
96 "text/xml",
97 "text/vnd.wap.wml",
98 "text/vnd.wap.si",
99 "text/vnd.wap.sl",
100 "text/vnd.wap.co",
101 "text/vnd.wap.emn",
102 "application/3gpp-ims+xml",
103 "application/atom+xml",
104 "application/auth-policy+xml",
105 "application/ccmp+xml",
106 "application/conference-info+xml", /*RFC4575*/
107 "application/cpim-pidf+xml",
108 "application/cpl+xml",
109 "application/dds-web+xml",
110 "application/im-iscomposing+xml", /*RFC3994*/
111 "application/load-control+xml", /*RFC7200*/
112 "application/mathml+xml",
113 "application/media_control+xml",
114 "application/note+xml",
115 "application/pidf+xml",
116 "application/pidf-diff+xml",
117 "application/poc-settings+xml",
118 "application/rdf+xml",
119 "application/reginfo+xml",
120 "application/resource-lists+xml",
121 "application/rlmi+xml",
122 "application/rls-services+xml",
123 "application/rss+xml",
124 "application/rs-metadata+xml",
125 "application/smil",
126 "application/simple-filter+xml",
127 "application/simple-message-summary+xml", /*RFC3842*/
128 "application/simservs+xml",
129 "application/soap+xml",
130 "application/vnd.etsi.aoc+xml",
131 "application/vnd.etsi.cug+xml",
132 "application/vnd.etsi.iptvcommand+xml",
133 "application/vnd.etsi.iptvdiscovery+xml",
134 "application/vnd.etsi.iptvprofile+xml",
135 "application/vnd.etsi.iptvsad-bc+xml",
136 "application/vnd.etsi.iptvsad-cod+xml",
137 "application/vnd.etsi.iptvsad-npvr+xml",
138 "application/vnd.etsi.iptvservice+xml",
139 "application/vnd.etsi.iptvsync+xml",
140 "application/vnd.etsi.iptvueprofile+xml",
141 "application/vnd.etsi.mcid+xml",
142 "application/vnd.etsi.overload-control-policy-dataset+xml",
143 "application/vnd.etsi.pstn+xml",
144 "application/vnd.etsi.sci+xml",
145 "application/vnd.etsi.simservs+xml",
146 "application/vnd.etsi.tsl+xml",
147 "application/vnd.oma.xdm-apd+xml",
148 "application/vnd.oma.fnl+xml",
149 "application/vnd.oma.access-permissions-list+xml",
150 "application/vnd.oma.alias-principals-list+xml",
151 "application/upp-directory+xml", /*OMA-ERELD-XDM-V2_2_1-20170124-A*/
152 "application/vnd.oma.xdm-hi+xml",
153 "application/vnd.oma.xdm-rhi+xml",
154 "application/vnd.oma.xdm-prefs+xml",
155 "application/vnd.oma.xdcp+xml",
156 "application/vnd.oma.bcast.associated-procedure-parameter+xml",
157 "application/vnd.oma.bcast.drm-trigger+xml",
158 "application/vnd.oma.bcast.imd+xml",
159 "application/vnd.oma.bcast.notification+xml",
160 "application/vnd.oma.bcast.sgdd+xml",
161 "application/vnd.oma.bcast.smartcard-trigger+xml",
162 "application/vnd.oma.bcast.sprov+xml",
163 "application/vnd.oma.cab-address-book+xml",
164 "application/vnd.oma.cab-feature-handler+xml",
165 "application/vnd.oma.cab-pcc+xml",
166 "application/vnd.oma.cab-subs-invite+xml",
167 "application/vnd.oma.cab-user-prefs+xml",
168 "application/vnd.oma.dd2+xml",
169 "application/vnd.oma.drm.risd+xml",
170 "application/vnd.oma.group-usage-list+xml",
171 "application/vnd.oma.pal+xml",
172 "application/vnd.oma.poc.detailed-progress-report+xml",
173 "application/vnd.oma.poc.final-report+xml",
174 "application/vnd.oma.poc.groups+xml",
175 "application/vnd.oma.poc.invocation-descriptor+xml",
176 "application/vnd.oma.poc.optimized-progress-report+xml",
177 "application/vnd.oma.scidm.messages+xml",
178 "application/vnd.oma.suppnot+xml", /*OMA-ERELD-Presence_SIMPLE-V2_0-20120710-A*/
179 "application/vnd.oma.xcap-directory+xml",
180 "application/vnd.omads-email+xml",
181 "application/vnd.omads-file+xml",
182 "application/vnd.omads-folder+xml",
183 "application/vnd.3gpp.access-transfer-events+xml",
184 "application/vnd.3gpp.bsf+xml",
185 "application/vnd.3gpp.comm-div-info+xml", /*3GPP TS 24.504 version 8.19.0*/
186 "application/vnd.3gpp.cw+xml",
187 "application/vnd.3gpp.iut+xml", /*3GPP TS 24.337*/
188 "application/vnc.3gpp.iut-config+xml", /*3GPP TS 24.337*/
189 "application/vnd.3gpp.mcptt-info+xml", /*3GPP TS 24.379 version 17.6.0*/
190 "application/vnd.3gpp.mcptt-mbms-usage-info+xml", /*3GPP TS 24.379 version 17.6.0*/
191 "application/vnd.3gpp.mcptt-location-info+xml", /*3GPP TS 24.379 version 17.6.0*/
192 "application/vnd.3gpp.mcptt-affiliation-command+xml", /*3GPP TS 24.379 version 17.6.0*/
193 "application/vnd.3gpp.mcptt-floor-request+xml", /*3GPP TS 24.379 version 17.6.0*/
194 "application/vnd.3gpp.mcptt-signed+xml", /*3GPP TS 24.379 version 17.6.0*/
195 "application/vnd.3gpp.mcptt-regroup+xml", /*3GPP TS 24.379 version 17.6.0*/
196 "application/vnd.3gpp.mcdata-info+xml", /*3GPP TS 24.282 version 17.6.2*/
197 "application/vnd.3gpp.mcdata-mbms-usage-info+xml", /*3GPP TS 24.282 version 17.6.2*/
198 "application/vnd.3gpp.mcdata-location-info+xml", /*3GPP TS 24.282 version 17.6.2*/
199 "application/vnd.3gpp.mcdata-affiliation-command+xml", /*3GPP TS 24.282 version 17.6.2*/
200 "application/vnd.3gpp.mcdata-regroup+xml", /*3GPP TS 24.282 version 17.6.2*/
201 "application/vnd.3gpp.mcvideo-info+xml", /*3GPP TS 24.281 version 17.6.0*/
202 "application/vnd.3gpp.mcvideo-mbms-usage-info+xml", /*3GPP TS 24.281 version 17.6.0*/
203 "application/vnd.3gpp.mcvideo-location-info+xml", /*3GPP TS 24.281 version 17.6.0*/
204 "application/vnd.3gpp.mcvideo-affiliation-command+xml",/*3GPP TS 24.281 version 17.6.0*/
205 "application/vnd.3gpp.transmission-request+xml", /*3GPP TS 24.281 version 17.6.0*/
206 "application/vnd.3gpp.mcptt-ue-init-config+xml", /*3GPP TS 24.484 version 17.5.0*/
207 "application/vnd.3gpp.mcptt-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
208 "application/vnd.3gpp.mcptt-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
209 "application/vnd.3gpp.mcptt-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
210 "application/vnd.3gpp.mcdata-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
211 "application/vnd.3gpp.mcvideo-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
212 "application/vnd.3gpp.mcvideo-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
213 "application/vnd.3gpp.mcvideo-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
214 "application/vnd.3gpp.mcdata-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
215 "application/vnd.3gpp.mcdata-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
216 "application/vnd.3gpp.mid-call+xml",
217 "application/vnd.3gpp-prose-pc3ch+xml",
218 "application/vnd.3gpp-prose+xml",
219 "application/vnd.3gpp.replication+xml", /*3GPP TS 24.337*/
220 "application/vnd.3gpp.sms+xml",
221 "application/vnd.3gpp.srvcc-info+xml",
222 "application/vnd.3gpp.srvcc-ext+xml",
223 "application/vnd.3gpp.state-and-event-info+xml",
224 "application/vnd.3gpp.ussd+xml",
225 "application/vnd.3gpp2.bcmcsinfo+xml",
226 "application/vnd.wv.csp+xml",
227 "application/vnd.wv.csp.xml",
228 "application/watcherinfo+xml",
229 "application/xcap-att+xml",
230 "application/xcap-caps+xml",
231 "application/xcap-diff+xml",
232 "application/xcap-el+xml",
233 "application/xcap-error+xml",
234 "application/xcap-ns+xml",
235 "application/xml",
236 "application/xml-dtd",
237 "application/xpidf+xml",
238 "application/xslt+xml",
239 "application/x-crd+xml",
240 "application/x-wms-logconnectstats",
241 "application/x-wms-logplaystats",
242 "application/x-wms-sendevent",
243 "image/svg+xml",
244 "message/imdn+xml", /*RFC5438*/
245 };
246
247 static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child)
248 {
249 new_child->first_child = NULL;
250 new_child->last_child = NULL;
251
252 new_child->parent = parent;
253 new_child->next_sibling = NULL;
254 new_child->prev_sibling = NULL;
255 if (parent == NULL) return; /* root */
256
257 if (parent->first_child == NULL) { /* the 1st child */
258 parent->first_child = new_child;
259 } else { /* following children */
260 parent->last_child->next_sibling = new_child;
261 new_child->prev_sibling = parent->last_child;
262 }
263 parent->last_child = new_child;
264 }
265
266 /* Try to get the 'encoding' attribute from XML declaration, and convert it to
267 * Wireshark character encoding.
268 */
269 static unsigned
270 get_char_encoding(tvbuff_t* tvb, packet_info* pinfo, char** ret_encoding_name) {
271 uint32_t iana_charset_id;
272 unsigned ws_encoding_id;
273 char* encoding_str;
274 GMatchInfo* match_info;
275 const char* xmldecl = (char*)tvb_get_string_enc(pinfo->pool, tvb, 0,
276 MIN(100, tvb_captured_length(tvb)), ENC_UTF_8);
277
278 g_regex_match(encoding_pattern, xmldecl, 0, &match_info);
279 if (g_match_info_matches(match_info)) {
280 char* match_ret = g_match_info_fetch(match_info, 1);
281 encoding_str = ascii_strup_inplace(wmem_strdup(pinfo->pool, match_ret));
282 g_free(match_ret);
283 /* Get the iana charset enum number by the name of the charset. */
284 iana_charset_id = str_to_val(encoding_str,
285 VALUE_STRING_EXT_VS_P(&mibenum_vals_character_sets_ext), IANA_CS_US_ASCII);
286 } else {
287 /* Use default encoding preference if this xml does not contains 'encoding' attribute. */
288 iana_charset_id = pref_default_encoding;
289 encoding_str = val_to_str_ext_wmem(pinfo->pool, iana_charset_id,
290 &mibenum_vals_character_sets_ext, "UNKNOWN");
291 }
292 g_match_info_free(match_info);
293
294 ws_encoding_id = mibenum_charset_to_encoding((unsigned)iana_charset_id);
295
296 /* UTF-8 compatible with ASCII */
297 if (ws_encoding_id == (ENC_NA | ENC_ASCII)) {
298 ws_encoding_id = ENC_UTF_8;
299 *ret_encoding_name = wmem_strdup(pinfo->pool, "UTF-8");
300 } else {
301 *ret_encoding_name = encoding_str;
302 }
303
304 return ws_encoding_id;
305 }
306
307 static int
308 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree, void* data _U_)
309 {
310 tvbparse_t *tt;
311 static GPtrArray *stack;
312 xml_frame_t *current_frame;
313 const char *colinfo_str;
314 tvbuff_t *decoded;
315 uint16_t try_bom;
316
317 if (stack != NULL)
318 g_ptr_array_free(stack, true);
319
320 stack = g_ptr_array_new();
321 current_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
322 current_frame->type = XML_FRAME_ROOT;
323 current_frame->name = NULL;
324 current_frame->name_orig_case = NULL;
325 current_frame->value = NULL;
326 current_frame->pinfo = pinfo;
327 insert_xml_frame(NULL, current_frame);
328 g_ptr_array_add(stack, current_frame);
329
330 /* Detect and act on possible byte-order mark (BOM) */
331 try_bom = tvb_get_ntohs(tvb, 0);
332 if (try_bom == 0xFEFF) {
333 /* UTF-16BE */
334 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), ENC_UTF_16|ENC_BIG_ENDIAN);
335 size_t l = strlen(data_str);
336 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
337 add_new_data_source(pinfo, decoded, "Decoded UTF-16BE text");
338 }
339 else if(try_bom == 0xFFFE) {
340 /* UTF-16LE (or possibly UTF-32LE, but Wireshark doesn't support UTF-32) */
341 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), ENC_UTF_16|ENC_LITTLE_ENDIAN);
342 size_t l = strlen(data_str);
343 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
344 add_new_data_source(pinfo, decoded, "Decoded UTF-16LE text");
345 }
346 /* Could also test if try_bom is 0xnn00 or 0x00nn to guess endianness if we wanted */
347 else {
348 /* Get character encoding according to XML declaration or preference. */
349 char* encoding_name;
350 unsigned encoding = get_char_encoding(tvb, pinfo, &encoding_name);
351
352 /* Encoding string with encoding, either with or without BOM */
353 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), encoding);
354 size_t l = strlen(data_str);
355 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
356 add_new_data_source(pinfo, decoded, wmem_strdup_printf(pinfo->pool, "Decoded %s text", encoding_name));
357 }
358
359 tt = tvbparse_init(pinfo->pool, decoded, 0, -1, stack, want_ignore);
360 current_frame->start_offset = 0;
361 current_frame->length = tvb_captured_length(decoded);
362
363 current_frame->decryption_keys = wmem_map_new(wmem_packet_scope(), g_str_hash, g_str_equal);
364
365 root_ns = NULL;
366
367 if (pinfo->match_string)
368 root_ns = (xml_ns_t *)wmem_map_lookup(media_types, pinfo->match_string);
369
370 if (! root_ns ) {
371 root_ns = &xml_ns;
372 colinfo_str = "/XML";
373 } else {
374 char *colinfo_str_buf;
375 colinfo_str_buf = wmem_strconcat(wmem_packet_scope(), "/", root_ns->name, NULL);
376 ascii_strup_inplace(colinfo_str_buf);
377 colinfo_str = colinfo_str_buf;
378 }
379
380 col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
381
382 current_frame->ns = root_ns;
383
384 current_frame->item = proto_tree_add_item(tree, current_frame->ns->hf_tag, decoded, 0, -1, ENC_UTF_8|ENC_NA);
385 current_frame->tree = proto_item_add_subtree(current_frame->item, current_frame->ns->ett);
386 current_frame->last_item = current_frame->item;
387
388 while(tvbparse_get(tt, want)) ;
389
390 /* Save XML structure in case it is useful for the caller */
391 p_add_proto_data(pinfo->pool, pinfo, xml_ns.hf_tag, 0, current_frame);
392
393 return tvb_captured_length(tvb);
394 }
395
396 static bool dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree, void *data)
397 {
398 if (tvbparse_peek(tvbparse_init(pinfo->pool, tvb, 0, -1, NULL, want_ignore), want_heur)) {
399 dissect_xml(tvb, pinfo, tree, data);
400 return true;
401 } else if (pref_heuristic_unicode) {
402 const uint8_t *data_str;
403 tvbuff_t *unicode_tvb;
404 uint16_t try_bom;
405 /* XXX - UCS-2, or UTF-16? */
406 int enc = ENC_UCS_2|ENC_LITTLE_ENDIAN;
407 size_t l;
408
409 try_bom = tvb_get_ntohs(tvb, 0);
410 if (try_bom == 0xFEFF) {
411 enc = ENC_UTF_16|ENC_BIG_ENDIAN;
412 }
413 else if(try_bom == 0xFFFE) {
414 enc = ENC_UTF_16|ENC_LITTLE_ENDIAN;
415 }
416
417 data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), enc);
418 l = strlen(data_str);
419 unicode_tvb = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
420 if (tvbparse_peek(tvbparse_init(pinfo->pool, unicode_tvb, 0, -1, NULL, want_ignore), want_heur)) {
421 add_new_data_source(pinfo, unicode_tvb, "UTF8");
422 dissect_xml(unicode_tvb, pinfo, tree, data);
423 return true;
424 }
425 }
426 return false;
427 }
428
429 xml_frame_t *xml_get_tag(xml_frame_t *frame, const char *name)
430 {
431 xml_frame_t *tag = NULL;
432
433 xml_frame_t *xml_item = frame->first_child;
434 while (xml_item) {
435 if (xml_item->type == XML_FRAME_TAG) {
436 if (!name) { /* get the 1st tag */
437 tag = xml_item;
438 break;
439 } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
440 tag = xml_item;
441 break;
442 }
443 }
444 xml_item = xml_item->next_sibling;
445 }
446
447 return tag;
448 }
449
450 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const char *name)
451 {
452 xml_frame_t *attr = NULL;
453
454 xml_frame_t *xml_item = frame->first_child;
455 while (xml_item) {
456 if ((xml_item->type == XML_FRAME_ATTRIB) &&
457 xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
458 attr = xml_item;
459 break;
460 }
461 xml_item = xml_item->next_sibling;
462 }
463
464 return attr;
465 }
466
467 xml_frame_t *xml_get_cdata(xml_frame_t *frame)
468 {
469 xml_frame_t *cdata = NULL;
470
471 xml_frame_t *xml_item = frame->first_child;
472 while (xml_item) {
473 if (xml_item->type == XML_FRAME_CDATA) {
474 cdata = xml_item;
475 break;
476 }
477 xml_item = xml_item->next_sibling;
478 }
479
480 return cdata;
481 }
482
483 static void after_token(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
484 {
485 GPtrArray *stack = (GPtrArray *)tvbparse_data;
486 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
487 int hfid;
488 bool is_cdata = false;
489 proto_item *pi;
490 xml_frame_t *new_frame = NULL;
491 char *text = NULL;
492
493 if (tok->id == XML_CDATA) {
494 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
495 is_cdata = true;
496 } else if ( tok->id > 0) {
497 hfid = tok->id;
498 } else {
499 hfid = xml_ns.hf_cdata;
500 }
501
502 pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
503
504 text = tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len);
505 proto_item_set_text(pi, "%s", text);
506
507 if (is_cdata) {
508 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
509 new_frame->type = XML_FRAME_CDATA;
510 new_frame->name = NULL;
511 new_frame->name_orig_case = NULL;
512 new_frame->value = tvb_new_subset_length(tok->tvb, tok->offset, tok->len);
513 insert_xml_frame(current_frame, new_frame);
514 new_frame->item = pi;
515 new_frame->last_item = pi;
516 new_frame->tree = NULL;
517 new_frame->start_offset = tok->offset;
518 new_frame->length = tok->len;
519 new_frame->ns = NULL;
520 new_frame->pinfo = current_frame->pinfo;
521 }
522
523 if (new_frame != NULL &&
524 current_frame != NULL &&
525 current_frame->name_orig_case != NULL &&
526 strcmp(current_frame->name_orig_case, "BinarySecurityToken") == 0)
527 {
528 xml_frame_t *value_type = NULL;
529
530 value_type = xml_get_attrib(current_frame, "ValueType");
531 if (value_type != NULL) {
532 const char *s = "http://docs.oasis-open.org/wss/oasis-wss-kerberos-token-profile-1.1#GSS_Kerberosv5_AP_REQ";
533 size_t l = strlen(s);
534 int c;
535 c = tvb_strneql(value_type->value, 0, s, l);
536 if (c == 0) {
537 tvbuff_t *ssp_tvb = base64_to_tvb(new_frame->value, text);
538 add_new_data_source(current_frame->pinfo, ssp_tvb, "GSSAPI Data");
539 call_dissector(gssapi_handle, ssp_tvb,
540 current_frame->pinfo, current_frame->tree);
541 }
542 }
543 }
544 }
545
546 static void before_xmpli(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
547 {
548 GPtrArray *stack = (GPtrArray *)tvbparse_data;
549 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
550 proto_item *pi;
551 proto_tree *pt;
552 tvbparse_elem_t *name_tok = tok->sub->next;
553 char *name = tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb, name_tok->offset, name_tok->len, ENC_ASCII);
554 xml_ns_t *ns = (xml_ns_t *)wmem_map_lookup(xmpli_names, name);
555 xml_frame_t *new_frame;
556
557 int hf_tag;
558 int ett;
559
560 ascii_strdown_inplace(name);
561 if (!ns) {
562 hf_tag = hf_xmlpi;
563 ett = ett_xmpli;
564 } else {
565 hf_tag = ns->hf_tag;
566 ett = ns->ett;
567 }
568
569 pi = proto_tree_add_item(current_frame->tree, hf_tag, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
570
571 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, (name_tok->offset - tok->offset) + name_tok->len));
572
573 pt = proto_item_add_subtree(pi, ett);
574
575 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
576 new_frame->type = XML_FRAME_XMPLI;
577 new_frame->name = name;
578 new_frame->name_orig_case = name;
579 new_frame->value = NULL;
580 insert_xml_frame(current_frame, new_frame);
581 new_frame->item = pi;
582 new_frame->last_item = pi;
583 new_frame->tree = pt;
584 new_frame->start_offset = tok->offset;
585 new_frame->length = tok->len;
586 new_frame->ns = ns;
587 new_frame->pinfo = current_frame->pinfo;
588
589 g_ptr_array_add(stack, new_frame);
590
591 }
592
593 static void after_xmlpi(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
594 {
595 GPtrArray *stack = (GPtrArray *)tvbparse_data;
596 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
597
598 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
599
600 if (stack->len > 1) {
601 g_ptr_array_remove_index_fast(stack, stack->len - 1);
602 } else {
603 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_xmpli_tag,
604 tok->tvb, tok->offset, tok->len);
605 }
606 }
607
608 static void before_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
609 {
610 GPtrArray *stack = (GPtrArray *)tvbparse_data;
611 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
612 tvbparse_elem_t *name_tok = tok->sub->next;
613 char *root_name;
614 char *name = NULL, *name_orig_case = NULL;
615 xml_ns_t *ns;
616 xml_frame_t *new_frame;
617 proto_item *pi;
618 proto_tree *pt;
619
620 if (name_tok->sub->id == XML_SCOPED_NAME) {
621 tvbparse_elem_t *root_tok = name_tok->sub->sub;
622 tvbparse_elem_t *leaf_tok = name_tok->sub->sub->next->next;
623 xml_ns_t *nameroot_ns;
624
625 root_name = (char *)tvb_get_string_enc(wmem_packet_scope(), root_tok->tvb, root_tok->offset, root_tok->len, ENC_ASCII);
626 name = (char *)tvb_get_string_enc(wmem_packet_scope(), leaf_tok->tvb, leaf_tok->offset, leaf_tok->len, ENC_ASCII);
627 name_orig_case = name;
628
629 nameroot_ns = (xml_ns_t *)wmem_map_lookup(xml_ns.elements, root_name);
630
631 if(nameroot_ns) {
632 ns = (xml_ns_t *)wmem_map_lookup(nameroot_ns->elements, name);
633 if (!ns) {
634 ns = &unknown_ns;
635 }
636 } else {
637 ns = &unknown_ns;
638 }
639
640 } else {
641 name = tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb, name_tok->offset, name_tok->len, ENC_ASCII);
642 name_orig_case = wmem_strdup(wmem_packet_scope(), name);
643 ascii_strdown_inplace(name);
644
645 if(current_frame->ns) {
646 ns = (xml_ns_t *)wmem_map_lookup(current_frame->ns->elements, name);
647
648 if (!ns) {
649 if (! ( ns = (xml_ns_t *)wmem_map_lookup(root_ns->elements, name) ) ) {
650 ns = &unknown_ns;
651 }
652 }
653 } else {
654 ns = &unknown_ns;
655 }
656 }
657
658 pi = proto_tree_add_item(current_frame->tree, ns->hf_tag, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
659 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb,
660 tok->offset,
661 (name_tok->offset - tok->offset) + name_tok->len));
662
663 pt = proto_item_add_subtree(pi, ns->ett);
664
665 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
666 new_frame->type = XML_FRAME_TAG;
667 new_frame->name = name;
668 new_frame->name_orig_case = name_orig_case;
669 new_frame->value = NULL;
670 insert_xml_frame(current_frame, new_frame);
671 new_frame->item = pi;
672 new_frame->last_item = pi;
673 new_frame->tree = pt;
674 new_frame->start_offset = tok->offset;
675 new_frame->length = tok->len;
676 new_frame->ns = ns;
677 new_frame->pinfo = current_frame->pinfo;
678
679 g_ptr_array_add(stack, new_frame);
680
681 }
682
683 static void after_open_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
684 {
685 GPtrArray *stack = (GPtrArray *)tvbparse_data;
686 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
687
688 proto_item_append_text(current_frame->last_item, ">");
689 }
690
691 static void after_closed_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
692 {
693 GPtrArray *stack = (GPtrArray *)tvbparse_data;
694 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
695
696 proto_item_append_text(current_frame->last_item, "/>");
697
698 if (stack->len > 1) {
699 g_ptr_array_remove_index_fast(stack, stack->len - 1);
700 } else {
701 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
702 tok->tvb, tok->offset, tok->len);
703 }
704 }
705
706 #ifdef HAVE_KERBEROS
707 struct decryption_key {
708 char *id;
709 size_t key_length;
710 uint8_t key[HASH_SHA1_LENGTH];
711 };
712
713 static void P_SHA1(const uint8_t *Secret, size_t Secret_len,
714 const uint8_t *Seed, size_t Seed_len,
715 uint8_t Result[HASH_SHA1_LENGTH])
716 {
717 gcry_md_hd_t hd = NULL;
718 uint8_t *digest = NULL;
719
720 /*
721 * https://social.microsoft.com/Forums/en-US/c485d98b-6e0b-49e7-ab34-8ecf8d694d31/signing-soap-message-request-via-adfs?forum=crmdevelopment#6cee9fa8-dc24-4524-a5a2-c3d17e05d50e
722 */
723 gcry_md_open(&hd, GCRY_MD_SHA1, GCRY_MD_FLAG_HMAC);
724 gcry_md_setkey(hd, Secret, Secret_len);
725 gcry_md_write(hd, Seed, Seed_len);
726 digest = gcry_md_read(hd, GCRY_MD_SHA1);
727 memcpy(Result, digest, HASH_SHA1_LENGTH);
728
729 gcry_md_close(hd);
730 }
731 #endif /* HAVE_KERBEROS */
732
733 static void after_untag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
734 {
735 GPtrArray *stack = (GPtrArray *)tvbparse_data;
736 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
737 #ifdef HAVE_KERBEROS
738 xml_frame_t *top_frame = (xml_frame_t *)g_ptr_array_index(stack, 0);
739 #endif /* HAVE_KERBEROS */
740
741 proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
742 current_frame->length = (tok->offset - current_frame->start_offset) + tok->len;
743
744 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
745
746 if (stack->len > 1) {
747 g_ptr_array_remove_index_fast(stack, stack->len - 1);
748 } else {
749 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
750 tok->tvb, tok->offset, tok->len);
751 }
752
753 #ifdef HAVE_KERBEROS
754 if (current_frame->name_orig_case == NULL) {
755 return;
756 }
757
758 if (strcmp(current_frame->name_orig_case, "DerivedKeyToken") == 0) {
759 xml_frame_t *id_frame = xml_get_attrib(current_frame, "u:Id");
760 xml_frame_t *nonce_frame = xml_get_tag(current_frame, "Nonce");
761 xml_frame_t *nonce_cdata = NULL;
762 tvbuff_t *nonce_tvb = NULL;
763 enc_key_t *ek = NULL;
764 uint8_t seed[64];
765 size_t seed_length = 16; // TODO
766 const size_t key_length = 16; //TODO
767
768 if (id_frame != NULL && nonce_frame != NULL) {
769 nonce_cdata = xml_get_cdata(nonce_frame);
770 }
771 if (nonce_cdata != NULL) {
772 char *text = tvb_format_text(wmem_packet_scope(), nonce_cdata->value, 0,
773 tvb_reported_length(nonce_cdata->value));
774 nonce_tvb = base64_to_tvb(nonce_cdata->value, text);
775 }
776 if (nonce_tvb != NULL) {
777 seed_length = tvb_reported_length(nonce_tvb);
778 seed_length = MIN(seed_length, sizeof(seed));
779 tvb_memcpy(nonce_tvb, seed, 0, seed_length);
780
781 if (krb_decrypt) {
782 read_keytab_file_from_preferences();
783 }
784
785 for (ek=enc_key_list;ek;ek=ek->next) {
786 if (ek->fd_num == (int)current_frame->pinfo->num) {
787 break;
788 }
789 }
790 }
791 if (ek != NULL) {
792 struct decryption_key *key;
793 char *id_str;
794
795 id_str = tvb_format_text(wmem_packet_scope(),
796 id_frame->value, 0,
797 tvb_reported_length(id_frame->value));
798
799 key = wmem_new0(wmem_packet_scope(), struct decryption_key);
800 key->id = wmem_strdup_printf(wmem_packet_scope(), "#%s", id_str);
801 P_SHA1(ek->keyvalue, ek->keylength, seed, seed_length, key->key);
802 key->key_length = key_length;
803
804 wmem_map_insert(top_frame->decryption_keys, key->id, key);
805 }
806 }
807 if (strcmp(current_frame->name_orig_case, "CipherValue") == 0) {
808 xml_frame_t *encrypted_frame = current_frame->parent->parent;
809 xml_frame_t *key_info_frame = NULL;
810 xml_frame_t *token_frame = NULL;
811 xml_frame_t *reference_frame = NULL;
812 xml_frame_t *uri_frame = NULL;
813 const struct decryption_key *key = NULL;
814 xml_frame_t *cdata_frame = NULL;
815 tvbuff_t *crypt_tvb = NULL;
816 tvbuff_t *plain_tvb = NULL;
817
818 key_info_frame = xml_get_tag(encrypted_frame, "KeyInfo");
819 if (key_info_frame != NULL) {
820 token_frame = xml_get_tag(key_info_frame, "SecurityTokenReference");
821 }
822 if (token_frame != NULL) {
823 reference_frame = xml_get_tag(token_frame, "Reference");
824 }
825 if (reference_frame != NULL) {
826 uri_frame = xml_get_attrib(reference_frame, "URI");
827 }
828
829 if (uri_frame != NULL) {
830 char *key_id = tvb_format_text(wmem_packet_scope(), uri_frame->value, 0,
831 tvb_reported_length(uri_frame->value));
832
833 key = (const struct decryption_key *)wmem_map_lookup(top_frame->decryption_keys, key_id);
834 }
835 if (key != NULL) {
836 cdata_frame = xml_get_cdata(current_frame);
837 }
838 if (cdata_frame != NULL) {
839 char *text = tvb_format_text(wmem_packet_scope(), cdata_frame->value, 0,
840 tvb_reported_length(cdata_frame->value));
841 crypt_tvb = base64_to_tvb(cdata_frame->value, text);
842 }
843 if (crypt_tvb != NULL) {
844 gcry_cipher_hd_t cipher_hd = NULL;
845 uint8_t *data = NULL;
846 unsigned data_length = tvb_reported_length(crypt_tvb);
847
848 data = (uint8_t *)tvb_memdup(wmem_packet_scope(),
849 crypt_tvb, 0, data_length);
850
851 /* Open the cipher. */
852 gcry_cipher_open(&cipher_hd, GCRY_CIPHER_AES128, GCRY_CIPHER_MODE_CBC, 0);
853
854 gcry_cipher_setkey(cipher_hd, key->key, key->key_length);
855 gcry_cipher_encrypt(cipher_hd, data, data_length, NULL, 0);
856 gcry_cipher_close(cipher_hd);
857
858 plain_tvb = tvb_new_child_real_data(crypt_tvb, data,
859 data_length, data_length);
860 add_new_data_source(current_frame->pinfo, plain_tvb, "Decrypted Data");
861 }
862 }
863 #endif /* HAVE_KERBEROS */
864 }
865
866 static void before_dtd_doctype(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
867 {
868 GPtrArray *stack = (GPtrArray *)tvbparse_data;
869 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
870 xml_frame_t *new_frame;
871 tvbparse_elem_t *name_tok = tok->sub->next->next->next->sub->sub;
872 proto_tree *dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
873 name_tok->tvb, name_tok->offset,
874 name_tok->len, ENC_ASCII);
875
876 proto_item_set_text(dtd_item, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len));
877
878 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
879 new_frame->type = XML_FRAME_DTD_DOCTYPE;
880 new_frame->name = (char *)tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb,
881 name_tok->offset,
882 name_tok->len, ENC_ASCII);
883 new_frame->name_orig_case = new_frame->name;
884 new_frame->value = NULL;
885 insert_xml_frame(current_frame, new_frame);
886 new_frame->item = dtd_item;
887 new_frame->last_item = dtd_item;
888 new_frame->tree = proto_item_add_subtree(dtd_item, ett_dtd);
889 new_frame->start_offset = tok->offset;
890 new_frame->length = tok->len;
891 new_frame->ns = NULL;
892 new_frame->pinfo = current_frame->pinfo;
893
894 g_ptr_array_add(stack, new_frame);
895 }
896
897 static void pop_stack(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
898 {
899 GPtrArray *stack = (GPtrArray *)tvbparse_data;
900 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
901
902 if (stack->len > 1) {
903 g_ptr_array_remove_index_fast(stack, stack->len - 1);
904 } else {
905 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
906 tok->tvb, tok->offset, tok->len);
907 }
908 }
909
910 static void after_dtd_close(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
911 {
912 GPtrArray *stack = (GPtrArray *)tvbparse_data;
913 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
914
915 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
916 if (stack->len > 1) {
917 g_ptr_array_remove_index_fast(stack, stack->len - 1);
918 } else {
919 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
920 tok->tvb, tok->offset, tok->len);
921 }
922 }
923
924 static void get_attrib_value(void *tvbparse_data _U_, const void *wanted_data _U_, tvbparse_elem_t *tok)
925 {
926 tok->data = tok->sub;
927 }
928
929 static void after_attrib(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
930 {
931 GPtrArray *stack = (GPtrArray *)tvbparse_data;
932 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
933 char *name, *name_orig_case;
934 tvbparse_elem_t *value;
935 tvbparse_elem_t *value_part = (tvbparse_elem_t *)tok->sub->next->next->data;
936 int *hfidp;
937 int hfid;
938 proto_item *pi;
939 xml_frame_t *new_frame;
940
941 name = tvb_get_string_enc(wmem_packet_scope(), tok->sub->tvb, tok->sub->offset, tok->sub->len, ENC_ASCII);
942 name_orig_case = wmem_strdup(wmem_packet_scope(), name);
943 ascii_strdown_inplace(name);
944
945 if(current_frame->ns && (hfidp = (int *)wmem_map_lookup(current_frame->ns->attributes, name) )) {
946 hfid = *hfidp;
947 value = value_part;
948 } else {
949 hfid = hf_unknowwn_attrib;
950 value = tok;
951 }
952
953 pi = proto_tree_add_item(current_frame->tree, hfid, value->tvb, value->offset, value->len, ENC_UTF_8|ENC_NA);
954 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len));
955
956 current_frame->last_item = pi;
957
958 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
959 new_frame->type = XML_FRAME_ATTRIB;
960 new_frame->name = name;
961 new_frame->name_orig_case = name_orig_case;
962 new_frame->value = tvb_new_subset_length(value_part->tvb, value_part->offset,
963 value_part->len);
964 insert_xml_frame(current_frame, new_frame);
965 new_frame->item = pi;
966 new_frame->last_item = pi;
967 new_frame->tree = NULL;
968 new_frame->start_offset = tok->offset;
969 new_frame->length = tok->len;
970 new_frame->ns = NULL;
971 new_frame->pinfo = current_frame->pinfo;
972
973 }
974
975 static void unrecognized_token(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
976 {
977 GPtrArray *stack = (GPtrArray *)tvbparse_data;
978 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
979
980 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_unrecognized_text,
981 tok->tvb, tok->offset, tok->len);
982
983 }
984
985
986
987 static void init_xml_parser(void)
988 {
989 tvbparse_wanted_t *want_name =
990 tvbparse_chars(-1, 1, 0,
991 "abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",
992 NULL, NULL, NULL);
993 tvbparse_wanted_t *want_attr_name =
994 tvbparse_chars(-1, 1, 0,
995 "abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",
996 NULL, NULL, NULL);
997
998 tvbparse_wanted_t *want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
999 want_name,
1000 tvbparse_char(-1, ":", NULL, NULL, NULL),
1001 want_name,
1002 NULL);
1003
1004 tvbparse_wanted_t *want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
1005 want_scoped_name,
1006 want_name,
1007 NULL);
1008
1009 tvbparse_wanted_t *want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
1010 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb, '\"', '\\'),
1011 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb, '\'', '\\'),
1012 tvbparse_chars(-1, 1, 0, "0123456789", NULL, NULL, NULL),
1013 want_name,
1014 NULL);
1015
1016 tvbparse_wanted_t *want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
1017 tvbparse_set_seq(-1, NULL, NULL, after_attrib,
1018 want_attr_name,
1019 tvbparse_char(-1, "=", NULL, NULL, NULL),
1020 want_attrib_value,
1021 NULL));
1022
1023 tvbparse_wanted_t *want_stoptag = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1024 tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
1025 tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
1026 NULL);
1027
1028 tvbparse_wanted_t *want_stopxmlpi = tvbparse_string(-1, "?>", NULL, NULL, after_xmlpi);
1029
1030 tvbparse_wanted_t *want_comment = tvbparse_set_seq(hf_comment, NULL, NULL, after_token,
1031 tvbparse_string(-1, "<!--", NULL, NULL, NULL),
1032 tvbparse_until(-1, NULL, NULL, NULL,
1033 tvbparse_string(-1, "-->", NULL, NULL, NULL),
1034 TP_UNTIL_INCLUDE),
1035 NULL);
1036
1037 tvbparse_wanted_t *want_cdatasection = tvbparse_set_seq(hf_cdatasection, NULL, NULL, after_token,
1038 tvbparse_string(-1, "<![CDATA[", NULL, NULL, NULL),
1039 tvbparse_until(-1, NULL, NULL, NULL,
1040 tvbparse_string(-1, "]]>", NULL, NULL, NULL),
1041 TP_UNTIL_INCLUDE),
1042 NULL);
1043
1044 tvbparse_wanted_t *want_xmlpi = tvbparse_set_seq(hf_xmlpi, NULL, before_xmpli, NULL,
1045 tvbparse_string(-1, "<?", NULL, NULL, NULL),
1046 want_name,
1047 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1048 want_stopxmlpi,
1049 tvbparse_set_seq(-1, NULL, NULL, NULL,
1050 want_attributes,
1051 want_stopxmlpi,
1052 NULL),
1053 NULL),
1054 NULL);
1055
1056 tvbparse_wanted_t *want_closing_tag = tvbparse_set_seq(0, NULL, NULL, after_untag,
1057 tvbparse_char(-1, "<", NULL, NULL, NULL),
1058 tvbparse_char(-1, "/", NULL, NULL, NULL),
1059 want_tag_name,
1060 tvbparse_char(-1, ">", NULL, NULL, NULL),
1061 NULL);
1062
1063 tvbparse_wanted_t *want_doctype_start = tvbparse_set_seq(-1, NULL, before_dtd_doctype, NULL,
1064 tvbparse_char(-1, "<", NULL, NULL, NULL),
1065 tvbparse_char(-1, "!", NULL, NULL, NULL),
1066 tvbparse_casestring(-1, "DOCTYPE", NULL, NULL, NULL),
1067 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1068 tvbparse_set_seq(-1, NULL, NULL, NULL,
1069 want_name,
1070 tvbparse_char(-1, "[", NULL, NULL, NULL),
1071 NULL),
1072 tvbparse_set_seq(-1, NULL, NULL, pop_stack,
1073 want_name,
1074 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1075 tvbparse_casestring(-1, "PUBLIC", NULL, NULL, NULL),
1076 tvbparse_casestring(-1, "SYSTEM", NULL, NULL, NULL),
1077 NULL),
1078 tvbparse_until(-1, NULL, NULL, NULL,
1079 tvbparse_char(-1, ">", NULL, NULL, NULL),
1080 TP_UNTIL_INCLUDE),
1081 NULL),
1082 NULL),
1083 NULL);
1084
1085 tvbparse_wanted_t *want_dtd_tag = tvbparse_set_seq(hf_dtd_tag, NULL, NULL, after_token,
1086 tvbparse_char(-1, "<", NULL, NULL, NULL),
1087 tvbparse_char(-1, "!", NULL, NULL, NULL),
1088 tvbparse_until(-1, NULL, NULL, NULL,
1089 tvbparse_char(-1, ">", NULL, NULL, NULL),
1090 TP_UNTIL_INCLUDE),
1091 NULL);
1092
1093 tvbparse_wanted_t *want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
1094 tvbparse_char(-1, "<", NULL, NULL, NULL),
1095 want_tag_name,
1096 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1097 tvbparse_set_seq(-1, NULL, NULL, NULL,
1098 want_attributes,
1099 want_stoptag,
1100 NULL),
1101 want_stoptag,
1102 NULL),
1103 NULL);
1104
1105 tvbparse_wanted_t *want_dtd_close = tvbparse_set_seq(-1, NULL, NULL, after_dtd_close,
1106 tvbparse_char(-1, "]", NULL, NULL, NULL),
1107 tvbparse_char(-1, ">", NULL, NULL, NULL),
1108 NULL);
1109
1110 want_ignore = tvbparse_chars(-1, 1, 0, " \t\r\n", NULL, NULL, NULL);
1111
1112
1113 want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1114 want_comment,
1115 want_cdatasection,
1116 want_xmlpi,
1117 want_closing_tag,
1118 want_doctype_start,
1119 want_dtd_close,
1120 want_dtd_tag,
1121 want_tag,
1122 tvbparse_not_chars(XML_CDATA, 1, 0, "<", NULL, NULL, after_token),
1123 tvbparse_not_chars(-1, 1, 0, " \t\r\n", NULL, NULL, unrecognized_token),
1124 NULL);
1125
1126 want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1127 want_comment,
1128 want_cdatasection,
1129 want_xmlpi,
1130 want_doctype_start,
1131 want_dtd_tag,
1132 want_tag,
1133 NULL);
1134
1135 }
1136
1137
1138 static xml_ns_t *xml_new_namespace(wmem_map_t *hash, const char *name, ...)
1139 {
1140 xml_ns_t *ns = wmem_new(wmem_epan_scope(), xml_ns_t);
1141 va_list ap;
1142 char *attr_name;
1143
1144 ns->name = wmem_strdup(wmem_epan_scope(), name);
1145 ns->hf_tag = -1;
1146 ns->hf_cdata = -1;
1147 ns->ett = -1;
1148 ns->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1149 ns->elements = NULL;
1150
1151 va_start(ap, name);
1152
1153 while(( attr_name = va_arg(ap, char *) )) {
1154 int *hfp = wmem_new(wmem_epan_scope(), int);
1155 *hfp = -1;
1156 wmem_map_insert(ns->attributes, wmem_strdup(wmem_epan_scope(), attr_name), hfp);
1157 };
1158
1159 va_end(ap);
1160
1161 wmem_map_insert(hash, ns->name, ns);
1162
1163 return ns;
1164 }
1165
1166
1167 static void add_xml_field(wmem_array_t *hfs, int *p_id, const char *name, const char *fqn)
1168 {
1169 hf_register_info hfri;
1170
1171 hfri.p_id = p_id;
1172 hfri.hfinfo.name = name;
1173 hfri.hfinfo.abbrev = fqn;
1174 hfri.hfinfo.type = FT_STRING;
1175 hfri.hfinfo.display = BASE_NONE;
1176 hfri.hfinfo.strings = NULL;
1177 hfri.hfinfo.bitmask = 0x0;
1178 hfri.hfinfo.blurb = NULL;
1179 HFILL_INIT(hfri);
1180
1181 wmem_array_append_one(hfs, hfri);
1182 }
1183
1184 static void add_xml_attribute_names(void *k, void *v, void *p)
1185 {
1186 struct _attr_reg_data *d = (struct _attr_reg_data *)p;
1187 const char *basename = wmem_strconcat(wmem_epan_scope(), d->basename, ".", (char *)k, NULL);
1188
1189 add_xml_field(d->hf, (int*) v, (char *)k, basename);
1190 }
1191
1192
1193 static void add_xmlpi_namespace(void *k _U_, void *v, void *p)
1194 {
1195 xml_ns_t *ns = (xml_ns_t *)v;
1196 const char *basename = wmem_strconcat(wmem_epan_scope(), (char *)p, ".", ns->name, NULL);
1197 int *ett_p = &(ns->ett);
1198 struct _attr_reg_data d;
1199
1200 add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
1201
1202 g_array_append_val(ett_arr, ett_p);
1203
1204 d.basename = basename;
1205 d.hf = hf_arr;
1206
1207 wmem_map_foreach(ns->attributes, add_xml_attribute_names, &d);
1208
1209 }
1210
1211 static void destroy_dtd_data(dtd_build_data_t *dtd_data)
1212 {
1213 g_free(dtd_data->proto_name);
1214 g_free(dtd_data->media_type);
1215 g_free(dtd_data->description);
1216 g_free(dtd_data->proto_root);
1217
1218 g_string_free(dtd_data->error, true);
1219
1220 while(dtd_data->elements->len) {
1221 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index_fast(dtd_data->elements, 0);
1222 g_ptr_array_free(nl->list, true);
1223 g_free(nl->name);
1224 g_free(nl);
1225 }
1226
1227 g_ptr_array_free(dtd_data->elements, true);
1228
1229 while(dtd_data->attributes->len) {
1230 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index_fast(dtd_data->attributes, 0);
1231 g_ptr_array_free(nl->list, true);
1232 g_free(nl->name);
1233 g_free(nl);
1234 }
1235
1236 g_ptr_array_free(dtd_data->attributes, true);
1237
1238 g_free(dtd_data);
1239 }
1240
1241 static void copy_attrib_item(void *k, void *v _U_, void *p)
1242 {
1243 char *key = (char *)wmem_strdup(wmem_epan_scope(), (const char *)k);
1244 int *value = wmem_new(wmem_epan_scope(), int);
1245 wmem_map_t *dst = (wmem_map_t *)p;
1246
1247 *value = -1;
1248 wmem_map_insert(dst, key, value);
1249
1250 }
1251
1252 static wmem_map_t *copy_attributes_hash(wmem_map_t *src)
1253 {
1254 wmem_map_t *dst = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1255
1256 wmem_map_foreach(src, copy_attrib_item, dst);
1257
1258 return dst;
1259 }
1260
1261 static xml_ns_t *duplicate_element(xml_ns_t *orig)
1262 {
1263 xml_ns_t *new_item = wmem_new(wmem_epan_scope(), xml_ns_t);
1264 unsigned i;
1265
1266 new_item->name = wmem_strdup(wmem_epan_scope(), orig->name);
1267 new_item->hf_tag = -1;
1268 new_item->hf_cdata = -1;
1269 new_item->ett = -1;
1270 new_item->attributes = copy_attributes_hash(orig->attributes);
1271 new_item->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1272 new_item->element_names = g_ptr_array_new();
1273
1274 for(i=0; i < orig->element_names->len; i++) {
1275 g_ptr_array_add(new_item->element_names,
1276 g_ptr_array_index(orig->element_names, i));
1277 }
1278
1279 return new_item;
1280 }
1281
1282 static char *fully_qualified_name(GPtrArray *hier, char *name, char *proto_name)
1283 {
1284 unsigned i;
1285 wmem_strbuf_t *s = wmem_strbuf_new(wmem_epan_scope(), proto_name);
1286
1287 wmem_strbuf_append(s, ".");
1288
1289 for (i = 1; i < hier->len; i++) {
1290 wmem_strbuf_append_printf(s, "%s.", (char *)g_ptr_array_index(hier, i));
1291 }
1292
1293 wmem_strbuf_append(s, name);
1294
1295 return wmem_strbuf_finalize(s);
1296 }
1297
1298
1299 // NOLINTNEXTLINE(misc-no-recursion)
1300 static xml_ns_t *make_xml_hier(char *elem_name,
1301 xml_ns_t *root,
1302 wmem_map_t *elements,
1303 GPtrArray *hier,
1304 GString *error,
1305 wmem_array_t *hfs,
1306 GArray *etts,
1307 char *proto_name)
1308 {
1309 xml_ns_t *fresh;
1310 xml_ns_t *orig;
1311 char *fqn;
1312 int *ett_p;
1313 bool recurred = false;
1314 unsigned i;
1315 struct _attr_reg_data d;
1316
1317 if ( g_str_equal(elem_name, root->name) ) {
1318 return NULL;
1319 }
1320
1321 if (! ( orig = (xml_ns_t *)wmem_map_lookup(elements, elem_name) )) {
1322 g_string_append_printf(error, "element '%s' is not defined\n", elem_name);
1323 return NULL;
1324 }
1325
1326 if (hier->len >= prefs.gui_max_tree_depth) {
1327 g_string_append_printf(error, "hierarchy too deep: %u\n", hier->len);
1328 return NULL;
1329 }
1330
1331 for (i = 0; i < hier->len; i++) {
1332 if( (elem_name) && (strcmp(elem_name, (char *) g_ptr_array_index(hier, i) ) == 0 )) {
1333 recurred = true;
1334 }
1335 }
1336
1337 if (recurred) {
1338 return NULL;
1339 }
1340
1341 fqn = fully_qualified_name(hier, elem_name, proto_name);
1342
1343 fresh = duplicate_element(orig);
1344 fresh->fqn = fqn;
1345
1346 add_xml_field(hfs, &(fresh->hf_tag), wmem_strdup(wmem_epan_scope(), elem_name), fqn);
1347 add_xml_field(hfs, &(fresh->hf_cdata), wmem_strdup(wmem_epan_scope(), elem_name), fqn);
1348
1349 ett_p = &fresh->ett;
1350 g_array_append_val(etts, ett_p);
1351
1352 d.basename = fqn;
1353 d.hf = hfs;
1354
1355 wmem_map_foreach(fresh->attributes, add_xml_attribute_names, &d);
1356
1357 while(fresh->element_names->len) {
1358 char *child_name = (char *)g_ptr_array_remove_index(fresh->element_names, 0);
1359 xml_ns_t *child_element = NULL;
1360
1361 g_ptr_array_add(hier, elem_name);
1362 child_element = make_xml_hier(child_name, root, elements, hier, error, hfs, etts, proto_name);
1363 g_ptr_array_remove_index_fast(hier, hier->len - 1);
1364
1365 if (child_element) {
1366 wmem_map_insert(fresh->elements, child_element->name, child_element);
1367 }
1368 }
1369
1370 g_ptr_array_free(fresh->element_names, true);
1371 fresh->element_names = NULL;
1372 return fresh;
1373 }
1374
1375 static void free_elements(void *k _U_, void *v, void *p _U_)
1376 {
1377 xml_ns_t *e = (xml_ns_t *)v;
1378
1379 while (e->element_names->len) {
1380 g_free(g_ptr_array_remove_index(e->element_names, 0));
1381 }
1382
1383 g_ptr_array_free(e->element_names, true);
1384 }
1385
1386 static void register_dtd(dtd_build_data_t *dtd_data, GString *errors)
1387 {
1388 wmem_map_t *elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1389 char *root_name = NULL;
1390 xml_ns_t *root_element = NULL;
1391 wmem_array_t *hfs;
1392 GArray *etts;
1393 GPtrArray *hier;
1394 char *curr_name;
1395 GPtrArray *element_names = g_ptr_array_new();
1396
1397 /* we first populate elements with the those coming from the parser */
1398 while(dtd_data->elements->len) {
1399 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index(dtd_data->elements, 0);
1400 xml_ns_t *element = wmem_new(wmem_epan_scope(), xml_ns_t);
1401
1402 /* we will use the first element found as root in case no other one was given. */
1403 if (root_name == NULL)
1404 root_name = wmem_strdup(wmem_epan_scope(), nl->name);
1405
1406 element->name = wmem_strdup(wmem_epan_scope(), nl->name);
1407 element->element_names = nl->list;
1408 element->hf_tag = -1;
1409 element->hf_cdata = -1;
1410 element->ett = -1;
1411 element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1412 element->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1413
1414 if( wmem_map_lookup(elements, element->name) ) {
1415 g_string_append_printf(errors, "element %s defined more than once\n", element->name);
1416 free_elements(NULL, element, NULL);
1417 } else {
1418 wmem_map_insert(elements, element->name, element);
1419 g_ptr_array_add(element_names, wmem_strdup(wmem_epan_scope(), element->name));
1420 }
1421
1422 g_free(nl->name);
1423 g_free(nl);
1424 }
1425
1426 /* then we add the attributes to its relative elements */
1427 while(dtd_data->attributes->len) {
1428 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index(dtd_data->attributes, 0);
1429 xml_ns_t *element = (xml_ns_t *)wmem_map_lookup(elements, nl->name);
1430
1431 if (element) {
1432 while(nl->list->len) {
1433 char *name = (char *)g_ptr_array_remove_index(nl->list, 0);
1434 int *id_p = wmem_new(wmem_epan_scope(), int);
1435
1436 *id_p = -1;
1437 wmem_map_insert(element->attributes, wmem_strdup(wmem_epan_scope(), name), id_p);
1438 g_free(name); }
1439 }
1440 else {
1441 g_string_append_printf(errors, "element %s is not defined\n", nl->name);
1442 }
1443
1444 g_free(nl->name);
1445 g_ptr_array_free(nl->list, true);
1446 g_free(nl);
1447 }
1448
1449 /* if a proto_root is defined in the dtd we'll use that as root */
1450 if( dtd_data->proto_root ) {
1451 wmem_free(wmem_epan_scope(), root_name);
1452 root_name = wmem_strdup(wmem_epan_scope(), dtd_data->proto_root);
1453 }
1454
1455 /* we use a stack with the names to avoid recurring infinitely */
1456 hier = g_ptr_array_new();
1457
1458 /*
1459 * if a proto name was given in the dtd the dtd will be used as a protocol
1460 * or else the dtd will be loaded as a branch of the xml namespace
1461 */
1462 if( ! dtd_data->proto_name ) {
1463 hfs = hf_arr;
1464 etts = ett_arr;
1465 g_ptr_array_add(hier, wmem_strdup(wmem_epan_scope(), "xml"));
1466 } else {
1467 /*
1468 * if we were given a proto_name the namespace will be registered
1469 * as an independent protocol with its own hf and ett arrays.
1470 */
1471 hfs = wmem_array_new(wmem_epan_scope(), sizeof(hf_register_info));
1472 etts = g_array_new(false, false, sizeof(int *));
1473 }
1474
1475 /* the root element of the dtd's namespace */
1476 root_element = wmem_new(wmem_epan_scope(), xml_ns_t);
1477 root_element->name = wmem_strdup(wmem_epan_scope(), root_name);
1478 root_element->fqn = dtd_data->proto_name ? wmem_strdup(wmem_epan_scope(), dtd_data->proto_name) : root_element->name;
1479 root_element->hf_tag = -1;
1480 root_element->hf_cdata = -1;
1481 root_element->ett = -1;
1482 root_element->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1483 root_element->element_names = element_names;
1484
1485 /*
1486 * we can either create a namespace as a flat namespace
1487 * in which all the elements are at the root level
1488 * or we can create a recursive namespace
1489 */
1490 if (dtd_data->recursion) {
1491 xml_ns_t *orig_root;
1492
1493 make_xml_hier(root_name, root_element, elements, hier, errors, hfs, etts, dtd_data->proto_name);
1494
1495 wmem_map_insert(root_element->elements, (void *)root_element->name, root_element);
1496
1497 orig_root = (xml_ns_t *)wmem_map_lookup(elements, root_name);
1498
1499 /* if the root element was defined copy its attrlist to the child */
1500 if(orig_root) {
1501 struct _attr_reg_data d;
1502
1503 d.basename = dtd_data->proto_name;
1504 d.hf = hfs;
1505
1506 root_element->attributes = copy_attributes_hash(orig_root->attributes);
1507 wmem_map_foreach(root_element->attributes, add_xml_attribute_names, &d);
1508 } else {
1509 root_element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1510 }
1511
1512 /* we then create all the sub hierarchies to catch the recurred cases */
1513 g_ptr_array_add(hier, root_name);
1514
1515 while(root_element->element_names->len) {
1516 curr_name = (char *)g_ptr_array_remove_index(root_element->element_names, 0);
1517
1518 if( ! wmem_map_lookup(root_element->elements, curr_name) ) {
1519 xml_ns_t *fresh = make_xml_hier(curr_name, root_element, elements, hier, errors,
1520 hfs, etts, dtd_data->proto_name);
1521 wmem_map_insert(root_element->elements, (void *)fresh->name, fresh);
1522 }
1523 }
1524
1525 } else {
1526 /* a flat namespace */
1527 g_ptr_array_add(hier, root_name);
1528
1529 root_element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1530
1531 while(root_element->element_names->len) {
1532 xml_ns_t *fresh;
1533 int *ett_p;
1534 struct _attr_reg_data d;
1535
1536 curr_name = (char *)g_ptr_array_remove_index(root_element->element_names, 0);
1537 fresh = duplicate_element((xml_ns_t *)wmem_map_lookup(elements, curr_name));
1538 fresh->fqn = fully_qualified_name(hier, curr_name, root_name);
1539
1540 add_xml_field(hfs, &(fresh->hf_tag), curr_name, fresh->fqn);
1541 add_xml_field(hfs, &(fresh->hf_cdata), curr_name, fresh->fqn);
1542
1543 d.basename = fresh->fqn;
1544 d.hf = hfs;
1545
1546 wmem_map_foreach(fresh->attributes, add_xml_attribute_names, &d);
1547
1548 ett_p = &fresh->ett;
1549 g_array_append_val(etts, ett_p);
1550
1551 g_ptr_array_free(fresh->element_names, true);
1552
1553 wmem_map_insert(root_element->elements, (void *)fresh->name, fresh);
1554 }
1555 }
1556
1557 g_ptr_array_free(element_names, true);
1558
1559 g_ptr_array_free(hier, true);
1560
1561 /*
1562 * if we were given a proto_name the namespace will be registered
1563 * as an independent protocol.
1564 */
1565 if( dtd_data->proto_name ) {
1566 int *ett_p;
1567 char *full_name, *short_name;
1568
1569 if (dtd_data->description) {
1570 full_name = wmem_strdup(wmem_epan_scope(), dtd_data->description);
1571 } else {
1572 full_name = wmem_strdup(wmem_epan_scope(), root_name);
1573 }
1574 short_name = wmem_strdup_printf(wmem_epan_scope(),"%s_%s", dtd_data->proto_name, "dtd");
1575
1576 ett_p = &root_element->ett;
1577 g_array_append_val(etts, ett_p);
1578
1579 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1580
1581 root_element->hf_tag = proto_register_protocol(full_name, short_name, short_name);
1582 proto_register_field_array(root_element->hf_tag, (hf_register_info*)wmem_array_get_raw(hfs), wmem_array_get_count(hfs));
1583 proto_register_subtree_array((int **)etts->data, etts->len);
1584
1585 if (dtd_data->media_type) {
1586 char* media_type = wmem_strdup(wmem_epan_scope(), dtd_data->media_type);
1587 wmem_map_insert(media_types, media_type, root_element);
1588 }
1589
1590 g_array_free(etts, true);
1591 }
1592
1593 wmem_map_insert(xml_ns.elements, root_element->name, root_element);
1594 wmem_map_foreach(elements, free_elements, NULL);
1595
1596 destroy_dtd_data(dtd_data);
1597 wmem_free(wmem_epan_scope(), root_name);
1598 }
1599
1600 # define DIRECTORY_T GDir
1601 # define FILE_T char
1602 # define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1603 # define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1604 # define GETFNAME_OP(file) (file);
1605 # define CLOSEDIR_OP(dir) g_dir_close(dir)
1606
1607 static void init_xml_names(void)
1608 {
1609 unsigned i;
1610 DIRECTORY_T *dir;
1611 const FILE_T *file;
1612 const char *filename;
1613 char *dirname;
1614
1615 GError **dummy = wmem_new(wmem_epan_scope(), GError *);
1616 *dummy = NULL;
1617
1618 xmpli_names = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1619 media_types = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1620
1621 unknown_ns.elements = xml_ns.elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1622 unknown_ns.attributes = xml_ns.attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1623
1624 xml_new_namespace(xmpli_names, "xml", "version", "encoding", "standalone", NULL);
1625
1626 dirname = get_persconffile_path("dtds", false);
1627
1628 if (test_for_directory(dirname) != EISDIR) {
1629 /* Although dir isn't a directory it may still use memory */
1630 g_free(dirname);
1631 dirname = get_datafile_path("dtds");
1632 }
1633
1634 if (test_for_directory(dirname) == EISDIR) {
1635 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1636 GString *errors = g_string_new("");
1637
1638 while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1639 unsigned namelen;
1640 filename = GETFNAME_OP(file);
1641
1642 namelen = (int)strlen(filename);
1643 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4), ".dtd") == 0 ) ) {
1644 GString *preparsed;
1645 dtd_build_data_t *dtd_data;
1646
1647 g_string_truncate(errors, 0);
1648 preparsed = dtd_preparse(dirname, filename, errors);
1649
1650 if (errors->len) {
1651 report_failure("Dtd Preparser in file %s%c%s: %s",
1652 dirname, G_DIR_SEPARATOR, filename, errors->str);
1653 continue;
1654 }
1655
1656 dtd_data = dtd_parse(preparsed);
1657
1658 g_string_free(preparsed, true);
1659
1660 if (dtd_data->error->len) {
1661 report_failure("Dtd Parser in file %s%c%s: %s",
1662 dirname, G_DIR_SEPARATOR, filename, dtd_data->error->str);
1663 destroy_dtd_data(dtd_data);
1664 continue;
1665 }
1666
1667 register_dtd(dtd_data, errors);
1668
1669 if (errors->len) {
1670 report_failure("Dtd Registration in file: %s%c%s: %s",
1671 dirname, G_DIR_SEPARATOR, filename, errors->str);
1672 continue;
1673 }
1674 }
1675 }
1676 g_string_free(errors, true);
1677
1678 CLOSEDIR_OP(dir);
1679 }
1680 }
1681
1682 g_free(dirname);
1683
1684 for(i=0;i<array_length(default_media_types);i++) {
1685 if( ! wmem_map_lookup(media_types, default_media_types[i]) ) {
1686 wmem_map_insert(media_types, (void *)default_media_types[i], &xml_ns);
1687 }
1688 }
1689
1690 wmem_map_foreach(xmpli_names, add_xmlpi_namespace, (void *)"xml.xmlpi");
1691
1692 wmem_free(wmem_epan_scope(), dummy);
1693 }
1694
1695 static void
1696 xml_init_protocol(void)
1697 {
1698 // The longest encoding at https://www.iana.org/assignments/character-sets/character-sets.xml
1699 // is 45 characters (Extended_UNIX_Code_Packed_Format_for_Japanese).
1700 encoding_pattern = g_regex_new("^\\s*<[?]xml\\s+version\\s*=\\s*[\"']\\s*(?U:.+)\\s*[\"']\\s+encoding\\s*=\\s*[\"']\\s*((?U).{1,50})\\s*[\"']", G_REGEX_CASELESS, 0, 0);
1701 }
1702
1703 static void
1704 xml_cleanup_protocol(void) {
1705 g_regex_unref(encoding_pattern);
1706 }
1707
1708 void
1709 proto_register_xml(void)
1710 {
1711 static int *ett_base[] = {
1712 &unknown_ns.ett,
1713 &xml_ns.ett,
1714 &ett_dtd,
1715 &ett_xmpli
1716 };
1717
1718 static hf_register_info hf_base[] = {
1719 { &hf_xmlpi,
1720 {"XMLPI", "xml.xmlpi",
1721 FT_STRING, BASE_NONE, NULL, 0,
1722 NULL, HFILL }
1723 },
1724 { &hf_cdatasection,
1725 {"CDATASection", "xml.cdatasection",
1726 FT_STRING, BASE_NONE, NULL, 0,
1727 NULL, HFILL }
1728 },
1729 { &hf_comment,
1730 {"Comment", "xml.comment",
1731 FT_STRING, BASE_NONE, NULL, 0,
1732 NULL, HFILL }
1733 },
1734 { &hf_unknowwn_attrib,
1735 {"Attribute", "xml.attribute",
1736 FT_STRING, BASE_NONE, NULL, 0,
1737 NULL, HFILL }
1738 },
1739 { &hf_doctype,
1740 {"Doctype", "xml.doctype",
1741 FT_STRING, BASE_NONE, NULL, 0,
1742 NULL, HFILL }
1743 },
1744 { &hf_dtd_tag,
1745 {"DTD Tag", "xml.dtdtag",
1746 FT_STRING, BASE_NONE, NULL, 0,
1747 NULL, HFILL }
1748 },
1749 { &unknown_ns.hf_cdata,
1750 {"CDATA", "xml.cdata",
1751 FT_STRING, BASE_NONE, NULL, 0, NULL,
1752 HFILL }
1753 },
1754 { &unknown_ns.hf_tag,
1755 {"Tag", "xml.tag",
1756 FT_STRING, BASE_NONE, NULL, 0,
1757 NULL, HFILL }
1758 },
1759 { &xml_ns.hf_cdata,
1760 {"Unknown", "xml.unknown",
1761 FT_STRING, BASE_NONE, NULL, 0,
1762 NULL, HFILL }
1763 }
1764 };
1765
1766 static ei_register_info ei[] = {
1767 { &ei_xml_closing_unopened_tag, { "xml.closing_unopened_tag", PI_MALFORMED, PI_ERROR, "Closing an unopened tag", EXPFILL }},
1768 { &ei_xml_closing_unopened_xmpli_tag, { "xml.closing_unopened_xmpli_tag", PI_MALFORMED, PI_ERROR, "Closing an unopened xmpli tag", EXPFILL }},
1769 { &ei_xml_unrecognized_text, { "xml.unrecognized_text", PI_PROTOCOL, PI_WARN, "Unrecognized text", EXPFILL }},
1770 };
1771
1772 module_t *xml_module;
1773 expert_module_t* expert_xml;
1774
1775 hf_arr = wmem_array_new(wmem_epan_scope(), sizeof(hf_register_info));
1776 ett_arr = g_array_new(false, false, sizeof(int *));
1777
1778 wmem_array_append(hf_arr, hf_base, array_length(hf_base));
1779 g_array_append_vals(ett_arr, ett_base, array_length(ett_base));
1780
1781 init_xml_names();
1782
1783 xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1784
1785 proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)wmem_array_get_raw(hf_arr), wmem_array_get_count(hf_arr));
1786 proto_register_subtree_array((int **)ett_arr->data, ett_arr->len);
1787 expert_xml = expert_register_protocol(xml_ns.hf_tag);
1788 expert_register_field_array(expert_xml, ei, array_length(ei));
1789
1790 xml_module = prefs_register_protocol(xml_ns.hf_tag, NULL);
1791 prefs_register_obsolete_preference(xml_module, "heuristic");
1792 prefs_register_obsolete_preference(xml_module, "heuristic_tcp");
1793 prefs_register_obsolete_preference(xml_module, "heuristic_udp");
1794 /* XXX - UCS-2, or UTF-16? */
1795 prefs_register_bool_preference(xml_module, "heuristic_unicode", "Use Unicode in heuristics",
1796 "Try to recognize XML encoded in Unicode (UCS-2BE)",
1797 &pref_heuristic_unicode);
1798
1799 prefs_register_enum_preference(xml_module, "default_encoding", "Default character encoding",
1800 "Use this charset if the 'encoding' attribute of XML declaration is missing."
1801 "Unsupported encoding will be replaced by the default UTF-8.",
1802 &pref_default_encoding, ws_supported_mibenum_vals_character_sets_ev_array, false);
1803
1804 g_array_free(ett_arr, true);
1805
1806 register_init_routine(&xml_init_protocol);
1807 register_cleanup_routine(&xml_cleanup_protocol);
1808
1809 xml_handle = register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1810
1811 init_xml_parser();
1812 }
1813
1814 static void
1815 add_dissector_media(void *k, void *v _U_, void *p _U_)
1816 {
1817 dissector_add_string("media_type", (char *)k, xml_handle);
1818 }
1819
1820 void
1821 proto_reg_handoff_xml(void)
1822 {
1823 wmem_map_foreach(media_types, add_dissector_media, NULL);
1824 dissector_add_string("media_type.suffix", "xml", xml_handle); /* RFC 7303 9.6 */
1825 dissector_add_uint_range_with_preference("tcp.port", "", xml_handle);
1826 dissector_add_uint_range_with_preference("udp.port", XML_UDP_PORT_RANGE, xml_handle);
1827
1828 gssapi_handle = find_dissector_add_dependency("gssapi", xml_ns.hf_tag);
1829
1830 heur_dissector_add("http", dissect_xml_heur, "XML in HTTP", "xml_http", xml_ns.hf_tag, HEURISTIC_DISABLE);
1831 heur_dissector_add("sip", dissect_xml_heur, "XML in SIP", "xml_sip", xml_ns.hf_tag, HEURISTIC_DISABLE);
1832 heur_dissector_add("media", dissect_xml_heur, "XML in media", "xml_media", xml_ns.hf_tag, HEURISTIC_DISABLE);
1833 heur_dissector_add("tcp", dissect_xml_heur, "XML over TCP", "xml_tcp", xml_ns.hf_tag, HEURISTIC_DISABLE);
1834 heur_dissector_add("udp", dissect_xml_heur, "XML over UDP", "xml_udp", xml_ns.hf_tag, HEURISTIC_DISABLE);
1835
1836 heur_dissector_add("wtap_file", dissect_xml_heur, "XML file", "xml_wtap", xml_ns.hf_tag, HEURISTIC_ENABLE);
1837
1838 dissector_add_uint("acdr.tls_application", TLS_APP_XML, xml_handle);
1839 }
1840
1841 /*
1842 * Editor modelines - https://www.wireshark.org/tools/modelines.html
1843 *
1844 * Local variables:
1845 * c-basic-offset: 4
1846 * tab-width: 8
1847 * indent-tabs-mode: nil
1848 * End:
1849 *
1850 * vi: set shiftwidth=4 tabstop=8 expandtab:
1851 * :indentSize=4:tabSize=8:noTabs=true:
1852 */
Metzes wireshark repo