?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
epan/dissectors/pidl/ C99 drsuapi
[wireshark-sm.git] / epan / dissectors / packet-xml.c
blob6a1264541a3d89f253858fb97a36413c499be12a
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 "packet-kerberos.h"
34 #include "read_keytab_file.h"
35
36 #include "packet-xml.h"
37 #include "packet-acdr.h"
38
39 void proto_register_xml(void);
40 void proto_reg_handoff_xml(void);
41
42 struct _attr_reg_data {
43 wmem_array_t *hf;
44 const char *basename;
45 };
46
47
48 static int ett_dtd;
49 static int ett_xmpli;
50
51 static int hf_unknowwn_attrib;
52 static int hf_comment;
53 static int hf_xmlpi;
54 static int hf_dtd_tag;
55 static int hf_doctype;
56 static int hf_cdatasection;
57
58 static expert_field ei_xml_closing_unopened_tag;
59 static expert_field ei_xml_closing_unopened_xmpli_tag;
60 static expert_field ei_xml_unrecognized_text;
61
62 /* dissector handles */
63 static dissector_handle_t xml_handle;
64 static dissector_handle_t gssapi_handle;
65
66 /* Port 3702 is IANA-registered for Web Service Discovery, which uses
67 * SOAP-over-UDP to send XML */
68 #define XML_UDP_PORT_RANGE "3702"
69
70 /* parser definitions */
71 static tvbparse_wanted_t *want;
72 static tvbparse_wanted_t *want_ignore;
73 static tvbparse_wanted_t *want_heur;
74
75 static wmem_map_t *xmpli_names;
76 static wmem_map_t *media_types;
77
78 static xml_ns_t xml_ns = {"xml", "/", -1, -1, -1, NULL, NULL, NULL};
79 static xml_ns_t unknown_ns = {"unknown", "?", -1, -1, -1, NULL, NULL, NULL};
80 static xml_ns_t *root_ns;
81
82 static bool pref_heuristic_unicode;
83 static int pref_default_encoding = IANA_CS_UTF_8;
84
85
86 #define XML_CDATA -1000
87 #define XML_SCOPED_NAME -1001
88
89
90 static wmem_array_t *hf_arr;
91 static GArray *ett_arr;
92 static GRegex* encoding_pattern;
93
94 static const char *default_media_types[] = {
95 "text/xml",
96 "text/vnd.wap.wml",
97 "text/vnd.wap.si",
98 "text/vnd.wap.sl",
99 "text/vnd.wap.co",
100 "text/vnd.wap.emn",
101 "application/3gpp-ims+xml",
102 "application/atom+xml",
103 "application/auth-policy+xml",
104 "application/ccmp+xml",
105 "application/conference-info+xml", /*RFC4575*/
106 "application/cpim-pidf+xml",
107 "application/cpl+xml",
108 "application/dds-web+xml",
109 "application/im-iscomposing+xml", /*RFC3994*/
110 "application/load-control+xml", /*RFC7200*/
111 "application/mathml+xml",
112 "application/media_control+xml",
113 "application/note+xml",
114 "application/pidf+xml",
115 "application/pidf-diff+xml",
116 "application/poc-settings+xml",
117 "application/rdf+xml",
118 "application/reginfo+xml",
119 "application/resource-lists+xml",
120 "application/rlmi+xml",
121 "application/rls-services+xml",
122 "application/rss+xml",
123 "application/rs-metadata+xml",
124 "application/smil",
125 "application/simple-filter+xml",
126 "application/simple-message-summary+xml", /*RFC3842*/
127 "application/simservs+xml",
128 "application/soap+xml",
129 "application/vnd.etsi.aoc+xml",
130 "application/vnd.etsi.cug+xml",
131 "application/vnd.etsi.iptvcommand+xml",
132 "application/vnd.etsi.iptvdiscovery+xml",
133 "application/vnd.etsi.iptvprofile+xml",
134 "application/vnd.etsi.iptvsad-bc+xml",
135 "application/vnd.etsi.iptvsad-cod+xml",
136 "application/vnd.etsi.iptvsad-npvr+xml",
137 "application/vnd.etsi.iptvservice+xml",
138 "application/vnd.etsi.iptvsync+xml",
139 "application/vnd.etsi.iptvueprofile+xml",
140 "application/vnd.etsi.mcid+xml",
141 "application/vnd.etsi.overload-control-policy-dataset+xml",
142 "application/vnd.etsi.pstn+xml",
143 "application/vnd.etsi.sci+xml",
144 "application/vnd.etsi.simservs+xml",
145 "application/vnd.etsi.tsl+xml",
146 "application/vnd.oma.xdm-apd+xml",
147 "application/vnd.oma.fnl+xml",
148 "application/vnd.oma.access-permissions-list+xml",
149 "application/vnd.oma.alias-principals-list+xml",
150 "application/upp-directory+xml", /*OMA-ERELD-XDM-V2_2_1-20170124-A*/
151 "application/vnd.oma.xdm-hi+xml",
152 "application/vnd.oma.xdm-rhi+xml",
153 "application/vnd.oma.xdm-prefs+xml",
154 "application/vnd.oma.xdcp+xml",
155 "application/vnd.oma.bcast.associated-procedure-parameter+xml",
156 "application/vnd.oma.bcast.drm-trigger+xml",
157 "application/vnd.oma.bcast.imd+xml",
158 "application/vnd.oma.bcast.notification+xml",
159 "application/vnd.oma.bcast.sgdd+xml",
160 "application/vnd.oma.bcast.smartcard-trigger+xml",
161 "application/vnd.oma.bcast.sprov+xml",
162 "application/vnd.oma.cab-address-book+xml",
163 "application/vnd.oma.cab-feature-handler+xml",
164 "application/vnd.oma.cab-pcc+xml",
165 "application/vnd.oma.cab-subs-invite+xml",
166 "application/vnd.oma.cab-user-prefs+xml",
167 "application/vnd.oma.dd2+xml",
168 "application/vnd.oma.drm.risd+xml",
169 "application/vnd.oma.group-usage-list+xml",
170 "application/vnd.oma.pal+xml",
171 "application/vnd.oma.poc.detailed-progress-report+xml",
172 "application/vnd.oma.poc.final-report+xml",
173 "application/vnd.oma.poc.groups+xml",
174 "application/vnd.oma.poc.invocation-descriptor+xml",
175 "application/vnd.oma.poc.optimized-progress-report+xml",
176 "application/vnd.oma.scidm.messages+xml",
177 "application/vnd.oma.suppnot+xml", /*OMA-ERELD-Presence_SIMPLE-V2_0-20120710-A*/
178 "application/vnd.oma.xcap-directory+xml",
179 "application/vnd.omads-email+xml",
180 "application/vnd.omads-file+xml",
181 "application/vnd.omads-folder+xml",
182 "application/vnd.3gpp.access-transfer-events+xml",
183 "application/vnd.3gpp.bsf+xml",
184 "application/vnd.3gpp.comm-div-info+xml", /*3GPP TS 24.504 version 8.19.0*/
185 "application/vnd.3gpp.cw+xml",
186 "application/vnd.3gpp.iut+xml", /*3GPP TS 24.337*/
187 "application/vnc.3gpp.iut-config+xml", /*3GPP TS 24.337*/
188 "application/vnd.3gpp.mcptt-info+xml", /*3GPP TS 24.379 version 17.6.0*/
189 "application/vnd.3gpp.mcptt-mbms-usage-info+xml", /*3GPP TS 24.379 version 17.6.0*/
190 "application/vnd.3gpp.mcptt-location-info+xml", /*3GPP TS 24.379 version 17.6.0*/
191 "application/vnd.3gpp.mcptt-affiliation-command+xml", /*3GPP TS 24.379 version 17.6.0*/
192 "application/vnd.3gpp.mcptt-floor-request+xml", /*3GPP TS 24.379 version 17.6.0*/
193 "application/vnd.3gpp.mcptt-signed+xml", /*3GPP TS 24.379 version 17.6.0*/
194 "application/vnd.3gpp.mcptt-regroup+xml", /*3GPP TS 24.379 version 17.6.0*/
195 "application/vnd.3gpp.mcdata-info+xml", /*3GPP TS 24.282 version 17.6.2*/
196 "application/vnd.3gpp.mcdata-mbms-usage-info+xml", /*3GPP TS 24.282 version 17.6.2*/
197 "application/vnd.3gpp.mcdata-location-info+xml", /*3GPP TS 24.282 version 17.6.2*/
198 "application/vnd.3gpp.mcdata-affiliation-command+xml", /*3GPP TS 24.282 version 17.6.2*/
199 "application/vnd.3gpp.mcdata-regroup+xml", /*3GPP TS 24.282 version 17.6.2*/
200 "application/vnd.3gpp.mcvideo-info+xml", /*3GPP TS 24.281 version 17.6.0*/
201 "application/vnd.3gpp.mcvideo-mbms-usage-info+xml", /*3GPP TS 24.281 version 17.6.0*/
202 "application/vnd.3gpp.mcvideo-location-info+xml", /*3GPP TS 24.281 version 17.6.0*/
203 "application/vnd.3gpp.mcvideo-affiliation-command+xml",/*3GPP TS 24.281 version 17.6.0*/
204 "application/vnd.3gpp.transmission-request+xml", /*3GPP TS 24.281 version 17.6.0*/
205 "application/vnd.3gpp.mcptt-ue-init-config+xml", /*3GPP TS 24.484 version 17.5.0*/
206 "application/vnd.3gpp.mcptt-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
207 "application/vnd.3gpp.mcptt-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
208 "application/vnd.3gpp.mcptt-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
209 "application/vnd.3gpp.mcdata-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
210 "application/vnd.3gpp.mcvideo-service-config+xml", /*3GPP TS 24.484 version 17.5.0*/
211 "application/vnd.3gpp.mcvideo-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
212 "application/vnd.3gpp.mcvideo-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
213 "application/vnd.3gpp.mcdata-ue-config+xml", /*3GPP TS 24.484 version 17.5.0*/
214 "application/vnd.3gpp.mcdata-user-profile+xml", /*3GPP TS 24.484 version 17.5.0*/
215 "application/vnd.3gpp.mid-call+xml",
216 "application/vnd.3gpp-prose-pc3ch+xml",
217 "application/vnd.3gpp-prose+xml",
218 "application/vnd.3gpp.replication+xml", /*3GPP TS 24.337*/
219 "application/vnd.3gpp.sms+xml",
220 "application/vnd.3gpp.srvcc-info+xml",
221 "application/vnd.3gpp.srvcc-ext+xml",
222 "application/vnd.3gpp.state-and-event-info+xml",
223 "application/vnd.3gpp.ussd+xml",
224 "application/vnd.3gpp2.bcmcsinfo+xml",
225 "application/vnd.wv.csp+xml",
226 "application/vnd.wv.csp.xml",
227 "application/watcherinfo+xml",
228 "application/xcap-att+xml",
229 "application/xcap-caps+xml",
230 "application/xcap-diff+xml",
231 "application/xcap-el+xml",
232 "application/xcap-error+xml",
233 "application/xcap-ns+xml",
234 "application/xml",
235 "application/xml-dtd",
236 "application/xpidf+xml",
237 "application/xslt+xml",
238 "application/x-crd+xml",
239 "application/x-wms-logconnectstats",
240 "application/x-wms-logplaystats",
241 "application/x-wms-sendevent",
242 "image/svg+xml",
243 "message/imdn+xml", /*RFC5438*/
244 };
245
246 static void insert_xml_frame(xml_frame_t *parent, xml_frame_t *new_child)
247 {
248 new_child->first_child = NULL;
249 new_child->last_child = NULL;
250
251 new_child->parent = parent;
252 new_child->next_sibling = NULL;
253 new_child->prev_sibling = NULL;
254 if (parent == NULL) return; /* root */
255
256 if (parent->first_child == NULL) { /* the 1st child */
257 parent->first_child = new_child;
258 } else { /* following children */
259 parent->last_child->next_sibling = new_child;
260 new_child->prev_sibling = parent->last_child;
261 }
262 parent->last_child = new_child;
263 }
264
265 /* Try to get the 'encoding' attribute from XML declaration, and convert it to
266 * Wireshark character encoding.
267 */
268 static unsigned
269 get_char_encoding(tvbuff_t* tvb, packet_info* pinfo, char** ret_encoding_name) {
270 uint32_t iana_charset_id;
271 unsigned ws_encoding_id;
272 char* encoding_str;
273 GMatchInfo* match_info;
274 const char* xmldecl = (char*)tvb_get_string_enc(pinfo->pool, tvb, 0,
275 MIN(100, tvb_captured_length(tvb)), ENC_UTF_8);
276
277 g_regex_match(encoding_pattern, xmldecl, 0, &match_info);
278 if (g_match_info_matches(match_info)) {
279 char* match_ret = g_match_info_fetch(match_info, 1);
280 encoding_str = ascii_strup_inplace(wmem_strdup(pinfo->pool, match_ret));
281 g_free(match_ret);
282 /* Get the iana charset enum number by the name of the charset. */
283 iana_charset_id = str_to_val(encoding_str,
284 VALUE_STRING_EXT_VS_P(&mibenum_vals_character_sets_ext), IANA_CS_US_ASCII);
285 } else {
286 /* Use default encoding preference if this xml does not contains 'encoding' attribute. */
287 iana_charset_id = pref_default_encoding;
288 encoding_str = val_to_str_ext_wmem(pinfo->pool, iana_charset_id,
289 &mibenum_vals_character_sets_ext, "UNKNOWN");
290 }
291 g_match_info_free(match_info);
292
293 ws_encoding_id = mibenum_charset_to_encoding((unsigned)iana_charset_id);
294
295 /* UTF-8 compatible with ASCII */
296 if (ws_encoding_id == (ENC_NA | ENC_ASCII)) {
297 ws_encoding_id = ENC_UTF_8;
298 *ret_encoding_name = wmem_strdup(pinfo->pool, "UTF-8");
299 } else {
300 *ret_encoding_name = encoding_str;
301 }
302
303 return ws_encoding_id;
304 }
305
306 static int
307 dissect_xml(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree, void* data _U_)
308 {
309 tvbparse_t *tt;
310 static GPtrArray *stack;
311 xml_frame_t *current_frame;
312 const char *colinfo_str;
313 tvbuff_t *decoded;
314 uint16_t try_bom;
315
316 if (stack != NULL)
317 g_ptr_array_free(stack, true);
318
319 stack = g_ptr_array_new();
320 current_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
321 current_frame->type = XML_FRAME_ROOT;
322 current_frame->name = NULL;
323 current_frame->name_orig_case = NULL;
324 current_frame->value = NULL;
325 current_frame->pinfo = pinfo;
326 insert_xml_frame(NULL, current_frame);
327 g_ptr_array_add(stack, current_frame);
328
329 /* Detect and act on possible byte-order mark (BOM) */
330 try_bom = tvb_get_ntohs(tvb, 0);
331 if (try_bom == 0xFEFF) {
332 /* UTF-16BE */
333 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), ENC_UTF_16|ENC_BIG_ENDIAN);
334 size_t l = strlen(data_str);
335 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
336 add_new_data_source(pinfo, decoded, "Decoded UTF-16BE text");
337 }
338 else if(try_bom == 0xFFFE) {
339 /* UTF-16LE (or possibly UTF-32LE, but Wireshark doesn't support UTF-32) */
340 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), ENC_UTF_16|ENC_LITTLE_ENDIAN);
341 size_t l = strlen(data_str);
342 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
343 add_new_data_source(pinfo, decoded, "Decoded UTF-16LE text");
344 }
345 /* Could also test if try_bom is 0xnn00 or 0x00nn to guess endianness if we wanted */
346 else {
347 /* Get character encoding according to XML declaration or preference. */
348 char* encoding_name;
349 unsigned encoding = get_char_encoding(tvb, pinfo, &encoding_name);
350
351 /* Encoding string with encoding, either with or without BOM */
352 const uint8_t *data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), encoding);
353 size_t l = strlen(data_str);
354 decoded = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
355 add_new_data_source(pinfo, decoded, wmem_strdup_printf(pinfo->pool, "Decoded %s text", encoding_name));
356 }
357
358 tt = tvbparse_init(pinfo->pool, decoded, 0, -1, stack, want_ignore);
359 current_frame->start_offset = 0;
360 current_frame->length = tvb_captured_length(decoded);
361
362 current_frame->decryption_keys = wmem_map_new(wmem_packet_scope(), g_str_hash, g_str_equal);
363
364 root_ns = NULL;
365
366 if (pinfo->match_string)
367 root_ns = (xml_ns_t *)wmem_map_lookup(media_types, pinfo->match_string);
368
369 if (! root_ns ) {
370 root_ns = &xml_ns;
371 colinfo_str = "/XML";
372 } else {
373 char *colinfo_str_buf;
374 colinfo_str_buf = wmem_strconcat(wmem_packet_scope(), "/", root_ns->name, NULL);
375 ascii_strup_inplace(colinfo_str_buf);
376 colinfo_str = colinfo_str_buf;
377 }
378
379 col_append_str(pinfo->cinfo, COL_PROTOCOL, colinfo_str);
380
381 current_frame->ns = root_ns;
382
383 current_frame->item = proto_tree_add_item(tree, current_frame->ns->hf_tag, decoded, 0, -1, ENC_UTF_8|ENC_NA);
384 current_frame->tree = proto_item_add_subtree(current_frame->item, current_frame->ns->ett);
385 current_frame->last_item = current_frame->item;
386
387 while(tvbparse_get(tt, want)) ;
388
389 /* Save XML structure in case it is useful for the caller */
390 p_add_proto_data(pinfo->pool, pinfo, xml_ns.hf_tag, 0, current_frame);
391
392 return tvb_captured_length(tvb);
393 }
394
395 static bool dissect_xml_heur(tvbuff_t *tvb, packet_info *pinfo, proto_tree *tree, void *data)
396 {
397 if (tvbparse_peek(tvbparse_init(pinfo->pool, tvb, 0, -1, NULL, want_ignore), want_heur)) {
398 dissect_xml(tvb, pinfo, tree, data);
399 return true;
400 } else if (pref_heuristic_unicode) {
401 const uint8_t *data_str;
402 tvbuff_t *unicode_tvb;
403 uint16_t try_bom;
404 /* XXX - UCS-2, or UTF-16? */
405 int enc = ENC_UCS_2|ENC_LITTLE_ENDIAN;
406 size_t l;
407
408 try_bom = tvb_get_ntohs(tvb, 0);
409 if (try_bom == 0xFEFF) {
410 enc = ENC_UTF_16|ENC_BIG_ENDIAN;
411 }
412 else if(try_bom == 0xFFFE) {
413 enc = ENC_UTF_16|ENC_LITTLE_ENDIAN;
414 }
415
416 data_str = tvb_get_string_enc(pinfo->pool, tvb, 0, tvb_captured_length(tvb), enc);
417 l = strlen(data_str);
418 unicode_tvb = tvb_new_child_real_data(tvb, data_str, (unsigned)l, (int)l);
419 if (tvbparse_peek(tvbparse_init(pinfo->pool, unicode_tvb, 0, -1, NULL, want_ignore), want_heur)) {
420 add_new_data_source(pinfo, unicode_tvb, "UTF8");
421 dissect_xml(unicode_tvb, pinfo, tree, data);
422 return true;
423 }
424 }
425 return false;
426 }
427
428 xml_frame_t *xml_get_tag(xml_frame_t *frame, const char *name)
429 {
430 xml_frame_t *tag = NULL;
431
432 xml_frame_t *xml_item = frame->first_child;
433 while (xml_item) {
434 if (xml_item->type == XML_FRAME_TAG) {
435 if (!name) { /* get the 1st tag */
436 tag = xml_item;
437 break;
438 } else if (xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
439 tag = xml_item;
440 break;
441 }
442 }
443 xml_item = xml_item->next_sibling;
444 }
445
446 return tag;
447 }
448
449 xml_frame_t *xml_get_attrib(xml_frame_t *frame, const char *name)
450 {
451 xml_frame_t *attr = NULL;
452
453 xml_frame_t *xml_item = frame->first_child;
454 while (xml_item) {
455 if ((xml_item->type == XML_FRAME_ATTRIB) &&
456 xml_item->name_orig_case && !strcmp(xml_item->name_orig_case, name)) {
457 attr = xml_item;
458 break;
459 }
460 xml_item = xml_item->next_sibling;
461 }
462
463 return attr;
464 }
465
466 xml_frame_t *xml_get_cdata(xml_frame_t *frame)
467 {
468 xml_frame_t *cdata = NULL;
469
470 xml_frame_t *xml_item = frame->first_child;
471 while (xml_item) {
472 if (xml_item->type == XML_FRAME_CDATA) {
473 cdata = xml_item;
474 break;
475 }
476 xml_item = xml_item->next_sibling;
477 }
478
479 return cdata;
480 }
481
482 static void after_token(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
483 {
484 GPtrArray *stack = (GPtrArray *)tvbparse_data;
485 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
486 int hfid;
487 bool is_cdata = false;
488 proto_item *pi;
489 xml_frame_t *new_frame = NULL;
490 char *text = NULL;
491
492 if (tok->id == XML_CDATA) {
493 hfid = current_frame->ns ? current_frame->ns->hf_cdata : xml_ns.hf_cdata;
494 is_cdata = true;
495 } else if ( tok->id > 0) {
496 hfid = tok->id;
497 } else {
498 hfid = xml_ns.hf_cdata;
499 }
500
501 pi = proto_tree_add_item(current_frame->tree, hfid, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
502
503 text = tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len);
504 proto_item_set_text(pi, "%s", text);
505
506 if (is_cdata) {
507 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
508 new_frame->type = XML_FRAME_CDATA;
509 new_frame->name = NULL;
510 new_frame->name_orig_case = NULL;
511 new_frame->value = tvb_new_subset_length(tok->tvb, tok->offset, tok->len);
512 insert_xml_frame(current_frame, new_frame);
513 new_frame->item = pi;
514 new_frame->last_item = pi;
515 new_frame->tree = NULL;
516 new_frame->start_offset = tok->offset;
517 new_frame->length = tok->len;
518 new_frame->ns = NULL;
519 new_frame->pinfo = current_frame->pinfo;
520 }
521
522 if (new_frame != NULL &&
523 current_frame != NULL &&
524 current_frame->name_orig_case != NULL &&
525 strcmp(current_frame->name_orig_case, "BinarySecurityToken") == 0)
526 {
527 xml_frame_t *value_type = NULL;
528
529 value_type = xml_get_attrib(current_frame, "ValueType");
530 if (value_type != NULL) {
531 const char *s = "http://docs.oasis-open.org/wss/oasis-wss-kerberos-token-profile-1.1#GSS_Kerberosv5_AP_REQ";
532 size_t l = strlen(s);
533 int c;
534 c = tvb_strneql(value_type->value, 0, s, l);
535 if (c == 0) {
536 tvbuff_t *ssp_tvb = base64_to_tvb(new_frame->value, text);
537 add_new_data_source(current_frame->pinfo, ssp_tvb, "GSSAPI Data");
538 call_dissector(gssapi_handle, ssp_tvb,
539 current_frame->pinfo, current_frame->tree);
540 }
541 }
542 }
543 }
544
545 static void before_xmpli(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
546 {
547 GPtrArray *stack = (GPtrArray *)tvbparse_data;
548 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
549 proto_item *pi;
550 proto_tree *pt;
551 tvbparse_elem_t *name_tok = tok->sub->next;
552 char *name = tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb, name_tok->offset, name_tok->len, ENC_ASCII);
553 xml_ns_t *ns = (xml_ns_t *)wmem_map_lookup(xmpli_names, name);
554 xml_frame_t *new_frame;
555
556 int hf_tag;
557 int ett;
558
559 ascii_strdown_inplace(name);
560 if (!ns) {
561 hf_tag = hf_xmlpi;
562 ett = ett_xmpli;
563 } else {
564 hf_tag = ns->hf_tag;
565 ett = ns->ett;
566 }
567
568 pi = proto_tree_add_item(current_frame->tree, hf_tag, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
569
570 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, (name_tok->offset - tok->offset) + name_tok->len));
571
572 pt = proto_item_add_subtree(pi, ett);
573
574 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
575 new_frame->type = XML_FRAME_XMPLI;
576 new_frame->name = name;
577 new_frame->name_orig_case = name;
578 new_frame->value = NULL;
579 insert_xml_frame(current_frame, new_frame);
580 new_frame->item = pi;
581 new_frame->last_item = pi;
582 new_frame->tree = pt;
583 new_frame->start_offset = tok->offset;
584 new_frame->length = tok->len;
585 new_frame->ns = ns;
586 new_frame->pinfo = current_frame->pinfo;
587
588 g_ptr_array_add(stack, new_frame);
589
590 }
591
592 static void after_xmlpi(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
593 {
594 GPtrArray *stack = (GPtrArray *)tvbparse_data;
595 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
596
597 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
598
599 if (stack->len > 1) {
600 g_ptr_array_remove_index_fast(stack, stack->len - 1);
601 } else {
602 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_xmpli_tag,
603 tok->tvb, tok->offset, tok->len);
604 }
605 }
606
607 static void before_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
608 {
609 GPtrArray *stack = (GPtrArray *)tvbparse_data;
610 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
611 tvbparse_elem_t *name_tok = tok->sub->next;
612 char *root_name;
613 char *name = NULL, *name_orig_case = NULL;
614 xml_ns_t *ns;
615 xml_frame_t *new_frame;
616 proto_item *pi;
617 proto_tree *pt;
618
619 if (name_tok->sub->id == XML_SCOPED_NAME) {
620 tvbparse_elem_t *root_tok = name_tok->sub->sub;
621 tvbparse_elem_t *leaf_tok = name_tok->sub->sub->next->next;
622 xml_ns_t *nameroot_ns;
623
624 root_name = (char *)tvb_get_string_enc(wmem_packet_scope(), root_tok->tvb, root_tok->offset, root_tok->len, ENC_ASCII);
625 name = (char *)tvb_get_string_enc(wmem_packet_scope(), leaf_tok->tvb, leaf_tok->offset, leaf_tok->len, ENC_ASCII);
626 name_orig_case = name;
627
628 nameroot_ns = (xml_ns_t *)wmem_map_lookup(xml_ns.elements, root_name);
629
630 if(nameroot_ns) {
631 ns = (xml_ns_t *)wmem_map_lookup(nameroot_ns->elements, name);
632 if (!ns) {
633 ns = &unknown_ns;
634 }
635 } else {
636 ns = &unknown_ns;
637 }
638
639 } else {
640 name = tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb, name_tok->offset, name_tok->len, ENC_ASCII);
641 name_orig_case = wmem_strdup(wmem_packet_scope(), name);
642 ascii_strdown_inplace(name);
643
644 if(current_frame->ns) {
645 ns = (xml_ns_t *)wmem_map_lookup(current_frame->ns->elements, name);
646
647 if (!ns) {
648 if (! ( ns = (xml_ns_t *)wmem_map_lookup(root_ns->elements, name) ) ) {
649 ns = &unknown_ns;
650 }
651 }
652 } else {
653 ns = &unknown_ns;
654 }
655 }
656
657 pi = proto_tree_add_item(current_frame->tree, ns->hf_tag, tok->tvb, tok->offset, tok->len, ENC_UTF_8|ENC_NA);
658 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb,
659 tok->offset,
660 (name_tok->offset - tok->offset) + name_tok->len));
661
662 pt = proto_item_add_subtree(pi, ns->ett);
663
664 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
665 new_frame->type = XML_FRAME_TAG;
666 new_frame->name = name;
667 new_frame->name_orig_case = name_orig_case;
668 new_frame->value = NULL;
669 insert_xml_frame(current_frame, new_frame);
670 new_frame->item = pi;
671 new_frame->last_item = pi;
672 new_frame->tree = pt;
673 new_frame->start_offset = tok->offset;
674 new_frame->length = tok->len;
675 new_frame->ns = ns;
676 new_frame->pinfo = current_frame->pinfo;
677
678 g_ptr_array_add(stack, new_frame);
679
680 }
681
682 static void after_open_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
683 {
684 GPtrArray *stack = (GPtrArray *)tvbparse_data;
685 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
686
687 proto_item_append_text(current_frame->last_item, ">");
688 }
689
690 static void after_closed_tag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
691 {
692 GPtrArray *stack = (GPtrArray *)tvbparse_data;
693 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
694
695 proto_item_append_text(current_frame->last_item, "/>");
696
697 if (stack->len > 1) {
698 g_ptr_array_remove_index_fast(stack, stack->len - 1);
699 } else {
700 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
701 tok->tvb, tok->offset, tok->len);
702 }
703 }
704
705 #ifdef HAVE_KERBEROS
706 struct decryption_key {
707 char *id;
708 size_t key_length;
709 uint8_t key[HASH_SHA1_LENGTH];
710 };
711
712 static void P_SHA1(const uint8_t *Secret, size_t Secret_len,
713 const uint8_t *Seed, size_t Seed_len,
714 uint8_t Result[HASH_SHA1_LENGTH])
715 {
716 gcry_md_hd_t hd = NULL;
717 uint8_t *digest = NULL;
718
719 /*
720 * 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
721 */
722 gcry_md_open(&hd, GCRY_MD_SHA1, GCRY_MD_FLAG_HMAC);
723 gcry_md_setkey(hd, Secret, Secret_len);
724 gcry_md_write(hd, Seed, Seed_len);
725 digest = gcry_md_read(hd, GCRY_MD_SHA1);
726 memcpy(Result, digest, HASH_SHA1_LENGTH);
727
728 gcry_md_close(hd);
729 }
730 #endif /* HAVE_KERBEROS */
731
732 static void after_untag(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
733 {
734 GPtrArray *stack = (GPtrArray *)tvbparse_data;
735 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
736 #ifdef HAVE_KERBEROS
737 xml_frame_t *top_frame = (xml_frame_t *)g_ptr_array_index(stack, 0);
738 #endif /* HAVE_KERBEROS */
739
740 proto_item_set_len(current_frame->item, (tok->offset - current_frame->start_offset) + tok->len);
741 current_frame->length = (tok->offset - current_frame->start_offset) + tok->len;
742
743 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
744
745 if (stack->len > 1) {
746 g_ptr_array_remove_index_fast(stack, stack->len - 1);
747 } else {
748 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
749 tok->tvb, tok->offset, tok->len);
750 }
751
752 #ifdef HAVE_KERBEROS
753 if (current_frame->name_orig_case == NULL) {
754 return;
755 }
756
757 if (strcmp(current_frame->name_orig_case, "DerivedKeyToken") == 0) {
758 xml_frame_t *id_frame = xml_get_attrib(current_frame, "u:Id");
759 xml_frame_t *nonce_frame = xml_get_tag(current_frame, "Nonce");
760 xml_frame_t *nonce_cdata = NULL;
761 tvbuff_t *nonce_tvb = NULL;
762 enc_key_t *ek = NULL;
763 uint8_t seed[64];
764 size_t seed_length = 16; // TODO
765 const size_t key_length = 16; //TODO
766
767 if (id_frame != NULL && nonce_frame != NULL) {
768 nonce_cdata = xml_get_cdata(nonce_frame);
769 }
770 if (nonce_cdata != NULL) {
771 char *text = tvb_format_text(wmem_packet_scope(), nonce_cdata->value, 0,
772 tvb_reported_length(nonce_cdata->value));
773 nonce_tvb = base64_to_tvb(nonce_cdata->value, text);
774 }
775 if (nonce_tvb != NULL) {
776 seed_length = tvb_reported_length(nonce_tvb);
777 seed_length = MIN(seed_length, sizeof(seed));
778 tvb_memcpy(nonce_tvb, seed, 0, seed_length);
779
780 if (krb_decrypt) {
781 read_keytab_file_from_preferences();
782 }
783
784 for (ek=enc_key_list;ek;ek=ek->next) {
785 if (ek->fd_num == (int)current_frame->pinfo->num) {
786 break;
787 }
788 }
789 }
790 if (ek != NULL) {
791 struct decryption_key *key;
792 char *id_str;
793
794 id_str = tvb_format_text(wmem_packet_scope(),
795 id_frame->value, 0,
796 tvb_reported_length(id_frame->value));
797
798 key = wmem_new0(wmem_packet_scope(), struct decryption_key);
799 key->id = wmem_strdup_printf(wmem_packet_scope(), "#%s", id_str);
800 P_SHA1(ek->keyvalue, ek->keylength, seed, seed_length, key->key);
801 key->key_length = key_length;
802
803 wmem_map_insert(top_frame->decryption_keys, key->id, key);
804 }
805 }
806 if (strcmp(current_frame->name_orig_case, "CipherValue") == 0) {
807 xml_frame_t *encrypted_frame = current_frame->parent->parent;
808 xml_frame_t *key_info_frame = NULL;
809 xml_frame_t *token_frame = NULL;
810 xml_frame_t *reference_frame = NULL;
811 xml_frame_t *uri_frame = NULL;
812 const struct decryption_key *key = NULL;
813 xml_frame_t *cdata_frame = NULL;
814 tvbuff_t *crypt_tvb = NULL;
815 tvbuff_t *plain_tvb = NULL;
816
817 key_info_frame = xml_get_tag(encrypted_frame, "KeyInfo");
818 if (key_info_frame != NULL) {
819 token_frame = xml_get_tag(key_info_frame, "SecurityTokenReference");
820 }
821 if (token_frame != NULL) {
822 reference_frame = xml_get_tag(token_frame, "Reference");
823 }
824 if (reference_frame != NULL) {
825 uri_frame = xml_get_attrib(reference_frame, "URI");
826 }
827
828 if (uri_frame != NULL) {
829 char *key_id = tvb_format_text(wmem_packet_scope(), uri_frame->value, 0,
830 tvb_reported_length(uri_frame->value));
831
832 key = (const struct decryption_key *)wmem_map_lookup(top_frame->decryption_keys, key_id);
833 }
834 if (key != NULL) {
835 cdata_frame = xml_get_cdata(current_frame);
836 }
837 if (cdata_frame != NULL) {
838 char *text = tvb_format_text(wmem_packet_scope(), cdata_frame->value, 0,
839 tvb_reported_length(cdata_frame->value));
840 crypt_tvb = base64_to_tvb(cdata_frame->value, text);
841 }
842 if (crypt_tvb != NULL) {
843 gcry_cipher_hd_t cipher_hd = NULL;
844 uint8_t *data = NULL;
845 unsigned data_length = tvb_reported_length(crypt_tvb);
846
847 data = (uint8_t *)tvb_memdup(wmem_packet_scope(),
848 crypt_tvb, 0, data_length);
849
850 /* Open the cipher. */
851 gcry_cipher_open(&cipher_hd, GCRY_CIPHER_AES128, GCRY_CIPHER_MODE_CBC, 0);
852
853 gcry_cipher_setkey(cipher_hd, key->key, key->key_length);
854 gcry_cipher_encrypt(cipher_hd, data, data_length, NULL, 0);
855 gcry_cipher_close(cipher_hd);
856
857 plain_tvb = tvb_new_child_real_data(crypt_tvb, data,
858 data_length, data_length);
859 add_new_data_source(current_frame->pinfo, plain_tvb, "Decrypted Data");
860 }
861 }
862 #endif /* HAVE_KERBEROS */
863 }
864
865 static void before_dtd_doctype(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
866 {
867 GPtrArray *stack = (GPtrArray *)tvbparse_data;
868 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
869 xml_frame_t *new_frame;
870 tvbparse_elem_t *name_tok = tok->sub->next->next->next->sub->sub;
871 proto_tree *dtd_item = proto_tree_add_item(current_frame->tree, hf_doctype,
872 name_tok->tvb, name_tok->offset,
873 name_tok->len, ENC_ASCII);
874
875 proto_item_set_text(dtd_item, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len));
876
877 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
878 new_frame->type = XML_FRAME_DTD_DOCTYPE;
879 new_frame->name = (char *)tvb_get_string_enc(wmem_packet_scope(), name_tok->tvb,
880 name_tok->offset,
881 name_tok->len, ENC_ASCII);
882 new_frame->name_orig_case = new_frame->name;
883 new_frame->value = NULL;
884 insert_xml_frame(current_frame, new_frame);
885 new_frame->item = dtd_item;
886 new_frame->last_item = dtd_item;
887 new_frame->tree = proto_item_add_subtree(dtd_item, ett_dtd);
888 new_frame->start_offset = tok->offset;
889 new_frame->length = tok->len;
890 new_frame->ns = NULL;
891 new_frame->pinfo = current_frame->pinfo;
892
893 g_ptr_array_add(stack, new_frame);
894 }
895
896 static void pop_stack(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
897 {
898 GPtrArray *stack = (GPtrArray *)tvbparse_data;
899 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
900
901 if (stack->len > 1) {
902 g_ptr_array_remove_index_fast(stack, stack->len - 1);
903 } else {
904 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
905 tok->tvb, tok->offset, tok->len);
906 }
907 }
908
909 static void after_dtd_close(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
910 {
911 GPtrArray *stack = (GPtrArray *)tvbparse_data;
912 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
913
914 proto_tree_add_format_text(current_frame->tree, tok->tvb, tok->offset, tok->len);
915 if (stack->len > 1) {
916 g_ptr_array_remove_index_fast(stack, stack->len - 1);
917 } else {
918 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_closing_unopened_tag,
919 tok->tvb, tok->offset, tok->len);
920 }
921 }
922
923 static void get_attrib_value(void *tvbparse_data _U_, const void *wanted_data _U_, tvbparse_elem_t *tok)
924 {
925 tok->data = tok->sub;
926 }
927
928 static void after_attrib(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok)
929 {
930 GPtrArray *stack = (GPtrArray *)tvbparse_data;
931 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
932 char *name, *name_orig_case;
933 tvbparse_elem_t *value;
934 tvbparse_elem_t *value_part = (tvbparse_elem_t *)tok->sub->next->next->data;
935 int *hfidp;
936 int hfid;
937 proto_item *pi;
938 xml_frame_t *new_frame;
939
940 name = tvb_get_string_enc(wmem_packet_scope(), tok->sub->tvb, tok->sub->offset, tok->sub->len, ENC_ASCII);
941 name_orig_case = wmem_strdup(wmem_packet_scope(), name);
942 ascii_strdown_inplace(name);
943
944 if(current_frame->ns && (hfidp = (int *)wmem_map_lookup(current_frame->ns->attributes, name) )) {
945 hfid = *hfidp;
946 value = value_part;
947 } else {
948 hfid = hf_unknowwn_attrib;
949 value = tok;
950 }
951
952 pi = proto_tree_add_item(current_frame->tree, hfid, value->tvb, value->offset, value->len, ENC_UTF_8|ENC_NA);
953 proto_item_set_text(pi, "%s", tvb_format_text(wmem_packet_scope(), tok->tvb, tok->offset, tok->len));
954
955 current_frame->last_item = pi;
956
957 new_frame = wmem_new(wmem_packet_scope(), xml_frame_t);
958 new_frame->type = XML_FRAME_ATTRIB;
959 new_frame->name = name;
960 new_frame->name_orig_case = name_orig_case;
961 new_frame->value = tvb_new_subset_length(value_part->tvb, value_part->offset,
962 value_part->len);
963 insert_xml_frame(current_frame, new_frame);
964 new_frame->item = pi;
965 new_frame->last_item = pi;
966 new_frame->tree = NULL;
967 new_frame->start_offset = tok->offset;
968 new_frame->length = tok->len;
969 new_frame->ns = NULL;
970 new_frame->pinfo = current_frame->pinfo;
971
972 }
973
974 static void unrecognized_token(void *tvbparse_data, const void *wanted_data _U_, tvbparse_elem_t *tok _U_)
975 {
976 GPtrArray *stack = (GPtrArray *)tvbparse_data;
977 xml_frame_t *current_frame = (xml_frame_t *)g_ptr_array_index(stack, stack->len - 1);
978
979 proto_tree_add_expert(current_frame->tree, current_frame->pinfo, &ei_xml_unrecognized_text,
980 tok->tvb, tok->offset, tok->len);
981
982 }
983
984
985
986 static void init_xml_parser(void)
987 {
988 tvbparse_wanted_t *want_name =
989 tvbparse_chars(-1, 1, 0,
990 "abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789",
991 NULL, NULL, NULL);
992 tvbparse_wanted_t *want_attr_name =
993 tvbparse_chars(-1, 1, 0,
994 "abcdefghijklmnopqrstuvwxyz.-_ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789:",
995 NULL, NULL, NULL);
996
997 tvbparse_wanted_t *want_scoped_name = tvbparse_set_seq(XML_SCOPED_NAME, NULL, NULL, NULL,
998 want_name,
999 tvbparse_char(-1, ":", NULL, NULL, NULL),
1000 want_name,
1001 NULL);
1002
1003 tvbparse_wanted_t *want_tag_name = tvbparse_set_oneof(0, NULL, NULL, NULL,
1004 want_scoped_name,
1005 want_name,
1006 NULL);
1007
1008 tvbparse_wanted_t *want_attrib_value = tvbparse_set_oneof(0, NULL, NULL, get_attrib_value,
1009 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb, '\"', '\\'),
1010 tvbparse_quoted(-1, NULL, NULL, tvbparse_shrink_token_cb, '\'', '\\'),
1011 tvbparse_chars(-1, 1, 0, "0123456789", NULL, NULL, NULL),
1012 want_name,
1013 NULL);
1014
1015 tvbparse_wanted_t *want_attributes = tvbparse_one_or_more(-1, NULL, NULL, NULL,
1016 tvbparse_set_seq(-1, NULL, NULL, after_attrib,
1017 want_attr_name,
1018 tvbparse_char(-1, "=", NULL, NULL, NULL),
1019 want_attrib_value,
1020 NULL));
1021
1022 tvbparse_wanted_t *want_stoptag = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1023 tvbparse_char(-1, ">", NULL, NULL, after_open_tag),
1024 tvbparse_string(-1, "/>", NULL, NULL, after_closed_tag),
1025 NULL);
1026
1027 tvbparse_wanted_t *want_stopxmlpi = tvbparse_string(-1, "?>", NULL, NULL, after_xmlpi);
1028
1029 tvbparse_wanted_t *want_comment = tvbparse_set_seq(hf_comment, NULL, NULL, after_token,
1030 tvbparse_string(-1, "<!--", NULL, NULL, NULL),
1031 tvbparse_until(-1, NULL, NULL, NULL,
1032 tvbparse_string(-1, "-->", NULL, NULL, NULL),
1033 TP_UNTIL_INCLUDE),
1034 NULL);
1035
1036 tvbparse_wanted_t *want_cdatasection = tvbparse_set_seq(hf_cdatasection, NULL, NULL, after_token,
1037 tvbparse_string(-1, "<![CDATA[", NULL, NULL, NULL),
1038 tvbparse_until(-1, NULL, NULL, NULL,
1039 tvbparse_string(-1, "]]>", NULL, NULL, NULL),
1040 TP_UNTIL_INCLUDE),
1041 NULL);
1042
1043 tvbparse_wanted_t *want_xmlpi = tvbparse_set_seq(hf_xmlpi, NULL, before_xmpli, NULL,
1044 tvbparse_string(-1, "<?", NULL, NULL, NULL),
1045 want_name,
1046 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1047 want_stopxmlpi,
1048 tvbparse_set_seq(-1, NULL, NULL, NULL,
1049 want_attributes,
1050 want_stopxmlpi,
1051 NULL),
1052 NULL),
1053 NULL);
1054
1055 tvbparse_wanted_t *want_closing_tag = tvbparse_set_seq(0, NULL, NULL, after_untag,
1056 tvbparse_char(-1, "<", NULL, NULL, NULL),
1057 tvbparse_char(-1, "/", NULL, NULL, NULL),
1058 want_tag_name,
1059 tvbparse_char(-1, ">", NULL, NULL, NULL),
1060 NULL);
1061
1062 tvbparse_wanted_t *want_doctype_start = tvbparse_set_seq(-1, NULL, before_dtd_doctype, NULL,
1063 tvbparse_char(-1, "<", NULL, NULL, NULL),
1064 tvbparse_char(-1, "!", NULL, NULL, NULL),
1065 tvbparse_casestring(-1, "DOCTYPE", NULL, NULL, NULL),
1066 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1067 tvbparse_set_seq(-1, NULL, NULL, NULL,
1068 want_name,
1069 tvbparse_char(-1, "[", NULL, NULL, NULL),
1070 NULL),
1071 tvbparse_set_seq(-1, NULL, NULL, pop_stack,
1072 want_name,
1073 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1074 tvbparse_casestring(-1, "PUBLIC", NULL, NULL, NULL),
1075 tvbparse_casestring(-1, "SYSTEM", NULL, NULL, NULL),
1076 NULL),
1077 tvbparse_until(-1, NULL, NULL, NULL,
1078 tvbparse_char(-1, ">", NULL, NULL, NULL),
1079 TP_UNTIL_INCLUDE),
1080 NULL),
1081 NULL),
1082 NULL);
1083
1084 tvbparse_wanted_t *want_dtd_tag = tvbparse_set_seq(hf_dtd_tag, NULL, NULL, after_token,
1085 tvbparse_char(-1, "<", NULL, NULL, NULL),
1086 tvbparse_char(-1, "!", NULL, NULL, NULL),
1087 tvbparse_until(-1, NULL, NULL, NULL,
1088 tvbparse_char(-1, ">", NULL, NULL, NULL),
1089 TP_UNTIL_INCLUDE),
1090 NULL);
1091
1092 tvbparse_wanted_t *want_tag = tvbparse_set_seq(-1, NULL, before_tag, NULL,
1093 tvbparse_char(-1, "<", NULL, NULL, NULL),
1094 want_tag_name,
1095 tvbparse_set_oneof(-1, NULL, NULL, NULL,
1096 tvbparse_set_seq(-1, NULL, NULL, NULL,
1097 want_attributes,
1098 want_stoptag,
1099 NULL),
1100 want_stoptag,
1101 NULL),
1102 NULL);
1103
1104 tvbparse_wanted_t *want_dtd_close = tvbparse_set_seq(-1, NULL, NULL, after_dtd_close,
1105 tvbparse_char(-1, "]", NULL, NULL, NULL),
1106 tvbparse_char(-1, ">", NULL, NULL, NULL),
1107 NULL);
1108
1109 want_ignore = tvbparse_chars(-1, 1, 0, " \t\r\n", NULL, NULL, NULL);
1110
1111
1112 want = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1113 want_comment,
1114 want_cdatasection,
1115 want_xmlpi,
1116 want_closing_tag,
1117 want_doctype_start,
1118 want_dtd_close,
1119 want_dtd_tag,
1120 want_tag,
1121 tvbparse_not_chars(XML_CDATA, 1, 0, "<", NULL, NULL, after_token),
1122 tvbparse_not_chars(-1, 1, 0, " \t\r\n", NULL, NULL, unrecognized_token),
1123 NULL);
1124
1125 want_heur = tvbparse_set_oneof(-1, NULL, NULL, NULL,
1126 want_comment,
1127 want_cdatasection,
1128 want_xmlpi,
1129 want_doctype_start,
1130 want_dtd_tag,
1131 want_tag,
1132 NULL);
1133
1134 }
1135
1136
1137 static xml_ns_t *xml_new_namespace(wmem_map_t *hash, const char *name, ...)
1138 {
1139 xml_ns_t *ns = wmem_new(wmem_epan_scope(), xml_ns_t);
1140 va_list ap;
1141 char *attr_name;
1142
1143 ns->name = wmem_strdup(wmem_epan_scope(), name);
1144 ns->hf_tag = -1;
1145 ns->hf_cdata = -1;
1146 ns->ett = -1;
1147 ns->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1148 ns->elements = NULL;
1149
1150 va_start(ap, name);
1151
1152 while(( attr_name = va_arg(ap, char *) )) {
1153 int *hfp = wmem_new(wmem_epan_scope(), int);
1154 *hfp = -1;
1155 wmem_map_insert(ns->attributes, wmem_strdup(wmem_epan_scope(), attr_name), hfp);
1156 };
1157
1158 va_end(ap);
1159
1160 wmem_map_insert(hash, ns->name, ns);
1161
1162 return ns;
1163 }
1164
1165
1166 static void add_xml_field(wmem_array_t *hfs, int *p_id, const char *name, const char *fqn)
1167 {
1168 hf_register_info hfri;
1169
1170 hfri.p_id = p_id;
1171 hfri.hfinfo.name = name;
1172 hfri.hfinfo.abbrev = fqn;
1173 hfri.hfinfo.type = FT_STRING;
1174 hfri.hfinfo.display = BASE_NONE;
1175 hfri.hfinfo.strings = NULL;
1176 hfri.hfinfo.bitmask = 0x0;
1177 hfri.hfinfo.blurb = NULL;
1178 HFILL_INIT(hfri);
1179
1180 wmem_array_append_one(hfs, hfri);
1181 }
1182
1183 static void add_xml_attribute_names(void *k, void *v, void *p)
1184 {
1185 struct _attr_reg_data *d = (struct _attr_reg_data *)p;
1186 const char *basename = wmem_strconcat(wmem_epan_scope(), d->basename, ".", (char *)k, NULL);
1187
1188 add_xml_field(d->hf, (int*) v, (char *)k, basename);
1189 }
1190
1191
1192 static void add_xmlpi_namespace(void *k _U_, void *v, void *p)
1193 {
1194 xml_ns_t *ns = (xml_ns_t *)v;
1195 const char *basename = wmem_strconcat(wmem_epan_scope(), (char *)p, ".", ns->name, NULL);
1196 int *ett_p = &(ns->ett);
1197 struct _attr_reg_data d;
1198
1199 add_xml_field(hf_arr, &(ns->hf_tag), basename, basename);
1200
1201 g_array_append_val(ett_arr, ett_p);
1202
1203 d.basename = basename;
1204 d.hf = hf_arr;
1205
1206 wmem_map_foreach(ns->attributes, add_xml_attribute_names, &d);
1207
1208 }
1209
1210 static void destroy_dtd_data(dtd_build_data_t *dtd_data)
1211 {
1212 g_free(dtd_data->proto_name);
1213 g_free(dtd_data->media_type);
1214 g_free(dtd_data->description);
1215 g_free(dtd_data->proto_root);
1216
1217 g_string_free(dtd_data->error, true);
1218
1219 while(dtd_data->elements->len) {
1220 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index_fast(dtd_data->elements, 0);
1221 g_ptr_array_free(nl->list, true);
1222 g_free(nl->name);
1223 g_free(nl);
1224 }
1225
1226 g_ptr_array_free(dtd_data->elements, true);
1227
1228 while(dtd_data->attributes->len) {
1229 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index_fast(dtd_data->attributes, 0);
1230 g_ptr_array_free(nl->list, true);
1231 g_free(nl->name);
1232 g_free(nl);
1233 }
1234
1235 g_ptr_array_free(dtd_data->attributes, true);
1236
1237 g_free(dtd_data);
1238 }
1239
1240 static void copy_attrib_item(void *k, void *v _U_, void *p)
1241 {
1242 char *key = (char *)wmem_strdup(wmem_epan_scope(), (const char *)k);
1243 int *value = wmem_new(wmem_epan_scope(), int);
1244 wmem_map_t *dst = (wmem_map_t *)p;
1245
1246 *value = -1;
1247 wmem_map_insert(dst, key, value);
1248
1249 }
1250
1251 static wmem_map_t *copy_attributes_hash(wmem_map_t *src)
1252 {
1253 wmem_map_t *dst = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1254
1255 wmem_map_foreach(src, copy_attrib_item, dst);
1256
1257 return dst;
1258 }
1259
1260 static xml_ns_t *duplicate_element(xml_ns_t *orig)
1261 {
1262 xml_ns_t *new_item = wmem_new(wmem_epan_scope(), xml_ns_t);
1263 unsigned i;
1264
1265 new_item->name = wmem_strdup(wmem_epan_scope(), orig->name);
1266 new_item->hf_tag = -1;
1267 new_item->hf_cdata = -1;
1268 new_item->ett = -1;
1269 new_item->attributes = copy_attributes_hash(orig->attributes);
1270 new_item->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1271 new_item->element_names = g_ptr_array_new();
1272
1273 for(i=0; i < orig->element_names->len; i++) {
1274 g_ptr_array_add(new_item->element_names,
1275 g_ptr_array_index(orig->element_names, i));
1276 }
1277
1278 return new_item;
1279 }
1280
1281 static char *fully_qualified_name(GPtrArray *hier, char *name, char *proto_name)
1282 {
1283 unsigned i;
1284 wmem_strbuf_t *s = wmem_strbuf_new(wmem_epan_scope(), proto_name);
1285
1286 wmem_strbuf_append(s, ".");
1287
1288 for (i = 1; i < hier->len; i++) {
1289 wmem_strbuf_append_printf(s, "%s.", (char *)g_ptr_array_index(hier, i));
1290 }
1291
1292 wmem_strbuf_append(s, name);
1293
1294 return wmem_strbuf_finalize(s);
1295 }
1296
1297
1298 // NOLINTNEXTLINE(misc-no-recursion)
1299 static xml_ns_t *make_xml_hier(char *elem_name,
1300 xml_ns_t *root,
1301 wmem_map_t *elements,
1302 GPtrArray *hier,
1303 GString *error,
1304 wmem_array_t *hfs,
1305 GArray *etts,
1306 char *proto_name)
1307 {
1308 xml_ns_t *fresh;
1309 xml_ns_t *orig;
1310 char *fqn;
1311 int *ett_p;
1312 bool recurred = false;
1313 unsigned i;
1314 struct _attr_reg_data d;
1315
1316 if ( g_str_equal(elem_name, root->name) ) {
1317 return NULL;
1318 }
1319
1320 if (! ( orig = (xml_ns_t *)wmem_map_lookup(elements, elem_name) )) {
1321 g_string_append_printf(error, "element '%s' is not defined\n", elem_name);
1322 return NULL;
1323 }
1324
1325 if (hier->len >= prefs.gui_max_tree_depth) {
1326 g_string_append_printf(error, "hierarchy too deep: %u\n", hier->len);
1327 return NULL;
1328 }
1329
1330 for (i = 0; i < hier->len; i++) {
1331 if( (elem_name) && (strcmp(elem_name, (char *) g_ptr_array_index(hier, i) ) == 0 )) {
1332 recurred = true;
1333 }
1334 }
1335
1336 if (recurred) {
1337 return NULL;
1338 }
1339
1340 fqn = fully_qualified_name(hier, elem_name, proto_name);
1341
1342 fresh = duplicate_element(orig);
1343 fresh->fqn = fqn;
1344
1345 add_xml_field(hfs, &(fresh->hf_tag), wmem_strdup(wmem_epan_scope(), elem_name), fqn);
1346 add_xml_field(hfs, &(fresh->hf_cdata), wmem_strdup(wmem_epan_scope(), elem_name), fqn);
1347
1348 ett_p = &fresh->ett;
1349 g_array_append_val(etts, ett_p);
1350
1351 d.basename = fqn;
1352 d.hf = hfs;
1353
1354 wmem_map_foreach(fresh->attributes, add_xml_attribute_names, &d);
1355
1356 while(fresh->element_names->len) {
1357 char *child_name = (char *)g_ptr_array_remove_index(fresh->element_names, 0);
1358 xml_ns_t *child_element = NULL;
1359
1360 g_ptr_array_add(hier, elem_name);
1361 child_element = make_xml_hier(child_name, root, elements, hier, error, hfs, etts, proto_name);
1362 g_ptr_array_remove_index_fast(hier, hier->len - 1);
1363
1364 if (child_element) {
1365 wmem_map_insert(fresh->elements, child_element->name, child_element);
1366 }
1367 }
1368
1369 g_ptr_array_free(fresh->element_names, true);
1370 fresh->element_names = NULL;
1371 return fresh;
1372 }
1373
1374 static void free_elements(void *k _U_, void *v, void *p _U_)
1375 {
1376 xml_ns_t *e = (xml_ns_t *)v;
1377
1378 while (e->element_names->len) {
1379 g_free(g_ptr_array_remove_index(e->element_names, 0));
1380 }
1381
1382 g_ptr_array_free(e->element_names, true);
1383 }
1384
1385 static void register_dtd(dtd_build_data_t *dtd_data, GString *errors)
1386 {
1387 wmem_map_t *elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1388 char *root_name = NULL;
1389 xml_ns_t *root_element = NULL;
1390 wmem_array_t *hfs;
1391 GArray *etts;
1392 GPtrArray *hier;
1393 char *curr_name;
1394 GPtrArray *element_names = g_ptr_array_new();
1395
1396 /* we first populate elements with the those coming from the parser */
1397 while(dtd_data->elements->len) {
1398 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index(dtd_data->elements, 0);
1399 xml_ns_t *element = wmem_new(wmem_epan_scope(), xml_ns_t);
1400
1401 /* we will use the first element found as root in case no other one was given. */
1402 if (root_name == NULL)
1403 root_name = wmem_strdup(wmem_epan_scope(), nl->name);
1404
1405 element->name = wmem_strdup(wmem_epan_scope(), nl->name);
1406 element->element_names = nl->list;
1407 element->hf_tag = -1;
1408 element->hf_cdata = -1;
1409 element->ett = -1;
1410 element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1411 element->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1412
1413 if( wmem_map_lookup(elements, element->name) ) {
1414 g_string_append_printf(errors, "element %s defined more than once\n", element->name);
1415 free_elements(NULL, element, NULL);
1416 } else {
1417 wmem_map_insert(elements, element->name, element);
1418 g_ptr_array_add(element_names, wmem_strdup(wmem_epan_scope(), element->name));
1419 }
1420
1421 g_free(nl->name);
1422 g_free(nl);
1423 }
1424
1425 /* then we add the attributes to its relative elements */
1426 while(dtd_data->attributes->len) {
1427 dtd_named_list_t *nl = (dtd_named_list_t *)g_ptr_array_remove_index(dtd_data->attributes, 0);
1428 xml_ns_t *element = (xml_ns_t *)wmem_map_lookup(elements, nl->name);
1429
1430 if (element) {
1431 while(nl->list->len) {
1432 char *name = (char *)g_ptr_array_remove_index(nl->list, 0);
1433 int *id_p = wmem_new(wmem_epan_scope(), int);
1434
1435 *id_p = -1;
1436 wmem_map_insert(element->attributes, wmem_strdup(wmem_epan_scope(), name), id_p);
1437 g_free(name); }
1438 }
1439 else {
1440 g_string_append_printf(errors, "element %s is not defined\n", nl->name);
1441 }
1442
1443 g_free(nl->name);
1444 g_ptr_array_free(nl->list, true);
1445 g_free(nl);
1446 }
1447
1448 /* if a proto_root is defined in the dtd we'll use that as root */
1449 if( dtd_data->proto_root ) {
1450 wmem_free(wmem_epan_scope(), root_name);
1451 root_name = wmem_strdup(wmem_epan_scope(), dtd_data->proto_root);
1452 }
1453
1454 /* we use a stack with the names to avoid recurring infinitely */
1455 hier = g_ptr_array_new();
1456
1457 /*
1458 * if a proto name was given in the dtd the dtd will be used as a protocol
1459 * or else the dtd will be loaded as a branch of the xml namespace
1460 */
1461 if( ! dtd_data->proto_name ) {
1462 hfs = hf_arr;
1463 etts = ett_arr;
1464 g_ptr_array_add(hier, wmem_strdup(wmem_epan_scope(), "xml"));
1465 } else {
1466 /*
1467 * if we were given a proto_name the namespace will be registered
1468 * as an independent protocol with its own hf and ett arrays.
1469 */
1470 hfs = wmem_array_new(wmem_epan_scope(), sizeof(hf_register_info));
1471 etts = g_array_new(false, false, sizeof(int *));
1472 }
1473
1474 /* the root element of the dtd's namespace */
1475 root_element = wmem_new(wmem_epan_scope(), xml_ns_t);
1476 root_element->name = wmem_strdup(wmem_epan_scope(), root_name);
1477 root_element->fqn = dtd_data->proto_name ? wmem_strdup(wmem_epan_scope(), dtd_data->proto_name) : root_element->name;
1478 root_element->hf_tag = -1;
1479 root_element->hf_cdata = -1;
1480 root_element->ett = -1;
1481 root_element->elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1482 root_element->element_names = element_names;
1483
1484 /*
1485 * we can either create a namespace as a flat namespace
1486 * in which all the elements are at the root level
1487 * or we can create a recursive namespace
1488 */
1489 if (dtd_data->recursion) {
1490 xml_ns_t *orig_root;
1491
1492 make_xml_hier(root_name, root_element, elements, hier, errors, hfs, etts, dtd_data->proto_name);
1493
1494 wmem_map_insert(root_element->elements, (void *)root_element->name, root_element);
1495
1496 orig_root = (xml_ns_t *)wmem_map_lookup(elements, root_name);
1497
1498 /* if the root element was defined copy its attrlist to the child */
1499 if(orig_root) {
1500 struct _attr_reg_data d;
1501
1502 d.basename = dtd_data->proto_name;
1503 d.hf = hfs;
1504
1505 root_element->attributes = copy_attributes_hash(orig_root->attributes);
1506 wmem_map_foreach(root_element->attributes, add_xml_attribute_names, &d);
1507 } else {
1508 root_element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1509 }
1510
1511 /* we then create all the sub hierarchies to catch the recurred cases */
1512 g_ptr_array_add(hier, root_name);
1513
1514 while(root_element->element_names->len) {
1515 curr_name = (char *)g_ptr_array_remove_index(root_element->element_names, 0);
1516
1517 if( ! wmem_map_lookup(root_element->elements, curr_name) ) {
1518 xml_ns_t *fresh = make_xml_hier(curr_name, root_element, elements, hier, errors,
1519 hfs, etts, dtd_data->proto_name);
1520 wmem_map_insert(root_element->elements, (void *)fresh->name, fresh);
1521 }
1522 }
1523
1524 } else {
1525 /* a flat namespace */
1526 g_ptr_array_add(hier, root_name);
1527
1528 root_element->attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1529
1530 while(root_element->element_names->len) {
1531 xml_ns_t *fresh;
1532 int *ett_p;
1533 struct _attr_reg_data d;
1534
1535 curr_name = (char *)g_ptr_array_remove_index(root_element->element_names, 0);
1536 fresh = duplicate_element((xml_ns_t *)wmem_map_lookup(elements, curr_name));
1537 fresh->fqn = fully_qualified_name(hier, curr_name, root_name);
1538
1539 add_xml_field(hfs, &(fresh->hf_tag), curr_name, fresh->fqn);
1540 add_xml_field(hfs, &(fresh->hf_cdata), curr_name, fresh->fqn);
1541
1542 d.basename = fresh->fqn;
1543 d.hf = hfs;
1544
1545 wmem_map_foreach(fresh->attributes, add_xml_attribute_names, &d);
1546
1547 ett_p = &fresh->ett;
1548 g_array_append_val(etts, ett_p);
1549
1550 g_ptr_array_free(fresh->element_names, true);
1551
1552 wmem_map_insert(root_element->elements, (void *)fresh->name, fresh);
1553 }
1554 }
1555
1556 g_ptr_array_free(element_names, true);
1557
1558 g_ptr_array_free(hier, true);
1559
1560 /*
1561 * if we were given a proto_name the namespace will be registered
1562 * as an independent protocol.
1563 */
1564 if( dtd_data->proto_name ) {
1565 int *ett_p;
1566 char *full_name, *short_name;
1567
1568 if (dtd_data->description) {
1569 full_name = wmem_strdup(wmem_epan_scope(), dtd_data->description);
1570 } else {
1571 full_name = wmem_strdup(wmem_epan_scope(), root_name);
1572 }
1573 short_name = wmem_strdup(wmem_epan_scope(), dtd_data->proto_name);
1574
1575 ett_p = &root_element->ett;
1576 g_array_append_val(etts, ett_p);
1577
1578 add_xml_field(hfs, &root_element->hf_cdata, root_element->name, root_element->fqn);
1579
1580 root_element->hf_tag = proto_register_protocol(full_name, short_name, short_name);
1581 proto_register_field_array(root_element->hf_tag, (hf_register_info*)wmem_array_get_raw(hfs), wmem_array_get_count(hfs));
1582 proto_register_subtree_array((int **)etts->data, etts->len);
1583
1584 if (dtd_data->media_type) {
1585 char* media_type = wmem_strdup(wmem_epan_scope(), dtd_data->media_type);
1586 wmem_map_insert(media_types, media_type, root_element);
1587 }
1588
1589 g_array_free(etts, true);
1590 }
1591
1592 wmem_map_insert(xml_ns.elements, root_element->name, root_element);
1593 wmem_map_foreach(elements, free_elements, NULL);
1594
1595 destroy_dtd_data(dtd_data);
1596 wmem_free(wmem_epan_scope(), root_name);
1597 }
1598
1599 # define DIRECTORY_T GDir
1600 # define FILE_T char
1601 # define OPENDIR_OP(name) g_dir_open(name, 0, dummy)
1602 # define DIRGETNEXT_OP(dir) g_dir_read_name(dir)
1603 # define GETFNAME_OP(file) (file);
1604 # define CLOSEDIR_OP(dir) g_dir_close(dir)
1605
1606 static void init_xml_names(void)
1607 {
1608 unsigned i;
1609 DIRECTORY_T *dir;
1610 const FILE_T *file;
1611 const char *filename;
1612 char *dirname;
1613
1614 GError **dummy = wmem_new(wmem_epan_scope(), GError *);
1615 *dummy = NULL;
1616
1617 xmpli_names = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1618 media_types = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1619
1620 unknown_ns.elements = xml_ns.elements = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1621 unknown_ns.attributes = xml_ns.attributes = wmem_map_new(wmem_epan_scope(), g_str_hash, g_str_equal);
1622
1623 xml_new_namespace(xmpli_names, "xml", "version", "encoding", "standalone", NULL);
1624
1625 dirname = get_persconffile_path("dtds", false);
1626
1627 if (test_for_directory(dirname) != EISDIR) {
1628 /* Although dir isn't a directory it may still use memory */
1629 g_free(dirname);
1630 dirname = get_datafile_path("dtds");
1631 }
1632
1633 if (test_for_directory(dirname) == EISDIR) {
1634 if ((dir = OPENDIR_OP(dirname)) != NULL) {
1635 GString *errors = g_string_new("");
1636
1637 while ((file = DIRGETNEXT_OP(dir)) != NULL) {
1638 unsigned namelen;
1639 filename = GETFNAME_OP(file);
1640
1641 namelen = (int)strlen(filename);
1642 if ( namelen > 4 && ( g_ascii_strcasecmp(filename+(namelen-4), ".dtd") == 0 ) ) {
1643 GString *preparsed;
1644 dtd_build_data_t *dtd_data;
1645
1646 g_string_truncate(errors, 0);
1647 preparsed = dtd_preparse(dirname, filename, errors);
1648
1649 if (errors->len) {
1650 report_failure("Dtd Preparser in file %s%c%s: %s",
1651 dirname, G_DIR_SEPARATOR, filename, errors->str);
1652 continue;
1653 }
1654
1655 dtd_data = dtd_parse(preparsed);
1656
1657 g_string_free(preparsed, true);
1658
1659 if (dtd_data->error->len) {
1660 report_failure("Dtd Parser in file %s%c%s: %s",
1661 dirname, G_DIR_SEPARATOR, filename, dtd_data->error->str);
1662 destroy_dtd_data(dtd_data);
1663 continue;
1664 }
1665
1666 register_dtd(dtd_data, errors);
1667
1668 if (errors->len) {
1669 report_failure("Dtd Registration in file: %s%c%s: %s",
1670 dirname, G_DIR_SEPARATOR, filename, errors->str);
1671 continue;
1672 }
1673 }
1674 }
1675 g_string_free(errors, true);
1676
1677 CLOSEDIR_OP(dir);
1678 }
1679 }
1680
1681 g_free(dirname);
1682
1683 for(i=0;i<array_length(default_media_types);i++) {
1684 if( ! wmem_map_lookup(media_types, default_media_types[i]) ) {
1685 wmem_map_insert(media_types, (void *)default_media_types[i], &xml_ns);
1686 }
1687 }
1688
1689 wmem_map_foreach(xmpli_names, add_xmlpi_namespace, (void *)"xml.xmlpi");
1690
1691 wmem_free(wmem_epan_scope(), dummy);
1692 }
1693
1694 static void
1695 xml_init_protocol(void)
1696 {
1697 // The longest encoding at https://www.iana.org/assignments/character-sets/character-sets.xml
1698 // is 45 characters (Extended_UNIX_Code_Packed_Format_for_Japanese).
1699 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);
1700 }
1701
1702 static void
1703 xml_cleanup_protocol(void) {
1704 g_regex_unref(encoding_pattern);
1705 }
1706
1707 void
1708 proto_register_xml(void)
1709 {
1710 static int *ett_base[] = {
1711 &unknown_ns.ett,
1712 &xml_ns.ett,
1713 &ett_dtd,
1714 &ett_xmpli
1715 };
1716
1717 static hf_register_info hf_base[] = {
1718 { &hf_xmlpi,
1719 {"XMLPI", "xml.xmlpi",
1720 FT_STRING, BASE_NONE, NULL, 0,
1721 NULL, HFILL }
1722 },
1723 { &hf_cdatasection,
1724 {"CDATASection", "xml.cdatasection",
1725 FT_STRING, BASE_NONE, NULL, 0,
1726 NULL, HFILL }
1727 },
1728 { &hf_comment,
1729 {"Comment", "xml.comment",
1730 FT_STRING, BASE_NONE, NULL, 0,
1731 NULL, HFILL }
1732 },
1733 { &hf_unknowwn_attrib,
1734 {"Attribute", "xml.attribute",
1735 FT_STRING, BASE_NONE, NULL, 0,
1736 NULL, HFILL }
1737 },
1738 { &hf_doctype,
1739 {"Doctype", "xml.doctype",
1740 FT_STRING, BASE_NONE, NULL, 0,
1741 NULL, HFILL }
1742 },
1743 { &hf_dtd_tag,
1744 {"DTD Tag", "xml.dtdtag",
1745 FT_STRING, BASE_NONE, NULL, 0,
1746 NULL, HFILL }
1747 },
1748 { &unknown_ns.hf_cdata,
1749 {"CDATA", "xml.cdata",
1750 FT_STRING, BASE_NONE, NULL, 0, NULL,
1751 HFILL }
1752 },
1753 { &unknown_ns.hf_tag,
1754 {"Tag", "xml.tag",
1755 FT_STRING, BASE_NONE, NULL, 0,
1756 NULL, HFILL }
1757 },
1758 { &xml_ns.hf_cdata,
1759 {"Unknown", "xml.unknown",
1760 FT_STRING, BASE_NONE, NULL, 0,
1761 NULL, HFILL }
1762 }
1763 };
1764
1765 static ei_register_info ei[] = {
1766 { &ei_xml_closing_unopened_tag, { "xml.closing_unopened_tag", PI_MALFORMED, PI_ERROR, "Closing an unopened tag", EXPFILL }},
1767 { &ei_xml_closing_unopened_xmpli_tag, { "xml.closing_unopened_xmpli_tag", PI_MALFORMED, PI_ERROR, "Closing an unopened xmpli tag", EXPFILL }},
1768 { &ei_xml_unrecognized_text, { "xml.unrecognized_text", PI_PROTOCOL, PI_WARN, "Unrecognized text", EXPFILL }},
1769 };
1770
1771 module_t *xml_module;
1772 expert_module_t* expert_xml;
1773
1774 hf_arr = wmem_array_new(wmem_epan_scope(), sizeof(hf_register_info));
1775 ett_arr = g_array_new(false, false, sizeof(int *));
1776
1777 wmem_array_append(hf_arr, hf_base, array_length(hf_base));
1778 g_array_append_vals(ett_arr, ett_base, array_length(ett_base));
1779
1780 init_xml_names();
1781
1782 xml_ns.hf_tag = proto_register_protocol("eXtensible Markup Language", "XML", xml_ns.name);
1783
1784 proto_register_field_array(xml_ns.hf_tag, (hf_register_info*)wmem_array_get_raw(hf_arr), wmem_array_get_count(hf_arr));
1785 proto_register_subtree_array((int **)ett_arr->data, ett_arr->len);
1786 expert_xml = expert_register_protocol(xml_ns.hf_tag);
1787 expert_register_field_array(expert_xml, ei, array_length(ei));
1788
1789 xml_module = prefs_register_protocol(xml_ns.hf_tag, NULL);
1790 prefs_register_obsolete_preference(xml_module, "heuristic");
1791 prefs_register_obsolete_preference(xml_module, "heuristic_tcp");
1792 prefs_register_obsolete_preference(xml_module, "heuristic_udp");
1793 /* XXX - UCS-2, or UTF-16? */
1794 prefs_register_bool_preference(xml_module, "heuristic_unicode", "Use Unicode in heuristics",
1795 "Try to recognize XML encoded in Unicode (UCS-2BE)",
1796 &pref_heuristic_unicode);
1797
1798 prefs_register_enum_preference(xml_module, "default_encoding", "Default character encoding",
1799 "Use this charset if the 'encoding' attribute of XML declaration is missing."
1800 "Unsupported encoding will be replaced by the default UTF-8.",
1801 &pref_default_encoding, ws_supported_mibenum_vals_character_sets_ev_array, false);
1802
1803 g_array_free(ett_arr, true);
1804
1805 register_init_routine(&xml_init_protocol);
1806 register_cleanup_routine(&xml_cleanup_protocol);
1807
1808 xml_handle = register_dissector("xml", dissect_xml, xml_ns.hf_tag);
1809
1810 init_xml_parser();
1811 }
1812
1813 static void
1814 add_dissector_media(void *k, void *v _U_, void *p _U_)
1815 {
1816 dissector_add_string("media_type", (char *)k, xml_handle);
1817 }
1818
1819 void
1820 proto_reg_handoff_xml(void)
1821 {
1822 wmem_map_foreach(media_types, add_dissector_media, NULL);
1823 dissector_add_string("media_type.suffix", "xml", xml_handle); /* RFC 7303 9.6 */
1824 dissector_add_uint_range_with_preference("tcp.port", "", xml_handle);
1825 dissector_add_uint_range_with_preference("udp.port", XML_UDP_PORT_RANGE, xml_handle);
1826
1827 gssapi_handle = find_dissector_add_dependency("gssapi", xml_ns.hf_tag);
1828
1829 heur_dissector_add("http", dissect_xml_heur, "XML in HTTP", "xml_http", xml_ns.hf_tag, HEURISTIC_DISABLE);
1830 heur_dissector_add("sip", dissect_xml_heur, "XML in SIP", "xml_sip", xml_ns.hf_tag, HEURISTIC_DISABLE);
1831 heur_dissector_add("media", dissect_xml_heur, "XML in media", "xml_media", xml_ns.hf_tag, HEURISTIC_DISABLE);
1832 heur_dissector_add("tcp", dissect_xml_heur, "XML over TCP", "xml_tcp", xml_ns.hf_tag, HEURISTIC_DISABLE);
1833 heur_dissector_add("udp", dissect_xml_heur, "XML over UDP", "xml_udp", xml_ns.hf_tag, HEURISTIC_DISABLE);
1834
1835 heur_dissector_add("wtap_file", dissect_xml_heur, "XML file", "xml_wtap", xml_ns.hf_tag, HEURISTIC_ENABLE);
1836
1837 dissector_add_uint("acdr.tls_application", TLS_APP_XML, xml_handle);
1838 }
1839
1840 /*
1841 * Editor modelines - https://www.wireshark.org/tools/modelines.html
1842 *
1843 * Local variables:
1844 * c-basic-offset: 4
1845 * tab-width: 8
1846 * indent-tabs-mode: nil
1847 * End:
1848 *
1849 * vi: set shiftwidth=4 tabstop=8 expandtab:
1850 * :indentSize=4:tabSize=8:noTabs=true:
1851 */
Metzes wireshark repo