1 /*
2 * "Canonical XML" implementation
3 * http://www.w3.org/TR/xml-c14n
4 *
5 * "Exclusive XML Canonicalization" implementation
6 * http://www.w3.org/TR/xml-exc-c14n
7 *
8 * See Copyright for the status of this software.
9 *
10 * Author: Aleksey Sanin <aleksey@aleksey.com>
11 */
12 #define IN_LIBXML
13 #include "libxml.h"
14 #ifdef LIBXML_C14N_ENABLED
15 #ifdef LIBXML_OUTPUT_ENABLED
16
17 #ifdef HAVE_STDLIB_H
18 #include <stdlib.h>
19 #endif
20 #include <string.h>
21
22 #include <libxml/tree.h>
23 #include <libxml/parser.h>
24 #include <libxml/uri.h>
25 #include <libxml/xmlerror.h>
26 #include <libxml/globals.h>
27 #include <libxml/xpathInternals.h>
28 #include <libxml/c14n.h>
29
30 /************************************************************************
31 * *
32 * Some declaration better left private ATM *
33 * *
34 ************************************************************************/
35
36 typedef enum {
37 XMLC14N_BEFORE_DOCUMENT_ELEMENT = 0,
38 XMLC14N_INSIDE_DOCUMENT_ELEMENT = 1,
39 XMLC14N_AFTER_DOCUMENT_ELEMENT = 2
40 } xmlC14NPosition;
41
42 typedef struct _xmlC14NVisibleNsStack {
43 int nsCurEnd; /* number of nodes in the set */
44 int nsPrevStart; /* the begginning of the stack for previous visible node */
45 int nsPrevEnd; /* the end of the stack for previous visible node */
46 int nsMax; /* size of the array as allocated */
47 xmlNsPtr *nsTab; /* array of ns in no particular order */
48 xmlNodePtr *nodeTab; /* array of nodes in no particular order */
49 } xmlC14NVisibleNsStack, *xmlC14NVisibleNsStackPtr;
50
51 typedef struct _xmlC14NCtx {
52 /* input parameters */
53 xmlDocPtr doc;
54 xmlC14NIsVisibleCallback is_visible_callback;
55 void* user_data;
56 int with_comments;
57 xmlOutputBufferPtr buf;
58
59 /* position in the XML document */
60 xmlC14NPosition pos;
61 int parent_is_doc;
62 xmlC14NVisibleNsStackPtr ns_rendered;
63
64 /* C14N mode */
65 xmlC14NMode mode;
66
67 /* exclusive canonicalization */
68 xmlChar **inclusive_ns_prefixes;
69
70 /* error number */
71 int error;
72 } xmlC14NCtx, *xmlC14NCtxPtr;
73
74 static xmlC14NVisibleNsStackPtr xmlC14NVisibleNsStackCreate (void);
75 static void xmlC14NVisibleNsStackDestroy (xmlC14NVisibleNsStackPtr cur);
76 static void xmlC14NVisibleNsStackAdd (xmlC14NVisibleNsStackPtr cur,
77 xmlNsPtr ns,
78 xmlNodePtr node);
79 static void xmlC14NVisibleNsStackSave (xmlC14NVisibleNsStackPtr cur,
80 xmlC14NVisibleNsStackPtr state);
81 static void xmlC14NVisibleNsStackRestore (xmlC14NVisibleNsStackPtr cur,
82 xmlC14NVisibleNsStackPtr state);
83 static void xmlC14NVisibleNsStackShift (xmlC14NVisibleNsStackPtr cur);
84 static int xmlC14NVisibleNsStackFind (xmlC14NVisibleNsStackPtr cur,
85 xmlNsPtr ns);
86 static int xmlExcC14NVisibleNsStackFind (xmlC14NVisibleNsStackPtr cur,
87 xmlNsPtr ns,
88 xmlC14NCtxPtr ctx);
89
90 static int xmlC14NIsNodeInNodeset (xmlNodeSetPtr nodes,
91 xmlNodePtr node,
92 xmlNodePtr parent);
93
94
95
96 static int xmlC14NProcessNode(xmlC14NCtxPtr ctx, xmlNodePtr cur);
97 static int xmlC14NProcessNodeList(xmlC14NCtxPtr ctx, xmlNodePtr cur);
98 typedef enum {
99 XMLC14N_NORMALIZE_ATTR = 0,
100 XMLC14N_NORMALIZE_COMMENT = 1,
101 XMLC14N_NORMALIZE_PI = 2,
102 XMLC14N_NORMALIZE_TEXT = 3
103 } xmlC14NNormalizationMode;
104
105 static xmlChar *xmlC11NNormalizeString(const xmlChar * input,
106 xmlC14NNormalizationMode mode);
107
108 #define xmlC11NNormalizeAttr( a ) \
109 xmlC11NNormalizeString((a), XMLC14N_NORMALIZE_ATTR)
110 #define xmlC11NNormalizeComment( a ) \
111 xmlC11NNormalizeString((a), XMLC14N_NORMALIZE_COMMENT)
112 #define xmlC11NNormalizePI( a ) \
113 xmlC11NNormalizeString((a), XMLC14N_NORMALIZE_PI)
114 #define xmlC11NNormalizeText( a ) \
115 xmlC11NNormalizeString((a), XMLC14N_NORMALIZE_TEXT)
116
117 #define xmlC14NIsVisible( ctx, node, parent ) \
118 (((ctx)->is_visible_callback != NULL) ? \
119 (ctx)->is_visible_callback((ctx)->user_data, \
120 (xmlNodePtr)(node), (xmlNodePtr)(parent)) : 1)
121
122 #define xmlC14NIsExclusive( ctx ) \
123 ( (ctx)->mode == XML_C14N_EXCLUSIVE_1_0 )
124
125 /************************************************************************
126 * *
127 * Some factorized error routines *
128 * *
129 ************************************************************************/
130
131 /**
132 * xmlC14NErrMemory:
133 * @extra: extra informations
134 *
135 * Handle a redefinition of memory error
136 */
137 static void
xmlC14NErrMemory(const char * extra)138 xmlC14NErrMemory(const char *extra)
139 {
140 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
141 XML_ERR_NO_MEMORY, XML_ERR_ERROR, NULL, 0, extra,
142 NULL, NULL, 0, 0,
143 "Memory allocation failed : %s\n", extra);
144 }
145
146 /**
147 * xmlC14NErrParam:
148 * @extra: extra informations
149 *
150 * Handle a redefinition of param error
151 */
152 static void
xmlC14NErrParam(const char * extra)153 xmlC14NErrParam(const char *extra)
154 {
155 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
156 XML_ERR_INTERNAL_ERROR, XML_ERR_ERROR, NULL, 0, extra,
157 NULL, NULL, 0, 0,
158 "Invalid parameter : %s\n", extra);
159 }
160
161 /**
162 * xmlC14NErrInternal:
163 * @extra: extra informations
164 *
165 * Handle a redefinition of internal error
166 */
167 static void
xmlC14NErrInternal(const char * extra)168 xmlC14NErrInternal(const char *extra)
169 {
170 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
171 XML_ERR_INTERNAL_ERROR, XML_ERR_ERROR, NULL, 0, extra,
172 NULL, NULL, 0, 0,
173 "Internal error : %s\n", extra);
174 }
175
176 /**
177 * xmlC14NErrInvalidNode:
178 * @extra: extra informations
179 *
180 * Handle a redefinition of invalid node error
181 */
182 static void
xmlC14NErrInvalidNode(const char * node_type,const char * extra)183 xmlC14NErrInvalidNode(const char *node_type, const char *extra)
184 {
185 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
186 XML_C14N_INVALID_NODE, XML_ERR_ERROR, NULL, 0, extra,
187 NULL, NULL, 0, 0,
188 "Node %s is invalid here : %s\n", node_type, extra);
189 }
190
191 /**
192 * xmlC14NErrUnknownNode:
193 * @extra: extra informations
194 *
195 * Handle a redefinition of unknown node error
196 */
197 static void
xmlC14NErrUnknownNode(int node_type,const char * extra)198 xmlC14NErrUnknownNode(int node_type, const char *extra)
199 {
200 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
201 XML_C14N_UNKNOW_NODE, XML_ERR_ERROR, NULL, 0, extra,
202 NULL, NULL, 0, 0,
203 "Unknown node type %d found : %s\n", node_type, extra);
204 }
205
206 /**
207 * xmlC14NErrRelativeNamespace:
208 * @extra: extra informations
209 *
210 * Handle a redefinition of relative namespace error
211 */
212 static void
xmlC14NErrRelativeNamespace(const char * ns_uri)213 xmlC14NErrRelativeNamespace(const char *ns_uri)
214 {
215 __xmlRaiseError(NULL, NULL, NULL, NULL, NULL, XML_FROM_C14N,
216 XML_C14N_RELATIVE_NAMESPACE, XML_ERR_ERROR, NULL, 0, NULL,
217 NULL, NULL, 0, 0,
218 "Relative namespace UR is invalid here : %s\n", ns_uri);
219 }
220
221
222
223 /**
224 * xmlC14NErr:
225 * @ctxt: a C14N evaluation context
226 * @node: the context node
227 * @error: the erorr code
228 * @msg: the message
229 * @extra: extra informations
230 *
231 * Handle a redefinition of attribute error
232 */
233 static void
xmlC14NErr(xmlC14NCtxPtr ctxt,xmlNodePtr node,int error,const char * msg)234 xmlC14NErr(xmlC14NCtxPtr ctxt, xmlNodePtr node, int error,
235 const char * msg)
236 {
237 if (ctxt != NULL)
238 ctxt->error = error;
239 __xmlRaiseError(NULL, NULL, NULL,
240 ctxt, node, XML_FROM_C14N, error,
241 XML_ERR_ERROR, NULL, 0,
242 NULL, NULL, NULL, 0, 0, "%s", msg);
243 }
244
245 /************************************************************************
246 * *
247 * The implementation internals *
248 * *
249 ************************************************************************/
250 #define XML_NAMESPACES_DEFAULT 16
251
252 static int
xmlC14NIsNodeInNodeset(xmlNodeSetPtr nodes,xmlNodePtr node,xmlNodePtr parent)253 xmlC14NIsNodeInNodeset(xmlNodeSetPtr nodes, xmlNodePtr node, xmlNodePtr parent) {
254 if((nodes != NULL) && (node != NULL)) {
255 if(node->type != XML_NAMESPACE_DECL) {
256 return(xmlXPathNodeSetContains(nodes, node));
257 } else {
258 xmlNs ns;
259
260 memcpy(&ns, node, sizeof(ns));
261
262 /* this is a libxml hack! check xpath.c for details */
263 if((parent != NULL) && (parent->type == XML_ATTRIBUTE_NODE)) {
264 ns.next = (xmlNsPtr)parent->parent;
265 } else {
266 ns.next = (xmlNsPtr)parent;
267 }
268
269 /*
270 * If the input is an XPath node-set, then the node-set must explicitly
271 * contain every node to be rendered to the canonical form.
272 */
273 return(xmlXPathNodeSetContains(nodes, (xmlNodePtr)&ns));
274 }
275 }
276 return(1);
277 }
278
279 static xmlC14NVisibleNsStackPtr
xmlC14NVisibleNsStackCreate(void)280 xmlC14NVisibleNsStackCreate(void) {
281 xmlC14NVisibleNsStackPtr ret;
282
283 ret = (xmlC14NVisibleNsStackPtr) xmlMalloc(sizeof(xmlC14NVisibleNsStack));
284 if (ret == NULL) {
285 xmlC14NErrMemory("creating namespaces stack");
286 return(NULL);
287 }
288 memset(ret, 0 , (size_t) sizeof(xmlC14NVisibleNsStack));
289 return(ret);
290 }
291
292 static void
xmlC14NVisibleNsStackDestroy(xmlC14NVisibleNsStackPtr cur)293 xmlC14NVisibleNsStackDestroy(xmlC14NVisibleNsStackPtr cur) {
294 if(cur == NULL) {
295 xmlC14NErrParam("destroying namespaces stack");
296 return;
297 }
298 if(cur->nsTab != NULL) {
299 memset(cur->nsTab, 0, cur->nsMax * sizeof(xmlNsPtr));
300 xmlFree(cur->nsTab);
301 }
302 if(cur->nodeTab != NULL) {
303 memset(cur->nodeTab, 0, cur->nsMax * sizeof(xmlNodePtr));
304 xmlFree(cur->nodeTab);
305 }
306 memset(cur, 0, sizeof(xmlC14NVisibleNsStack));
307 xmlFree(cur);
308
309 }
310
311 static void
xmlC14NVisibleNsStackAdd(xmlC14NVisibleNsStackPtr cur,xmlNsPtr ns,xmlNodePtr node)312 xmlC14NVisibleNsStackAdd(xmlC14NVisibleNsStackPtr cur, xmlNsPtr ns, xmlNodePtr node) {
313 if((cur == NULL) ||
314 ((cur->nsTab == NULL) && (cur->nodeTab != NULL)) ||
315 ((cur->nsTab != NULL) && (cur->nodeTab == NULL))) {
316 xmlC14NErrParam("adding namespace to stack");
317 return;
318 }
319
320 if ((cur->nsTab == NULL) && (cur->nodeTab == NULL)) {
321 cur->nsTab = (xmlNsPtr*) xmlMalloc(XML_NAMESPACES_DEFAULT * sizeof(xmlNsPtr));
322 cur->nodeTab = (xmlNodePtr*) xmlMalloc(XML_NAMESPACES_DEFAULT * sizeof(xmlNodePtr));
323 if ((cur->nsTab == NULL) || (cur->nodeTab == NULL)) {
324 xmlC14NErrMemory("adding node to stack");
325 return;
326 }
327 memset(cur->nsTab, 0 , XML_NAMESPACES_DEFAULT * sizeof(xmlNsPtr));
328 memset(cur->nodeTab, 0 , XML_NAMESPACES_DEFAULT * sizeof(xmlNodePtr));
329 cur->nsMax = XML_NAMESPACES_DEFAULT;
330 } else if(cur->nsMax == cur->nsCurEnd) {
331 void *tmp;
332 int tmpSize;
333
334 tmpSize = 2 * cur->nsMax;
335 tmp = xmlRealloc(cur->nsTab, tmpSize * sizeof(xmlNsPtr));
336 if (tmp == NULL) {
337 xmlC14NErrMemory("adding node to stack");
338 return;
339 }
340 cur->nsTab = (xmlNsPtr*)tmp;
341
342 tmp = xmlRealloc(cur->nodeTab, tmpSize * sizeof(xmlNodePtr));
343 if (tmp == NULL) {
344 xmlC14NErrMemory("adding node to stack");
345 return;
346 }
347 cur->nodeTab = (xmlNodePtr*)tmp;
348
349 cur->nsMax = tmpSize;
350 }
351 cur->nsTab[cur->nsCurEnd] = ns;
352 cur->nodeTab[cur->nsCurEnd] = node;
353
354 ++cur->nsCurEnd;
355 }
356
357 static void
xmlC14NVisibleNsStackSave(xmlC14NVisibleNsStackPtr cur,xmlC14NVisibleNsStackPtr state)358 xmlC14NVisibleNsStackSave(xmlC14NVisibleNsStackPtr cur, xmlC14NVisibleNsStackPtr state) {
359 if((cur == NULL) || (state == NULL)) {
360 xmlC14NErrParam("saving namespaces stack");
361 return;
362 }
363
364 state->nsCurEnd = cur->nsCurEnd;
365 state->nsPrevStart = cur->nsPrevStart;
366 state->nsPrevEnd = cur->nsPrevEnd;
367 }
368
369 static void
xmlC14NVisibleNsStackRestore(xmlC14NVisibleNsStackPtr cur,xmlC14NVisibleNsStackPtr state)370 xmlC14NVisibleNsStackRestore(xmlC14NVisibleNsStackPtr cur, xmlC14NVisibleNsStackPtr state) {
371 if((cur == NULL) || (state == NULL)) {
372 xmlC14NErrParam("restoring namespaces stack");
373 return;
374 }
375 cur->nsCurEnd = state->nsCurEnd;
376 cur->nsPrevStart = state->nsPrevStart;
377 cur->nsPrevEnd = state->nsPrevEnd;
378 }
379
380 static void
xmlC14NVisibleNsStackShift(xmlC14NVisibleNsStackPtr cur)381 xmlC14NVisibleNsStackShift(xmlC14NVisibleNsStackPtr cur) {
382 if(cur == NULL) {
383 xmlC14NErrParam("shifting namespaces stack");
384 return;
385 }
386 cur->nsPrevStart = cur->nsPrevEnd;
387 cur->nsPrevEnd = cur->nsCurEnd;
388 }
389
390 static int
xmlC14NStrEqual(const xmlChar * str1,const xmlChar * str2)391 xmlC14NStrEqual(const xmlChar *str1, const xmlChar *str2) {
392 if (str1 == str2) return(1);
393 if (str1 == NULL) return((*str2) == '\0');
394 if (str2 == NULL) return((*str1) == '\0');
395 do {
396 if (*str1++ != *str2) return(0);
397 } while (*str2++);
398 return(1);
399 }
400
401 /**
402 * xmlC14NVisibleNsStackFind:
403 * @ctx: the C14N context
404 * @ns: the namespace to check
405 *
406 * Checks whether the given namespace was already rendered or not
407 *
408 * Returns 1 if we already wrote this namespace or 0 otherwise
409 */
410 static int
xmlC14NVisibleNsStackFind(xmlC14NVisibleNsStackPtr cur,xmlNsPtr ns)411 xmlC14NVisibleNsStackFind(xmlC14NVisibleNsStackPtr cur, xmlNsPtr ns)
412 {
413 int i;
414 const xmlChar *prefix;
415 const xmlChar *href;
416 int has_empty_ns;
417
418 if(cur == NULL) {
419 xmlC14NErrParam("searching namespaces stack (c14n)");
420 return (0);
421 }
422
423 /*
424 * if the default namespace xmlns="" is not defined yet then
425 * we do not want to print it out
426 */
427 prefix = ((ns == NULL) || (ns->prefix == NULL)) ? BAD_CAST "" : ns->prefix;
428 href = ((ns == NULL) || (ns->href == NULL)) ? BAD_CAST "" : ns->href;
429 has_empty_ns = (xmlC14NStrEqual(prefix, NULL) && xmlC14NStrEqual(href, NULL));
430
431 if (cur->nsTab != NULL) {
432 int start = (has_empty_ns) ? 0 : cur->nsPrevStart;
433 for (i = cur->nsCurEnd - 1; i >= start; --i) {
434 xmlNsPtr ns1 = cur->nsTab[i];
435
436 if(xmlC14NStrEqual(prefix, (ns1 != NULL) ? ns1->prefix : NULL)) {
437 return(xmlC14NStrEqual(href, (ns1 != NULL) ? ns1->href : NULL));
438 }
439 }
440 }
441 return(has_empty_ns);
442 }
443
444 static int
xmlExcC14NVisibleNsStackFind(xmlC14NVisibleNsStackPtr cur,xmlNsPtr ns,xmlC14NCtxPtr ctx)445 xmlExcC14NVisibleNsStackFind(xmlC14NVisibleNsStackPtr cur, xmlNsPtr ns, xmlC14NCtxPtr ctx) {
446 int i;
447 const xmlChar *prefix;
448 const xmlChar *href;
449 int has_empty_ns;
450
451 if(cur == NULL) {
452 xmlC14NErrParam("searching namespaces stack (exc c14n)");
453 return (0);
454 }
455
456 /*
457 * if the default namespace xmlns="" is not defined yet then
458 * we do not want to print it out
459 */
460 prefix = ((ns == NULL) || (ns->prefix == NULL)) ? BAD_CAST "" : ns->prefix;
461 href = ((ns == NULL) || (ns->href == NULL)) ? BAD_CAST "" : ns->href;
462 has_empty_ns = (xmlC14NStrEqual(prefix, NULL) && xmlC14NStrEqual(href, NULL));
463
464 if (cur->nsTab != NULL) {
465 int start = 0;
466 for (i = cur->nsCurEnd - 1; i >= start; --i) {
467 xmlNsPtr ns1 = cur->nsTab[i];
468
469 if(xmlC14NStrEqual(prefix, (ns1 != NULL) ? ns1->prefix : NULL)) {
470 if(xmlC14NStrEqual(href, (ns1 != NULL) ? ns1->href : NULL)) {
471 return(xmlC14NIsVisible(ctx, ns1, cur->nodeTab[i]));
472 } else {
473 return(0);
474 }
475 }
476 }
477 }
478 return(has_empty_ns);
479 }
480
481
482
483
484 /**
485 * xmlC14NIsXmlNs:
486 * @ns: the namespace to check
487 *
488 * Checks whether the given namespace is a default "xml:" namespace
489 * with href="http://www.w3.org/XML/1998/namespace"
490 *
491 * Returns 1 if the node is default or 0 otherwise
492 */
493
494 /* todo: make it a define? */
495 static int
xmlC14NIsXmlNs(xmlNsPtr ns)496 xmlC14NIsXmlNs(xmlNsPtr ns)
497 {
498 return ((ns != NULL) &&
499 (xmlStrEqual(ns->prefix, BAD_CAST "xml")) &&
500 (xmlStrEqual(ns->href, XML_XML_NAMESPACE)));
501 }
502
503
504 /**
505 * xmlC14NNsCompare:
506 * @ns1: the pointer to first namespace
507 * @ns2: the pointer to second namespace
508 *
509 * Compares the namespaces by names (prefixes).
510 *
511 * Returns -1 if ns1 < ns2, 0 if ns1 == ns2 or 1 if ns1 > ns2.
512 */
513 static int
xmlC14NNsCompare(xmlNsPtr ns1,xmlNsPtr ns2)514 xmlC14NNsCompare(xmlNsPtr ns1, xmlNsPtr ns2)
515 {
516 if (ns1 == ns2)
517 return (0);
518 if (ns1 == NULL)
519 return (-1);
520 if (ns2 == NULL)
521 return (1);
522
523 return (xmlStrcmp(ns1->prefix, ns2->prefix));
524 }
525
526
527 /**
528 * xmlC14NPrintNamespaces:
529 * @ns: the pointer to namespace
530 * @ctx: the C14N context
531 *
532 * Prints the given namespace to the output buffer from C14N context.
533 *
534 * Returns 1 on success or 0 on fail.
535 */
536 static int
xmlC14NPrintNamespaces(const xmlNsPtr ns,xmlC14NCtxPtr ctx)537 xmlC14NPrintNamespaces(const xmlNsPtr ns, xmlC14NCtxPtr ctx)
538 {
539
540 if ((ns == NULL) || (ctx == NULL)) {
541 xmlC14NErrParam("writing namespaces");
542 return 0;
543 }
544
545 if (ns->prefix != NULL) {
546 xmlOutputBufferWriteString(ctx->buf, " xmlns:");
547 xmlOutputBufferWriteString(ctx->buf, (const char *) ns->prefix);
548 xmlOutputBufferWriteString(ctx->buf, "=\"");
549 } else {
550 xmlOutputBufferWriteString(ctx->buf, " xmlns=\"");
551 }
552 if(ns->href != NULL) {
553 xmlOutputBufferWriteString(ctx->buf, (const char *) ns->href);
554 }
555 xmlOutputBufferWriteString(ctx->buf, "\"");
556 return (1);
557 }
558
559 /**
560 * xmlC14NProcessNamespacesAxis:
561 * @ctx: the C14N context
562 * @node: the current node
563 *
564 * Prints out canonical namespace axis of the current node to the
565 * buffer from C14N context as follows
566 *
567 * Canonical XML v 1.0 (http://www.w3.org/TR/xml-c14n)
568 *
569 * Namespace Axis
570 * Consider a list L containing only namespace nodes in the
571 * axis and in the node-set in lexicographic order (ascending). To begin
572 * processing L, if the first node is not the default namespace node (a node
573 * with no namespace URI and no local name), then generate a space followed
574 * by xmlns="" if and only if the following conditions are met:
575 * - the element E that owns the axis is in the node-set
576 * - The nearest ancestor element of E in the node-set has a default
577 * namespace node in the node-set (default namespace nodes always
578 * have non-empty values in XPath)
579 * The latter condition eliminates unnecessary occurrences of xmlns="" in
580 * the canonical form since an element only receives an xmlns="" if its
581 * default namespace is empty and if it has an immediate parent in the
582 * canonical form that has a non-empty default namespace. To finish
583 * processing L, simply process every namespace node in L, except omit
584 * namespace node with local name xml, which defines the xml prefix,
585 * if its string value is http://www.w3.org/XML/1998/namespace.
586 *
587 * Exclusive XML Canonicalization v 1.0 (http://www.w3.org/TR/xml-exc-c14n)
588 * Canonical XML applied to a document subset requires the search of the
589 * ancestor nodes of each orphan element node for attributes in the xml
590 * namespace, such as xml:lang and xml:space. These are copied into the
591 * element node except if a declaration of the same attribute is already
592 * in the attribute axis of the element (whether or not it is included in
593 * the document subset). This search and copying are omitted from the
594 * Exclusive XML Canonicalization method.
595 *
596 * Returns 0 on success or -1 on fail.
597 */
598 static int
xmlC14NProcessNamespacesAxis(xmlC14NCtxPtr ctx,xmlNodePtr cur,int visible)599 xmlC14NProcessNamespacesAxis(xmlC14NCtxPtr ctx, xmlNodePtr cur, int visible)
600 {
601 xmlNodePtr n;
602 xmlNsPtr ns, tmp;
603 xmlListPtr list;
604 int already_rendered;
605 int has_empty_ns = 0;
606
607 if ((ctx == NULL) || (cur == NULL) || (cur->type != XML_ELEMENT_NODE)) {
608 xmlC14NErrParam("processing namespaces axis (c14n)");
609 return (-1);
610 }
611
612 /*
613 * Create a sorted list to store element namespaces
614 */
615 list = xmlListCreate(NULL, (xmlListDataCompare) xmlC14NNsCompare);
616 if (list == NULL) {
617 xmlC14NErrInternal("creating namespaces list (c14n)");
618 return (-1);
619 }
620
621 /* check all namespaces */
622 for(n = cur; n != NULL; n = n->parent) {
623 for(ns = n->nsDef; ns != NULL; ns = ns->next) {
624 tmp = xmlSearchNs(cur->doc, cur, ns->prefix);
625
626 if((tmp == ns) && !xmlC14NIsXmlNs(ns) && xmlC14NIsVisible(ctx, ns, cur)) {
627 already_rendered = xmlC14NVisibleNsStackFind(ctx->ns_rendered, ns);
628 if(visible) {
629 xmlC14NVisibleNsStackAdd(ctx->ns_rendered, ns, cur);
630 }
631 if(!already_rendered) {
632 xmlListInsert(list, ns);
633 }
634 if(xmlStrlen(ns->prefix) == 0) {
635 has_empty_ns = 1;
636 }
637 }
638 }
639 }
640
641 /**
642 * if the first node is not the default namespace node (a node with no
643 * namespace URI and no local name), then generate a space followed by
644 * xmlns="" if and only if the following conditions are met:
645 * - the element E that owns the axis is in the node-set
646 * - the nearest ancestor element of E in the node-set has a default
647 * namespace node in the node-set (default namespace nodes always
648 * have non-empty values in XPath)
649 */
650 if(visible && !has_empty_ns) {
651 static xmlNs ns_default;
652
653 memset(&ns_default, 0, sizeof(ns_default));
654 if(!xmlC14NVisibleNsStackFind(ctx->ns_rendered, &ns_default)) {
655 xmlC14NPrintNamespaces(&ns_default, ctx);
656 }
657 }
658
659
660 /*
661 * print out all elements from list
662 */
663 xmlListWalk(list, (xmlListWalker) xmlC14NPrintNamespaces, (const void *) ctx);
664
665 /*
666 * Cleanup
667 */
668 xmlListDelete(list);
669 return (0);
670 }
671
672
673 /**
674 * xmlExcC14NProcessNamespacesAxis:
675 * @ctx: the C14N context
676 * @node: the current node
677 *
678 * Prints out exclusive canonical namespace axis of the current node to the
679 * buffer from C14N context as follows
680 *
681 * Exclusive XML Canonicalization
682 * http://www.w3.org/TR/xml-exc-c14n
683 *
684 * If the element node is in the XPath subset then output the node in
685 * accordance with Canonical XML except for namespace nodes which are
686 * rendered as follows:
687 *
688 * 1. Render each namespace node iff:
689 * * it is visibly utilized by the immediate parent element or one of
690 * its attributes, or is present in InclusiveNamespaces PrefixList, and
691 * * its prefix and value do not appear in ns_rendered. ns_rendered is
692 * obtained by popping the state stack in order to obtain a list of
693 * prefixes and their values which have already been rendered by
694 * an output ancestor of the namespace node's parent element.
695 * 2. Append the rendered namespace node to the list ns_rendered of namespace
696 * nodes rendered by output ancestors. Push ns_rendered on state stack and
697 * recurse.
698 * 3. After the recursion returns, pop thestate stack.
699 *
700 *
701 * Returns 0 on success or -1 on fail.
702 */
703 static int
xmlExcC14NProcessNamespacesAxis(xmlC14NCtxPtr ctx,xmlNodePtr cur,int visible)704 xmlExcC14NProcessNamespacesAxis(xmlC14NCtxPtr ctx, xmlNodePtr cur, int visible)
705 {
706 xmlNsPtr ns;
707 xmlListPtr list;
708 xmlAttrPtr attr;
709 int already_rendered;
710 int has_empty_ns = 0;
711 int has_visibly_utilized_empty_ns = 0;
712 int has_empty_ns_in_inclusive_list = 0;
713
714 if ((ctx == NULL) || (cur == NULL) || (cur->type != XML_ELEMENT_NODE)) {
715 xmlC14NErrParam("processing namespaces axis (exc c14n)");
716 return (-1);
717 }
718
719 if(!xmlC14NIsExclusive(ctx)) {
720 xmlC14NErrParam("processing namespaces axis (exc c14n)");
721 return (-1);
722
723 }
724
725 /*
726 * Create a sorted list to store element namespaces
727 */
728 list = xmlListCreate(NULL, (xmlListDataCompare) xmlC14NNsCompare);
729 if (list == NULL) {
730 xmlC14NErrInternal("creating namespaces list (exc c14n)");
731 return (-1);
732 }
733
734 /*
735 * process inclusive namespaces:
736 * All namespace nodes appearing on inclusive ns list are
737 * handled as provided in Canonical XML
738 */
739 if(ctx->inclusive_ns_prefixes != NULL) {
740 xmlChar *prefix;
741 int i;
742
743 for (i = 0; ctx->inclusive_ns_prefixes[i] != NULL; ++i) {
744 prefix = ctx->inclusive_ns_prefixes[i];
745 /*
746 * Special values for namespace with empty prefix
747 */
748 if (xmlStrEqual(prefix, BAD_CAST "#default")
749 || xmlStrEqual(prefix, BAD_CAST "")) {
750 prefix = NULL;
751 has_empty_ns_in_inclusive_list = 1;
752 }
753
754 ns = xmlSearchNs(cur->doc, cur, prefix);
755 if((ns != NULL) && !xmlC14NIsXmlNs(ns) && xmlC14NIsVisible(ctx, ns, cur)) {
756 already_rendered = xmlC14NVisibleNsStackFind(ctx->ns_rendered, ns);
757 if(visible) {
758 xmlC14NVisibleNsStackAdd(ctx->ns_rendered, ns, cur);
759 }
760 if(!already_rendered) {
761 xmlListInsert(list, ns);
762 }
763 if(xmlStrlen(ns->prefix) == 0) {
764 has_empty_ns = 1;
765 }
766 }
767 }
768 }
769
770 /* add node namespace */
771 if(cur->ns != NULL) {
772 ns = cur->ns;
773 } else {
774 ns = xmlSearchNs(cur->doc, cur, NULL);
775 has_visibly_utilized_empty_ns = 1;
776 }
777 if((ns != NULL) && !xmlC14NIsXmlNs(ns)) {
778 if(visible && xmlC14NIsVisible(ctx, ns, cur)) {
779 if(!xmlExcC14NVisibleNsStackFind(ctx->ns_rendered, ns, ctx)) {
780 xmlListInsert(list, ns);
781 }
782 }
783 if(visible) {
784 xmlC14NVisibleNsStackAdd(ctx->ns_rendered, ns, cur);
785 }
786 if(xmlStrlen(ns->prefix) == 0) {
787 has_empty_ns = 1;
788 }
789 }
790
791
792 /* add attributes */
793 for(attr = cur->properties; attr != NULL; attr = attr->next) {
794 /*
795 * we need to check that attribute is visible and has non
796 * default namespace (XML Namespaces: "default namespaces
797 * do not apply directly to attributes")
798 */
799 if((attr->ns != NULL) && !xmlC14NIsXmlNs(attr->ns) && xmlC14NIsVisible(ctx, attr, cur)) {
800 already_rendered = xmlExcC14NVisibleNsStackFind(ctx->ns_rendered, attr->ns, ctx);
801 xmlC14NVisibleNsStackAdd(ctx->ns_rendered, attr->ns, cur);
802 if(!already_rendered && visible) {
803 xmlListInsert(list, attr->ns);
804 }
805 if(xmlStrlen(attr->ns->prefix) == 0) {
806 has_empty_ns = 1;
807 }
808 } else if((attr->ns != NULL) && (xmlStrlen(attr->ns->prefix) == 0) && (xmlStrlen(attr->ns->href) == 0)) {
809 has_visibly_utilized_empty_ns = 1;
810 }
811 }
812
813 /*
814 * Process xmlns=""
815 */
816 if(visible && has_visibly_utilized_empty_ns &&
817 !has_empty_ns && !has_empty_ns_in_inclusive_list) {
818 static xmlNs ns_default;
819
820 memset(&ns_default, 0, sizeof(ns_default));
821
822 already_rendered = xmlExcC14NVisibleNsStackFind(ctx->ns_rendered, &ns_default, ctx);
823 if(!already_rendered) {
824 xmlC14NPrintNamespaces(&ns_default, ctx);
825 }
826 } else if(visible && !has_empty_ns && has_empty_ns_in_inclusive_list) {
827 static xmlNs ns_default;
828
829 memset(&ns_default, 0, sizeof(ns_default));
830 if(!xmlC14NVisibleNsStackFind(ctx->ns_rendered, &ns_default)) {
831 xmlC14NPrintNamespaces(&ns_default, ctx);
832 }
833 }
834
835
836
837 /*
838 * print out all elements from list
839 */
840 xmlListWalk(list, (xmlListWalker) xmlC14NPrintNamespaces, (const void *) ctx);
841
842 /*
843 * Cleanup
844 */
845 xmlListDelete(list);
846 return (0);
847 }
848
849
850 /**
851 * xmlC14NIsXmlAttr:
852 * @attr: the attr to check
853 *
854 * Checks whether the given attribute is a default "xml:" namespace
855 * with href="http://www.w3.org/XML/1998/namespace"
856 *
857 * Returns 1 if the node is default or 0 otherwise
858 */
859
860 /* todo: make it a define? */
861 static int
xmlC14NIsXmlAttr(xmlAttrPtr attr)862 xmlC14NIsXmlAttr(xmlAttrPtr attr)
863 {
864 return ((attr->ns != NULL) &&
865 (xmlC14NIsXmlNs(attr->ns) != 0));
866 }
867
868
869 /**
870 * xmlC14NAttrsCompare:
871 * @attr1: the pointer tls o first attr
872 * @attr2: the pointer to second attr
873 *
874 * Prints the given attribute to the output buffer from C14N context.
875 *
876 * Returns -1 if attr1 < attr2, 0 if attr1 == attr2 or 1 if attr1 > attr2.
877 */
878 static int
xmlC14NAttrsCompare(xmlAttrPtr attr1,xmlAttrPtr attr2)879 xmlC14NAttrsCompare(xmlAttrPtr attr1, xmlAttrPtr attr2)
880 {
881 int ret = 0;
882
883 /*
884 * Simple cases
885 */
886 if (attr1 == attr2)
887 return (0);
888 if (attr1 == NULL)
889 return (-1);
890 if (attr2 == NULL)
891 return (1);
892 if (attr1->ns == attr2->ns) {
893 return (xmlStrcmp(attr1->name, attr2->name));
894 }
895
896 /*
897 * Attributes in the default namespace are first
898 * because the default namespace is not applied to
899 * unqualified attributes
900 */
901 if (attr1->ns == NULL)
902 return (-1);
903 if (attr2->ns == NULL)
904 return (1);
905 if (attr1->ns->prefix == NULL)
906 return (-1);
907 if (attr2->ns->prefix == NULL)
908 return (1);
909
910 ret = xmlStrcmp(attr1->ns->href, attr2->ns->href);
911 if (ret == 0) {
912 ret = xmlStrcmp(attr1->name, attr2->name);
913 }
914 return (ret);
915 }
916
917
918 /**
919 * xmlC14NPrintAttrs:
920 * @attr: the pointer to attr
921 * @ctx: the C14N context
922 *
923 * Prints out canonical attribute urrent node to the
924 * buffer from C14N context as follows
925 *
926 * Canonical XML v 1.0 (http://www.w3.org/TR/xml-c14n)
927 *
928 * Returns 1 on success or 0 on fail.
929 */
930 static int
xmlC14NPrintAttrs(const xmlAttrPtr attr,xmlC14NCtxPtr ctx)931 xmlC14NPrintAttrs(const xmlAttrPtr attr, xmlC14NCtxPtr ctx)
932 {
933 xmlChar *value;
934 xmlChar *buffer;
935
936 if ((attr == NULL) || (ctx == NULL)) {
937 xmlC14NErrParam("writing attributes");
938 return (0);
939 }
940
941 xmlOutputBufferWriteString(ctx->buf, " ");
942 if (attr->ns != NULL && xmlStrlen(attr->ns->prefix) > 0) {
943 xmlOutputBufferWriteString(ctx->buf,
944 (const char *) attr->ns->prefix);
945 xmlOutputBufferWriteString(ctx->buf, ":");
946 }
947 xmlOutputBufferWriteString(ctx->buf, (const char *) attr->name);
948 xmlOutputBufferWriteString(ctx->buf, "=\"");
949
950 value = xmlNodeListGetString(ctx->doc, attr->children, 1);
951 /* todo: should we log an error if value==NULL ? */
952 if (value != NULL) {
953 buffer = xmlC11NNormalizeAttr(value);
954 xmlFree(value);
955 if (buffer != NULL) {
956 xmlOutputBufferWriteString(ctx->buf, (const char *) buffer);
957 xmlFree(buffer);
958 } else {
959 xmlC14NErrInternal("normalizing attributes axis");
960 return (0);
961 }
962 }
963 xmlOutputBufferWriteString(ctx->buf, "\"");
964 return (1);
965 }
966
967 /**
968 * xmlC14NFindHiddenParentAttr:
969 *
970 * Finds an attribute in a hidden parent node.
971 *
972 * Returns a pointer to the attribute node (if found) or NULL otherwise.
973 */
974 static xmlAttrPtr
xmlC14NFindHiddenParentAttr(xmlC14NCtxPtr ctx,xmlNodePtr cur,const xmlChar * name,const xmlChar * ns)975 xmlC14NFindHiddenParentAttr(xmlC14NCtxPtr ctx, xmlNodePtr cur, const xmlChar * name, const xmlChar * ns)
976 {
977 xmlAttrPtr res;
978 while((cur != NULL) && (!xmlC14NIsVisible(ctx, cur, cur->parent))) {
979 res = xmlHasNsProp(cur, name, ns);
980 if(res != NULL) {
981 return res;
982 }
983
984 cur = cur->parent;
985 }
986
987 return NULL;
988 }
989
990 /**
991 * xmlC14NFixupBaseAttr:
992 *
993 * Fixes up the xml:base attribute
994 *
995 * Returns the newly created attribute or NULL
996 */
997 static xmlAttrPtr
xmlC14NFixupBaseAttr(xmlC14NCtxPtr ctx,xmlAttrPtr xml_base_attr)998 xmlC14NFixupBaseAttr(xmlC14NCtxPtr ctx, xmlAttrPtr xml_base_attr)
999 {
1000 xmlChar * res = NULL;
1001 xmlNodePtr cur;
1002 xmlAttrPtr attr;
1003 xmlChar * tmp_str;
1004 xmlChar * tmp_str2;
1005 int tmp_str_len;
1006
1007 if ((ctx == NULL) || (xml_base_attr == NULL) || (xml_base_attr->parent == NULL)) {
1008 xmlC14NErrParam("processing xml:base attribute");
1009 return (NULL);
1010 }
1011
1012 /* start from current value */
1013 res = xmlNodeListGetString(ctx->doc, xml_base_attr->children, 1);
1014 if(res == NULL) {
1015 xmlC14NErrInternal("processing xml:base attribute - can't get attr value");
1016 return (NULL);
1017 }
1018
1019 /* go up the stack until we find a node that we rendered already */
1020 cur = xml_base_attr->parent->parent;
1021 while((cur != NULL) && (!xmlC14NIsVisible(ctx, cur, cur->parent))) {
1022 attr = xmlHasNsProp(cur, BAD_CAST "base", XML_XML_NAMESPACE);
1023 if(attr != NULL) {
1024 /* get attr value */
1025 tmp_str = xmlNodeListGetString(ctx->doc, attr->children, 1);
1026 if(tmp_str == NULL) {
1027 xmlFree(res);
1028
1029 xmlC14NErrInternal("processing xml:base attribute - can't get attr value");
1030 return (NULL);
1031 }
1032
1033 /* we need to add '/' if our current base uri ends with '..' or '.'
1034 to ensure that we are forced to go "up" all the time */
1035 tmp_str_len = xmlStrlen(tmp_str);
1036 if(tmp_str_len > 1 && tmp_str[tmp_str_len - 2] == '.') {
1037 tmp_str2 = xmlStrcat(tmp_str, BAD_CAST "/");
1038 if(tmp_str2 == NULL) {
1039 xmlFree(tmp_str);
1040 xmlFree(res);
1041
1042 xmlC14NErrInternal("processing xml:base attribute - can't modify uri");
1043 return (NULL);
1044 }
1045
1046 tmp_str = tmp_str2;
1047 }
1048
1049 /* build uri */
1050 tmp_str2 = xmlBuildURI(res, tmp_str);
1051 if(tmp_str2 == NULL) {
1052 xmlFree(tmp_str);
1053 xmlFree(res);
1054
1055 xmlC14NErrInternal("processing xml:base attribute - can't construct uri");
1056 return (NULL);
1057 }
1058
1059 /* cleanup and set the new res */
1060 xmlFree(tmp_str);
1061 xmlFree(res);
1062 res = tmp_str2;
1063 }
1064
1065 /* next */
1066 cur = cur->parent;
1067 }
1068
1069 /* check if result uri is empty or not */
1070 if((res == NULL) || xmlStrEqual(res, BAD_CAST "")) {
1071 xmlFree(res);
1072 return (NULL);
1073 }
1074
1075 /* create and return the new attribute node */
1076 attr = xmlNewNsProp(NULL, xml_base_attr->ns, BAD_CAST "base", res);
1077 if(attr == NULL) {
1078 xmlFree(res);
1079
1080 xmlC14NErrInternal("processing xml:base attribute - can't construct attribute");
1081 return (NULL);
1082 }
1083
1084 /* done */
1085 xmlFree(res);
1086 return (attr);
1087 }
1088
1089 /**
1090 * xmlC14NProcessAttrsAxis:
1091 * @ctx: the C14N context
1092 * @cur: the current node
1093 * @parent_visible: the visibility of parent node
1094 * @all_parents_visible: the visibility of all parent nodes
1095 *
1096 * Prints out canonical attribute axis of the current node to the
1097 * buffer from C14N context as follows
1098 *
1099 * Canonical XML v 1.0 (http://www.w3.org/TR/xml-c14n)
1100 *
1101 * Attribute Axis
1102 * In lexicographic order (ascending), process each node that
1103 * is in the element's attribute axis and in the node-set.
1104 *
1105 * The processing of an element node E MUST be modified slightly
1106 * when an XPath node-set is given as input and the element's
1107 * parent is omitted from the node-set.
1108 *
1109 *
1110 * Exclusive XML Canonicalization v 1.0 (http://www.w3.org/TR/xml-exc-c14n)
1111 *
1112 * Canonical XML applied to a document subset requires the search of the
1113 * ancestor nodes of each orphan element node for attributes in the xml
1114 * namespace, such as xml:lang and xml:space. These are copied into the
1115 * element node except if a declaration of the same attribute is already
1116 * in the attribute axis of the element (whether or not it is included in
1117 * the document subset). This search and copying are omitted from the
1118 * Exclusive XML Canonicalization method.
1119 *
1120 * Returns 0 on success or -1 on fail.
1121 */
1122 static int
xmlC14NProcessAttrsAxis(xmlC14NCtxPtr ctx,xmlNodePtr cur,int parent_visible)1123 xmlC14NProcessAttrsAxis(xmlC14NCtxPtr ctx, xmlNodePtr cur, int parent_visible)
1124 {
1125 xmlAttrPtr attr;
1126 xmlListPtr list;
1127 xmlAttrPtr attrs_to_delete = NULL;
1128
1129 /* special processing for 1.1 spec */
1130 xmlAttrPtr xml_base_attr = NULL;
1131 xmlAttrPtr xml_lang_attr = NULL;
1132 xmlAttrPtr xml_space_attr = NULL;
1133
1134 if ((ctx == NULL) || (cur == NULL) || (cur->type != XML_ELEMENT_NODE)) {
1135 xmlC14NErrParam("processing attributes axis");
1136 return (-1);
1137 }
1138
1139 /*
1140 * Create a sorted list to store element attributes
1141 */
1142 list = xmlListCreate(NULL, (xmlListDataCompare) xmlC14NAttrsCompare);
1143 if (list == NULL) {
1144 xmlC14NErrInternal("creating attributes list");
1145 return (-1);
1146 }
1147
1148 switch(ctx->mode) {
1149 case XML_C14N_1_0:
1150 /* The processing of an element node E MUST be modified slightly when an XPath node-set is
1151 * given as input and the element's parent is omitted from the node-set. The method for processing
1152 * the attribute axis of an element E in the node-set is enhanced. All element nodes along E's
1153 * ancestor axis are examined for nearest occurrences of attributes in the xml namespace, such
1154 * as xml:lang and xml:space (whether or not they are in the node-set). From this list of attributes,
1155 * remove any that are in E's attribute axis (whether or not they are in the node-set). Then,
1156 * lexicographically merge this attribute list with the nodes of E's attribute axis that are in
1157 * the node-set. The result of visiting the attribute axis is computed by processing the attribute
1158 * nodes in this merged attribute list.
1159 */
1160
1161 /*
1162 * Add all visible attributes from current node.
1163 */
1164 attr = cur->properties;
1165 while (attr != NULL) {
1166 /* check that attribute is visible */
1167 if (xmlC14NIsVisible(ctx, attr, cur)) {
1168 xmlListInsert(list, attr);
1169 }
1170 attr = attr->next;
1171 }
1172
1173 /*
1174 * Handle xml attributes
1175 */
1176 if (parent_visible && (cur->parent != NULL) &&
1177 (!xmlC14NIsVisible(ctx, cur->parent, cur->parent->parent)))
1178 {
1179 xmlNodePtr tmp;
1180
1181 /*
1182 * If XPath node-set is not specified then the parent is always
1183 * visible!
1184 */
1185 tmp = cur->parent;
1186 while (tmp != NULL) {
1187 attr = tmp->properties;
1188 while (attr != NULL) {
1189 if (xmlC14NIsXmlAttr(attr) != 0) {
1190 if (xmlListSearch(list, attr) == NULL) {
1191 xmlListInsert(list, attr);
1192 }
1193 }
1194 attr = attr->next;
1195 }
1196 tmp = tmp->parent;
1197 }
1198 }
1199
1200 /* done */
1201 break;
1202 case XML_C14N_EXCLUSIVE_1_0:
1203 /* attributes in the XML namespace, such as xml:lang and xml:space
1204 * are not imported into orphan nodes of the document subset
1205 */
1206
1207 /*
1208 * Add all visible attributes from current node.
1209 */
1210 attr = cur->properties;
1211 while (attr != NULL) {
1212 /* check that attribute is visible */
1213 if (xmlC14NIsVisible(ctx, attr, cur)) {
1214 xmlListInsert(list, attr);
1215 }
1216 attr = attr->next;
1217 }
1218
1219 /* do nothing special for xml attributes */
1220 break;
1221 case XML_C14N_1_1:
1222 /* The processing of an element node E MUST be modified slightly when an XPath node-set is
1223 * given as input and some of the element's ancestors are omitted from the node-set.
1224 *
1225 * Simple inheritable attributes are attributes that have a value that requires at most a simple
1226 * redeclaration. This redeclaration is done by supplying a new value in the child axis. The
1227 * redeclaration of a simple inheritable attribute A contained in one of E's ancestors is done
1228 * by supplying a value to an attribute Ae inside E with the same name. Simple inheritable attributes
1229 * are xml:lang and xml:space.
1230 *
1231 * The method for processing the attribute axis of an element E in the node-set is hence enhanced.
1232 * All element nodes along E's ancestor axis are examined for the nearest occurrences of simple
1233 * inheritable attributes in the xml namespace, such as xml:lang and xml:space (whether or not they
1234 * are in the node-set). From this list of attributes, any simple inheritable attributes that are
1235 * already in E's attribute axis (whether or not they are in the node-set) are removed. Then,
1236 * lexicographically merge this attribute list with the nodes of E's attribute axis that are in
1237 * the node-set. The result of visiting the attribute axis is computed by processing the attribute
1238 * nodes in this merged attribute list.
1239 *
1240 * The xml:id attribute is not a simple inheritable attribute and no processing of these attributes is
1241 * performed.
1242 *
1243 * The xml:base attribute is not a simple inheritable attribute and requires special processing beyond
1244 * a simple redeclaration.
1245 *
1246 * Attributes in the XML namespace other than xml:base, xml:id, xml:lang, and xml:space MUST be processed
1247 * as ordinary attributes.
1248 */
1249
1250 /*
1251 * Add all visible attributes from current node.
1252 */
1253 attr = cur->properties;
1254 while (attr != NULL) {
1255 /* special processing for XML attribute kiks in only when we have invisible parents */
1256 if ((!parent_visible) || (xmlC14NIsXmlAttr(attr) == 0)) {
1257 /* check that attribute is visible */
1258 if (xmlC14NIsVisible(ctx, attr, cur)) {
1259 xmlListInsert(list, attr);
1260 }
1261 } else {
1262 int matched = 0;
1263
1264 /* check for simple inheritance attributes */
1265 if((!matched) && (xml_lang_attr == NULL) && xmlStrEqual(attr->name, BAD_CAST "lang")) {
1266 xml_lang_attr = attr;
1267 matched = 1;
1268 }
1269 if((!matched) && (xml_space_attr == NULL) && xmlStrEqual(attr->name, BAD_CAST "space")) {
1270 xml_space_attr = attr;
1271 matched = 1;
1272 }
1273
1274 /* check for base attr */
1275 if((!matched) && (xml_base_attr == NULL) && xmlStrEqual(attr->name, BAD_CAST "base")) {
1276 xml_base_attr = attr;
1277 matched = 1;
1278 }
1279
1280 /* otherwise, it is a normal attribute, so just check if it is visible */
1281 if((!matched) && xmlC14NIsVisible(ctx, attr, cur)) {
1282 xmlListInsert(list, attr);
1283 }
1284 }
1285
1286 /* move to the next one */
1287 attr = attr->next;
1288 }
1289
1290 /* special processing for XML attribute kiks in only when we have invisible parents */
1291 if ((parent_visible)) {
1292
1293 /* simple inheritance attributes - copy */
1294 if(xml_lang_attr == NULL) {
1295 xml_lang_attr = xmlC14NFindHiddenParentAttr(ctx, cur->parent, BAD_CAST "lang", XML_XML_NAMESPACE);
1296 }
1297 if(xml_lang_attr != NULL) {
1298 xmlListInsert(list, xml_lang_attr);
1299 }
1300 if(xml_space_attr == NULL) {
1301 xml_space_attr = xmlC14NFindHiddenParentAttr(ctx, cur->parent, BAD_CAST "space", XML_XML_NAMESPACE);
1302 }
1303 if(xml_space_attr != NULL) {
1304 xmlListInsert(list, xml_space_attr);
1305 }
1306
1307 /* base uri attribute - fix up */
1308 if(xml_base_attr == NULL) {
1309 /* if we don't have base uri attribute, check if we have a "hidden" one above */
1310 xml_base_attr = xmlC14NFindHiddenParentAttr(ctx, cur->parent, BAD_CAST "base", XML_XML_NAMESPACE);
1311 }
1312 if(xml_base_attr != NULL) {
1313 xml_base_attr = xmlC14NFixupBaseAttr(ctx, xml_base_attr);
1314 if(xml_base_attr != NULL) {
1315 xmlListInsert(list, xml_base_attr);
1316
1317 /* note that we MUST delete returned attr node ourselves! */
1318 xml_base_attr->next = attrs_to_delete;
1319 attrs_to_delete = xml_base_attr;
1320 }
1321 }
1322 }
1323
1324 /* done */
1325 break;
1326 }
1327
1328 /*
1329 * print out all elements from list
1330 */
1331 xmlListWalk(list, (xmlListWalker) xmlC14NPrintAttrs, (const void *) ctx);
1332
1333 /*
1334 * Cleanup
1335 */
1336 xmlFreePropList(attrs_to_delete);
1337 xmlListDelete(list);
1338 return (0);
1339 }
1340
1341 /**
1342 * xmlC14NCheckForRelativeNamespaces:
1343 * @ctx: the C14N context
1344 * @cur: the current element node
1345 *
1346 * Checks that current element node has no relative namespaces defined
1347 *
1348 * Returns 0 if the node has no relative namespaces or -1 otherwise.
1349 */
1350 static int
xmlC14NCheckForRelativeNamespaces(xmlC14NCtxPtr ctx,xmlNodePtr cur)1351 xmlC14NCheckForRelativeNamespaces(xmlC14NCtxPtr ctx, xmlNodePtr cur)
1352 {
1353 xmlNsPtr ns;
1354
1355 if ((ctx == NULL) || (cur == NULL) || (cur->type != XML_ELEMENT_NODE)) {
1356 xmlC14NErrParam("checking for relative namespaces");
1357 return (-1);
1358 }
1359
1360 ns = cur->nsDef;
1361 while (ns != NULL) {
1362 if (xmlStrlen(ns->href) > 0) {
1363 xmlURIPtr uri;
1364
1365 uri = xmlParseURI((const char *) ns->href);
1366 if (uri == NULL) {
1367 xmlC14NErrInternal("parsing namespace uri");
1368 return (-1);
1369 }
1370 if (xmlStrlen((const xmlChar *) uri->scheme) == 0) {
1371 xmlC14NErrRelativeNamespace(uri->scheme);
1372 xmlFreeURI(uri);
1373 return (-1);
1374 }
1375 if ((xmlStrcasecmp((const xmlChar *) uri->scheme, BAD_CAST "urn") != 0)
1376 && (xmlStrcasecmp((const xmlChar *) uri->scheme, BAD_CAST "dav") !=0)
1377 && (xmlStrlen((const xmlChar *) uri->server) == 0)) {
1378 xmlC14NErrRelativeNamespace(uri->scheme);
1379 xmlFreeURI(uri);
1380 return (-1);
1381 }
1382 xmlFreeURI(uri);
1383 }
1384 ns = ns->next;
1385 }
1386 return (0);
1387 }
1388
1389 /**
1390 * xmlC14NProcessElementNode:
1391 * @ctx: the pointer to C14N context object
1392 * @cur: the node to process
1393 * @visible: this node is visible
1394 * @all_parents_visible: whether all the parents of this node are visible
1395 *
1396 * Canonical XML v 1.0 (http://www.w3.org/TR/xml-c14n)
1397 *
1398 * Element Nodes
1399 * If the element is not in the node-set, then the result is obtained
1400 * by processing the namespace axis, then the attribute axis, then
1401 * processing the child nodes of the element that are in the node-set
1402 * (in document order). If the element is in the node-set, then the result
1403 * is an open angle bracket (<), the element QName, the result of
1404 * processing the namespace axis, the result of processing the attribute
1405 * axis, a close angle bracket (>), the result of processing the child
1406 * nodes of the element that are in the node-set (in document order), an
1407 * open angle bracket, a forward slash (/), the element QName, and a close
1408 * angle bracket.
1409 *
1410 * Returns non-negative value on success or negative value on fail
1411 */
1412 static int
xmlC14NProcessElementNode(xmlC14NCtxPtr ctx,xmlNodePtr cur,int visible)1413 xmlC14NProcessElementNode(xmlC14NCtxPtr ctx, xmlNodePtr cur, int visible)
1414 {
1415 int ret;
1416 xmlC14NVisibleNsStack state;
1417 int parent_is_doc = 0;
1418
1419 if ((ctx == NULL) || (cur == NULL) || (cur->type != XML_ELEMENT_NODE)) {
1420 xmlC14NErrParam("processing element node");
1421 return (-1);
1422 }
1423
1424 /*
1425 * Check relative relative namespaces:
1426 * implementations of XML canonicalization MUST report an operation
1427 * failure on documents containing relative namespace URIs.
1428 */
1429 if (xmlC14NCheckForRelativeNamespaces(ctx, cur) < 0) {
1430 xmlC14NErrInternal("checking for relative namespaces");
1431 return (-1);
1432 }
1433
1434
1435 /*
1436 * Save ns_rendered stack position
1437 */
1438 memset(&state, 0, sizeof(state));
1439 xmlC14NVisibleNsStackSave(ctx->ns_rendered, &state);
1440
1441 if (visible) {
1442 if (ctx->parent_is_doc) {
1443 /* save this flag into the stack */
1444 parent_is_doc = ctx->parent_is_doc;
1445 ctx->parent_is_doc = 0;
1446 ctx->pos = XMLC14N_INSIDE_DOCUMENT_ELEMENT;
1447 }
1448 xmlOutputBufferWriteString(ctx->buf, "<");
1449
1450 if ((cur->ns != NULL) && (xmlStrlen(cur->ns->prefix) > 0)) {
1451 xmlOutputBufferWriteString(ctx->buf,
1452 (const char *) cur->ns->prefix);
1453 xmlOutputBufferWriteString(ctx->buf, ":");
1454 }
1455 xmlOutputBufferWriteString(ctx->buf, (const char *) cur->name);
1456 }
1457
1458 if (!xmlC14NIsExclusive(ctx)) {
1459 ret = xmlC14NProcessNamespacesAxis(ctx, cur, visible);
1460 } else {
1461 ret = xmlExcC14NProcessNamespacesAxis(ctx, cur, visible);
1462 }
1463 if (ret < 0) {
1464 xmlC14NErrInternal("processing namespaces axis");
1465 return (-1);
1466 }
1467 /* todo: shouldn't this go to "visible only"? */
1468 if(visible) {
1469 xmlC14NVisibleNsStackShift(ctx->ns_rendered);
1470 }
1471
1472 ret = xmlC14NProcessAttrsAxis(ctx, cur, visible);
1473 if (ret < 0) {
1474 xmlC14NErrInternal("processing attributes axis");
1475 return (-1);
1476 }
1477
1478 if (visible) {
1479 xmlOutputBufferWriteString(ctx->buf, ">");
1480 }
1481 if (cur->children != NULL) {
1482 ret = xmlC14NProcessNodeList(ctx, cur->children);
1483 if (ret < 0) {
1484 xmlC14NErrInternal("processing childrens list");
1485 return (-1);
1486 }
1487 }
1488 if (visible) {
1489 xmlOutputBufferWriteString(ctx->buf, "</");
1490 if ((cur->ns != NULL) && (xmlStrlen(cur->ns->prefix) > 0)) {
1491 xmlOutputBufferWriteString(ctx->buf,
1492 (const char *) cur->ns->prefix);
1493 xmlOutputBufferWriteString(ctx->buf, ":");
1494 }
1495 xmlOutputBufferWriteString(ctx->buf, (const char *) cur->name);
1496 xmlOutputBufferWriteString(ctx->buf, ">");
1497 if (parent_is_doc) {
1498 /* restore this flag from the stack for next node */
1499 ctx->parent_is_doc = parent_is_doc;
1500 ctx->pos = XMLC14N_AFTER_DOCUMENT_ELEMENT;
1501 }
1502 }
1503
1504 /*
1505 * Restore ns_rendered stack position
1506 */
1507 xmlC14NVisibleNsStackRestore(ctx->ns_rendered, &state);
1508 return (0);
1509 }
1510
1511 /**
1512 * xmlC14NProcessNode:
1513 * @ctx: the pointer to C14N context object
1514 * @cur: the node to process
1515 *
1516 * Processes the given node
1517 *
1518 * Returns non-negative value on success or negative value on fail
1519 */
1520 static int
xmlC14NProcessNode(xmlC14NCtxPtr ctx,xmlNodePtr cur)1521 xmlC14NProcessNode(xmlC14NCtxPtr ctx, xmlNodePtr cur)
1522 {
1523 int ret = 0;
1524 int visible;
1525
1526 if ((ctx == NULL) || (cur == NULL)) {
1527 xmlC14NErrParam("processing node");
1528 return (-1);
1529 }
1530
1531 visible = xmlC14NIsVisible(ctx, cur, cur->parent);
1532 switch (cur->type) {
1533 case XML_ELEMENT_NODE:
1534 ret = xmlC14NProcessElementNode(ctx, cur, visible);
1535 break;
1536 case XML_CDATA_SECTION_NODE:
1537 case XML_TEXT_NODE:
1538 /*
1539 * Text Nodes
1540 * the string value, except all ampersands are replaced
1541 * by &, all open angle brackets (<) are replaced by <, all closing
1542 * angle brackets (>) are replaced by >, and all #xD characters are
1543 * replaced by 
.
1544 */
1545 /* cdata sections are processed as text nodes */
1546 /* todo: verify that cdata sections are included in XPath nodes set */
1547 if ((visible) && (cur->content != NULL)) {
1548 xmlChar *buffer;
1549
1550 buffer = xmlC11NNormalizeText(cur->content);
1551 if (buffer != NULL) {
1552 xmlOutputBufferWriteString(ctx->buf,
1553 (const char *) buffer);
1554 xmlFree(buffer);
1555 } else {
1556 xmlC14NErrInternal("normalizing text node");
1557 return (-1);
1558 }
1559 }
1560 break;
1561 case XML_PI_NODE:
1562 /*
1563 * Processing Instruction (PI) Nodes-
1564 * The opening PI symbol (<?), the PI target name of the node,
1565 * a leading space and the string value if it is not empty, and
1566 * the closing PI symbol (?>). If the string value is empty,
1567 * then the leading space is not added. Also, a trailing #xA is
1568 * rendered after the closing PI symbol for PI children of the
1569 * root node with a lesser document order than the document
1570 * element, and a leading #xA is rendered before the opening PI
1571 * symbol of PI children of the root node with a greater document
1572 * order than the document element.
1573 */
1574 if (visible) {
1575 if (ctx->pos == XMLC14N_AFTER_DOCUMENT_ELEMENT) {
1576 xmlOutputBufferWriteString(ctx->buf, "\x0A<?");
1577 } else {
1578 xmlOutputBufferWriteString(ctx->buf, "<?");
1579 }
1580
1581 xmlOutputBufferWriteString(ctx->buf,
1582 (const char *) cur->name);
1583 if ((cur->content != NULL) && (*(cur->content) != '\0')) {
1584 xmlChar *buffer;
1585
1586 xmlOutputBufferWriteString(ctx->buf, " ");
1587
1588 /* todo: do we need to normalize pi? */
1589 buffer = xmlC11NNormalizePI(cur->content);
1590 if (buffer != NULL) {
1591 xmlOutputBufferWriteString(ctx->buf,
1592 (const char *) buffer);
1593 xmlFree(buffer);
1594 } else {
1595 xmlC14NErrInternal("normalizing pi node");
1596 return (-1);
1597 }
1598 }
1599
1600 if (ctx->pos == XMLC14N_BEFORE_DOCUMENT_ELEMENT) {
1601 xmlOutputBufferWriteString(ctx->buf, "?>\x0A");
1602 } else {
1603 xmlOutputBufferWriteString(ctx->buf, "?>");
1604 }
1605 }
1606 break;
1607 case XML_COMMENT_NODE:
1608 /*
1609 * Comment Nodes
1610 * Nothing if generating canonical XML without comments. For
1611 * canonical XML with comments, generate the opening comment
1612 * symbol (<!--), the string value of the node, and the
1613 * closing comment symbol (-->). Also, a trailing #xA is rendered
1614 * after the closing comment symbol for comment children of the
1615 * root node with a lesser document order than the document
1616 * element, and a leading #xA is rendered before the opening
1617 * comment symbol of comment children of the root node with a
1618 * greater document order than the document element. (Comment
1619 * children of the root node represent comments outside of the
1620 * top-level document element and outside of the document type
1621 * declaration).
1622 */
1623 if (visible && ctx->with_comments) {
1624 if (ctx->pos == XMLC14N_AFTER_DOCUMENT_ELEMENT) {
1625 xmlOutputBufferWriteString(ctx->buf, "\x0A<!--");
1626 } else {
1627 xmlOutputBufferWriteString(ctx->buf, "<!--");
1628 }
1629
1630 if (cur->content != NULL) {
1631 xmlChar *buffer;
1632
1633 /* todo: do we need to normalize comment? */
1634 buffer = xmlC11NNormalizeComment(cur->content);
1635 if (buffer != NULL) {
1636 xmlOutputBufferWriteString(ctx->buf,
1637 (const char *) buffer);
1638 xmlFree(buffer);
1639 } else {
1640 xmlC14NErrInternal("normalizing comment node");
1641 return (-1);
1642 }
1643 }
1644
1645 if (ctx->pos == XMLC14N_BEFORE_DOCUMENT_ELEMENT) {
1646 xmlOutputBufferWriteString(ctx->buf, "-->\x0A");
1647 } else {
1648 xmlOutputBufferWriteString(ctx->buf, "-->");
1649 }
1650 }
1651 break;
1652 case XML_DOCUMENT_NODE:
1653 case XML_DOCUMENT_FRAG_NODE: /* should be processed as document? */
1654 #ifdef LIBXML_DOCB_ENABLED
1655 case XML_DOCB_DOCUMENT_NODE: /* should be processed as document? */
1656 #endif
1657 #ifdef LIBXML_HTML_ENABLED
1658 case XML_HTML_DOCUMENT_NODE: /* should be processed as document? */
1659 #endif
1660 if (cur->children != NULL) {
1661 ctx->pos = XMLC14N_BEFORE_DOCUMENT_ELEMENT;
1662 ctx->parent_is_doc = 1;
1663 ret = xmlC14NProcessNodeList(ctx, cur->children);
1664 }
1665 break;
1666
1667 case XML_ATTRIBUTE_NODE:
1668 xmlC14NErrInvalidNode("XML_ATTRIBUTE_NODE", "processing node");
1669 return (-1);
1670 case XML_NAMESPACE_DECL:
1671 xmlC14NErrInvalidNode("XML_NAMESPACE_DECL", "processing node");
1672 return (-1);
1673 case XML_ENTITY_REF_NODE:
1674 xmlC14NErrInvalidNode("XML_ENTITY_REF_NODE", "processing node");
1675 return (-1);
1676 case XML_ENTITY_NODE:
1677 xmlC14NErrInvalidNode("XML_ENTITY_NODE", "processing node");
1678 return (-1);
1679
1680 case XML_DOCUMENT_TYPE_NODE:
1681 case XML_NOTATION_NODE:
1682 case XML_DTD_NODE:
1683 case XML_ELEMENT_DECL:
1684 case XML_ATTRIBUTE_DECL:
1685 case XML_ENTITY_DECL:
1686 #ifdef LIBXML_XINCLUDE_ENABLED
1687 case XML_XINCLUDE_START:
1688 case XML_XINCLUDE_END:
1689 #endif
1690 /*
1691 * should be ignored according to "W3C Canonical XML"
1692 */
1693 break;
1694 default:
1695 xmlC14NErrUnknownNode(cur->type, "processing node");
1696 return (-1);
1697 }
1698
1699 return (ret);
1700 }
1701
1702 /**
1703 * xmlC14NProcessNodeList:
1704 * @ctx: the pointer to C14N context object
1705 * @cur: the node to start from
1706 *
1707 * Processes all nodes in the row starting from cur.
1708 *
1709 * Returns non-negative value on success or negative value on fail
1710 */
1711 static int
xmlC14NProcessNodeList(xmlC14NCtxPtr ctx,xmlNodePtr cur)1712 xmlC14NProcessNodeList(xmlC14NCtxPtr ctx, xmlNodePtr cur)
1713 {
1714 int ret;
1715
1716 if (ctx == NULL) {
1717 xmlC14NErrParam("processing node list");
1718 return (-1);
1719 }
1720
1721 for (ret = 0; cur != NULL && ret >= 0; cur = cur->next) {
1722 ret = xmlC14NProcessNode(ctx, cur);
1723 }
1724 return (ret);
1725 }
1726
1727
1728 /**
1729 * xmlC14NFreeCtx:
1730 * @ctx: the pointer to C14N context object
1731 *
1732 * Cleanups the C14N context object.
1733 */
1734
1735 static void
xmlC14NFreeCtx(xmlC14NCtxPtr ctx)1736 xmlC14NFreeCtx(xmlC14NCtxPtr ctx)
1737 {
1738 if (ctx == NULL) {
1739 xmlC14NErrParam("freeing context");
1740 return;
1741 }
1742
1743 if (ctx->ns_rendered != NULL) {
1744 xmlC14NVisibleNsStackDestroy(ctx->ns_rendered);
1745 }
1746 xmlFree(ctx);
1747 }
1748
1749 /**
1750 * xmlC14NNewCtx:
1751 * @doc: the XML document for canonization
1752 * @is_visible_callback:the function to use to determine is node visible
1753 * or not
1754 * @user_data: the first parameter for @is_visible_callback function
1755 * (in most cases, it is nodes set)
1756 * @mode: the c14n mode (see @xmlC14NMode)
1757 * @inclusive_ns_prefixe the list of inclusive namespace prefixes
1758 * ended with a NULL or NULL if there is no
1759 * inclusive namespaces (only for `
1760 * canonicalization)
1761 * @with_comments: include comments in the result (!=0) or not (==0)
1762 * @buf: the output buffer to store canonical XML; this
1763 * buffer MUST have encoder==NULL because C14N requires
1764 * UTF-8 output
1765 *
1766 * Creates new C14N context object to store C14N parameters.
1767 *
1768 * Returns pointer to newly created object (success) or NULL (fail)
1769 */
1770 static xmlC14NCtxPtr
xmlC14NNewCtx(xmlDocPtr doc,xmlC14NIsVisibleCallback is_visible_callback,void * user_data,xmlC14NMode mode,xmlChar ** inclusive_ns_prefixes,int with_comments,xmlOutputBufferPtr buf)1771 xmlC14NNewCtx(xmlDocPtr doc,
1772 xmlC14NIsVisibleCallback is_visible_callback, void* user_data,
1773 xmlC14NMode mode, xmlChar ** inclusive_ns_prefixes,
1774 int with_comments, xmlOutputBufferPtr buf)
1775 {
1776 xmlC14NCtxPtr ctx = NULL;
1777
1778 if ((doc == NULL) || (buf == NULL)) {
1779 xmlC14NErrParam("creating new context");
1780 return (NULL);
1781 }
1782
1783 /*
1784 * Validate the encoding output buffer encoding
1785 */
1786 if (buf->encoder != NULL) {
1787 xmlC14NErr(ctx, (xmlNodePtr) doc, XML_C14N_REQUIRES_UTF8,
1788 "xmlC14NNewCtx: output buffer encoder != NULL but C14N requires UTF8 output\n");
1789 return (NULL);
1790 }
1791
1792 /*
1793 * Validate the XML document encoding value, if provided.
1794 */
1795 if (doc->charset != XML_CHAR_ENCODING_UTF8) {
1796 xmlC14NErr(ctx, (xmlNodePtr) doc, XML_C14N_REQUIRES_UTF8,
1797 "xmlC14NNewCtx: source document not in UTF8\n");
1798 return (NULL);
1799 }
1800
1801 /*
1802 * Allocate a new xmlC14NCtxPtr and fill the fields.
1803 */
1804 ctx = (xmlC14NCtxPtr) xmlMalloc(sizeof(xmlC14NCtx));
1805 if (ctx == NULL) {
1806 xmlC14NErrMemory("creating context");
1807 return (NULL);
1808 }
1809 memset(ctx, 0, sizeof(xmlC14NCtx));
1810
1811 /*
1812 * initialize C14N context
1813 */
1814 ctx->doc = doc;
1815 ctx->with_comments = with_comments;
1816 ctx->is_visible_callback = is_visible_callback;
1817 ctx->user_data = user_data;
1818 ctx->buf = buf;
1819 ctx->parent_is_doc = 1;
1820 ctx->pos = XMLC14N_BEFORE_DOCUMENT_ELEMENT;
1821 ctx->ns_rendered = xmlC14NVisibleNsStackCreate();
1822
1823 if(ctx->ns_rendered == NULL) {
1824 xmlC14NErr(ctx, (xmlNodePtr) doc, XML_C14N_CREATE_STACK,
1825 "xmlC14NNewCtx: xmlC14NVisibleNsStackCreate failed\n");
1826 xmlC14NFreeCtx(ctx);
1827 return (NULL);
1828 }
1829
1830 /*
1831 * Set "mode" flag and remember list of incluseve prefixes
1832 * for exclusive c14n
1833 */
1834 ctx->mode = mode;
1835 if(xmlC14NIsExclusive(ctx)) {
1836 ctx->inclusive_ns_prefixes = inclusive_ns_prefixes;
1837 }
1838 return (ctx);
1839 }
1840
1841 /**
1842 * xmlC14NExecute:
1843 * @doc: the XML document for canonization
1844 * @is_visible_callback:the function to use to determine is node visible
1845 * or not
1846 * @user_data: the first parameter for @is_visible_callback function
1847 * (in most cases, it is nodes set)
1848 * @mode: the c14n mode (see @xmlC14NMode)
1849 * @inclusive_ns_prefixes: the list of inclusive namespace prefixes
1850 * ended with a NULL or NULL if there is no
1851 * inclusive namespaces (only for exclusive
1852 * canonicalization, ignored otherwise)
1853 * @with_comments: include comments in the result (!=0) or not (==0)
1854 * @buf: the output buffer to store canonical XML; this
1855 * buffer MUST have encoder==NULL because C14N requires
1856 * UTF-8 output
1857 *
1858 * Dumps the canonized image of given XML document into the provided buffer.
1859 * For details see "Canonical XML" (http://www.w3.org/TR/xml-c14n) or
1860 * "Exclusive XML Canonicalization" (http://www.w3.org/TR/xml-exc-c14n)
1861 *
1862 * Returns non-negative value on success or a negative value on fail
1863 */
1864 int
xmlC14NExecute(xmlDocPtr doc,xmlC14NIsVisibleCallback is_visible_callback,void * user_data,int mode,xmlChar ** inclusive_ns_prefixes,int with_comments,xmlOutputBufferPtr buf)1865 xmlC14NExecute(xmlDocPtr doc, xmlC14NIsVisibleCallback is_visible_callback,
1866 void* user_data, int mode, xmlChar **inclusive_ns_prefixes,
1867 int with_comments, xmlOutputBufferPtr buf) {
1868
1869 xmlC14NCtxPtr ctx;
1870 xmlC14NMode c14n_mode = XML_C14N_1_0;
1871 int ret;
1872
1873 if ((buf == NULL) || (doc == NULL)) {
1874 xmlC14NErrParam("executing c14n");
1875 return (-1);
1876 }
1877
1878 /* for backward compatibility, we have to have "mode" as "int"
1879 and here we check that user gives valid value */
1880 switch(mode) {
1881 case XML_C14N_1_0:
1882 case XML_C14N_EXCLUSIVE_1_0:
1883 case XML_C14N_1_1:
1884 c14n_mode = (xmlC14NMode)mode;
1885 break;
1886 default:
1887 xmlC14NErrParam("invalid mode for executing c14n");
1888 return (-1);
1889 }
1890
1891 /*
1892 * Validate the encoding output buffer encoding
1893 */
1894 if (buf->encoder != NULL) {
1895 xmlC14NErr(NULL, (xmlNodePtr) doc, XML_C14N_REQUIRES_UTF8,
1896 "xmlC14NExecute: output buffer encoder != NULL but C14N requires UTF8 output\n");
1897 return (-1);
1898 }
1899
1900 ctx = xmlC14NNewCtx(doc, is_visible_callback, user_data,
1901 c14n_mode, inclusive_ns_prefixes,
1902 with_comments, buf);
1903 if (ctx == NULL) {
1904 xmlC14NErr(NULL, (xmlNodePtr) doc, XML_C14N_CREATE_CTXT,
1905 "xmlC14NExecute: unable to create C14N context\n");
1906 return (-1);
1907 }
1908
1909
1910
1911 /*
1912 * Root Node
1913 * The root node is the parent of the top-level document element. The
1914 * result of processing each of its child nodes that is in the node-set
1915 * in document order. The root node does not generate a byte order mark,
1916 * XML declaration, nor anything from within the document type
1917 * declaration.
1918 */
1919 if (doc->children != NULL) {
1920 ret = xmlC14NProcessNodeList(ctx, doc->children);
1921 if (ret < 0) {
1922 xmlC14NErrInternal("processing docs children list");
1923 xmlC14NFreeCtx(ctx);
1924 return (-1);
1925 }
1926 }
1927
1928 /*
1929 * Flush buffer to get number of bytes written
1930 */
1931 ret = xmlOutputBufferFlush(buf);
1932 if (ret < 0) {
1933 xmlC14NErrInternal("flushing output buffer");
1934 xmlC14NFreeCtx(ctx);
1935 return (-1);
1936 }
1937
1938 /*
1939 * Cleanup
1940 */
1941 xmlC14NFreeCtx(ctx);
1942 return (ret);
1943 }
1944
1945 /**
1946 * xmlC14NDocSaveTo:
1947 * @doc: the XML document for canonization
1948 * @nodes: the nodes set to be included in the canonized image
1949 * or NULL if all document nodes should be included
1950 * @mode: the c14n mode (see @xmlC14NMode)
1951 * @inclusive_ns_prefixes: the list of inclusive namespace prefixes
1952 * ended with a NULL or NULL if there is no
1953 * inclusive namespaces (only for exclusive
1954 * canonicalization, ignored otherwise)
1955 * @with_comments: include comments in the result (!=0) or not (==0)
1956 * @buf: the output buffer to store canonical XML; this
1957 * buffer MUST have encoder==NULL because C14N requires
1958 * UTF-8 output
1959 *
1960 * Dumps the canonized image of given XML document into the provided buffer.
1961 * For details see "Canonical XML" (http://www.w3.org/TR/xml-c14n) or
1962 * "Exclusive XML Canonicalization" (http://www.w3.org/TR/xml-exc-c14n)
1963 *
1964 * Returns non-negative value on success or a negative value on fail
1965 */
1966 int
xmlC14NDocSaveTo(xmlDocPtr doc,xmlNodeSetPtr nodes,int mode,xmlChar ** inclusive_ns_prefixes,int with_comments,xmlOutputBufferPtr buf)1967 xmlC14NDocSaveTo(xmlDocPtr doc, xmlNodeSetPtr nodes,
1968 int mode, xmlChar ** inclusive_ns_prefixes,
1969 int with_comments, xmlOutputBufferPtr buf) {
1970 return(xmlC14NExecute(doc,
1971 (xmlC14NIsVisibleCallback)xmlC14NIsNodeInNodeset,
1972 nodes,
1973 mode,
1974 inclusive_ns_prefixes,
1975 with_comments,
1976 buf));
1977 }
1978
1979
1980 /**
1981 * xmlC14NDocDumpMemory:
1982 * @doc: the XML document for canonization
1983 * @nodes: the nodes set to be included in the canonized image
1984 * or NULL if all document nodes should be included
1985 * @mode: the c14n mode (see @xmlC14NMode)
1986 * @inclusive_ns_prefixes: the list of inclusive namespace prefixes
1987 * ended with a NULL or NULL if there is no
1988 * inclusive namespaces (only for exclusive
1989 * canonicalization, ignored otherwise)
1990 * @with_comments: include comments in the result (!=0) or not (==0)
1991 * @doc_txt_ptr: the memory pointer for allocated canonical XML text;
1992 * the caller of this functions is responsible for calling
1993 * xmlFree() to free allocated memory
1994 *
1995 * Dumps the canonized image of given XML document into memory.
1996 * For details see "Canonical XML" (http://www.w3.org/TR/xml-c14n) or
1997 * "Exclusive XML Canonicalization" (http://www.w3.org/TR/xml-exc-c14n)
1998 *
1999 * Returns the number of bytes written on success or a negative value on fail
2000 */
2001 int
xmlC14NDocDumpMemory(xmlDocPtr doc,xmlNodeSetPtr nodes,int mode,xmlChar ** inclusive_ns_prefixes,int with_comments,xmlChar ** doc_txt_ptr)2002 xmlC14NDocDumpMemory(xmlDocPtr doc, xmlNodeSetPtr nodes,
2003 int mode, xmlChar ** inclusive_ns_prefixes,
2004 int with_comments, xmlChar ** doc_txt_ptr)
2005 {
2006 int ret;
2007 xmlOutputBufferPtr buf;
2008
2009 if (doc_txt_ptr == NULL) {
2010 xmlC14NErrParam("dumping doc to memory");
2011 return (-1);
2012 }
2013
2014 *doc_txt_ptr = NULL;
2015
2016 /*
2017 * create memory buffer with UTF8 (default) encoding
2018 */
2019 buf = xmlAllocOutputBuffer(NULL);
2020 if (buf == NULL) {
2021 xmlC14NErrMemory("creating output buffer");
2022 return (-1);
2023 }
2024
2025 /*
2026 * canonize document and write to buffer
2027 */
2028 ret = xmlC14NDocSaveTo(doc, nodes, mode, inclusive_ns_prefixes,
2029 with_comments, buf);
2030 if (ret < 0) {
2031 xmlC14NErrInternal("saving doc to output buffer");
2032 (void) xmlOutputBufferClose(buf);
2033 return (-1);
2034 }
2035
2036 ret = buf->buffer->use;
2037 if (ret > 0) {
2038 *doc_txt_ptr = xmlStrndup(buf->buffer->content, ret);
2039 }
2040 (void) xmlOutputBufferClose(buf);
2041
2042 if ((*doc_txt_ptr == NULL) && (ret > 0)) {
2043 xmlC14NErrMemory("coping canonicanized document");
2044 return (-1);
2045 }
2046 return (ret);
2047 }
2048
2049 /**
2050 * xmlC14NDocSave:
2051 * @doc: the XML document for canonization
2052 * @nodes: the nodes set to be included in the canonized image
2053 * or NULL if all document nodes should be included
2054 * @mode: the c14n mode (see @xmlC14NMode)
2055 * @inclusive_ns_prefixes: the list of inclusive namespace prefixes
2056 * ended with a NULL or NULL if there is no
2057 * inclusive namespaces (only for exclusive
2058 * canonicalization, ignored otherwise)
2059 * @with_comments: include comments in the result (!=0) or not (==0)
2060 * @filename: the filename to store canonical XML image
2061 * @compression: the compression level (zlib requred):
2062 * -1 - libxml default,
2063 * 0 - uncompressed,
2064 * >0 - compression level
2065 *
2066 * Dumps the canonized image of given XML document into the file.
2067 * For details see "Canonical XML" (http://www.w3.org/TR/xml-c14n) or
2068 * "Exclusive XML Canonicalization" (http://www.w3.org/TR/xml-exc-c14n)
2069 *
2070 * Returns the number of bytes written success or a negative value on fail
2071 */
2072 int
xmlC14NDocSave(xmlDocPtr doc,xmlNodeSetPtr nodes,int mode,xmlChar ** inclusive_ns_prefixes,int with_comments,const char * filename,int compression)2073 xmlC14NDocSave(xmlDocPtr doc, xmlNodeSetPtr nodes,
2074 int mode, xmlChar ** inclusive_ns_prefixes,
2075 int with_comments, const char *filename, int compression)
2076 {
2077 xmlOutputBufferPtr buf;
2078 int ret;
2079
2080 if (filename == NULL) {
2081 xmlC14NErrParam("saving doc");
2082 return (-1);
2083 }
2084 #ifdef HAVE_ZLIB_H
2085 if (compression < 0)
2086 compression = xmlGetCompressMode();
2087 #endif
2088
2089 /*
2090 * save the content to a temp buffer, use default UTF8 encoding.
2091 */
2092 buf = xmlOutputBufferCreateFilename(filename, NULL, compression);
2093 if (buf == NULL) {
2094 xmlC14NErrInternal("creating temporary filename");
2095 return (-1);
2096 }
2097
2098 /*
2099 * canonize document and write to buffer
2100 */
2101 ret = xmlC14NDocSaveTo(doc, nodes, mode, inclusive_ns_prefixes,
2102 with_comments, buf);
2103 if (ret < 0) {
2104 xmlC14NErrInternal("cannicanize document to buffer");
2105 (void) xmlOutputBufferClose(buf);
2106 return (-1);
2107 }
2108
2109 /*
2110 * get the numbers of bytes written
2111 */
2112 ret = xmlOutputBufferClose(buf);
2113 return (ret);
2114 }
2115
2116
2117
2118 /*
2119 * Macro used to grow the current buffer.
2120 */
2121 #define growBufferReentrant() { \
2122 buffer_size *= 2; \
2123 buffer = (xmlChar *) \
2124 xmlRealloc(buffer, buffer_size * sizeof(xmlChar)); \
2125 if (buffer == NULL) { \
2126 xmlC14NErrMemory("growing buffer"); \
2127 return(NULL); \
2128 } \
2129 }
2130
2131 /**
2132 * xmlC11NNormalizeString:
2133 * @input: the input string
2134 * @mode: the normalization mode (attribute, comment, PI or text)
2135 *
2136 * Converts a string to a canonical (normalized) format. The code is stolen
2137 * from xmlEncodeEntitiesReentrant(). Added normalization of \x09, \x0a, \x0A
2138 * and the @mode parameter
2139 *
2140 * Returns a normalized string (caller is responsible for calling xmlFree())
2141 * or NULL if an error occurs
2142 */
2143 static xmlChar *
xmlC11NNormalizeString(const xmlChar * input,xmlC14NNormalizationMode mode)2144 xmlC11NNormalizeString(const xmlChar * input,
2145 xmlC14NNormalizationMode mode)
2146 {
2147 const xmlChar *cur = input;
2148 xmlChar *buffer = NULL;
2149 xmlChar *out = NULL;
2150 int buffer_size = 0;
2151
2152 if (input == NULL)
2153 return (NULL);
2154
2155 /*
2156 * allocate an translation buffer.
2157 */
2158 buffer_size = 1000;
2159 buffer = (xmlChar *) xmlMallocAtomic(buffer_size * sizeof(xmlChar));
2160 if (buffer == NULL) {
2161 xmlC14NErrMemory("allocating buffer");
2162 return (NULL);
2163 }
2164 out = buffer;
2165
2166 while (*cur != '\0') {
2167 if ((out - buffer) > (buffer_size - 10)) {
2168 int indx = out - buffer;
2169
2170 growBufferReentrant();
2171 out = &buffer[indx];
2172 }
2173
2174 if ((*cur == '<') && ((mode == XMLC14N_NORMALIZE_ATTR) ||
2175 (mode == XMLC14N_NORMALIZE_TEXT))) {
2176 *out++ = '&';
2177 *out++ = 'l';
2178 *out++ = 't';
2179 *out++ = ';';
2180 } else if ((*cur == '>') && (mode == XMLC14N_NORMALIZE_TEXT)) {
2181 *out++ = '&';
2182 *out++ = 'g';
2183 *out++ = 't';
2184 *out++ = ';';
2185 } else if ((*cur == '&') && ((mode == XMLC14N_NORMALIZE_ATTR) ||
2186 (mode == XMLC14N_NORMALIZE_TEXT))) {
2187 *out++ = '&';
2188 *out++ = 'a';
2189 *out++ = 'm';
2190 *out++ = 'p';
2191 *out++ = ';';
2192 } else if ((*cur == '"') && (mode == XMLC14N_NORMALIZE_ATTR)) {
2193 *out++ = '&';
2194 *out++ = 'q';
2195 *out++ = 'u';
2196 *out++ = 'o';
2197 *out++ = 't';
2198 *out++ = ';';
2199 } else if ((*cur == '\x09') && (mode == XMLC14N_NORMALIZE_ATTR)) {
2200 *out++ = '&';
2201 *out++ = '#';
2202 *out++ = 'x';
2203 *out++ = '9';
2204 *out++ = ';';
2205 } else if ((*cur == '\x0A') && (mode == XMLC14N_NORMALIZE_ATTR)) {
2206 *out++ = '&';
2207 *out++ = '#';
2208 *out++ = 'x';
2209 *out++ = 'A';
2210 *out++ = ';';
2211 } else if ((*cur == '\x0D') && ((mode == XMLC14N_NORMALIZE_ATTR) ||
2212 (mode == XMLC14N_NORMALIZE_TEXT) ||
2213 (mode == XMLC14N_NORMALIZE_COMMENT) ||
2214 (mode == XMLC14N_NORMALIZE_PI))) {
2215 *out++ = '&';
2216 *out++ = '#';
2217 *out++ = 'x';
2218 *out++ = 'D';
2219 *out++ = ';';
2220 } else {
2221 /*
2222 * Works because on UTF-8, all extended sequences cannot
2223 * result in bytes in the ASCII range.
2224 */
2225 *out++ = *cur;
2226 }
2227 cur++;
2228 }
2229 *out = 0;
2230 return (buffer);
2231 }
2232 #endif /* LIBXML_OUTPUT_ENABLED */
2233 #define bottom_c14n
2234 #include "elfgcchack.h"
2235 #endif /* LIBXML_C14N_ENABLED */
2236