xml/apr_xml.c

   1 /* Licensed to the Apache Software Foundation (ASF) under one or more
   2  * contributor license agreements.  See the NOTICE file distributed with
   3  * this work for additional information regarding copyright ownership.
   4  * The ASF licenses this file to You under the Apache License, Version 2.0
   5  * (the "License"); you may not use this file except in compliance with
   6  * the License.  You may obtain a copy of the License at
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 #include "apr.h"
  18 #include "apr_strings.h"
  19
  20 #define APR_WANT_STDIO          /* for sprintf() */
  21 #define APR_WANT_STRFUNC
  22 #include "apr_want.h"
  23
  24 #include "apr_xml.h"
  25
  26 #include "apu_config.h"
  27
  28 #if defined(HAVE_XMLPARSE_XMLPARSE_H)
  29 #include <xmlparse/xmlparse.h>
  30 #elif defined(HAVE_XMLTOK_XMLPARSE_H)
  31 #include <xmltok/xmlparse.h>
  32 #elif defined(HAVE_XML_XMLPARSE_H)
  33 #include <xml/xmlparse.h>
  34 #else
  35 #include <expat.h>
  36 #endif
  37
  38 #define DEBUG_CR "\r\n"
  39
  40 static const char APR_KW_xmlns[] = { 0x78, 0x6D, 0x6C, 0x6E, 0x73, '\0' };
  41 static const char APR_KW_xmlns_lang[] = { 0x78, 0x6D, 0x6C, 0x3A, 0x6C, 0x61, 0x6E, 0x67, '\0' };
  42 static const char APR_KW_DAV[] = { 0x44, 0x41, 0x56, 0x3A, '\0' };
  43
  44 /* errors related to namespace processing */
  45 #define APR_XML_NS_ERROR_UNKNOWN_PREFIX (-1000)
  46 #define APR_XML_NS_ERROR_INVALID_DECL (-1001)
  47
  48 /* test for a namespace prefix that begins with [Xx][Mm][Ll] */
  49 #define APR_XML_NS_IS_RESERVED(name) \
  50         ( (name[0] == 0x58 || name[0] == 0x78) && \
  51           (name[1] == 0x4D || name[1] == 0x6D) && \
  52           (name[2] == 0x4C || name[2] == 0x6C) )
  53
  54
  55 /* the real (internal) definition of the parser context */
  56 struct apr_xml_parser {
  57     apr_xml_doc *doc;           /* the doc we're parsing */
  58     apr_pool_t *p;              /* the pool we allocate from */
  59     apr_xml_elem *cur_elem;     /* current element */
  60
  61     int error;                  /* an error has occurred */
  62 #define APR_XML_ERROR_EXPAT             1
  63 #define APR_XML_ERROR_PARSE_DONE        2
  64 /* also: public APR_XML_NS_ERROR_* values (if any) */
  65
  66     XML_Parser xp;              /* the actual (Expat) XML parser */
  67     enum XML_Error xp_err;      /* stored Expat error code */
  68 };
  69
  70 /* struct for scoping namespace declarations */
  71 typedef struct apr_xml_ns_scope {
  72     const char *prefix;         /* prefix used for this ns */
  73     int ns;                     /* index into namespace table */
  74     int emptyURI;               /* the namespace URI is the empty string */
  75     struct apr_xml_ns_scope *next;      /* next scoped namespace */
  76 } apr_xml_ns_scope;
  77
  78
  79 /* return namespace table index for a given prefix */
  80 static int find_prefix(apr_xml_parser *parser, const char *prefix)
  81 {
  82     apr_xml_elem *elem = parser->cur_elem;
  83
  84     /*
  85     ** Walk up the tree, looking for a namespace scope that defines this
  86     ** prefix.
  87     */
  88     for (; elem; elem = elem->parent) {
  89         apr_xml_ns_scope *ns_scope = elem->ns_scope;
  90
  91         for (ns_scope = elem->ns_scope; ns_scope; ns_scope = ns_scope->next) {
  92             if (strcmp(prefix, ns_scope->prefix) == 0) {
  93                 if (ns_scope->emptyURI) {
  94                     /*
  95                     ** It is possible to set the default namespace to an
  96                     ** empty URI string; this resets the default namespace
  97                     ** to mean "no namespace." We just found the prefix
  98                     ** refers to an empty URI, so return "no namespace."
  99                     */
 100                     return APR_XML_NS_NONE;
 101                 }
 102
 103                 return ns_scope->ns;
 104             }
 105         }
 106     }
 107
 108     /*
 109      * If the prefix is empty (""), this means that a prefix was not
 110      * specified in the element/attribute. The search that was performed
 111      * just above did not locate a default namespace URI (which is stored
 112      * into ns_scope with an empty prefix). This means the element/attribute
 113      * has "no namespace". We have a reserved value for this.
 114      */
 115     if (*prefix == '\0') {
 116         return APR_XML_NS_NONE;
 117     }
 118
 119     /* not found */
 120     return APR_XML_NS_ERROR_UNKNOWN_PREFIX;
 121 }
 122
 123 static void start_handler(void *userdata, const char *name, const char **attrs)
 124 {
 125     apr_xml_parser *parser = userdata;
 126     apr_xml_elem *elem;
 127     apr_xml_attr *attr;
 128     apr_xml_attr *prev;
 129     char *colon;
 130     const char *quoted;
 131     char *elem_name;
 132
 133     /* punt once we find an error */
 134     if (parser->error)
 135         return;
 136
 137     elem = apr_pcalloc(parser->p, sizeof(*elem));
 138
 139     /* prep the element */
 140     elem->name = elem_name = apr_pstrdup(parser->p, name);
 141
 142     /* fill in the attributes (note: ends up in reverse order) */
 143     while (*attrs) {
 144         attr = apr_palloc(parser->p, sizeof(*attr));
 145         attr->name = apr_pstrdup(parser->p, *attrs++);
 146         attr->value = apr_pstrdup(parser->p, *attrs++);
 147         attr->next = elem->attr;
 148         elem->attr = attr;
 149     }
 150
 151     /* hook the element into the tree */
 152     if (parser->cur_elem == NULL) {
 153         /* no current element; this also becomes the root */
 154         parser->cur_elem = parser->doc->root = elem;
 155     }
 156     else {
 157         /* this element appeared within the current elem */
 158         elem->parent = parser->cur_elem;
 159
 160         /* set up the child/sibling links */
 161         if (elem->parent->last_child == NULL) {
 162             /* no first child either */
 163             elem->parent->first_child = elem->parent->last_child = elem;
 164         }
 165         else {
 166             /* hook onto the end of the parent's children */
 167             elem->parent->last_child->next = elem;
 168             elem->parent->last_child = elem;
 169         }
 170
 171         /* this element is now the current element */
 172         parser->cur_elem = elem;
 173     }
 174
 175     /* scan the attributes for namespace declarations */
 176     for (prev = NULL, attr = elem->attr;
 177          attr;
 178          attr = attr->next) {
 179         if (strncmp(attr->name, APR_KW_xmlns, 5) == 0) {
 180             const char *prefix = &attr->name[5];
 181             apr_xml_ns_scope *ns_scope;
 182
 183             /* test for xmlns:foo= form and xmlns= form */
 184             if (*prefix == 0x3A) {
 185                 /* a namespace prefix declaration must have a
 186                    non-empty value. */
 187                 if (attr->value[0] == '\0') {
 188                     parser->error = APR_XML_NS_ERROR_INVALID_DECL;
 189                     return;
 190                 }
 191                 ++prefix;
 192             }
 193             else if (*prefix != '\0') {
 194                 /* advance "prev" since "attr" is still present */
 195                 prev = attr;
 196                 continue;
 197             }
 198
 199             /* quote the URI before we ever start working with it */
 200             quoted = apr_xml_quote_string(parser->p, attr->value, 1);
 201
 202             /* build and insert the new scope */
 203             ns_scope = apr_pcalloc(parser->p, sizeof(*ns_scope));
 204             ns_scope->prefix = prefix;
 205             ns_scope->ns = apr_xml_insert_uri(parser->doc->namespaces, quoted);
 206             ns_scope->emptyURI = *quoted == '\0';
 207             ns_scope->next = elem->ns_scope;
 208             elem->ns_scope = ns_scope;
 209
 210             /* remove this attribute from the element */
 211             if (prev == NULL)
 212                 elem->attr = attr->next;
 213             else
 214                 prev->next = attr->next;
 215
 216             /* Note: prev will not be advanced since we just removed "attr" */
 217         }
 218         else if (strcmp(attr->name, APR_KW_xmlns_lang) == 0) {
 219             /* save away the language (in quoted form) */
 220             elem->lang = apr_xml_quote_string(parser->p, attr->value, 1);
 221
 222             /* remove this attribute from the element */
 223             if (prev == NULL)
 224                 elem->attr = attr->next;
 225             else
 226                 prev->next = attr->next;
 227
 228             /* Note: prev will not be advanced since we just removed "attr" */
 229         }
 230         else {
 231             /* advance "prev" since "attr" is still present */
 232             prev = attr;
 233         }
 234     }
 235
 236     /*
 237     ** If an xml:lang attribute didn't exist (lang==NULL), then copy the
 238     ** language from the parent element (if present).
 239     **
 240     ** NOTE: elem_size() *depends* upon this pointer equality.
 241     */
 242     if (elem->lang == NULL && elem->parent != NULL)
 243         elem->lang = elem->parent->lang;
 244
 245     /* adjust the element's namespace */
 246     colon = strchr(elem_name, 0x3A);
 247     if (colon == NULL) {
 248         /*
 249          * The element is using the default namespace, which will always
 250          * be found. Either it will be "no namespace", or a default
 251          * namespace URI has been specified at some point.
 252          */
 253         elem->ns = find_prefix(parser, "");
 254     }
 255     else if (APR_XML_NS_IS_RESERVED(elem->name)) {
 256         elem->ns = APR_XML_NS_NONE;
 257     }
 258     else {
 259         *colon = '\0';
 260         elem->ns = find_prefix(parser, elem->name);
 261         elem->name = colon + 1;
 262
 263         if (APR_XML_NS_IS_ERROR(elem->ns)) {
 264             parser->error = elem->ns;
 265             return;
 266         }
 267     }
 268
 269     /* adjust all remaining attributes' namespaces */
 270     for (attr = elem->attr; attr; attr = attr->next) {
 271         /*
 272          * apr_xml_attr defines this as "const" but we dup'd it, so we
 273          * know that we can change it. a bit hacky, but the existing
 274          * structure def is best.
 275          */
 276         char *attr_name = (char *)attr->name;
 277
 278         colon = strchr(attr_name, 0x3A);
 279         if (colon == NULL) {
 280             /*
 281              * Attributes do NOT use the default namespace. Therefore,
 282              * we place them into the "no namespace" category.
 283              */
 284             attr->ns = APR_XML_NS_NONE;
 285         }
 286         else if (APR_XML_NS_IS_RESERVED(attr->name)) {
 287             attr->ns = APR_XML_NS_NONE;
 288         }
 289         else {
 290             *colon = '\0';
 291             attr->ns = find_prefix(parser, attr->name);
 292             attr->name = colon + 1;
 293
 294             if (APR_XML_NS_IS_ERROR(attr->ns)) {
 295                 parser->error = attr->ns;
 296                 return;
 297             }
 298         }
 299     }
 300 }
 301
 302 static void end_handler(void *userdata, const char *name)
 303 {
 304     apr_xml_parser *parser = userdata;
 305
 306     /* punt once we find an error */
 307     if (parser->error)
 308         return;
 309
 310     /* pop up one level */
 311     parser->cur_elem = parser->cur_elem->parent;
 312 }
 313
 314 static void cdata_handler(void *userdata, const char *data, int len)
 315 {
 316     apr_xml_parser *parser = userdata;
 317     apr_xml_elem *elem;
 318     apr_text_header *hdr;
 319     const char *s;
 320
 321     /* punt once we find an error */
 322     if (parser->error)
 323         return;
 324
 325     elem = parser->cur_elem;
 326     s = apr_pstrndup(parser->p, data, len);
 327
 328     if (elem->last_child == NULL) {
 329         /* no children yet. this cdata follows the start tag */
 330         hdr = &elem->first_cdata;
 331     }
 332     else {
 333         /* child elements exist. this cdata follows the last child. */
 334         hdr = &elem->last_child->following_cdata;
 335     }
 336
 337     apr_text_append(parser->p, hdr, s);
 338 }
 339
 340 static apr_status_t cleanup_parser(void *ctx)
 341 {
 342     apr_xml_parser *parser = ctx;
 343
 344     XML_ParserFree(parser->xp);
 345     parser->xp = NULL;
 346
 347     return APR_SUCCESS;
 348 }
 349
 350 APU_DECLARE(apr_xml_parser *) apr_xml_parser_create(apr_pool_t *pool)
 351 {
 352     apr_xml_parser *parser = apr_pcalloc(pool, sizeof(*parser));
 353
 354     parser->p = pool;
 355     parser->doc = apr_pcalloc(pool, sizeof(*parser->doc));
 356
 357     parser->doc->namespaces = apr_array_make(pool, 5, sizeof(const char *));
 358
 359     /* ### is there a way to avoid hard-coding this? */
 360     apr_xml_insert_uri(parser->doc->namespaces, APR_KW_DAV);
 361
 362     parser->xp = XML_ParserCreate(NULL);
 363     if (parser->xp == NULL) {
 364         (*apr_pool_abort_get(pool))(APR_ENOMEM);
 365         return NULL;
 366     }
 367
 368     apr_pool_cleanup_register(pool, parser, cleanup_parser,
 369                               apr_pool_cleanup_null);
 370
 371     XML_SetUserData(parser->xp, parser);
 372     XML_SetElementHandler(parser->xp, start_handler, end_handler);
 373     XML_SetCharacterDataHandler(parser->xp, cdata_handler);
 374
 375     return parser;
 376 }
 377
 378 static apr_status_t do_parse(apr_xml_parser *parser,
 379                              const char *data, apr_size_t len,
 380                              int is_final)
 381 {
 382     if (parser->xp == NULL) {
 383         parser->error = APR_XML_ERROR_PARSE_DONE;
 384     }
 385     else {
 386         int rv = XML_Parse(parser->xp, data, len, is_final);
 387
 388         if (rv == 0) {
 389             parser->error = APR_XML_ERROR_EXPAT;
 390             parser->xp_err = XML_GetErrorCode(parser->xp);
 391         }
 392     }
 393
 394     /* ### better error code? */
 395     return parser->error ? APR_EGENERAL : APR_SUCCESS;
 396 }
 397
 398 APU_DECLARE(apr_status_t) apr_xml_parser_feed(apr_xml_parser *parser,
 399                                               const char *data,
 400                                               apr_size_t len)
 401 {
 402     return do_parse(parser, data, len, 0 /* is_final */);
 403 }
 404
 405 APU_DECLARE(apr_status_t) apr_xml_parser_done(apr_xml_parser *parser,
 406                                               apr_xml_doc **pdoc)
 407 {
 408     char end;
 409     apr_status_t status = do_parse(parser, &end, 0, 1 /* is_final */);
 410
 411     /* get rid of the parser */
 412     (void) apr_pool_cleanup_run(parser->p, parser, cleanup_parser);
 413
 414     if (status)
 415         return status;
 416
 417     if (pdoc != NULL)
 418         *pdoc = parser->doc;
 419     return APR_SUCCESS;
 420 }
 421
 422 APU_DECLARE(char *) apr_xml_parser_geterror(apr_xml_parser *parser,
 423                                             char *errbuf,
 424                                             apr_size_t errbufsize)
 425 {
 426     int error = parser->error;
 427     const char *msg;
 428
 429     /* clear our record of an error */
 430     parser->error = 0;
 431
 432     switch (error) {
 433     case 0:
 434         msg = "No error.";
 435         break;
 436
 437     case APR_XML_NS_ERROR_UNKNOWN_PREFIX:
 438         msg = "An undefined namespace prefix was used.";
 439         break;
 440
 441     case APR_XML_NS_ERROR_INVALID_DECL:
 442         msg = "A namespace prefix was defined with an empty URI.";
 443         break;
 444
 445     case APR_XML_ERROR_EXPAT:
 446         (void) apr_snprintf(errbuf, errbufsize,
 447                             "XML parser error code: %s (%d)",
 448                             XML_ErrorString(parser->xp_err), parser->xp_err);
 449         return errbuf;
 450
 451     case APR_XML_ERROR_PARSE_DONE:
 452         msg = "The parser is not active.";
 453         break;
 454
 455     default:
 456         msg = "There was an unknown error within the XML body.";
 457         break;
 458     }
 459
 460     (void) apr_cpystrn(errbuf, msg, errbufsize);
 461     return errbuf;
 462 }
 463
 464 APU_DECLARE(apr_status_t) apr_xml_parse_file(apr_pool_t *p,
 465                                              apr_xml_parser **parser,
 466                                              apr_xml_doc **ppdoc,
 467                                              apr_file_t *xmlfd,
 468                                              apr_size_t buffer_length)
 469 {
 470     apr_status_t rv;
 471     char *buffer;
 472     apr_size_t length;
 473
 474     *parser = apr_xml_parser_create(p);
 475     if (*parser == NULL) {
 476         /* FIXME: returning an error code would be nice,
 477          * but we dont get one ;( */
 478         return APR_EGENERAL;
 479     }
 480     buffer = apr_palloc(p, buffer_length);
 481     length = buffer_length;
 482
 483     rv = apr_file_read(xmlfd, buffer, &length);
 484
 485     while (rv == APR_SUCCESS) {
 486         rv = apr_xml_parser_feed(*parser, buffer, length);
 487         if (rv != APR_SUCCESS) {
 488             return rv;
 489         }
 490
 491         length = buffer_length;
 492         rv = apr_file_read(xmlfd, buffer, &length);
 493     }
 494     if (rv != APR_EOF) {
 495         return rv;
 496     }
 497     rv = apr_xml_parser_done(*parser, ppdoc);
 498     *parser = NULL;
 499     return rv;
 500 }
 501
 502 APU_DECLARE(void) apr_text_append(apr_pool_t * p, apr_text_header *hdr,
 503                                   const char *text)
 504 {
 505     apr_text *t = apr_palloc(p, sizeof(*t));
 506
 507     t->text = text;
 508     t->next = NULL;
 509
 510     if (hdr->first == NULL) {
 511         /* no text elements yet */
 512         hdr->first = hdr->last = t;
 513     }
 514     else {
 515         /* append to the last text element */
 516         hdr->last->next = t;
 517         hdr->last = t;
 518     }
 519 }
 520
 521
 522 /* ---------------------------------------------------------------
 523 **
 524 ** XML UTILITY FUNCTIONS
 525 */
 526
 527 /*
 528 ** apr_xml_quote_string: quote an XML string
 529 **
 530 ** Replace '<', '>', and '&' with '&lt;', '&gt;', and '&amp;'.
 531 ** If quotes is true, then replace '"' with '&quot;'.
 532 **
 533 ** quotes is typically set to true for XML strings that will occur within
 534 ** double quotes -- attribute values.
 535 */
 536 APU_DECLARE(const char *) apr_xml_quote_string(apr_pool_t *p, const char *s,
 537                                                int quotes)
 538 {
 539     const char *scan;
 540     apr_size_t len = 0;
 541     apr_size_t extra = 0;
 542     char *qstr;
 543     char *qscan;
 544     char c;
 545
 546     for (scan = s; (c = *scan) != '\0'; ++scan, ++len) {
 547         if (c == '<' || c == '>')
 548             extra += 3;         /* &lt; or &gt; */
 549         else if (c == '&')
 550             extra += 4;         /* &amp; */
 551         else if (quotes && c == '"')
 552             extra += 5;         /* &quot; */
 553     }
 554
 555     /* nothing to do? */
 556     if (extra == 0)
 557         return s;
 558
 559     qstr = apr_palloc(p, len + extra + 1);
 560     for (scan = s, qscan = qstr; (c = *scan) != '\0'; ++scan) {
 561         if (c == '<') {
 562             *qscan++ = '&';
 563             *qscan++ = 'l';
 564             *qscan++ = 't';
 565             *qscan++ = ';';
 566         }
 567         else if (c == '>') {
 568             *qscan++ = '&';
 569             *qscan++ = 'g';
 570             *qscan++ = 't';
 571             *qscan++ = ';';
 572         }
 573         else if (c == '&') {
 574             *qscan++ = '&';
 575             *qscan++ = 'a';
 576             *qscan++ = 'm';
 577             *qscan++ = 'p';
 578             *qscan++ = ';';
 579         }
 580         else if (quotes && c == '"') {
 581             *qscan++ = '&';
 582             *qscan++ = 'q';
 583             *qscan++ = 'u';
 584             *qscan++ = 'o';
 585             *qscan++ = 't';
 586             *qscan++ = ';';
 587         }
 588         else {
 589             *qscan++ = c;
 590         }
 591     }
 592
 593     *qscan = '\0';
 594     return qstr;
 595 }
 596
 597 /* how many characters for the given integer? */
 598 #define APR_XML_NS_LEN(ns) ((ns) < 10 ? 1 : (ns) < 100 ? 2 : (ns) < 1000 ? 3 : \
 599                             (ns) < 10000 ? 4 : (ns) < 100000 ? 5 : \
 600                             (ns) < 1000000 ? 6 : (ns) < 10000000 ? 7 : \
 601                             (ns) < 100000000 ? 8 : (ns) < 1000000000 ? 9 : 10)
 602
 603 static apr_size_t text_size(const apr_text *t)
 604 {
 605     apr_size_t size = 0;
 606
 607     for (; t; t = t->next)
 608         size += strlen(t->text);
 609     return size;
 610 }
 611
 612 static apr_size_t elem_size(const apr_xml_elem *elem, int style,
 613                             apr_array_header_t *namespaces, int *ns_map)
 614 {
 615     apr_size_t size;
 616
 617     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 618         const apr_xml_attr *attr;
 619
 620         size = 0;
 621
 622         if (style == APR_XML_X2T_FULL_NS_LANG) {
 623             int i;
 624
 625             /*
 626             ** The outer element will contain xmlns:ns%d="%s" attributes
 627             ** and an xml:lang attribute, if applicable.
 628             */
 629
 630             for (i = namespaces->nelts; i--;) {
 631                 /* compute size of: ' xmlns:ns%d="%s"' */
 632                 size += (9 + APR_XML_NS_LEN(i) + 2 +
 633                          strlen(APR_XML_GET_URI_ITEM(namespaces, i)) + 1);
 634             }
 635
 636             if (elem->lang != NULL) {
 637                 /* compute size of: ' xml:lang="%s"' */
 638                 size += 11 + strlen(elem->lang) + 1;
 639             }
 640         }
 641
 642         if (elem->ns == APR_XML_NS_NONE) {
 643             /* compute size of: <%s> */
 644             size += 1 + strlen(elem->name) + 1;
 645         }
 646         else {
 647             int ns = ns_map ? ns_map[elem->ns] : elem->ns;
 648
 649             /* compute size of: <ns%d:%s> */
 650             size += 3 + APR_XML_NS_LEN(ns) + 1 + strlen(elem->name) + 1;
 651         }
 652
 653         if (APR_XML_ELEM_IS_EMPTY(elem)) {
 654             /* insert a closing "/" */
 655             size += 1;
 656         }
 657         else {
 658             /*
 659              * two of above plus "/":
 660              *     <ns%d:%s> ... </ns%d:%s>
 661              * OR  <%s> ... </%s>
 662              */
 663             size = 2 * size + 1;
 664         }
 665
 666         for (attr = elem->attr; attr; attr = attr->next) {
 667             if (attr->ns == APR_XML_NS_NONE) {
 668                 /* compute size of: ' %s="%s"' */
 669                 size += 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
 670             }
 671             else {
 672                 /* compute size of: ' ns%d:%s="%s"' */
 673                 int ns = ns_map ? ns_map[attr->ns] : attr->ns;
 674                 size += 3 + APR_XML_NS_LEN(ns) + 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
 675             }
 676         }
 677
 678         /*
 679         ** If the element has an xml:lang value that is *different* from
 680         ** its parent, then add the thing in: ' xml:lang="%s"'.
 681         **
 682         ** NOTE: we take advantage of the pointer equality established by
 683         ** the parsing for "inheriting" the xml:lang values from parents.
 684         */
 685         if (elem->lang != NULL &&
 686             (elem->parent == NULL || elem->lang != elem->parent->lang)) {
 687             size += 11 + strlen(elem->lang) + 1;
 688         }
 689     }
 690     else if (style == APR_XML_X2T_LANG_INNER) {
 691         /*
 692          * This style prepends the xml:lang value plus a null terminator.
 693          * If a lang value is not present, then we insert a null term.
 694          */
 695         size = elem->lang ? strlen(elem->lang) + 1 : 1;
 696     }
 697     else
 698         size = 0;
 699
 700     size += text_size(elem->first_cdata.first);
 701
 702     for (elem = elem->first_child; elem; elem = elem->next) {
 703         /* the size of the child element plus the CDATA that follows it */
 704         size += (elem_size(elem, APR_XML_X2T_FULL, NULL, ns_map) +
 705                  text_size(elem->following_cdata.first));
 706     }
 707
 708     return size;
 709 }
 710
 711 static char *write_text(char *s, const apr_text *t)
 712 {
 713     for (; t; t = t->next) {
 714         apr_size_t len = strlen(t->text);
 715         memcpy(s, t->text, len);
 716         s += len;
 717     }
 718     return s;
 719 }
 720
 721 static char *write_elem(char *s, const apr_xml_elem *elem, int style,
 722                         apr_array_header_t *namespaces, int *ns_map)
 723 {
 724     const apr_xml_elem *child;
 725     apr_size_t len;
 726     int ns;
 727
 728     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 729         int empty = APR_XML_ELEM_IS_EMPTY(elem);
 730         const apr_xml_attr *attr;
 731
 732         if (elem->ns == APR_XML_NS_NONE) {
 733             len = sprintf(s, "<%s", elem->name);
 734         }
 735         else {
 736             ns = ns_map ? ns_map[elem->ns] : elem->ns;
 737             len = sprintf(s, "<ns%d:%s", ns, elem->name);
 738         }
 739         s += len;
 740
 741         for (attr = elem->attr; attr; attr = attr->next) {
 742             if (attr->ns == APR_XML_NS_NONE)
 743                 len = sprintf(s, " %s=\"%s\"", attr->name, attr->value);
 744             else {
 745                 ns = ns_map ? ns_map[attr->ns] : attr->ns;
 746                 len = sprintf(s, " ns%d:%s=\"%s\"", ns, attr->name, attr->value);
 747             }
 748             s += len;
 749         }
 750
 751         /* add the xml:lang value if necessary */
 752         if (elem->lang != NULL &&
 753             (style == APR_XML_X2T_FULL_NS_LANG ||
 754              elem->parent == NULL ||
 755              elem->lang != elem->parent->lang)) {
 756             len = sprintf(s, " xml:lang=\"%s\"", elem->lang);
 757             s += len;
 758         }
 759
 760         /* add namespace definitions, if required */
 761         if (style == APR_XML_X2T_FULL_NS_LANG) {
 762             int i;
 763
 764             for (i = namespaces->nelts; i--;) {
 765                 len = sprintf(s, " xmlns:ns%d=\"%s\"", i,
 766                               APR_XML_GET_URI_ITEM(namespaces, i));
 767                 s += len;
 768             }
 769         }
 770
 771         /* no more to do. close it up and go. */
 772         if (empty) {
 773             *s++ = '/';
 774             *s++ = '>';
 775             return s;
 776         }
 777
 778         /* just close it */
 779         *s++ = '>';
 780     }
 781     else if (style == APR_XML_X2T_LANG_INNER) {
 782         /* prepend the xml:lang value */
 783         if (elem->lang != NULL) {
 784             len = strlen(elem->lang);
 785             memcpy(s, elem->lang, len);
 786             s += len;
 787         }
 788         *s++ = '\0';
 789     }
 790
 791     s = write_text(s, elem->first_cdata.first);
 792
 793     for (child = elem->first_child; child; child = child->next) {
 794         s = write_elem(s, child, APR_XML_X2T_FULL, NULL, ns_map);
 795         s = write_text(s, child->following_cdata.first);
 796     }
 797
 798     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 799         if (elem->ns == APR_XML_NS_NONE) {
 800             len = sprintf(s, "</%s>", elem->name);
 801         }
 802         else {
 803             ns = ns_map ? ns_map[elem->ns] : elem->ns;
 804             len = sprintf(s, "</ns%d:%s>", ns, elem->name);
 805         }
 806         s += len;
 807     }
 808
 809     return s;
 810 }
 811
 812 APU_DECLARE(void) apr_xml_quote_elem(apr_pool_t *p, apr_xml_elem *elem)
 813 {
 814     apr_text *scan_txt;
 815     apr_xml_attr *scan_attr;
 816     apr_xml_elem *scan_elem;
 817
 818     /* convert the element's text */
 819     for (scan_txt = elem->first_cdata.first;
 820          scan_txt != NULL;
 821          scan_txt = scan_txt->next) {
 822         scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
 823     }
 824     for (scan_txt = elem->following_cdata.first;
 825          scan_txt != NULL;
 826          scan_txt = scan_txt->next) {
 827         scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
 828     }
 829
 830     /* convert the attribute values */
 831     for (scan_attr = elem->attr;
 832          scan_attr != NULL;
 833          scan_attr = scan_attr->next) {
 834         scan_attr->value = apr_xml_quote_string(p, scan_attr->value, 1);
 835     }
 836
 837     /* convert the child elements */
 838     for (scan_elem = elem->first_child;
 839          scan_elem != NULL;
 840          scan_elem = scan_elem->next) {
 841         apr_xml_quote_elem(p, scan_elem);
 842     }
 843 }
 844
 845 /* convert an element to a text string */
 846 APU_DECLARE(void) apr_xml_to_text(apr_pool_t * p, const apr_xml_elem *elem,
 847                                   int style, apr_array_header_t *namespaces,
 848                                   int *ns_map, const char **pbuf,
 849                                   apr_size_t *psize)
 850 {
 851     /* get the exact size, plus a null terminator */
 852     apr_size_t size = elem_size(elem, style, namespaces, ns_map) + 1;
 853     char *s = apr_palloc(p, size);
 854
 855     (void) write_elem(s, elem, style, namespaces, ns_map);
 856     s[size - 1] = '\0';
 857
 858     *pbuf = s;
 859     if (psize)
 860         *psize = size;
 861 }
 862
 863 APU_DECLARE(const char *) apr_xml_empty_elem(apr_pool_t * p,
 864                                              const apr_xml_elem *elem)
 865 {
 866     if (elem->ns == APR_XML_NS_NONE) {
 867         /*
 868          * The prefix (xml...) is already within the prop name, or
 869          * the element simply has no prefix.
 870          */
 871         return apr_psprintf(p, "<%s/>" DEBUG_CR, elem->name);
 872     }
 873
 874     return apr_psprintf(p, "<ns%d:%s/>" DEBUG_CR, elem->ns, elem->name);
 875 }
 876
 877 /* return the URI's (existing) index, or insert it and return a new index */
 878 APU_DECLARE(int) apr_xml_insert_uri(apr_array_header_t *uri_array,
 879                                     const char *uri)
 880 {
 881     int i;
 882     const char **pelt;
 883
 884     /* never insert an empty URI; this index is always APR_XML_NS_NONE */
 885     if (*uri == '\0')
 886         return APR_XML_NS_NONE;
 887
 888     for (i = uri_array->nelts; i--;) {
 889         if (strcmp(uri, APR_XML_GET_URI_ITEM(uri_array, i)) == 0)
 890             return i;
 891     }
 892
 893     pelt = apr_array_push(uri_array);
 894     *pelt = uri;                /* assume uri is const or in a pool */
 895     return uri_array->nelts - 1;
 896 }
 897
 898 /* convert the element to EBCDIC */
 899 #if APR_CHARSET_EBCDIC
 900 static apr_status_t apr_xml_parser_convert_elem(apr_xml_elem *e,
 901                                                 apr_xlate_t *convset)
 902 {
 903     apr_xml_attr *a;
 904     apr_xml_elem *ec;
 905     apr_text *t;
 906     apr_size_t inbytes_left, outbytes_left;
 907     apr_status_t status;
 908
 909     inbytes_left = outbytes_left = strlen(e->name);
 910     status = apr_xlate_conv_buffer(convset, e->name,  &inbytes_left, (char *) e->name, &outbytes_left);
 911     if (status) {
 912         return status;
 913     }
 914
 915     for (t = e->first_cdata.first; t != NULL; t = t->next) {
 916         inbytes_left = outbytes_left = strlen(t->text);
 917         status = apr_xlate_conv_buffer(convset, t->text, &inbytes_left, (char *) t->text, &outbytes_left);
 918         if (status) {
 919             return status;
 920         }
 921     }
 922
 923     for (t = e->following_cdata.first;  t != NULL; t = t->next) {
 924         inbytes_left = outbytes_left = strlen(t->text);
 925         status = apr_xlate_conv_buffer(convset, t->text, &inbytes_left, (char *) t->text, &outbytes_left);
 926         if (status) {
 927             return status;
 928         }
 929     }
 930
 931     for (a = e->attr; a != NULL; a = a->next) {
 932         inbytes_left = outbytes_left = strlen(a->name);
 933         status = apr_xlate_conv_buffer(convset, a->name, &inbytes_left, (char *) a->name, &outbytes_left);
 934         if (status) {
 935             return status;
 936         }
 937         inbytes_left = outbytes_left = strlen(a->value);
 938         status = apr_xlate_conv_buffer(convset, a->value, &inbytes_left, (char *) a->value, &outbytes_left);
 939         if (status) {
 940             return status;
 941         }
 942     }
 943
 944     for (ec = e->first_child; ec != NULL; ec = ec->next) {
 945         status = apr_xml_parser_convert_elem(ec, convset);
 946         if (status) {
 947             return status;
 948         }
 949     }
 950     return APR_SUCCESS;
 951 }
 952
 953 /* convert the whole document to EBCDIC */
 954 APU_DECLARE(apr_status_t) apr_xml_parser_convert_doc(apr_pool_t *pool,
 955                                                      apr_xml_doc *pdoc,
 956                                                      apr_xlate_t *convset)
 957 {
 958     apr_status_t status;
 959     /* Don't convert the namespaces: they are constant! */
 960     if (pdoc->namespaces != NULL) {
 961         int i;
 962         apr_array_header_t *namespaces;
 963         namespaces = apr_array_make(pool, pdoc->namespaces->nelts, sizeof(const char *));
 964         if (namespaces == NULL)
 965             return APR_ENOMEM;
 966         for (i = 0; i < pdoc->namespaces->nelts; i++) {
 967             apr_size_t inbytes_left, outbytes_left;
 968             char *ptr = (char *) APR_XML_GET_URI_ITEM(pdoc->namespaces, i);
 969             ptr = apr_pstrdup(pool, ptr);
 970             if ( ptr == NULL)
 971                 return APR_ENOMEM;
 972             inbytes_left = outbytes_left = strlen(ptr);
 973             status = apr_xlate_conv_buffer(convset, ptr, &inbytes_left, ptr, &outbytes_left);
 974             if (status) {
 975                 return status;
 976             }
 977             apr_xml_insert_uri(namespaces, ptr);
 978         }
 979         pdoc->namespaces = namespaces;
 980     }
 981     return apr_xml_parser_convert_elem(pdoc->root, convset);
 982 }
 983 #endif