xml/apr_xml.c

   1 /* Copyright 2000-2005 The Apache Software Foundation or its licensors, as
   2  * applicable.
   3  *
   4  * Licensed under the Apache License, Version 2.0 (the "License");
   5  * you may not use this file except in compliance with the License.
   6  * 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 #ifdef APR_HAVE_OLD_EXPAT
  29 #include "xmlparse.h"
  30 #else
  31 #include "expat.h"
  32 #endif
  33
  34 #define DEBUG_CR "\r\n"
  35
  36 static const char APR_KW_xmlns[] = { 0x78, 0x6D, 0x6C, 0x6E, 0x73, '\0' };
  37 static const char APR_KW_xmlns_lang[] = { 0x78, 0x6D, 0x6C, 0x3A, 0x6C, 0x61, 0x6E, 0x67, '\0' };
  38 static const char APR_KW_DAV[] = { 0x44, 0x41, 0x56, 0x3A, '\0' };
  39
  40 /* errors related to namespace processing */
  41 #define APR_XML_NS_ERROR_UNKNOWN_PREFIX (-1000)
  42 #define APR_XML_NS_ERROR_INVALID_DECL (-1001)
  43
  44 /* test for a namespace prefix that begins with [Xx][Mm][Ll] */
  45 #define APR_XML_NS_IS_RESERVED(name) \
  46         ( (name[0] == 0x58 || name[0] == 0x78) && \
  47           (name[1] == 0x4D || name[1] == 0x6D) && \
  48           (name[2] == 0x4C || name[2] == 0x6C) )
  49
  50
  51 /* the real (internal) definition of the parser context */
  52 struct apr_xml_parser {
  53     apr_xml_doc *doc;           /* the doc we're parsing */
  54     apr_pool_t *p;              /* the pool we allocate from */
  55     apr_xml_elem *cur_elem;     /* current element */
  56
  57     int error;                  /* an error has occurred */
  58 #define APR_XML_ERROR_EXPAT             1
  59 #define APR_XML_ERROR_PARSE_DONE        2
  60 /* also: public APR_XML_NS_ERROR_* values (if any) */
  61
  62     XML_Parser xp;              /* the actual (Expat) XML parser */
  63     enum XML_Error xp_err;      /* stored Expat error code */
  64 };
  65
  66 /* struct for scoping namespace declarations */
  67 typedef struct apr_xml_ns_scope {
  68     const char *prefix;         /* prefix used for this ns */
  69     int ns;                     /* index into namespace table */
  70     int emptyURI;               /* the namespace URI is the empty string */
  71     struct apr_xml_ns_scope *next;      /* next scoped namespace */
  72 } apr_xml_ns_scope;
  73
  74
  75 /* return namespace table index for a given prefix */
  76 static int find_prefix(apr_xml_parser *parser, const char *prefix)
  77 {
  78     apr_xml_elem *elem = parser->cur_elem;
  79
  80     /*
  81     ** Walk up the tree, looking for a namespace scope that defines this
  82     ** prefix.
  83     */
  84     for (; elem; elem = elem->parent) {
  85         apr_xml_ns_scope *ns_scope = elem->ns_scope;
  86
  87         for (ns_scope = elem->ns_scope; ns_scope; ns_scope = ns_scope->next) {
  88             if (strcmp(prefix, ns_scope->prefix) == 0) {
  89                 if (ns_scope->emptyURI) {
  90                     /*
  91                     ** It is possible to set the default namespace to an
  92                     ** empty URI string; this resets the default namespace
  93                     ** to mean "no namespace." We just found the prefix
  94                     ** refers to an empty URI, so return "no namespace."
  95                     */
  96                     return APR_XML_NS_NONE;
  97                 }
  98
  99                 return ns_scope->ns;
 100             }
 101         }
 102     }
 103
 104     /*
 105      * If the prefix is empty (""), this means that a prefix was not
 106      * specified in the element/attribute. The search that was performed
 107      * just above did not locate a default namespace URI (which is stored
 108      * into ns_scope with an empty prefix). This means the element/attribute
 109      * has "no namespace". We have a reserved value for this.
 110      */
 111     if (*prefix == '\0') {
 112         return APR_XML_NS_NONE;
 113     }
 114
 115     /* not found */
 116     return APR_XML_NS_ERROR_UNKNOWN_PREFIX;
 117 }
 118
 119 static void start_handler(void *userdata, const char *name, const char **attrs)
 120 {
 121     apr_xml_parser *parser = userdata;
 122     apr_xml_elem *elem;
 123     apr_xml_attr *attr;
 124     apr_xml_attr *prev;
 125     char *colon;
 126     const char *quoted;
 127     char *elem_name;
 128
 129     /* punt once we find an error */
 130     if (parser->error)
 131         return;
 132
 133     elem = apr_pcalloc(parser->p, sizeof(*elem));
 134
 135     /* prep the element */
 136     elem->name = elem_name = apr_pstrdup(parser->p, name);
 137
 138     /* fill in the attributes (note: ends up in reverse order) */
 139     while (*attrs) {
 140         attr = apr_palloc(parser->p, sizeof(*attr));
 141         attr->name = apr_pstrdup(parser->p, *attrs++);
 142         attr->value = apr_pstrdup(parser->p, *attrs++);
 143         attr->next = elem->attr;
 144         elem->attr = attr;
 145     }
 146
 147     /* hook the element into the tree */
 148     if (parser->cur_elem == NULL) {
 149         /* no current element; this also becomes the root */
 150         parser->cur_elem = parser->doc->root = elem;
 151     }
 152     else {
 153         /* this element appeared within the current elem */
 154         elem->parent = parser->cur_elem;
 155
 156         /* set up the child/sibling links */
 157         if (elem->parent->last_child == NULL) {
 158             /* no first child either */
 159             elem->parent->first_child = elem->parent->last_child = elem;
 160         }
 161         else {
 162             /* hook onto the end of the parent's children */
 163             elem->parent->last_child->next = elem;
 164             elem->parent->last_child = elem;
 165         }
 166
 167         /* this element is now the current element */
 168         parser->cur_elem = elem;
 169     }
 170
 171     /* scan the attributes for namespace declarations */
 172     for (prev = NULL, attr = elem->attr;
 173          attr;
 174          attr = attr->next) {
 175         if (strncmp(attr->name, APR_KW_xmlns, 5) == 0) {
 176             const char *prefix = &attr->name[5];
 177             apr_xml_ns_scope *ns_scope;
 178
 179             /* test for xmlns:foo= form and xmlns= form */
 180             if (*prefix == 0x3A) {
 181                 /* a namespace prefix declaration must have a
 182                    non-empty value. */
 183                 if (attr->value[0] == '\0') {
 184                     parser->error = APR_XML_NS_ERROR_INVALID_DECL;
 185                     return;
 186                 }
 187                 ++prefix;
 188             }
 189             else if (*prefix != '\0') {
 190                 /* advance "prev" since "attr" is still present */
 191                 prev = attr;
 192                 continue;
 193             }
 194
 195             /* quote the URI before we ever start working with it */
 196             quoted = apr_xml_quote_string(parser->p, attr->value, 1);
 197
 198             /* build and insert the new scope */
 199             ns_scope = apr_pcalloc(parser->p, sizeof(*ns_scope));
 200             ns_scope->prefix = prefix;
 201             ns_scope->ns = apr_xml_insert_uri(parser->doc->namespaces, quoted);
 202             ns_scope->emptyURI = *quoted == '\0';
 203             ns_scope->next = elem->ns_scope;
 204             elem->ns_scope = ns_scope;
 205
 206             /* remove this attribute from the element */
 207             if (prev == NULL)
 208                 elem->attr = attr->next;
 209             else
 210                 prev->next = attr->next;
 211
 212             /* Note: prev will not be advanced since we just removed "attr" */
 213         }
 214         else if (strcmp(attr->name, APR_KW_xmlns_lang) == 0) {
 215             /* save away the language (in quoted form) */
 216             elem->lang = apr_xml_quote_string(parser->p, attr->value, 1);
 217
 218             /* remove this attribute from the element */
 219             if (prev == NULL)
 220                 elem->attr = attr->next;
 221             else
 222                 prev->next = attr->next;
 223
 224             /* Note: prev will not be advanced since we just removed "attr" */
 225         }
 226         else {
 227             /* advance "prev" since "attr" is still present */
 228             prev = attr;
 229         }
 230     }
 231
 232     /*
 233     ** If an xml:lang attribute didn't exist (lang==NULL), then copy the
 234     ** language from the parent element (if present).
 235     **
 236     ** NOTE: elem_size() *depends* upon this pointer equality.
 237     */
 238     if (elem->lang == NULL && elem->parent != NULL)
 239         elem->lang = elem->parent->lang;
 240
 241     /* adjust the element's namespace */
 242     colon = strchr(elem_name, 0x3A);
 243     if (colon == NULL) {
 244         /*
 245          * The element is using the default namespace, which will always
 246          * be found. Either it will be "no namespace", or a default
 247          * namespace URI has been specified at some point.
 248          */
 249         elem->ns = find_prefix(parser, "");
 250     }
 251     else if (APR_XML_NS_IS_RESERVED(elem->name)) {
 252         elem->ns = APR_XML_NS_NONE;
 253     }
 254     else {
 255         *colon = '\0';
 256         elem->ns = find_prefix(parser, elem->name);
 257         elem->name = colon + 1;
 258
 259         if (APR_XML_NS_IS_ERROR(elem->ns)) {
 260             parser->error = elem->ns;
 261             return;
 262         }
 263     }
 264
 265     /* adjust all remaining attributes' namespaces */
 266     for (attr = elem->attr; attr; attr = attr->next) {
 267         /*
 268          * apr_xml_attr defines this as "const" but we dup'd it, so we
 269          * know that we can change it. a bit hacky, but the existing
 270          * structure def is best.
 271          */
 272         char *attr_name = (char *)attr->name;
 273
 274         colon = strchr(attr_name, 0x3A);
 275         if (colon == NULL) {
 276             /*
 277              * Attributes do NOT use the default namespace. Therefore,
 278              * we place them into the "no namespace" category.
 279              */
 280             attr->ns = APR_XML_NS_NONE;
 281         }
 282         else if (APR_XML_NS_IS_RESERVED(attr->name)) {
 283             attr->ns = APR_XML_NS_NONE;
 284         }
 285         else {
 286             *colon = '\0';
 287             attr->ns = find_prefix(parser, attr->name);
 288             attr->name = colon + 1;
 289
 290             if (APR_XML_NS_IS_ERROR(attr->ns)) {
 291                 parser->error = attr->ns;
 292                 return;
 293             }
 294         }
 295     }
 296 }
 297
 298 static void end_handler(void *userdata, const char *name)
 299 {
 300     apr_xml_parser *parser = userdata;
 301
 302     /* punt once we find an error */
 303     if (parser->error)
 304         return;
 305
 306     /* pop up one level */
 307     parser->cur_elem = parser->cur_elem->parent;
 308 }
 309
 310 static void cdata_handler(void *userdata, const char *data, int len)
 311 {
 312     apr_xml_parser *parser = userdata;
 313     apr_xml_elem *elem;
 314     apr_text_header *hdr;
 315     const char *s;
 316
 317     /* punt once we find an error */
 318     if (parser->error)
 319         return;
 320
 321     elem = parser->cur_elem;
 322     s = apr_pstrndup(parser->p, data, len);
 323
 324     if (elem->last_child == NULL) {
 325         /* no children yet. this cdata follows the start tag */
 326         hdr = &elem->first_cdata;
 327     }
 328     else {
 329         /* child elements exist. this cdata follows the last child. */
 330         hdr = &elem->last_child->following_cdata;
 331     }
 332
 333     apr_text_append(parser->p, hdr, s);
 334 }
 335
 336 static apr_status_t cleanup_parser(void *ctx)
 337 {
 338     apr_xml_parser *parser = ctx;
 339
 340     XML_ParserFree(parser->xp);
 341     parser->xp = NULL;
 342
 343     return APR_SUCCESS;
 344 }
 345
 346 APU_DECLARE(apr_xml_parser *) apr_xml_parser_create(apr_pool_t *pool)
 347 {
 348     apr_xml_parser *parser = apr_pcalloc(pool, sizeof(*parser));
 349
 350     parser->p = pool;
 351     parser->doc = apr_pcalloc(pool, sizeof(*parser->doc));
 352
 353     parser->doc->namespaces = apr_array_make(pool, 5, sizeof(const char *));
 354
 355     /* ### is there a way to avoid hard-coding this? */
 356     apr_xml_insert_uri(parser->doc->namespaces, APR_KW_DAV);
 357
 358     parser->xp = XML_ParserCreate(NULL);
 359     if (parser->xp == NULL) {
 360         (*apr_pool_abort_get(pool))(APR_ENOMEM);
 361         return NULL;
 362     }
 363
 364     apr_pool_cleanup_register(pool, parser, cleanup_parser,
 365                               apr_pool_cleanup_null);
 366
 367     XML_SetUserData(parser->xp, parser);
 368     XML_SetElementHandler(parser->xp, start_handler, end_handler);
 369     XML_SetCharacterDataHandler(parser->xp, cdata_handler);
 370
 371     return parser;
 372 }
 373
 374 static apr_status_t do_parse(apr_xml_parser *parser,
 375                              const char *data, apr_size_t len,
 376                              int is_final)
 377 {
 378     if (parser->xp == NULL) {
 379         parser->error = APR_XML_ERROR_PARSE_DONE;
 380     }
 381     else {
 382         int rv = XML_Parse(parser->xp, data, len, is_final);
 383
 384         if (rv == 0) {
 385             parser->error = APR_XML_ERROR_EXPAT;
 386             parser->xp_err = XML_GetErrorCode(parser->xp);
 387         }
 388     }
 389
 390     /* ### better error code? */
 391     return parser->error ? APR_EGENERAL : APR_SUCCESS;
 392 }
 393
 394 APU_DECLARE(apr_status_t) apr_xml_parser_feed(apr_xml_parser *parser,
 395                                               const char *data,
 396                                               apr_size_t len)
 397 {
 398     return do_parse(parser, data, len, 0 /* is_final */);
 399 }
 400
 401 APU_DECLARE(apr_status_t) apr_xml_parser_done(apr_xml_parser *parser,
 402                                               apr_xml_doc **pdoc)
 403 {
 404     char end;
 405     apr_status_t status = do_parse(parser, &end, 0, 1 /* is_final */);
 406
 407     /* get rid of the parser */
 408     (void) apr_pool_cleanup_run(parser->p, parser, cleanup_parser);
 409
 410     if (status)
 411         return status;
 412
 413     if (pdoc != NULL)
 414         *pdoc = parser->doc;
 415     return APR_SUCCESS;
 416 }
 417
 418 APU_DECLARE(char *) apr_xml_parser_geterror(apr_xml_parser *parser,
 419                                             char *errbuf,
 420                                             apr_size_t errbufsize)
 421 {
 422     int error = parser->error;
 423     const char *msg;
 424
 425     /* clear our record of an error */
 426     parser->error = 0;
 427
 428     switch (error) {
 429     case 0:
 430         msg = "No error.";
 431         break;
 432
 433     case APR_XML_NS_ERROR_UNKNOWN_PREFIX:
 434         msg = "An undefined namespace prefix was used.";
 435         break;
 436
 437     case APR_XML_NS_ERROR_INVALID_DECL:
 438         msg = "A namespace prefix was defined with an empty URI.";
 439         break;
 440
 441     case APR_XML_ERROR_EXPAT:
 442         (void) apr_snprintf(errbuf, errbufsize,
 443                             "XML parser error code: %s (%d)",
 444                             XML_ErrorString(parser->xp_err), parser->xp_err);
 445         return errbuf;
 446
 447     case APR_XML_ERROR_PARSE_DONE:
 448         msg = "The parser is not active.";
 449         break;
 450
 451     default:
 452         msg = "There was an unknown error within the XML body.";
 453         break;
 454     }
 455
 456     (void) apr_cpystrn(errbuf, msg, errbufsize);
 457     return errbuf;
 458 }
 459
 460 APU_DECLARE(apr_status_t) apr_xml_parse_file(apr_pool_t *p,
 461                                              apr_xml_parser **parser,
 462                                              apr_xml_doc **ppdoc,
 463                                              apr_file_t *xmlfd,
 464                                              apr_size_t buffer_length)
 465 {
 466     apr_status_t rv;
 467     char *buffer;
 468     apr_size_t length;
 469
 470     *parser = apr_xml_parser_create(p);
 471     if (*parser == NULL) {
 472         /* FIXME: returning an error code would be nice,
 473          * but we dont get one ;( */
 474         return APR_EGENERAL;
 475     }
 476     buffer = apr_palloc(p, buffer_length);
 477     length = buffer_length;
 478
 479     rv = apr_file_read(xmlfd, buffer, &length);
 480
 481     while (rv == APR_SUCCESS) {
 482         rv = apr_xml_parser_feed(*parser, buffer, length);
 483         if (rv != APR_SUCCESS) {
 484             return rv;
 485         }
 486
 487         length = buffer_length;
 488         rv = apr_file_read(xmlfd, buffer, &length);
 489     }
 490     if (rv != APR_EOF) {
 491         return rv;
 492     }
 493     rv = apr_xml_parser_done(*parser, ppdoc);
 494     *parser = NULL;
 495     return rv;
 496 }
 497
 498 APU_DECLARE(void) apr_text_append(apr_pool_t * p, apr_text_header *hdr,
 499                                   const char *text)
 500 {
 501     apr_text *t = apr_palloc(p, sizeof(*t));
 502
 503     t->text = text;
 504     t->next = NULL;
 505
 506     if (hdr->first == NULL) {
 507         /* no text elements yet */
 508         hdr->first = hdr->last = t;
 509     }
 510     else {
 511         /* append to the last text element */
 512         hdr->last->next = t;
 513         hdr->last = t;
 514     }
 515 }
 516
 517
 518 /* ---------------------------------------------------------------
 519 **
 520 ** XML UTILITY FUNCTIONS
 521 */
 522
 523 /*
 524 ** apr_xml_quote_string: quote an XML string
 525 **
 526 ** Replace '<', '>', and '&' with '&lt;', '&gt;', and '&amp;'.
 527 ** If quotes is true, then replace '"' with '&quot;'.
 528 **
 529 ** quotes is typically set to true for XML strings that will occur within
 530 ** double quotes -- attribute values.
 531 */
 532 APU_DECLARE(const char *) apr_xml_quote_string(apr_pool_t *p, const char *s,
 533                                                int quotes)
 534 {
 535     const char *scan;
 536     apr_size_t len = 0;
 537     apr_size_t extra = 0;
 538     char *qstr;
 539     char *qscan;
 540     char c;
 541
 542     for (scan = s; (c = *scan) != '\0'; ++scan, ++len) {
 543         if (c == '<' || c == '>')
 544             extra += 3;         /* &lt; or &gt; */
 545         else if (c == '&')
 546             extra += 4;         /* &amp; */
 547         else if (quotes && c == '"')
 548             extra += 5;         /* &quot; */
 549     }
 550
 551     /* nothing to do? */
 552     if (extra == 0)
 553         return s;
 554
 555     qstr = apr_palloc(p, len + extra + 1);
 556     for (scan = s, qscan = qstr; (c = *scan) != '\0'; ++scan) {
 557         if (c == '<') {
 558             *qscan++ = '&';
 559             *qscan++ = 'l';
 560             *qscan++ = 't';
 561             *qscan++ = ';';
 562         }
 563         else if (c == '>') {
 564             *qscan++ = '&';
 565             *qscan++ = 'g';
 566             *qscan++ = 't';
 567             *qscan++ = ';';
 568         }
 569         else if (c == '&') {
 570             *qscan++ = '&';
 571             *qscan++ = 'a';
 572             *qscan++ = 'm';
 573             *qscan++ = 'p';
 574             *qscan++ = ';';
 575         }
 576         else if (quotes && c == '"') {
 577             *qscan++ = '&';
 578             *qscan++ = 'q';
 579             *qscan++ = 'u';
 580             *qscan++ = 'o';
 581             *qscan++ = 't';
 582             *qscan++ = ';';
 583         }
 584         else {
 585             *qscan++ = c;
 586         }
 587     }
 588
 589     *qscan = '\0';
 590     return qstr;
 591 }
 592
 593 /* how many characters for the given integer? */
 594 #define APR_XML_NS_LEN(ns) ((ns) < 10 ? 1 : (ns) < 100 ? 2 : (ns) < 1000 ? 3 : \
 595                             (ns) < 10000 ? 4 : (ns) < 100000 ? 5 : \
 596                             (ns) < 1000000 ? 6 : (ns) < 10000000 ? 7 : \
 597                             (ns) < 100000000 ? 8 : (ns) < 1000000000 ? 9 : 10)
 598
 599 static apr_size_t text_size(const apr_text *t)
 600 {
 601     apr_size_t size = 0;
 602
 603     for (; t; t = t->next)
 604         size += strlen(t->text);
 605     return size;
 606 }
 607
 608 static apr_size_t elem_size(const apr_xml_elem *elem, int style,
 609                             apr_array_header_t *namespaces, int *ns_map)
 610 {
 611     apr_size_t size;
 612
 613     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 614         const apr_xml_attr *attr;
 615
 616         size = 0;
 617
 618         if (style == APR_XML_X2T_FULL_NS_LANG) {
 619             int i;
 620
 621             /*
 622             ** The outer element will contain xmlns:ns%d="%s" attributes
 623             ** and an xml:lang attribute, if applicable.
 624             */
 625
 626             for (i = namespaces->nelts; i--;) {
 627                 /* compute size of: ' xmlns:ns%d="%s"' */
 628                 size += (9 + APR_XML_NS_LEN(i) + 2 +
 629                          strlen(APR_XML_GET_URI_ITEM(namespaces, i)) + 1);
 630             }
 631
 632             if (elem->lang != NULL) {
 633                 /* compute size of: ' xml:lang="%s"' */
 634                 size += 11 + strlen(elem->lang) + 1;
 635             }
 636         }
 637
 638         if (elem->ns == APR_XML_NS_NONE) {
 639             /* compute size of: <%s> */
 640             size += 1 + strlen(elem->name) + 1;
 641         }
 642         else {
 643             int ns = ns_map ? ns_map[elem->ns] : elem->ns;
 644
 645             /* compute size of: <ns%d:%s> */
 646             size += 3 + APR_XML_NS_LEN(ns) + 1 + strlen(elem->name) + 1;
 647         }
 648
 649         if (APR_XML_ELEM_IS_EMPTY(elem)) {
 650             /* insert a closing "/" */
 651             size += 1;
 652         }
 653         else {
 654             /*
 655              * two of above plus "/":
 656              *     <ns%d:%s> ... </ns%d:%s>
 657              * OR  <%s> ... </%s>
 658              */
 659             size = 2 * size + 1;
 660         }
 661
 662         for (attr = elem->attr; attr; attr = attr->next) {
 663             if (attr->ns == APR_XML_NS_NONE) {
 664                 /* compute size of: ' %s="%s"' */
 665                 size += 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
 666             }
 667             else {
 668                 /* compute size of: ' ns%d:%s="%s"' */
 669                 size += 3 + APR_XML_NS_LEN(attr->ns) + 1 + strlen(attr->name) + 2 + strlen(attr->value) + 1;
 670             }
 671         }
 672
 673         /*
 674         ** If the element has an xml:lang value that is *different* from
 675         ** its parent, then add the thing in: ' xml:lang="%s"'.
 676         **
 677         ** NOTE: we take advantage of the pointer equality established by
 678         ** the parsing for "inheriting" the xml:lang values from parents.
 679         */
 680         if (elem->lang != NULL &&
 681             (elem->parent == NULL || elem->lang != elem->parent->lang)) {
 682             size += 11 + strlen(elem->lang) + 1;
 683         }
 684     }
 685     else if (style == APR_XML_X2T_LANG_INNER) {
 686         /*
 687          * This style prepends the xml:lang value plus a null terminator.
 688          * If a lang value is not present, then we insert a null term.
 689          */
 690         size = elem->lang ? strlen(elem->lang) + 1 : 1;
 691     }
 692     else
 693         size = 0;
 694
 695     size += text_size(elem->first_cdata.first);
 696
 697     for (elem = elem->first_child; elem; elem = elem->next) {
 698         /* the size of the child element plus the CDATA that follows it */
 699         size += (elem_size(elem, APR_XML_X2T_FULL, NULL, ns_map) +
 700                  text_size(elem->following_cdata.first));
 701     }
 702
 703     return size;
 704 }
 705
 706 static char *write_text(char *s, const apr_text *t)
 707 {
 708     for (; t; t = t->next) {
 709         apr_size_t len = strlen(t->text);
 710         memcpy(s, t->text, len);
 711         s += len;
 712     }
 713     return s;
 714 }
 715
 716 static char *write_elem(char *s, const apr_xml_elem *elem, int style,
 717                         apr_array_header_t *namespaces, int *ns_map)
 718 {
 719     const apr_xml_elem *child;
 720     apr_size_t len;
 721     int ns;
 722
 723     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 724         int empty = APR_XML_ELEM_IS_EMPTY(elem);
 725         const apr_xml_attr *attr;
 726
 727         if (elem->ns == APR_XML_NS_NONE) {
 728             len = sprintf(s, "<%s", elem->name);
 729         }
 730         else {
 731             ns = ns_map ? ns_map[elem->ns] : elem->ns;
 732             len = sprintf(s, "<ns%d:%s", ns, elem->name);
 733         }
 734         s += len;
 735
 736         for (attr = elem->attr; attr; attr = attr->next) {
 737             if (attr->ns == APR_XML_NS_NONE)
 738                 len = sprintf(s, " %s=\"%s\"", attr->name, attr->value);
 739             else
 740                 len = sprintf(s, " ns%d:%s=\"%s\"", attr->ns, attr->name, attr->value);
 741             s += len;
 742         }
 743
 744         /* add the xml:lang value if necessary */
 745         if (elem->lang != NULL &&
 746             (style == APR_XML_X2T_FULL_NS_LANG ||
 747              elem->parent == NULL ||
 748              elem->lang != elem->parent->lang)) {
 749             len = sprintf(s, " xml:lang=\"%s\"", elem->lang);
 750             s += len;
 751         }
 752
 753         /* add namespace definitions, if required */
 754         if (style == APR_XML_X2T_FULL_NS_LANG) {
 755             int i;
 756
 757             for (i = namespaces->nelts; i--;) {
 758                 len = sprintf(s, " xmlns:ns%d=\"%s\"", i,
 759                               APR_XML_GET_URI_ITEM(namespaces, i));
 760                 s += len;
 761             }
 762         }
 763
 764         /* no more to do. close it up and go. */
 765         if (empty) {
 766             *s++ = '/';
 767             *s++ = '>';
 768             return s;
 769         }
 770
 771         /* just close it */
 772         *s++ = '>';
 773     }
 774     else if (style == APR_XML_X2T_LANG_INNER) {
 775         /* prepend the xml:lang value */
 776         if (elem->lang != NULL) {
 777             len = strlen(elem->lang);
 778             memcpy(s, elem->lang, len);
 779             s += len;
 780         }
 781         *s++ = '\0';
 782     }
 783
 784     s = write_text(s, elem->first_cdata.first);
 785
 786     for (child = elem->first_child; child; child = child->next) {
 787         s = write_elem(s, child, APR_XML_X2T_FULL, NULL, ns_map);
 788         s = write_text(s, child->following_cdata.first);
 789     }
 790
 791     if (style == APR_XML_X2T_FULL || style == APR_XML_X2T_FULL_NS_LANG) {
 792         if (elem->ns == APR_XML_NS_NONE) {
 793             len = sprintf(s, "</%s>", elem->name);
 794         }
 795         else {
 796             ns = ns_map ? ns_map[elem->ns] : elem->ns;
 797             len = sprintf(s, "</ns%d:%s>", ns, elem->name);
 798         }
 799         s += len;
 800     }
 801
 802     return s;
 803 }
 804
 805 APU_DECLARE(void) apr_xml_quote_elem(apr_pool_t *p, apr_xml_elem *elem)
 806 {
 807     apr_text *scan_txt;
 808     apr_xml_attr *scan_attr;
 809     apr_xml_elem *scan_elem;
 810
 811     /* convert the element's text */
 812     for (scan_txt = elem->first_cdata.first;
 813          scan_txt != NULL;
 814          scan_txt = scan_txt->next) {
 815         scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
 816     }
 817     for (scan_txt = elem->following_cdata.first;
 818          scan_txt != NULL;
 819          scan_txt = scan_txt->next) {
 820         scan_txt->text = apr_xml_quote_string(p, scan_txt->text, 0);
 821     }
 822
 823     /* convert the attribute values */
 824     for (scan_attr = elem->attr;
 825          scan_attr != NULL;
 826          scan_attr = scan_attr->next) {
 827         scan_attr->value = apr_xml_quote_string(p, scan_attr->value, 1);
 828     }
 829
 830     /* convert the child elements */
 831     for (scan_elem = elem->first_child;
 832          scan_elem != NULL;
 833          scan_elem = scan_elem->next) {
 834         apr_xml_quote_elem(p, scan_elem);
 835     }
 836 }
 837
 838 /* convert an element to a text string */
 839 APU_DECLARE(void) apr_xml_to_text(apr_pool_t * p, const apr_xml_elem *elem,
 840                                   int style, apr_array_header_t *namespaces,
 841                                   int *ns_map, const char **pbuf,
 842                                   apr_size_t *psize)
 843 {
 844     /* get the exact size, plus a null terminator */
 845     apr_size_t size = elem_size(elem, style, namespaces, ns_map) + 1;
 846     char *s = apr_palloc(p, size);
 847
 848     (void) write_elem(s, elem, style, namespaces, ns_map);
 849     s[size - 1] = '\0';
 850
 851     *pbuf = s;
 852     if (psize)
 853         *psize = size;
 854 }
 855
 856 APU_DECLARE(const char *) apr_xml_empty_elem(apr_pool_t * p,
 857                                              const apr_xml_elem *elem)
 858 {
 859     if (elem->ns == APR_XML_NS_NONE) {
 860         /*
 861          * The prefix (xml...) is already within the prop name, or
 862          * the element simply has no prefix.
 863          */
 864         return apr_psprintf(p, "<%s/>" DEBUG_CR, elem->name);
 865     }
 866
 867     return apr_psprintf(p, "<ns%d:%s/>" DEBUG_CR, elem->ns, elem->name);
 868 }
 869
 870 /* return the URI's (existing) index, or insert it and return a new index */
 871 APU_DECLARE(int) apr_xml_insert_uri(apr_array_header_t *uri_array,
 872                                     const char *uri)
 873 {
 874     int i;
 875     const char **pelt;
 876
 877     /* never insert an empty URI; this index is always APR_XML_NS_NONE */
 878     if (*uri == '\0')
 879         return APR_XML_NS_NONE;
 880
 881     for (i = uri_array->nelts; i--;) {
 882         if (strcmp(uri, APR_XML_GET_URI_ITEM(uri_array, i)) == 0)
 883             return i;
 884     }
 885
 886     pelt = apr_array_push(uri_array);
 887     *pelt = uri;                /* assume uri is const or in a pool */
 888     return uri_array->nelts - 1;
 889 }
 890
 891 /* convert the element to EBCDIC */
 892 #if APR_CHARSET_EBCDIC
 893 static apr_status_t apr_xml_parser_convert_elem(apr_xml_elem *e,
 894                                                 apr_xlate_t *convset)
 895 {
 896     apr_xml_attr *a;
 897     apr_xml_elem *ec;
 898     apr_text *t;
 899     apr_size_t inbytes_left, outbytes_left;
 900     apr_status_t status;
 901
 902     inbytes_left = outbytes_left = strlen(e->name);
 903     status = apr_xlate_conv_buffer(convset, e->name,  &inbytes_left, (char *) e->name, &outbytes_left);
 904     if (status) {
 905         return status;
 906     }
 907
 908     for (t = e->first_cdata.first; t != NULL; t = t->next) {
 909         inbytes_left = outbytes_left = strlen(t->text);
 910         status = apr_xlate_conv_buffer(convset, t->text, &inbytes_left, (char *) t->text, &outbytes_left);
 911         if (status) {
 912             return status;
 913         }
 914     }
 915
 916     for (t = e->following_cdata.first;  t != NULL; t = t->next) {
 917         inbytes_left = outbytes_left = strlen(t->text);
 918         status = apr_xlate_conv_buffer(convset, t->text, &inbytes_left, (char *) t->text, &outbytes_left);
 919         if (status) {
 920             return status;
 921         }
 922     }
 923
 924     for (a = e->attr; a != NULL; a = a->next) {
 925         inbytes_left = outbytes_left = strlen(a->name);
 926         status = apr_xlate_conv_buffer(convset, a->name, &inbytes_left, (char *) a->name, &outbytes_left);
 927         if (status) {
 928             return status;
 929         }
 930         inbytes_left = outbytes_left = strlen(a->value);
 931         status = apr_xlate_conv_buffer(convset, a->value, &inbytes_left, (char *) a->value, &outbytes_left);
 932         if (status) {
 933             return status;
 934         }
 935     }
 936
 937     for (ec = e->first_child; ec != NULL; ec = ec->next) {
 938         status = apr_xml_parser_convert_elem(ec, convset);
 939         if (status) {
 940             return status;
 941         }
 942     }
 943     return APR_SUCCESS;
 944 }
 945
 946 /* convert the whole document to EBCDIC */
 947 APU_DECLARE(apr_status_t) apr_xml_parser_convert_doc(apr_pool_t *pool,
 948                                                      apr_xml_doc *pdoc,
 949                                                      apr_xlate_t *convset)
 950 {
 951     apr_status_t status;
 952     /* Don't convert the namespaces: they are constant! */
 953     if (pdoc->namespaces != NULL) {
 954         int i;
 955         apr_array_header_t *namespaces;
 956         namespaces = apr_array_make(pool, pdoc->namespaces->nelts, sizeof(const char *));
 957         if (namespaces == NULL)
 958             return APR_ENOMEM;
 959         for (i = 0; i < pdoc->namespaces->nelts; i++) {
 960             apr_size_t inbytes_left, outbytes_left;
 961             char *ptr = (char *) APR_XML_GET_URI_ITEM(pdoc->namespaces, i);
 962             ptr = apr_pstrdup(pool, ptr);
 963             if ( ptr == NULL)
 964                 return APR_ENOMEM;
 965             inbytes_left = outbytes_left = strlen(ptr);
 966             status = apr_xlate_conv_buffer(convset, ptr, &inbytes_left, ptr, &outbytes_left);
 967             if (status) {
 968                 return status;
 969             }
 970             apr_xml_insert_uri(namespaces, ptr);
 971         }
 972         pdoc->namespaces = namespaces;
 973     }
 974     return apr_xml_parser_convert_elem(pdoc->root, convset);
 975 }
 976 #endif