wimlib.net Git - wimlib/blob - src/xml.c

   1 /*
   2  * xml.c
   3  *
   4  * Deals with the XML information in WIM files.  Uses the C library libxml2.
   5  */
   6
   7 /*
   8  * Copyright (C) 2012, 2013 Eric Biggers
   9  *
  10  * This file is part of wimlib, a library for working with WIM files.
  11  *
  12  * wimlib is free software; you can redistribute it and/or modify it under the
  13  * terms of the GNU General Public License as published by the Free
  14  * Software Foundation; either version 3 of the License, or (at your option)
  15  * any later version.
  16  *
  17  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  18  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  19  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  20  * details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  24  */
  25
  26 #include "dentry.h"
  27 #include "lookup_table.h"
  28 #include "timestamp.h"
  29 #include "wimlib_internal.h"
  30 #include "xml.h"
  31
  32 #include <string.h>
  33 #include <libxml/parser.h>
  34 #include <libxml/tree.h>
  35 #include <libxml/xmlwriter.h>
  36 #include <libxml/encoding.h>
  37 #include <limits.h>
  38
  39 /* Structures used to form an in-memory representation of the XML data (other
  40  * than the raw parse tree from libxml). */
  41
  42 struct windows_version {
  43         u64 major;
  44         u64 minor;
  45         u64 build;
  46         u64 sp_build;
  47         u64 sp_level;
  48 };
  49
  50 struct windows_info {
  51         u64      arch;
  52         tchar   *product_name;
  53         tchar   *edition_id;
  54         tchar   *installation_type;
  55         tchar   *hal;
  56         tchar   *product_type;
  57         tchar   *product_suite;
  58         tchar  **languages;
  59         tchar   *default_language;
  60         size_t   num_languages;
  61         tchar   *system_root;
  62         bool     windows_version_exists;
  63         struct   windows_version windows_version;
  64 };
  65
  66 struct image_info {
  67         int index;
  68         bool windows_info_exists;
  69         u64 dir_count;
  70         u64 file_count;
  71         u64 total_bytes;
  72         u64 hard_link_bytes;
  73         u64 creation_time;
  74         u64 last_modification_time;
  75         struct windows_info windows_info;
  76         tchar *name;
  77         tchar *description;
  78         tchar *display_name;
  79         tchar *display_description;
  80         tchar *flags;
  81         struct wim_lookup_table *lookup_table; /* Temporary field only */
  82 };
  83
  84 struct xml_string_spec {
  85         const utf8char *name;
  86         size_t offset;
  87 };
  88
  89 #define ELEM(STRING_NAME, MEMBER_NAME) \
  90         {STRING_NAME, offsetof(struct image_info, MEMBER_NAME)}
  91 static const struct xml_string_spec
  92 image_info_xml_string_specs[] = {
  93         ELEM("NAME", name),
  94         ELEM("DESCRIPTION", description),
  95         ELEM("DISPLAYNAME", display_name),
  96         ELEM("DISPLAYDESCRIPTION", display_description),
  97         ELEM("FLAGS", flags),
  98 };
  99 #undef ELEM
 100
 101 #define ELEM(STRING_NAME, MEMBER_NAME) \
 102         {STRING_NAME, offsetof(struct windows_info, MEMBER_NAME)}
 103 static const struct xml_string_spec
 104 windows_info_xml_string_specs[] = {
 105         ELEM("PRODUCTNAME", product_name),
 106         ELEM("EDITIONID", edition_id),
 107         ELEM("INSTALLATIONTYPE", installation_type),
 108         ELEM("HAL", hal),
 109         ELEM("PRODUCTTYPE", product_type),
 110         ELEM("PRODUCTSUITE", product_suite),
 111 };
 112 #undef ELEM
 113
 114
 115 /* Returns a statically allocated string that is a string representation of the
 116  * architecture number. */
 117 static const tchar *
 118 get_arch(int arch)
 119 {
 120         switch (arch) {
 121         case 0:
 122                 return T("x86");
 123         case 6:
 124                 return T("ia64");
 125         case 9:
 126                 return T("x86_64");
 127         /* XXX Are there other arch values? */
 128         default:
 129                 return NULL;
 130         }
 131 }
 132
 133
 134 /* Iterate through the children of an xmlNode. */
 135 #define for_node_child(parent, child)   \
 136         for (child = parent->children; child != NULL; child = child->next)
 137
 138 /* Utility functions for xmlNodes */
 139 static inline bool
 140 node_is_element(xmlNode *node)
 141 {
 142         return node->type == XML_ELEMENT_NODE;
 143 }
 144
 145 static inline bool
 146 node_is_text(xmlNode *node)
 147 {
 148         return node->type == XML_TEXT_NODE;
 149 }
 150
 151 static inline bool
 152 node_name_is(xmlNode *node, const utf8char *name)
 153 {
 154         /* For now, both upper case and lower case element names are accepted. */
 155         return strcasecmp((const char *)node->name, name) == 0;
 156 }
 157
 158 /* Finds the text node that is a child of an element node and returns its
 159  * content converted to a 64-bit unsigned integer.  Returns 0 if no text node is
 160  * found. */
 161 static u64
 162 node_get_u64(const xmlNode *u64_node)
 163 {
 164         xmlNode *child;
 165         for_node_child(u64_node, child)
 166                 if (node_is_text(child))
 167                         return strtoull((const char *)child->content, NULL, 10);
 168         return 0;
 169 }
 170
 171 /* Like node_get_u64(), but expects a number in base 16. */
 172 static u64
 173 node_get_hex_u64(const xmlNode *u64_node)
 174 {
 175         xmlNode *child;
 176         for_node_child(u64_node, child)
 177                 if (node_is_text(child))
 178                         return strtoull(child->content, NULL, 16);
 179         return 0;
 180 }
 181
 182 static int
 183 node_get_string(const xmlNode *string_node, tchar **tstr_ret)
 184 {
 185         xmlNode *child;
 186         tchar *tstr = NULL;
 187         int ret;
 188
 189         for_node_child(string_node, child) {
 190                 if (node_is_text(child) && child->content) {
 191                         ret = utf8_to_tstr_simple(child->content, &tstr);
 192                         if (ret)
 193                                 return ret;
 194                         break;
 195                 }
 196         }
 197         *tstr_ret = tstr;
 198         return 0;
 199 }
 200
 201 /* Returns the timestamp from a time node.  It has child elements <HIGHPART> and
 202  * <LOWPART> that are then used to construct a 64-bit timestamp. */
 203 static u64
 204 node_get_timestamp(const xmlNode *time_node)
 205 {
 206         u32 high_part = 0;
 207         u32 low_part = 0;
 208         xmlNode *child;
 209         for_node_child(time_node, child) {
 210                 if (!node_is_element(child))
 211                         continue;
 212                 if (node_name_is(child, "HIGHPART"))
 213                         high_part = node_get_hex_u64(child);
 214                 else if (node_name_is(child, "LOWPART"))
 215                         low_part = node_get_hex_u64(child);
 216         }
 217         return (u64)low_part | ((u64)high_part << 32);
 218 }
 219
 220 /* Used to sort an array of struct image_infos by their image indices. */
 221 static int
 222 sort_by_index(const void *p1, const void *p2)
 223 {
 224         int index_1 = ((const struct image_info*)p1)->index;
 225         int index_2 = ((const struct image_info*)p2)->index;
 226         if (index_1 < index_2)
 227                 return -1;
 228         else if (index_1 > index_2)
 229                 return 1;
 230         else
 231                 return 0;
 232 }
 233
 234
 235 /* Frees memory allocated inside a struct windows_info structure. */
 236 static void
 237 destroy_windows_info(struct windows_info *windows_info)
 238 {
 239         FREE(windows_info->product_name);
 240         FREE(windows_info->edition_id);
 241         FREE(windows_info->installation_type);
 242         FREE(windows_info->hal);
 243         FREE(windows_info->product_type);
 244         FREE(windows_info->product_suite);
 245         for (size_t i = 0; i < windows_info->num_languages; i++)
 246                 FREE(windows_info->languages[i]);
 247         FREE(windows_info->languages);
 248         FREE(windows_info->default_language);
 249         FREE(windows_info->system_root);
 250 }
 251
 252 /* Frees memory allocated inside a struct image_info structure. */
 253 static void
 254 destroy_image_info(struct image_info *image_info)
 255 {
 256         FREE(image_info->name);
 257         FREE(image_info->description);
 258         FREE(image_info->flags);
 259         FREE(image_info->display_name);
 260         FREE(image_info->display_description);
 261         destroy_windows_info(&image_info->windows_info);
 262         memset(image_info, 0, sizeof(struct image_info));
 263 }
 264
 265 void
 266 free_wim_info(struct wim_info *info)
 267 {
 268         if (info) {
 269                 if (info->images) {
 270                         for (int i = 0; i < info->num_images; i++)
 271                                 destroy_image_info(&info->images[i]);
 272                         FREE(info->images);
 273                 }
 274                 FREE(info);
 275         }
 276 }
 277
 278 /* Reads the information from a <VERSION> element inside the <WINDOWS> element.
 279  * */
 280 static void
 281 xml_read_windows_version(const xmlNode *version_node,
 282                          struct windows_version* windows_version)
 283 {
 284         xmlNode *child;
 285         for_node_child(version_node, child) {
 286                 if (!node_is_element(child))
 287                         continue;
 288                 if (node_name_is(child, "MAJOR"))
 289                         windows_version->major = node_get_u64(child);
 290                 else if (node_name_is(child, "MINOR"))
 291                         windows_version->minor = node_get_u64(child);
 292                 else if (node_name_is(child, "BUILD"))
 293                         windows_version->build = node_get_u64(child);
 294                 else if (node_name_is(child, "SPBUILD"))
 295                         windows_version->sp_build = node_get_u64(child);
 296                 else if (node_name_is(child, "SPLEVEL"))
 297                         windows_version->sp_level = node_get_u64(child);
 298         }
 299 }
 300
 301 /* Reads the information from a <LANGUAGE> element inside a <WINDOWS> element.
 302  * */
 303 static int
 304 xml_read_languages(const xmlNode *languages_node,
 305                    tchar ***languages_ret,
 306                    size_t *num_languages_ret,
 307                    tchar **default_language_ret)
 308 {
 309         xmlNode *child;
 310         size_t num_languages = 0;
 311         tchar **languages;
 312         int ret;
 313
 314         for_node_child(languages_node, child)
 315                 if (node_is_element(child) && node_name_is(child, "LANGUAGE"))
 316                         num_languages++;
 317
 318         languages = CALLOC(num_languages, sizeof(languages[0]));
 319         if (!languages)
 320                 return WIMLIB_ERR_NOMEM;
 321
 322         *languages_ret = languages;
 323         *num_languages_ret = num_languages;
 324
 325         ret = 0;
 326         for_node_child(languages_node, child) {
 327                 if (!node_is_element(child))
 328                         continue;
 329                 if (node_name_is(child, "LANGUAGE"))
 330                         ret = node_get_string(child, languages++);
 331                 else if (node_name_is(child, "DEFAULT"))
 332                         ret = node_get_string(child, default_language_ret);
 333                 if (ret != 0)
 334                         break;
 335         }
 336         return ret;
 337 }
 338
 339 /* Reads the information from a <WINDOWS> element inside an <IMAGE> element. */
 340 static int
 341 xml_read_windows_info(const xmlNode *windows_node,
 342                       struct windows_info *windows_info)
 343 {
 344         xmlNode *child;
 345         int ret = 0;
 346
 347         for_node_child(windows_node, child) {
 348                 if (!node_is_element(child))
 349                         continue;
 350                 if (node_name_is(child, "ARCH")) {
 351                         windows_info->arch = node_get_u64(child);
 352                 } else if (node_name_is(child, "PRODUCTNAME")) {
 353                         ret = node_get_string(child,
 354                                               &windows_info->product_name);
 355                 } else if (node_name_is(child, "EDITIONID")) {
 356                         ret = node_get_string(child,
 357                                               &windows_info->edition_id);
 358                 } else if (node_name_is(child, "INSTALLATIONTYPE")) {
 359                         ret = node_get_string(child,
 360                                               &windows_info->installation_type);
 361                 } else if (node_name_is(child, "PRODUCTTYPE")) {
 362                         ret = node_get_string(child,
 363                                               &windows_info->product_type);
 364                 } else if (node_name_is(child, "PRODUCTSUITE")) {
 365                         ret = node_get_string(child,
 366                                               &windows_info->product_suite);
 367                 } else if (node_name_is(child, "LANGUAGES")) {
 368                         ret = xml_read_languages(child,
 369                                                  &windows_info->languages,
 370                                                  &windows_info->num_languages,
 371                                                  &windows_info->default_language);
 372                 } else if (node_name_is(child, "VERSION")) {
 373                         xml_read_windows_version(child,
 374                                                 &windows_info->windows_version);
 375                         windows_info->windows_version_exists = true;
 376                 } else if (node_name_is(child, "SYSTEMROOT")) {
 377                         ret = node_get_string(child, &windows_info->system_root);
 378                 } else if (node_name_is(child, "HAL")) {
 379                         ret = node_get_string(child, &windows_info->hal);
 380                 }
 381                 if (ret != 0)
 382                         return ret;
 383         }
 384         return ret;
 385 }
 386
 387 /* Reads the information from an <IMAGE> element. */
 388 static int
 389 xml_read_image_info(xmlNode *image_node, struct image_info *image_info)
 390 {
 391         xmlNode *child;
 392         xmlChar *index_prop;
 393         int ret;
 394
 395         index_prop = xmlGetProp(image_node, "INDEX");
 396         if (index_prop) {
 397                 image_info->index = atoi(index_prop);
 398                 FREE(index_prop);
 399         } else {
 400                 image_info->index = 1;
 401         }
 402
 403         ret = 0;
 404         for_node_child(image_node, child) {
 405                 if (!node_is_element(child))
 406                         continue;
 407                 if (node_name_is(child, "DIRCOUNT"))
 408                         image_info->dir_count = node_get_u64(child);
 409                 else if (node_name_is(child, "FILECOUNT"))
 410                         image_info->file_count = node_get_u64(child);
 411                 else if (node_name_is(child, "TOTALBYTES"))
 412                         image_info->total_bytes = node_get_u64(child);
 413                 else if (node_name_is(child, "HARDLINKBYTES"))
 414                         image_info->hard_link_bytes = node_get_u64(child);
 415                 else if (node_name_is(child, "CREATIONTIME"))
 416                         image_info->creation_time = node_get_timestamp(child);
 417                 else if (node_name_is(child, "LASTMODIFICATIONTIME"))
 418                         image_info->last_modification_time = node_get_timestamp(child);
 419                 else if (node_name_is(child, "WINDOWS")) {
 420                         DEBUG("Found <WINDOWS> tag");
 421                         ret = xml_read_windows_info(child,
 422                                                     &image_info->windows_info);
 423                         image_info->windows_info_exists = true;
 424                 } else if (node_name_is(child, "NAME")) {
 425                         ret = node_get_string(child, &image_info->name);
 426                 } else if (node_name_is(child, "DESCRIPTION")) {
 427                         ret = node_get_string(child, &image_info->description);
 428                 } else if (node_name_is(child, "FLAGS")) {
 429                         ret = node_get_string(child, &image_info->flags);
 430                 } else if (node_name_is(child, "DISPLAYNAME")) {
 431                         ret = node_get_string(child, &image_info->display_name);
 432                 } else if (node_name_is(child, "DISPLAYDESCRIPTION")) {
 433                         ret = node_get_string(child, &image_info->display_description);
 434                 }
 435                 if (ret != 0)
 436                         return ret;
 437         }
 438         if (!image_info->name) {
 439                 tchar *empty_name;
 440                 WARNING("Image with index %d has no name", image_info->index);
 441                 empty_name = TMALLOC(1);
 442                 if (!empty_name)
 443                         return WIMLIB_ERR_NOMEM;
 444                 *empty_name = T('\0');
 445                 image_info->name = empty_name;
 446         }
 447         return ret;
 448 }
 449
 450 /* Reads the information from a <WIM> element, which should be the root element
 451  * of the XML tree. */
 452 static int
 453 xml_read_wim_info(const xmlNode *wim_node, struct wim_info **wim_info_ret)
 454 {
 455         struct wim_info *wim_info;
 456         xmlNode *child;
 457         int ret;
 458         int num_images;
 459         int i;
 460
 461         wim_info = CALLOC(1, sizeof(struct wim_info));
 462         if (!wim_info)
 463                 return WIMLIB_ERR_NOMEM;
 464
 465         /* Count how many images there are. */
 466         num_images = 0;
 467         for_node_child(wim_node, child) {
 468                 if (node_is_element(child) && node_name_is(child, "IMAGE")) {
 469                         if (num_images == INT_MAX) {
 470                                 return WIMLIB_ERR_IMAGE_COUNT;
 471                         }
 472                         num_images++;
 473                 }
 474         }
 475
 476         if (num_images > 0) {
 477                 /* Allocate the array of struct image_infos and fill them in. */
 478                 wim_info->images = CALLOC(num_images, sizeof(wim_info->images[0]));
 479                 if (!wim_info->images) {
 480                         ret = WIMLIB_ERR_NOMEM;
 481                         goto err;
 482                 }
 483                 wim_info->num_images = num_images;
 484                 i = 0;
 485                 for_node_child(wim_node, child) {
 486                         if (!node_is_element(child))
 487                                 continue;
 488                         if (node_name_is(child, "IMAGE")) {
 489                                 DEBUG("Found <IMAGE> tag");
 490                                 ret = xml_read_image_info(child,
 491                                                           &wim_info->images[i]);
 492                                 if (ret != 0)
 493                                         goto err;
 494                                 i++;
 495                         } else if (node_name_is(child, "TOTALBYTES")) {
 496                                 wim_info->total_bytes = node_get_u64(child);
 497                         }
 498                 }
 499
 500                 /* Sort the array of image info by image index. */
 501                 qsort(wim_info->images, num_images,
 502                       sizeof(struct image_info), sort_by_index);
 503
 504                 /* Make sure the image indices make sense */
 505                 for (i = 0; i < num_images; i++) {
 506                         if (wim_info->images[i].index != i + 1) {
 507                                 ERROR("WIM images are not indexed [1...%d] "
 508                                       "in XML data as expected",
 509                                       num_images);
 510                                 return WIMLIB_ERR_IMAGE_COUNT;
 511                         }
 512                 }
 513
 514         }
 515         *wim_info_ret = wim_info;
 516         return 0;
 517 err:
 518         free_wim_info(wim_info);
 519         return ret;
 520 }
 521
 522 /* Prints the information contained in a `struct windows_info'.
 523  *
 524  * Warning: any strings printed here are in UTF-8 encoding.  If the locale
 525  * character encoding is not UTF-8, the printed strings may be garbled. */
 526 static void
 527 print_windows_info(const struct windows_info *windows_info)
 528 {
 529         const struct windows_version *windows_version;
 530
 531         tprintf(T("Architecture:           %"TS"\n"),
 532                 get_arch(windows_info->arch) ?: T("unknown"));
 533
 534         if (windows_info->product_name) {
 535                 tprintf(T("Product Name:           %"TS"\n"),
 536                         windows_info->product_name);
 537         }
 538
 539         if (windows_info->edition_id) {
 540                 tprintf(T("Edition ID:             %"TS"\n"),
 541                         windows_info->edition_id);
 542         }
 543
 544         if (windows_info->installation_type) {
 545                 tprintf(T("Installation Type:      %"TS"\n"),
 546                         windows_info->installation_type);
 547         }
 548
 549         if (windows_info->hal) {
 550                 tprintf(T("HAL:                    %"TS"\n"),
 551                               windows_info->hal);
 552         }
 553
 554         if (windows_info->product_type) {
 555                 tprintf(T("Product Type:           %"TS"\n"),
 556                         windows_info->product_type);
 557         }
 558
 559         if (windows_info->product_suite) {
 560                 tprintf(T("Product Suite:          %"TS"\n"),
 561                         windows_info->product_suite);
 562         }
 563
 564         tprintf(T("Languages:              "));
 565         for (size_t i = 0; i < windows_info->num_languages; i++) {
 566
 567                 tfputs(windows_info->languages[i], stdout);
 568                 tputchar(T(' '));
 569         }
 570         tputchar(T('\n'));
 571         if (windows_info->default_language) {
 572                 tprintf(T("Default Language:       %"TS"\n"),
 573                         windows_info->default_language);
 574         }
 575         if (windows_info->system_root) {
 576                 tprintf(T("System Root:            %"TS"\n"),
 577                               windows_info->system_root);
 578         }
 579
 580         if (windows_info->windows_version_exists) {
 581                 windows_version = &windows_info->windows_version;
 582                 tprintf(T("Major Version:          %"PRIu64"\n"),
 583                         windows_version->major);
 584                 tprintf(T("Minor Version:          %"PRIu64"\n"),
 585                         windows_version->minor);
 586                 tprintf(T("Build:                  %"PRIu64"\n"),
 587                         windows_version->build);
 588                 tprintf(T("Service Pack Build:     %"PRIu64"\n"),
 589                         windows_version->sp_build);
 590                 tprintf(T("Service Pack Level:     %"PRIu64"\n"),
 591                         windows_version->sp_level);
 592         }
 593 }
 594
 595 static int
 596 xml_write_string(xmlTextWriter *writer, const char *name,
 597                  const tchar *tstr)
 598 {
 599         if (tstr) {
 600                 utf8char *utf8_str;
 601                 int rc = tstr_to_utf8_simple(tstr, &utf8_str);
 602                 if (rc)
 603                         return rc;
 604                 rc = xmlTextWriterWriteElement(writer, name, utf8_str);
 605                 FREE(utf8_str);
 606                 if (rc < 0)
 607                         return rc;
 608         }
 609         return 0;
 610 }
 611
 612 static int
 613 xml_write_strings_from_specs(xmlTextWriter *writer,
 614                              const void *struct_with_strings,
 615                              const struct xml_string_spec specs[],
 616                              size_t num_specs)
 617 {
 618         for (size_t i = 0; i < num_specs; i++) {
 619                 int rc = xml_write_string(writer, specs[i].name,
 620                                       *(const tchar * const *)
 621                                         (struct_with_strings + specs[i].offset));
 622                 if (rc)
 623                         return rc;
 624         }
 625         return 0;
 626 }
 627
 628 static int
 629 dup_strings_from_specs(const void *old_struct_with_strings,
 630                        void *new_struct_with_strings,
 631                        const struct xml_string_spec specs[],
 632                        size_t num_specs)
 633 {
 634         for (size_t i = 0; i < num_specs; i++) {
 635                 const tchar *old_str = *(const tchar * const *)
 636                                         ((const void*)old_struct_with_strings + specs[i].offset);
 637                 tchar **new_str_p = (tchar **)((void*)new_struct_with_strings + specs[i].offset);
 638                 if (old_str) {
 639                         *new_str_p = TSTRDUP(old_str);
 640                         if (!*new_str_p)
 641                                 return WIMLIB_ERR_NOMEM;
 642                 }
 643         }
 644         return 0;
 645 }
 646
 647 /* Writes the information contained in a `struct windows_version' to the XML
 648  * document being written.  This is the <VERSION> element inside the <WINDOWS>
 649  * element. */
 650 static int
 651 xml_write_windows_version(xmlTextWriter *writer,
 652                           const struct windows_version *version)
 653 {
 654         int rc;
 655         rc = xmlTextWriterStartElement(writer, "VERSION");
 656         if (rc < 0)
 657                 return rc;
 658
 659         rc = xmlTextWriterWriteFormatElement(writer, "MAJOR", "%"PRIu64,
 660                                              version->major);
 661         if (rc < 0)
 662                 return rc;
 663
 664         rc = xmlTextWriterWriteFormatElement(writer, "MINOR", "%"PRIu64,
 665                                              version->minor);
 666         if (rc < 0)
 667                 return rc;
 668
 669         rc = xmlTextWriterWriteFormatElement(writer, "BUILD", "%"PRIu64,
 670                                              version->build);
 671         if (rc < 0)
 672                 return rc;
 673
 674         rc = xmlTextWriterWriteFormatElement(writer, "SPBUILD", "%"PRIu64,
 675                                              version->sp_build);
 676         if (rc < 0)
 677                 return rc;
 678
 679         rc = xmlTextWriterWriteFormatElement(writer, "SPLEVEL", "%"PRIu64,
 680                                              version->sp_level);
 681         if (rc < 0)
 682                 return rc;
 683
 684         return xmlTextWriterEndElement(writer); /* </VERSION> */
 685 }
 686
 687 /* Writes the information contained in a `struct windows_info' to the XML
 688  * document being written. This is the <WINDOWS> element. */
 689 static int
 690 xml_write_windows_info(xmlTextWriter *writer,
 691                        const struct windows_info *windows_info)
 692 {
 693         int rc;
 694         rc = xmlTextWriterStartElement(writer, "WINDOWS");
 695         if (rc < 0)
 696                 return rc;
 697
 698         rc = xmlTextWriterWriteFormatElement(writer, "ARCH", "%"PRIu64,
 699                                              windows_info->arch);
 700         if (rc < 0)
 701                 return rc;
 702
 703         rc = xml_write_strings_from_specs(writer,
 704                                           windows_info,
 705                                           windows_info_xml_string_specs,
 706                                           ARRAY_LEN(windows_info_xml_string_specs));
 707         if (rc)
 708                 return rc;
 709
 710         if (windows_info->num_languages) {
 711                 rc = xmlTextWriterStartElement(writer, "LANGUAGES");
 712                 if (rc < 0)
 713                         return rc;
 714
 715                 for (size_t i = 0; i < windows_info->num_languages; i++) {
 716                         rc = xml_write_string(writer, "LANGUAGE",
 717                                               windows_info->languages[i]);
 718                         if (rc)
 719                                 return rc;
 720                 }
 721
 722                 rc = xml_write_string(writer, "DEFAULT",
 723                                       windows_info->default_language);
 724                 if (rc)
 725                         return rc;
 726
 727                 rc = xmlTextWriterEndElement(writer); /* </LANGUAGES> */
 728                 if (rc < 0)
 729                         return rc;
 730         }
 731
 732         if (windows_info->windows_version_exists) {
 733                 rc = xml_write_windows_version(writer, &windows_info->windows_version);
 734                 if (rc < 0)
 735                         return rc;
 736         }
 737
 738         rc = xml_write_string(writer, "SYSTEMROOT",
 739                               windows_info->system_root);
 740         if (rc)
 741                 return rc;
 742
 743         return xmlTextWriterEndElement(writer); /* </WINDOWS> */
 744 }
 745
 746 /* Writes a time element to the XML document being constructed in memory. */
 747 static int
 748 xml_write_time(xmlTextWriter *writer, const utf8char *element_name, u64 time)
 749 {
 750         int rc;
 751         rc = xmlTextWriterStartElement(writer, element_name);
 752         if (rc < 0)
 753                 return rc;
 754
 755         rc = xmlTextWriterWriteFormatElement(writer, "HIGHPART",
 756                                              "0x%08"PRIX32, (u32)(time >> 32));
 757         if (rc < 0)
 758                 return rc;
 759
 760         rc = xmlTextWriterWriteFormatElement(writer, "LOWPART",
 761                                              "0x%08"PRIX32, (u32)time);
 762         if (rc < 0)
 763                 return rc;
 764
 765         rc = xmlTextWriterEndElement(writer); /* </@element_name> */
 766         if (rc < 0)
 767                 return rc;
 768         return 0;
 769 }
 770
 771 /* Writes an <IMAGE> element to the XML document. */
 772 static int
 773 xml_write_image_info(xmlTextWriter *writer, const struct image_info *image_info)
 774 {
 775         int rc;
 776         rc = xmlTextWriterStartElement(writer, "IMAGE");
 777         if (rc < 0)
 778                 return rc;
 779
 780         rc = xmlTextWriterWriteFormatAttribute(writer, "INDEX", "%d",
 781                                                image_info->index);
 782         if (rc < 0)
 783                 return rc;
 784
 785         rc = xmlTextWriterWriteFormatElement(writer, "DIRCOUNT", "%"PRIu64,
 786                                              image_info->dir_count);
 787         if (rc < 0)
 788                 return rc;
 789
 790         rc = xmlTextWriterWriteFormatElement(writer, "FILECOUNT", "%"PRIu64,
 791                                              image_info->file_count);
 792         if (rc < 0)
 793                 return rc;
 794
 795         rc = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
 796                                              image_info->total_bytes);
 797         if (rc < 0)
 798                 return rc;
 799
 800         rc = xmlTextWriterWriteFormatElement(writer, "HARDLINKBYTES", "%"PRIu64,
 801                                              image_info->hard_link_bytes);
 802         if (rc < 0)
 803                 return rc;
 804
 805         rc = xml_write_time(writer, "CREATIONTIME", image_info->creation_time);
 806         if (rc < 0)
 807                 return rc;
 808
 809         rc = xml_write_time(writer, "LASTMODIFICATIONTIME",
 810                             image_info->last_modification_time);
 811         if (rc < 0)
 812                 return rc;
 813
 814         if (image_info->windows_info_exists) {
 815                 rc = xml_write_windows_info(writer, &image_info->windows_info);
 816                 if (rc < 0)
 817                         return rc;
 818         }
 819
 820         rc = xml_write_strings_from_specs(writer, image_info,
 821                                           image_info_xml_string_specs,
 822                                           ARRAY_LEN(image_info_xml_string_specs));
 823         if (rc)
 824                 return rc;
 825
 826         rc = xmlTextWriterEndElement(writer); /* </IMAGE> */
 827         if (rc < 0)
 828                 return rc;
 829         return 0;
 830 }
 831
 832
 833
 834 /* Makes space for another image in the XML information and return a pointer to
 835  * it.*/
 836 static struct image_info *
 837 add_image_info_struct(struct wim_info *wim_info)
 838 {
 839         struct image_info *images;
 840
 841         images = CALLOC(wim_info->num_images + 1, sizeof(struct image_info));
 842         if (!images)
 843                 return NULL;
 844         memcpy(images, wim_info->images,
 845                wim_info->num_images * sizeof(struct image_info));
 846         FREE(wim_info->images);
 847         wim_info->images = images;
 848         wim_info->num_images++;
 849         return &images[wim_info->num_images - 1];
 850 }
 851
 852 static int
 853 clone_windows_info(const struct windows_info *old, struct windows_info *new)
 854 {
 855         int ret;
 856
 857         ret = dup_strings_from_specs(old, new, windows_info_xml_string_specs,
 858                                      ARRAY_LEN(windows_info_xml_string_specs));
 859
 860         if (old->languages) {
 861                 new->languages = CALLOC(old->num_languages, sizeof(new->languages[0]));
 862                 if (!new->languages)
 863                         return WIMLIB_ERR_NOMEM;
 864                 new->num_languages = old->num_languages;
 865                 for (size_t i = 0; i < new->num_languages; i++) {
 866                         if (!old->languages[i])
 867                                 continue;
 868                         new->languages[i] = TSTRDUP(old->languages[i]);
 869                         if (!new->languages[i])
 870                                 return WIMLIB_ERR_NOMEM;
 871                 }
 872         }
 873         if (old->default_language &&
 874                         !(new->default_language = TSTRDUP(old->default_language)))
 875                 return WIMLIB_ERR_NOMEM;
 876         if (old->system_root && !(new->system_root = TSTRDUP(old->system_root)))
 877                 return WIMLIB_ERR_NOMEM;
 878         if (old->windows_version_exists) {
 879                 new->windows_version_exists = true;
 880                 memcpy(&new->windows_version, &old->windows_version,
 881                        sizeof(old->windows_version));
 882         }
 883         return 0;
 884 }
 885
 886 static int
 887 clone_image_info(const struct image_info *old, struct image_info *new)
 888 {
 889         int ret;
 890
 891         new->dir_count              = old->dir_count;
 892         new->file_count             = old->file_count;
 893         new->total_bytes            = old->total_bytes;
 894         new->hard_link_bytes        = old->hard_link_bytes;
 895         new->creation_time          = old->creation_time;
 896         new->last_modification_time = old->last_modification_time;
 897
 898         ret = dup_strings_from_specs(old, new,
 899                                      image_info_xml_string_specs,
 900                                      ARRAY_LEN(image_info_xml_string_specs));
 901         if (ret)
 902                 return ret;
 903
 904         if (old->windows_info_exists) {
 905                 new->windows_info_exists = true;
 906                 ret = clone_windows_info(&old->windows_info,
 907                                          &new->windows_info);
 908                 if (ret)
 909                         return ret;
 910         }
 911         return 0;
 912 }
 913
 914 /* Copies the XML information for an image between WIM files.
 915  *
 916  * @dest_image_name and @dest_image_description are ignored if they are NULL;
 917  * otherwise, they are used to override the image name and/or image description
 918  * from the XML data in the source WIM file.
 919  *
 920  * On failure, WIMLIB_ERR_NOMEM is returned and no changes are made.  Otherwise,
 921  * 0 is returned and the WIM information at *new_wim_info_p is modified.
 922  */
 923 int
 924 xml_export_image(const struct wim_info *old_wim_info,
 925                  int image,
 926                  struct wim_info **new_wim_info_p,
 927                  const tchar *dest_image_name,
 928                  const tchar *dest_image_description)
 929 {
 930         struct wim_info *new_wim_info;
 931         struct image_info *image_info;
 932         int ret;
 933
 934         DEBUG("Copying XML data between WIM files for source image %d.", image);
 935
 936         wimlib_assert(old_wim_info != NULL);
 937         wimlib_assert(image >= 1 && image <= old_wim_info->num_images);
 938
 939         if (*new_wim_info_p) {
 940                 new_wim_info = *new_wim_info_p;
 941         } else {
 942                 new_wim_info = CALLOC(1, sizeof(struct wim_info));
 943                 if (!new_wim_info)
 944                         goto err;
 945         }
 946
 947         image_info = add_image_info_struct(new_wim_info);
 948         if (!image_info)
 949                 goto err;
 950
 951         ret = clone_image_info(&old_wim_info->images[image - 1], image_info);
 952         if (ret != 0)
 953                 goto err_destroy_image_info;
 954
 955         image_info->index = new_wim_info->num_images;
 956
 957         if (dest_image_name) {
 958                 FREE(image_info->name);
 959                 image_info->name = TSTRDUP(dest_image_name);
 960                 if (!image_info->name)
 961                         goto err_destroy_image_info;
 962         }
 963         if (dest_image_description) {
 964                 FREE(image_info->description);
 965                 image_info->description = TSTRDUP(dest_image_description);
 966                 if (!image_info->description)
 967                         goto err_destroy_image_info;
 968         }
 969         *new_wim_info_p = new_wim_info;
 970         return 0;
 971 err_destroy_image_info:
 972         destroy_image_info(image_info);
 973 err:
 974         if (new_wim_info != *new_wim_info_p)
 975                 free_wim_info(new_wim_info);
 976         return WIMLIB_ERR_NOMEM;
 977 }
 978
 979 /* Removes an image from the XML information. */
 980 void
 981 xml_delete_image(struct wim_info **wim_info_p, int image)
 982 {
 983         struct wim_info *wim_info;
 984
 985         DEBUG("Deleting image %d from the XML data.", image);
 986
 987         wim_info = *wim_info_p;
 988
 989         destroy_image_info(&wim_info->images[image - 1]);
 990
 991         memmove(&wim_info->images[image - 1],
 992                 &wim_info->images[image],
 993                 (wim_info->num_images - image) * sizeof(struct image_info));
 994
 995         if (--wim_info->num_images == 0) {
 996                 free_wim_info(wim_info);
 997                 *wim_info_p = NULL;
 998         } else {
 999                 for (int i = image - 1; i < wim_info->num_images; i++)
1000                         wim_info->images[i].index--;
1001         }
1002 }
1003
1004 size_t
1005 xml_get_max_image_name_len(const WIMStruct *w)
1006 {
1007         size_t max_len = 0;
1008         if (w->wim_info) {
1009                 for (int i = 0; i < w->wim_info->num_images; i++) {
1010                         size_t len = tstrlen(w->wim_info->images[i].name);
1011                         if (len > max_len)
1012                                 max_len = len;
1013                 }
1014         }
1015         return max_len;
1016 }
1017
1018 #ifdef ENABLE_CUSTOM_MEMORY_ALLOCATOR
1019 void
1020 xml_set_memory_allocator(void *(*malloc_func)(size_t),
1021                          void (*free_func)(void *),
1022                          void *(*realloc_func)(void *, size_t))
1023 {
1024         xmlMemSetup(free_func, malloc_func, realloc_func, STRDUP);
1025 }
1026 #endif
1027
1028 static int
1029 calculate_dentry_statistics(struct wim_dentry *dentry, void *arg)
1030 {
1031         struct image_info *info = arg;
1032         struct wim_lookup_table *lookup_table = info->lookup_table;
1033         const struct wim_inode *inode = dentry->d_inode;
1034         struct wim_lookup_table_entry *lte;
1035
1036         /* Update directory count and file count.
1037          *
1038          * Each dentry counts as either a file or a directory, but not both.
1039          * The root directory is an exception: it is not counted at all.
1040          *
1041          * Symbolic links and junction points (and presumably other reparse
1042          * points) count as regular files.  This is despite the fact that
1043          * junction points have FILE_ATTRIBUTE_DIRECTORY set.
1044          */
1045         if (dentry_is_root(dentry))
1046                 return 0;
1047
1048         if (inode_is_directory(inode))
1049                 info->dir_count++;
1050         else
1051                 info->file_count++;
1052
1053         /*
1054          * Update total bytes and hard link bytes.
1055          *
1056          * Unfortunately there are some inconsistencies/bugs in the way this is
1057          * done.
1058          *
1059          * If there are no alternate data streams in the image, the "total
1060          * bytes" is the sum of the size of the un-named data stream of each
1061          * inode times the link count of that inode.  In other words, it would
1062          * be the total number of bytes of regular files you would have if you
1063          * extracted the full image without any hard-links.  The "hard link
1064          * bytes" is equal to the "total bytes" minus the size of the un-named
1065          * data stream of each inode.  In other words, the "hard link bytes"
1066          * counts the size of the un-named data stream for all the links to each
1067          * inode except the first one.
1068          *
1069          * Reparse points and directories don't seem to be counted in either the
1070          * total bytes or the hard link bytes.
1071          *
1072          * And now we get to the most confusing part, the alternate data
1073          * streams.  They are not counted in the "total bytes".  However, if the
1074          * link count of an inode with alternate data streams is 2 or greater,
1075          * the size of all the alternate data streams is included in the "hard
1076          * link bytes", and this size is multiplied by the link count (NOT one
1077          * less than the link count).
1078          */
1079         lte = inode_unnamed_lte(inode, info->lookup_table);
1080         if (lte) {
1081                 info->total_bytes += wim_resource_size(lte);
1082                 if (!dentry_is_first_in_inode(dentry))
1083                         info->hard_link_bytes += wim_resource_size(lte);
1084         }
1085
1086         if (inode->i_nlink >= 2 && dentry_is_first_in_inode(dentry)) {
1087                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
1088                         if (inode->i_ads_entries[i].stream_name_nbytes) {
1089                                 lte = inode_stream_lte(inode, i + 1, lookup_table);
1090                                 if (lte) {
1091                                         info->hard_link_bytes += inode->i_nlink *
1092                                                                  wim_resource_size(lte);
1093                                 }
1094                         }
1095                 }
1096         }
1097         return 0;
1098 }
1099
1100 /*
1101  * Calculate what to put in the <FILECOUNT>, <DIRCOUNT>, <TOTALBYTES>, and
1102  * <HARDLINKBYTES> elements of each <IMAGE>.
1103  *
1104  * Please note there is no official documentation for exactly how this is done.
1105  * But, see calculate_dentry_statistics().
1106  */
1107 void
1108 xml_update_image_info(WIMStruct *w, int image)
1109 {
1110         struct image_info *image_info;
1111
1112         DEBUG("Updating the image info for image %d", image);
1113
1114         image_info = &w->wim_info->images[image - 1];
1115
1116         image_info->file_count      = 0;
1117         image_info->dir_count       = 0;
1118         image_info->total_bytes     = 0;
1119         image_info->hard_link_bytes = 0;
1120
1121         image_info->lookup_table = w->lookup_table;
1122         for_dentry_in_tree(w->image_metadata[image - 1].root_dentry,
1123                            calculate_dentry_statistics,
1124                            image_info);
1125         image_info->last_modification_time = get_wim_timestamp();
1126 }
1127
1128 /* Adds an image to the XML information. */
1129 int
1130 xml_add_image(WIMStruct *w, const tchar *name)
1131 {
1132         struct wim_info *wim_info;
1133         struct image_info *image_info;
1134
1135         wimlib_assert(name != NULL);
1136
1137         /* If this is the first image, allocate the struct wim_info.  Otherwise
1138          * use the existing struct wim_info. */
1139         if (w->wim_info) {
1140                 wim_info = w->wim_info;
1141         } else {
1142                 wim_info = CALLOC(1, sizeof(struct wim_info));
1143                 if (!wim_info)
1144                         return WIMLIB_ERR_NOMEM;
1145         }
1146
1147         image_info = add_image_info_struct(wim_info);
1148         if (!image_info)
1149                 goto out_free_wim_info;
1150
1151         if (!(image_info->name = TSTRDUP(name)))
1152                 goto out_destroy_image_info;
1153
1154         w->wim_info = wim_info;
1155         image_info->index = wim_info->num_images;
1156         image_info->creation_time = get_wim_timestamp();
1157         xml_update_image_info(w, image_info->index);
1158         return 0;
1159
1160 out_destroy_image_info:
1161         destroy_image_info(image_info);
1162         wim_info->num_images--;
1163 out_free_wim_info:
1164         if (wim_info != w->wim_info)
1165                 FREE(wim_info);
1166         return WIMLIB_ERR_NOMEM;
1167 }
1168
1169 /* Prints information about the specified image from struct wim_info structure.
1170  * */
1171 void
1172 print_image_info(const struct wim_info *wim_info, int image)
1173 {
1174         const struct image_info *image_info;
1175         const tchar *desc;
1176         tchar buf[50];
1177
1178         wimlib_assert(image >= 1 && image <= wim_info->num_images);
1179
1180         image_info = &wim_info->images[image - 1];
1181
1182         tprintf(T("Index:                  %d\n"), image_info->index);
1183         tprintf(T("Name:                   %"TS"\n"), image_info->name);
1184
1185         /* Always print the Description: part even if there is no
1186          * description. */
1187         if (image_info->description)
1188                 desc = image_info->description;
1189         else
1190                 desc = T("");
1191         tprintf(T("Description:            %"TS"\n"), desc);
1192
1193         if (image_info->display_name) {
1194                 tprintf(T("Display Name:           %"TS"\n"),
1195                         image_info->display_name);
1196         }
1197
1198         if (image_info->display_description) {
1199                 tprintf(T("Display Description:    %"TS"\n"),
1200                         image_info->display_description);
1201         }
1202
1203         tprintf(T("Directory Count:        %"PRIu64"\n"), image_info->dir_count);
1204         tprintf(T("File Count:             %"PRIu64"\n"), image_info->file_count);
1205         tprintf(T("Total Bytes:            %"PRIu64"\n"), image_info->total_bytes);
1206         tprintf(T("Hard Link Bytes:        %"PRIu64"\n"), image_info->hard_link_bytes);
1207
1208         wim_timestamp_to_str(image_info->creation_time, buf, sizeof(buf));
1209         tprintf(T("Creation Time:          %"TS"\n"), buf);
1210
1211         wim_timestamp_to_str(image_info->creation_time, buf, sizeof(buf));
1212         tprintf(T("Last Modification Time: %"TS"\n"), buf);
1213         if (image_info->windows_info_exists)
1214                 print_windows_info(&image_info->windows_info);
1215         if (image_info->flags)
1216                 tprintf(T("Flags:                  %"TS"\n"), image_info->flags);
1217         tputchar('\n');
1218 }
1219
1220 void
1221 libxml_global_init()
1222 {
1223         xmlInitParser();
1224         xmlInitCharEncodingHandlers();
1225 }
1226
1227 void
1228 libxml_global_cleanup()
1229 {
1230         xmlCleanupParser();
1231         xmlCleanupCharEncodingHandlers();
1232 }
1233
1234 /*
1235  * Reads the XML data from a WIM file.
1236  */
1237 int
1238 read_xml_data(FILE *fp, const struct resource_entry *res_entry,
1239               utf16lechar **xml_data_ret, struct wim_info **info_ret)
1240 {
1241         utf16lechar *xml_data;
1242         xmlDoc *doc;
1243         xmlNode *root;
1244         int ret;
1245
1246         DEBUG("XML data is %"PRIu64" bytes at offset %"PRIu64"",
1247               (u64)res_entry->size, res_entry->offset);
1248
1249         if (resource_is_compressed(res_entry)) {
1250                 ERROR("XML data is supposed to be uncompressed");
1251                 ret = WIMLIB_ERR_XML;
1252                 goto out;
1253         }
1254
1255         if (res_entry->size < 2) {
1256                 ERROR("XML data must be at least 2 bytes long");
1257                 ret = WIMLIB_ERR_XML;
1258                 goto out;
1259         }
1260
1261         xml_data = MALLOC(res_entry->size + 3);
1262         if (!xml_data) {
1263                 ret = WIMLIB_ERR_NOMEM;
1264                 goto out;
1265         }
1266
1267         ret = read_uncompressed_resource(fp, res_entry->offset,
1268                                          res_entry->size, xml_data);
1269         if (ret != 0)
1270                 goto out_free_xml_data;
1271
1272         /* Null-terminate just in case */
1273         ((u8*)xml_data)[res_entry->size] = 0;
1274         ((u8*)xml_data)[res_entry->size + 1] = 0;
1275         ((u8*)xml_data)[res_entry->size + 2] = 0;
1276
1277         DEBUG("Parsing XML using libxml2 to create XML tree");
1278
1279         doc = xmlReadMemory((const char *)xml_data,
1280                             res_entry->size, "noname.xml", "UTF-16", 0);
1281
1282         if (!doc) {
1283                 ERROR("Failed to parse XML data");
1284                 ret = WIMLIB_ERR_XML;
1285                 goto out_free_xml_data;
1286         }
1287
1288         DEBUG("Constructing WIM information structure from XML tree.");
1289
1290         root = xmlDocGetRootElement(doc);
1291         if (!root) {
1292                 ERROR("WIM XML data is an empty XML document");
1293                 ret = WIMLIB_ERR_XML;
1294                 goto out_free_doc;
1295         }
1296
1297         if (!node_is_element(root) || !node_name_is(root, "WIM")) {
1298                 ERROR("Expected <WIM> for the root XML element (found <%s>)",
1299                       root->name);
1300                 ret = WIMLIB_ERR_XML;
1301                 goto out_free_doc;
1302         }
1303
1304         ret = xml_read_wim_info(root, info_ret);
1305         if (ret != 0)
1306                 goto out_free_doc;
1307
1308         *xml_data_ret = xml_data;
1309         xml_data = NULL;
1310 out_free_doc:
1311         DEBUG("Freeing XML tree.");
1312         xmlFreeDoc(doc);
1313 out_free_xml_data:
1314         FREE(xml_data);
1315 out:
1316         return ret;
1317 }
1318
1319 #define CHECK_RET  ({   if (ret < 0)  { \
1320                                 ERROR("Error writing XML data"); \
1321                                 ret = WIMLIB_ERR_WRITE; \
1322                                 goto out_free_text_writer; \
1323                         } })
1324
1325 /*
1326  * Writes XML data to a WIM file.
1327  *
1328  * If @total_bytes is non-zero, it specifies what to write to the TOTALBYTES
1329  * element in the XML data.  If zero, TOTALBYTES is given the default value of
1330  * the offset of the XML data.
1331  */
1332 int
1333 write_xml_data(const struct wim_info *wim_info, int image, FILE *out,
1334                u64 total_bytes, struct resource_entry *out_res_entry)
1335 {
1336         xmlCharEncodingHandler *encoding_handler;
1337         xmlOutputBuffer *out_buffer;
1338         xmlTextWriter *writer;
1339         int ret;
1340         off_t start_offset;
1341         off_t end_offset;
1342
1343         wimlib_assert(image == WIMLIB_ALL_IMAGES ||
1344                         (wim_info != NULL && image >= 1 &&
1345                          image <= wim_info->num_images));
1346
1347         start_offset = ftello(out);
1348         if (start_offset == -1)
1349                 return WIMLIB_ERR_WRITE;
1350
1351         DEBUG("Writing XML data for image %d at offset %"PRIu64,
1352               image, start_offset);
1353
1354         /* 2 bytes endianness marker for UTF-16LE.  This is _required_ for WIM
1355          * XML data. */
1356         if ((putc(0xff, out)) == EOF || (putc(0xfe, out) == EOF)) {
1357                 ERROR_WITH_ERRNO("Error writing XML data");
1358                 return WIMLIB_ERR_WRITE;
1359         }
1360
1361         /* The contents of the <TOTALBYTES> element in the XML data, under the
1362          * <WIM> element (not the <IMAGE> element), is for non-split WIMs the
1363          * size of the WIM file excluding the XML data and integrity table.
1364          * This should be equal to the current position in the output stream,
1365          * since the XML data and integrity table are the last elements of the
1366          * WIM.
1367          *
1368          * For split WIMs, <TOTALBYTES> takes into account the entire WIM, not
1369          * just the current part.  In that case, @total_bytes should be passed
1370          * in to this function. */
1371         if (total_bytes == 0)
1372                 total_bytes = start_offset;
1373
1374         /* The encoding of the XML data must be UTF-16LE. */
1375         encoding_handler = xmlGetCharEncodingHandler(XML_CHAR_ENCODING_UTF16LE);
1376         if (!encoding_handler) {
1377                 ERROR("Failed to get XML character encoding handler for UTF-16LE");
1378                 ret = WIMLIB_ERR_LIBXML_UTF16_HANDLER_NOT_AVAILABLE;
1379                 goto out;
1380         }
1381
1382         out_buffer = xmlOutputBufferCreateFile(out, encoding_handler);
1383         if (!out_buffer) {
1384                 ERROR("Failed to allocate xmlOutputBuffer");
1385                 ret = WIMLIB_ERR_NOMEM;
1386                 goto out;
1387         }
1388
1389         writer = xmlNewTextWriter(out_buffer);
1390         if (!writer) {
1391                 ERROR("Failed to allocate xmlTextWriter");
1392                 ret = WIMLIB_ERR_NOMEM;
1393                 goto out_output_buffer_close;
1394         }
1395
1396         DEBUG("Writing <WIM> element");
1397
1398         ret = xmlTextWriterStartElement(writer, "WIM");
1399         CHECK_RET;
1400
1401         ret = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
1402                                               total_bytes);
1403         CHECK_RET;
1404
1405         if (wim_info != NULL) {
1406                 int first, last;
1407                 if (image == WIMLIB_ALL_IMAGES) {
1408                         first = 1;
1409                         last = wim_info->num_images;
1410                 } else {
1411                         first = image;
1412                         last = image;
1413                 }
1414                 DEBUG("Writing %d <IMAGE> elements", last - first + 1);
1415                 for (int i = first; i <= last; i++) {
1416                         ret = xml_write_image_info(writer, &wim_info->images[i - 1]);
1417                         if (ret) {
1418                                 CHECK_RET;
1419                                 goto out_free_text_writer;
1420                         }
1421                 }
1422         }
1423
1424         ret = xmlTextWriterEndElement(writer);
1425         CHECK_RET;
1426
1427         ret = xmlTextWriterEndDocument(writer);
1428         CHECK_RET;
1429
1430         DEBUG("Ended XML document");
1431
1432         /* Call xmlFreeTextWriter() before ftello() because the former will
1433          * flush the file stream. */
1434         xmlFreeTextWriter(writer);
1435         writer = NULL;
1436
1437         end_offset = ftello(out);
1438         if (end_offset == -1) {
1439                 ret = WIMLIB_ERR_WRITE;
1440         } else {
1441                 ret = 0;
1442                 out_res_entry->offset        = start_offset;
1443                 out_res_entry->size          = end_offset - start_offset;
1444                 out_res_entry->original_size = end_offset - start_offset;
1445                 out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
1446         }
1447 out_free_text_writer:
1448         /* xmlFreeTextWriter will free the attached xmlOutputBuffer. */
1449         xmlFreeTextWriter(writer);
1450         out_buffer = NULL;
1451 out_output_buffer_close:
1452         if (out_buffer != NULL)
1453                 xmlOutputBufferClose(out_buffer);
1454 out:
1455         if (ret == 0)
1456                 DEBUG("Successfully wrote XML data");
1457         return ret;
1458 }
1459
1460 /* Returns the name of the specified image. */
1461 WIMLIBAPI const tchar *
1462 wimlib_get_image_name(const WIMStruct *w, int image)
1463 {
1464         if (image < 1 || image > w->hdr.image_count)
1465                 return NULL;
1466         return w->wim_info->images[image - 1].name;
1467 }
1468
1469 /* Returns the description of the specified image. */
1470 WIMLIBAPI const tchar *
1471 wimlib_get_image_description(const WIMStruct *w, int image)
1472 {
1473         if (image < 1 || image > w->hdr.image_count)
1474                 return NULL;
1475         return w->wim_info->images[image - 1].description;
1476 }
1477
1478 /* Determines if an image name is already used by some image in the WIM. */
1479 WIMLIBAPI bool
1480 wimlib_image_name_in_use(const WIMStruct *w, const tchar *name)
1481 {
1482         if (!name || !*name)
1483                 return false;
1484         for (int i = 1; i <= w->hdr.image_count; i++)
1485                 if (tstrcmp(w->wim_info->images[i - 1].name, name) == 0)
1486                         return true;
1487         return false;
1488 }
1489
1490
1491 /* Extracts the raw XML data to a file stream. */
1492 WIMLIBAPI int
1493 wimlib_extract_xml_data(WIMStruct *w, FILE *fp)
1494 {
1495         size_t bytes_written;
1496
1497         if (!w->xml_data)
1498                 return WIMLIB_ERR_INVALID_PARAM;
1499         bytes_written = fwrite(w->xml_data, 1, w->hdr.xml_res_entry.size, fp);
1500         if (bytes_written != w->hdr.xml_res_entry.size) {
1501                 ERROR_WITH_ERRNO("Failed to extract XML data");
1502                 return WIMLIB_ERR_WRITE;
1503         }
1504         return 0;
1505 }
1506
1507 /* Sets the name of an image in the WIM. */
1508 WIMLIBAPI int
1509 wimlib_set_image_name(WIMStruct *w, int image, const tchar *name)
1510 {
1511         tchar *p;
1512         int i;
1513
1514         DEBUG("Setting the name of image %d to %"TS, image, name);
1515
1516         if (!name || !*name) {
1517                 ERROR("Must specify a non-empty string for the image name");
1518                 return WIMLIB_ERR_INVALID_PARAM;
1519         }
1520
1521         if (image < 1 || image > w->hdr.image_count) {
1522                 ERROR("%d is not a valid image", image);
1523                 return WIMLIB_ERR_INVALID_IMAGE;
1524         }
1525
1526         for (i = 1; i <= w->hdr.image_count; i++) {
1527                 if (i == image)
1528                         continue;
1529                 if (tstrcmp(w->wim_info->images[i - 1].name, name) == 0) {
1530                         ERROR("The name \"%"TS"\" is already in use in the WIM!",
1531                               name);
1532                         return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1533                 }
1534         }
1535
1536         p = TSTRDUP(name);
1537         if (!p)
1538                 return WIMLIB_ERR_NOMEM;
1539
1540         FREE(w->wim_info->images[image - 1].name);
1541         w->wim_info->images[image - 1].name = p;
1542         return 0;
1543 }
1544
1545 static int
1546 do_set_image_info_str(WIMStruct *w, int image, const tchar *tstr,
1547                       size_t offset)
1548 {
1549         tchar *tstr_copy;
1550         tchar **dest_tstr_p;
1551
1552         if (image < 1 || image > w->hdr.image_count) {
1553                 ERROR("%d is not a valid image", image);
1554                 return WIMLIB_ERR_INVALID_IMAGE;
1555         }
1556         if (tstr) {
1557                 tstr_copy = TSTRDUP(tstr);
1558                 if (!tstr_copy)
1559                         return WIMLIB_ERR_NOMEM;
1560         } else {
1561                 tstr_copy = NULL;
1562         }
1563         dest_tstr_p = (tchar**)((void*)&w->wim_info->images[image - 1] + offset);
1564
1565         FREE(*dest_tstr_p);
1566         *dest_tstr_p = tstr_copy;
1567         return 0;
1568 }
1569
1570 /* Sets the description of an image in the WIM. */
1571 WIMLIBAPI int
1572 wimlib_set_image_descripton(WIMStruct *w, int image,
1573                             const tchar *description)
1574 {
1575         return do_set_image_info_str(w, image, description,
1576                                      offsetof(struct image_info, description));
1577 }
1578
1579 /* Set the <FLAGS> element of a WIM image */
1580 WIMLIBAPI int
1581 wimlib_set_image_flags(WIMStruct *w, int image, const tchar *flags)
1582 {
1583         return do_set_image_info_str(w, image, flags,
1584                                      offsetof(struct image_info, flags));
1585 }