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 #ifdef HAVE_CONFIG_H
  27 #  include "config.h"
  28 #endif
  29
  30 #include "wimlib/dentry.h"
  31 #include "wimlib/encoding.h"
  32 #include "wimlib/error.h"
  33 #include "wimlib/file_io.h"
  34 #include "wimlib/lookup_table.h"
  35 #include "wimlib/metadata.h"
  36 #include "wimlib/resource.h"
  37 #include "wimlib/timestamp.h"
  38 #include "wimlib/xml.h"
  39
  40 #include <libxml/encoding.h>
  41 #include <libxml/parser.h>
  42 #include <libxml/tree.h>
  43 #include <libxml/xmlwriter.h>
  44 #include <limits.h>
  45 #include <string.h>
  46
  47 /* Structures used to form an in-memory representation of the XML data (other
  48  * than the raw parse tree from libxml). */
  49
  50 struct windows_version {
  51         u64 major;
  52         u64 minor;
  53         u64 build;
  54         u64 sp_build;
  55         u64 sp_level;
  56 };
  57
  58 struct windows_info {
  59         u64      arch;
  60         tchar   *product_name;
  61         tchar   *edition_id;
  62         tchar   *installation_type;
  63         tchar   *hal;
  64         tchar   *product_type;
  65         tchar   *product_suite;
  66         tchar  **languages;
  67         tchar   *default_language;
  68         size_t   num_languages;
  69         tchar   *system_root;
  70         bool     windows_version_exists;
  71         struct   windows_version windows_version;
  72 };
  73
  74 struct image_info {
  75         int index;
  76         bool windows_info_exists;
  77         u64 dir_count;
  78         u64 file_count;
  79         u64 total_bytes;
  80         u64 hard_link_bytes;
  81         u64 creation_time;
  82         u64 last_modification_time;
  83         struct windows_info windows_info;
  84         tchar *name;
  85         tchar *description;
  86         tchar *display_name;
  87         tchar *display_description;
  88         tchar *flags;
  89         struct wim_lookup_table *lookup_table; /* temporary field */
  90 };
  91
  92 struct xml_string_spec {
  93         const char *name;
  94         size_t offset;
  95 };
  96
  97 #define ELEM(STRING_NAME, MEMBER_NAME) \
  98         {STRING_NAME, offsetof(struct image_info, MEMBER_NAME)}
  99 static const struct xml_string_spec
 100 image_info_xml_string_specs[] = {
 101         ELEM("NAME", name),
 102         ELEM("DESCRIPTION", description),
 103         ELEM("DISPLAYNAME", display_name),
 104         ELEM("DISPLAYDESCRIPTION", display_description),
 105         ELEM("FLAGS", flags),
 106 };
 107 #undef ELEM
 108
 109 #define ELEM(STRING_NAME, MEMBER_NAME) \
 110         {STRING_NAME, offsetof(struct windows_info, MEMBER_NAME)}
 111 static const struct xml_string_spec
 112 windows_info_xml_string_specs[] = {
 113         ELEM("PRODUCTNAME", product_name),
 114         ELEM("EDITIONID", edition_id),
 115         ELEM("INSTALLATIONTYPE", installation_type),
 116         ELEM("HAL", hal),
 117         ELEM("PRODUCTTYPE", product_type),
 118         ELEM("PRODUCTSUITE", product_suite),
 119 };
 120 #undef ELEM
 121
 122
 123 /* Returns a statically allocated string that is a string representation of the
 124  * architecture number. */
 125 static const tchar *
 126 get_arch(int arch)
 127 {
 128         switch (arch) {
 129         case 0:
 130                 return T("x86");
 131         case 6:
 132                 return T("ia64");
 133         case 9:
 134                 return T("x86_64");
 135         /* XXX Are there other arch values? */
 136         default:
 137                 return T("unknown");
 138         }
 139 }
 140
 141
 142 /* Iterate through the children of an xmlNode. */
 143 #define for_node_child(parent, child)   \
 144         for (child = parent->children; child != NULL; child = child->next)
 145
 146 /* Utility functions for xmlNodes */
 147 static inline bool
 148 node_is_element(xmlNode *node)
 149 {
 150         return node->type == XML_ELEMENT_NODE;
 151 }
 152
 153 static inline bool
 154 node_is_text(xmlNode *node)
 155 {
 156         return node->type == XML_TEXT_NODE;
 157 }
 158
 159 static inline bool
 160 node_name_is(xmlNode *node, const char *name)
 161 {
 162         /* For now, both upper case and lower case element names are accepted. */
 163         return strcasecmp((const char *)node->name, name) == 0;
 164 }
 165
 166 static u64
 167 node_get_number(const xmlNode *u64_node, int base)
 168 {
 169         xmlNode *child;
 170         for_node_child(u64_node, child)
 171                 if (node_is_text(child))
 172                         return strtoull(child->content, NULL, base);
 173         return 0;
 174 }
 175
 176 /* Finds the text node that is a child of an element node and returns its
 177  * content converted to a 64-bit unsigned integer.  Returns 0 if no text node is
 178  * found. */
 179 static u64
 180 node_get_u64(const xmlNode *u64_node)
 181 {
 182         return node_get_number(u64_node, 10);
 183 }
 184
 185 /* Like node_get_u64(), but expects a number in base 16. */
 186 static u64
 187 node_get_hex_u64(const xmlNode *u64_node)
 188 {
 189         return node_get_number(u64_node, 16);
 190 }
 191
 192 static int
 193 node_get_string(const xmlNode *string_node, tchar **tstr_ret)
 194 {
 195         xmlNode *child;
 196         tchar *tstr = NULL;
 197         int ret;
 198
 199         for_node_child(string_node, child) {
 200                 if (node_is_text(child) && child->content) {
 201                         ret = utf8_to_tstr_simple(child->content, &tstr);
 202                         if (ret)
 203                                 return ret;
 204                         break;
 205                 }
 206         }
 207         *tstr_ret = tstr;
 208         return 0;
 209 }
 210
 211 /* Returns the timestamp from a time node.  It has child elements <HIGHPART> and
 212  * <LOWPART> that are then used to construct a 64-bit timestamp. */
 213 static u64
 214 node_get_timestamp(const xmlNode *time_node)
 215 {
 216         u32 high_part = 0;
 217         u32 low_part = 0;
 218         xmlNode *child;
 219         for_node_child(time_node, child) {
 220                 if (!node_is_element(child))
 221                         continue;
 222                 if (node_name_is(child, "HIGHPART"))
 223                         high_part = node_get_hex_u64(child);
 224                 else if (node_name_is(child, "LOWPART"))
 225                         low_part = node_get_hex_u64(child);
 226         }
 227         return (u64)low_part | ((u64)high_part << 32);
 228 }
 229
 230 /* Used to sort an array of struct image_infos by their image indices. */
 231 static int
 232 sort_by_index(const void *p1, const void *p2)
 233 {
 234         int index_1 = ((const struct image_info*)p1)->index;
 235         int index_2 = ((const struct image_info*)p2)->index;
 236         if (index_1 < index_2)
 237                 return -1;
 238         else if (index_1 > index_2)
 239                 return 1;
 240         else
 241                 return 0;
 242 }
 243
 244
 245 /* Frees memory allocated inside a struct windows_info structure. */
 246 static void
 247 destroy_windows_info(struct windows_info *windows_info)
 248 {
 249         FREE(windows_info->product_name);
 250         FREE(windows_info->edition_id);
 251         FREE(windows_info->installation_type);
 252         FREE(windows_info->hal);
 253         FREE(windows_info->product_type);
 254         FREE(windows_info->product_suite);
 255         for (size_t i = 0; i < windows_info->num_languages; i++)
 256                 FREE(windows_info->languages[i]);
 257         FREE(windows_info->languages);
 258         FREE(windows_info->default_language);
 259         FREE(windows_info->system_root);
 260 }
 261
 262 /* Frees memory allocated inside a struct image_info structure. */
 263 static void
 264 destroy_image_info(struct image_info *image_info)
 265 {
 266         FREE(image_info->name);
 267         FREE(image_info->description);
 268         FREE(image_info->flags);
 269         FREE(image_info->display_name);
 270         FREE(image_info->display_description);
 271         destroy_windows_info(&image_info->windows_info);
 272         memset(image_info, 0, sizeof(struct image_info));
 273 }
 274
 275 void
 276 free_wim_info(struct wim_info *info)
 277 {
 278         if (info) {
 279                 if (info->images) {
 280                         for (int i = 0; i < info->num_images; i++)
 281                                 destroy_image_info(&info->images[i]);
 282                         FREE(info->images);
 283                 }
 284                 FREE(info);
 285         }
 286 }
 287
 288 /* Reads the information from a <VERSION> element inside the <WINDOWS> element.
 289  * */
 290 static void
 291 xml_read_windows_version(const xmlNode *version_node,
 292                          struct windows_version* windows_version)
 293 {
 294         xmlNode *child;
 295         for_node_child(version_node, child) {
 296                 if (!node_is_element(child))
 297                         continue;
 298                 if (node_name_is(child, "MAJOR"))
 299                         windows_version->major = node_get_u64(child);
 300                 else if (node_name_is(child, "MINOR"))
 301                         windows_version->minor = node_get_u64(child);
 302                 else if (node_name_is(child, "BUILD"))
 303                         windows_version->build = node_get_u64(child);
 304                 else if (node_name_is(child, "SPBUILD"))
 305                         windows_version->sp_build = node_get_u64(child);
 306                 else if (node_name_is(child, "SPLEVEL"))
 307                         windows_version->sp_level = node_get_u64(child);
 308         }
 309 }
 310
 311 /* Reads the information from a <LANGUAGE> element inside a <WINDOWS> element.
 312  * */
 313 static int
 314 xml_read_languages(const xmlNode *languages_node,
 315                    tchar ***languages_ret,
 316                    size_t *num_languages_ret,
 317                    tchar **default_language_ret)
 318 {
 319         xmlNode *child;
 320         size_t num_languages = 0;
 321         tchar **languages;
 322         int ret;
 323
 324         for_node_child(languages_node, child)
 325                 if (node_is_element(child) && node_name_is(child, "LANGUAGE"))
 326                         num_languages++;
 327
 328         languages = CALLOC(num_languages, sizeof(languages[0]));
 329         if (!languages)
 330                 return WIMLIB_ERR_NOMEM;
 331
 332         *languages_ret = languages;
 333         *num_languages_ret = num_languages;
 334
 335         ret = 0;
 336         for_node_child(languages_node, child) {
 337                 if (!node_is_element(child))
 338                         continue;
 339                 if (node_name_is(child, "LANGUAGE"))
 340                         ret = node_get_string(child, languages++);
 341                 else if (node_name_is(child, "DEFAULT"))
 342                         ret = node_get_string(child, default_language_ret);
 343                 if (ret != 0)
 344                         break;
 345         }
 346         return ret;
 347 }
 348
 349 /* Reads the information from a <WINDOWS> element inside an <IMAGE> element. */
 350 static int
 351 xml_read_windows_info(const xmlNode *windows_node,
 352                       struct windows_info *windows_info)
 353 {
 354         xmlNode *child;
 355         int ret = 0;
 356
 357         for_node_child(windows_node, child) {
 358                 if (!node_is_element(child))
 359                         continue;
 360                 if (node_name_is(child, "ARCH")) {
 361                         windows_info->arch = node_get_u64(child);
 362                 } else if (node_name_is(child, "PRODUCTNAME")) {
 363                         ret = node_get_string(child,
 364                                               &windows_info->product_name);
 365                 } else if (node_name_is(child, "EDITIONID")) {
 366                         ret = node_get_string(child,
 367                                               &windows_info->edition_id);
 368                 } else if (node_name_is(child, "INSTALLATIONTYPE")) {
 369                         ret = node_get_string(child,
 370                                               &windows_info->installation_type);
 371                 } else if (node_name_is(child, "PRODUCTTYPE")) {
 372                         ret = node_get_string(child,
 373                                               &windows_info->product_type);
 374                 } else if (node_name_is(child, "PRODUCTSUITE")) {
 375                         ret = node_get_string(child,
 376                                               &windows_info->product_suite);
 377                 } else if (node_name_is(child, "LANGUAGES")) {
 378                         ret = xml_read_languages(child,
 379                                                  &windows_info->languages,
 380                                                  &windows_info->num_languages,
 381                                                  &windows_info->default_language);
 382                 } else if (node_name_is(child, "VERSION")) {
 383                         xml_read_windows_version(child,
 384                                                 &windows_info->windows_version);
 385                         windows_info->windows_version_exists = true;
 386                 } else if (node_name_is(child, "SYSTEMROOT")) {
 387                         ret = node_get_string(child, &windows_info->system_root);
 388                 } else if (node_name_is(child, "HAL")) {
 389                         ret = node_get_string(child, &windows_info->hal);
 390                 }
 391                 if (ret != 0)
 392                         return ret;
 393         }
 394         return ret;
 395 }
 396
 397 /* Reads the information from an <IMAGE> element. */
 398 static int
 399 xml_read_image_info(xmlNode *image_node, struct image_info *image_info)
 400 {
 401         xmlNode *child;
 402         xmlChar *index_prop;
 403         int ret;
 404
 405         index_prop = xmlGetProp(image_node, "INDEX");
 406         if (index_prop) {
 407                 image_info->index = atoi(index_prop);
 408                 FREE(index_prop);
 409         } else {
 410                 image_info->index = 1;
 411         }
 412
 413         ret = 0;
 414         for_node_child(image_node, child) {
 415                 if (!node_is_element(child))
 416                         continue;
 417                 if (node_name_is(child, "DIRCOUNT"))
 418                         image_info->dir_count = node_get_u64(child);
 419                 else if (node_name_is(child, "FILECOUNT"))
 420                         image_info->file_count = node_get_u64(child);
 421                 else if (node_name_is(child, "TOTALBYTES"))
 422                         image_info->total_bytes = node_get_u64(child);
 423                 else if (node_name_is(child, "HARDLINKBYTES"))
 424                         image_info->hard_link_bytes = node_get_u64(child);
 425                 else if (node_name_is(child, "CREATIONTIME"))
 426                         image_info->creation_time = node_get_timestamp(child);
 427                 else if (node_name_is(child, "LASTMODIFICATIONTIME"))
 428                         image_info->last_modification_time = node_get_timestamp(child);
 429                 else if (node_name_is(child, "WINDOWS")) {
 430                         DEBUG("Found <WINDOWS> tag");
 431                         ret = xml_read_windows_info(child,
 432                                                     &image_info->windows_info);
 433                         image_info->windows_info_exists = true;
 434                 } else if (node_name_is(child, "NAME")) {
 435                         ret = node_get_string(child, &image_info->name);
 436                 } else if (node_name_is(child, "DESCRIPTION")) {
 437                         ret = node_get_string(child, &image_info->description);
 438                 } else if (node_name_is(child, "FLAGS")) {
 439                         ret = node_get_string(child, &image_info->flags);
 440                 } else if (node_name_is(child, "DISPLAYNAME")) {
 441                         ret = node_get_string(child, &image_info->display_name);
 442                 } else if (node_name_is(child, "DISPLAYDESCRIPTION")) {
 443                         ret = node_get_string(child, &image_info->display_description);
 444                 }
 445                 if (ret != 0)
 446                         return ret;
 447         }
 448         if (!image_info->name) {
 449                 tchar *empty_name;
 450                 WARNING("Image with index %d has no name", image_info->index);
 451                 empty_name = MALLOC(sizeof(tchar));
 452                 if (!empty_name)
 453                         return WIMLIB_ERR_NOMEM;
 454                 *empty_name = T('\0');
 455                 image_info->name = empty_name;
 456         }
 457         return ret;
 458 }
 459
 460 /* Reads the information from a <WIM> element, which should be the root element
 461  * of the XML tree. */
 462 static int
 463 xml_read_wim_info(const xmlNode *wim_node, struct wim_info **wim_info_ret)
 464 {
 465         struct wim_info *wim_info;
 466         xmlNode *child;
 467         int ret;
 468         int num_images;
 469         int i;
 470
 471         wim_info = CALLOC(1, sizeof(struct wim_info));
 472         if (!wim_info)
 473                 return WIMLIB_ERR_NOMEM;
 474
 475         /* Count how many images there are. */
 476         num_images = 0;
 477         for_node_child(wim_node, child) {
 478                 if (node_is_element(child) && node_name_is(child, "IMAGE")) {
 479                         if (num_images == INT_MAX) {
 480                                 return WIMLIB_ERR_IMAGE_COUNT;
 481                         }
 482                         num_images++;
 483                 }
 484         }
 485
 486         if (num_images > 0) {
 487                 /* Allocate the array of struct image_infos and fill them in. */
 488                 wim_info->images = CALLOC(num_images, sizeof(wim_info->images[0]));
 489                 if (!wim_info->images) {
 490                         ret = WIMLIB_ERR_NOMEM;
 491                         goto err;
 492                 }
 493                 wim_info->num_images = num_images;
 494                 i = 0;
 495                 for_node_child(wim_node, child) {
 496                         if (!node_is_element(child))
 497                                 continue;
 498                         if (node_name_is(child, "IMAGE")) {
 499                                 DEBUG("Found <IMAGE> tag");
 500                                 ret = xml_read_image_info(child,
 501                                                           &wim_info->images[i]);
 502                                 if (ret != 0)
 503                                         goto err;
 504                                 i++;
 505                         } else if (node_name_is(child, "TOTALBYTES")) {
 506                                 wim_info->total_bytes = node_get_u64(child);
 507                         }
 508                 }
 509
 510                 /* Sort the array of image info by image index. */
 511                 qsort(wim_info->images, num_images,
 512                       sizeof(struct image_info), sort_by_index);
 513
 514                 /* Make sure the image indices make sense */
 515                 for (i = 0; i < num_images; i++) {
 516                         if (wim_info->images[i].index != i + 1) {
 517                                 ERROR("WIM images are not indexed [1...%d] "
 518                                       "in XML data as expected",
 519                                       num_images);
 520                                 return WIMLIB_ERR_IMAGE_COUNT;
 521                         }
 522                 }
 523
 524         }
 525         *wim_info_ret = wim_info;
 526         return 0;
 527 err:
 528         free_wim_info(wim_info);
 529         return ret;
 530 }
 531
 532 /* Prints the information contained in a `struct windows_info'.
 533  *
 534  * Warning: any strings printed here are in UTF-8 encoding.  If the locale
 535  * character encoding is not UTF-8, the printed strings may be garbled. */
 536 static void
 537 print_windows_info(const struct windows_info *windows_info)
 538 {
 539         const struct windows_version *windows_version;
 540
 541         tprintf(T("Architecture:           %"TS"\n"),
 542                 get_arch(windows_info->arch));
 543
 544         if (windows_info->product_name) {
 545                 tprintf(T("Product Name:           %"TS"\n"),
 546                         windows_info->product_name);
 547         }
 548
 549         if (windows_info->edition_id) {
 550                 tprintf(T("Edition ID:             %"TS"\n"),
 551                         windows_info->edition_id);
 552         }
 553
 554         if (windows_info->installation_type) {
 555                 tprintf(T("Installation Type:      %"TS"\n"),
 556                         windows_info->installation_type);
 557         }
 558
 559         if (windows_info->hal) {
 560                 tprintf(T("HAL:                    %"TS"\n"),
 561                               windows_info->hal);
 562         }
 563
 564         if (windows_info->product_type) {
 565                 tprintf(T("Product Type:           %"TS"\n"),
 566                         windows_info->product_type);
 567         }
 568
 569         if (windows_info->product_suite) {
 570                 tprintf(T("Product Suite:          %"TS"\n"),
 571                         windows_info->product_suite);
 572         }
 573
 574         tprintf(T("Languages:              "));
 575         for (size_t i = 0; i < windows_info->num_languages; i++) {
 576
 577                 tfputs(windows_info->languages[i], stdout);
 578                 tputchar(T(' '));
 579         }
 580         tputchar(T('\n'));
 581         if (windows_info->default_language) {
 582                 tprintf(T("Default Language:       %"TS"\n"),
 583                         windows_info->default_language);
 584         }
 585         if (windows_info->system_root) {
 586                 tprintf(T("System Root:            %"TS"\n"),
 587                               windows_info->system_root);
 588         }
 589
 590         if (windows_info->windows_version_exists) {
 591                 windows_version = &windows_info->windows_version;
 592                 tprintf(T("Major Version:          %"PRIu64"\n"),
 593                         windows_version->major);
 594                 tprintf(T("Minor Version:          %"PRIu64"\n"),
 595                         windows_version->minor);
 596                 tprintf(T("Build:                  %"PRIu64"\n"),
 597                         windows_version->build);
 598                 tprintf(T("Service Pack Build:     %"PRIu64"\n"),
 599                         windows_version->sp_build);
 600                 tprintf(T("Service Pack Level:     %"PRIu64"\n"),
 601                         windows_version->sp_level);
 602         }
 603 }
 604
 605 static int
 606 xml_write_string(xmlTextWriter *writer, const char *name,
 607                  const tchar *tstr)
 608 {
 609         if (tstr) {
 610                 char *utf8_str;
 611                 int rc = tstr_to_utf8_simple(tstr, &utf8_str);
 612                 if (rc)
 613                         return rc;
 614                 rc = xmlTextWriterWriteElement(writer, name, utf8_str);
 615                 FREE(utf8_str);
 616                 if (rc < 0)
 617                         return rc;
 618         }
 619         return 0;
 620 }
 621
 622 static int
 623 xml_write_strings_from_specs(xmlTextWriter *writer,
 624                              const void *struct_with_strings,
 625                              const struct xml_string_spec specs[],
 626                              size_t num_specs)
 627 {
 628         for (size_t i = 0; i < num_specs; i++) {
 629                 int rc = xml_write_string(writer, specs[i].name,
 630                                       *(const tchar * const *)
 631                                         (struct_with_strings + specs[i].offset));
 632                 if (rc)
 633                         return rc;
 634         }
 635         return 0;
 636 }
 637
 638 static int
 639 dup_strings_from_specs(const void *old_struct_with_strings,
 640                        void *new_struct_with_strings,
 641                        const struct xml_string_spec specs[],
 642                        size_t num_specs)
 643 {
 644         for (size_t i = 0; i < num_specs; i++) {
 645                 const tchar *old_str = *(const tchar * const *)
 646                                         ((const void*)old_struct_with_strings + specs[i].offset);
 647                 tchar **new_str_p = (tchar **)((void*)new_struct_with_strings + specs[i].offset);
 648                 if (old_str) {
 649                         *new_str_p = TSTRDUP(old_str);
 650                         if (!*new_str_p)
 651                                 return WIMLIB_ERR_NOMEM;
 652                 }
 653         }
 654         return 0;
 655 }
 656
 657 /* Writes the information contained in a `struct windows_version' to the XML
 658  * document being written.  This is the <VERSION> element inside the <WINDOWS>
 659  * element. */
 660 static int
 661 xml_write_windows_version(xmlTextWriter *writer,
 662                           const struct windows_version *version)
 663 {
 664         int rc;
 665         rc = xmlTextWriterStartElement(writer, "VERSION");
 666         if (rc < 0)
 667                 return rc;
 668
 669         rc = xmlTextWriterWriteFormatElement(writer, "MAJOR", "%"PRIu64,
 670                                              version->major);
 671         if (rc < 0)
 672                 return rc;
 673
 674         rc = xmlTextWriterWriteFormatElement(writer, "MINOR", "%"PRIu64,
 675                                              version->minor);
 676         if (rc < 0)
 677                 return rc;
 678
 679         rc = xmlTextWriterWriteFormatElement(writer, "BUILD", "%"PRIu64,
 680                                              version->build);
 681         if (rc < 0)
 682                 return rc;
 683
 684         rc = xmlTextWriterWriteFormatElement(writer, "SPBUILD", "%"PRIu64,
 685                                              version->sp_build);
 686         if (rc < 0)
 687                 return rc;
 688
 689         rc = xmlTextWriterWriteFormatElement(writer, "SPLEVEL", "%"PRIu64,
 690                                              version->sp_level);
 691         if (rc < 0)
 692                 return rc;
 693
 694         return xmlTextWriterEndElement(writer); /* </VERSION> */
 695 }
 696
 697 /* Writes the information contained in a `struct windows_info' to the XML
 698  * document being written. This is the <WINDOWS> element. */
 699 static int
 700 xml_write_windows_info(xmlTextWriter *writer,
 701                        const struct windows_info *windows_info)
 702 {
 703         int rc;
 704         rc = xmlTextWriterStartElement(writer, "WINDOWS");
 705         if (rc < 0)
 706                 return rc;
 707
 708         rc = xmlTextWriterWriteFormatElement(writer, "ARCH", "%"PRIu64,
 709                                              windows_info->arch);
 710         if (rc < 0)
 711                 return rc;
 712
 713         rc = xml_write_strings_from_specs(writer,
 714                                           windows_info,
 715                                           windows_info_xml_string_specs,
 716                                           ARRAY_LEN(windows_info_xml_string_specs));
 717         if (rc)
 718                 return rc;
 719
 720         if (windows_info->num_languages) {
 721                 rc = xmlTextWriterStartElement(writer, "LANGUAGES");
 722                 if (rc < 0)
 723                         return rc;
 724
 725                 for (size_t i = 0; i < windows_info->num_languages; i++) {
 726                         rc = xml_write_string(writer, "LANGUAGE",
 727                                               windows_info->languages[i]);
 728                         if (rc)
 729                                 return rc;
 730                 }
 731
 732                 rc = xml_write_string(writer, "DEFAULT",
 733                                       windows_info->default_language);
 734                 if (rc)
 735                         return rc;
 736
 737                 rc = xmlTextWriterEndElement(writer); /* </LANGUAGES> */
 738                 if (rc < 0)
 739                         return rc;
 740         }
 741
 742         if (windows_info->windows_version_exists) {
 743                 rc = xml_write_windows_version(writer, &windows_info->windows_version);
 744                 if (rc < 0)
 745                         return rc;
 746         }
 747
 748         rc = xml_write_string(writer, "SYSTEMROOT",
 749                               windows_info->system_root);
 750         if (rc)
 751                 return rc;
 752
 753         return xmlTextWriterEndElement(writer); /* </WINDOWS> */
 754 }
 755
 756 /* Writes a time element to the XML document being constructed in memory. */
 757 static int
 758 xml_write_time(xmlTextWriter *writer, const char *element_name, u64 time)
 759 {
 760         int rc;
 761         rc = xmlTextWriterStartElement(writer, element_name);
 762         if (rc < 0)
 763                 return rc;
 764
 765         rc = xmlTextWriterWriteFormatElement(writer, "HIGHPART",
 766                                              "0x%08"PRIX32, (u32)(time >> 32));
 767         if (rc < 0)
 768                 return rc;
 769
 770         rc = xmlTextWriterWriteFormatElement(writer, "LOWPART",
 771                                              "0x%08"PRIX32, (u32)time);
 772         if (rc < 0)
 773                 return rc;
 774
 775         rc = xmlTextWriterEndElement(writer); /* </@element_name> */
 776         if (rc < 0)
 777                 return rc;
 778         return 0;
 779 }
 780
 781 /* Writes an <IMAGE> element to the XML document. */
 782 static int
 783 xml_write_image_info(xmlTextWriter *writer, const struct image_info *image_info)
 784 {
 785         int rc;
 786         rc = xmlTextWriterStartElement(writer, "IMAGE");
 787         if (rc < 0)
 788                 return rc;
 789
 790         rc = xmlTextWriterWriteFormatAttribute(writer, "INDEX", "%d",
 791                                                image_info->index);
 792         if (rc < 0)
 793                 return rc;
 794
 795         rc = xmlTextWriterWriteFormatElement(writer, "DIRCOUNT", "%"PRIu64,
 796                                              image_info->dir_count);
 797         if (rc < 0)
 798                 return rc;
 799
 800         rc = xmlTextWriterWriteFormatElement(writer, "FILECOUNT", "%"PRIu64,
 801                                              image_info->file_count);
 802         if (rc < 0)
 803                 return rc;
 804
 805         rc = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
 806                                              image_info->total_bytes);
 807         if (rc < 0)
 808                 return rc;
 809
 810         rc = xmlTextWriterWriteFormatElement(writer, "HARDLINKBYTES", "%"PRIu64,
 811                                              image_info->hard_link_bytes);
 812         if (rc < 0)
 813                 return rc;
 814
 815         rc = xml_write_time(writer, "CREATIONTIME", image_info->creation_time);
 816         if (rc < 0)
 817                 return rc;
 818
 819         rc = xml_write_time(writer, "LASTMODIFICATIONTIME",
 820                             image_info->last_modification_time);
 821         if (rc < 0)
 822                 return rc;
 823
 824         if (image_info->windows_info_exists) {
 825                 rc = xml_write_windows_info(writer, &image_info->windows_info);
 826                 if (rc)
 827                         return rc;
 828         }
 829
 830         rc = xml_write_strings_from_specs(writer, image_info,
 831                                           image_info_xml_string_specs,
 832                                           ARRAY_LEN(image_info_xml_string_specs));
 833         if (rc)
 834                 return rc;
 835
 836         rc = xmlTextWriterEndElement(writer); /* </IMAGE> */
 837         if (rc < 0)
 838                 return rc;
 839         return 0;
 840 }
 841
 842
 843
 844 /* Makes space for another image in the XML information and return a pointer to
 845  * it.*/
 846 static struct image_info *
 847 add_image_info_struct(struct wim_info *wim_info)
 848 {
 849         struct image_info *images;
 850
 851         images = CALLOC(wim_info->num_images + 1, sizeof(struct image_info));
 852         if (!images)
 853                 return NULL;
 854         memcpy(images, wim_info->images,
 855                wim_info->num_images * sizeof(struct image_info));
 856         FREE(wim_info->images);
 857         wim_info->images = images;
 858         wim_info->num_images++;
 859         return &images[wim_info->num_images - 1];
 860 }
 861
 862 static int
 863 clone_windows_info(const struct windows_info *old, struct windows_info *new)
 864 {
 865         int ret;
 866
 867         ret = dup_strings_from_specs(old, new, windows_info_xml_string_specs,
 868                                      ARRAY_LEN(windows_info_xml_string_specs));
 869         if (ret)
 870                 return ret;
 871
 872         if (old->languages) {
 873                 new->languages = CALLOC(old->num_languages, sizeof(new->languages[0]));
 874                 if (!new->languages)
 875                         return WIMLIB_ERR_NOMEM;
 876                 new->num_languages = old->num_languages;
 877                 for (size_t i = 0; i < new->num_languages; i++) {
 878                         if (!old->languages[i])
 879                                 continue;
 880                         new->languages[i] = TSTRDUP(old->languages[i]);
 881                         if (!new->languages[i])
 882                                 return WIMLIB_ERR_NOMEM;
 883                 }
 884         }
 885         if (old->default_language &&
 886                         !(new->default_language = TSTRDUP(old->default_language)))
 887                 return WIMLIB_ERR_NOMEM;
 888         if (old->system_root && !(new->system_root = TSTRDUP(old->system_root)))
 889                 return WIMLIB_ERR_NOMEM;
 890         if (old->windows_version_exists) {
 891                 new->windows_version_exists = true;
 892                 memcpy(&new->windows_version, &old->windows_version,
 893                        sizeof(old->windows_version));
 894         }
 895         return 0;
 896 }
 897
 898 static int
 899 clone_image_info(const struct image_info *old, struct image_info *new)
 900 {
 901         int ret;
 902
 903         new->dir_count              = old->dir_count;
 904         new->file_count             = old->file_count;
 905         new->total_bytes            = old->total_bytes;
 906         new->hard_link_bytes        = old->hard_link_bytes;
 907         new->creation_time          = old->creation_time;
 908         new->last_modification_time = old->last_modification_time;
 909
 910         ret = dup_strings_from_specs(old, new,
 911                                      image_info_xml_string_specs,
 912                                      ARRAY_LEN(image_info_xml_string_specs));
 913         if (ret)
 914                 return ret;
 915
 916         if (old->windows_info_exists) {
 917                 new->windows_info_exists = true;
 918                 ret = clone_windows_info(&old->windows_info,
 919                                          &new->windows_info);
 920                 if (ret)
 921                         return ret;
 922         }
 923         return 0;
 924 }
 925
 926 /* Copies the XML information for an image between WIM files.
 927  *
 928  * @dest_image_name and @dest_image_description are ignored if they are NULL;
 929  * otherwise, they are used to override the image name and/or image description
 930  * from the XML data in the source WIM file.
 931  *
 932  * On failure, WIMLIB_ERR_NOMEM is returned and no changes are made.  Otherwise,
 933  * 0 is returned and the WIM information at *new_wim_info_p is modified.
 934  */
 935 int
 936 xml_export_image(const struct wim_info *old_wim_info,
 937                  int image,
 938                  struct wim_info **new_wim_info_p,
 939                  const tchar *dest_image_name,
 940                  const tchar *dest_image_description)
 941 {
 942         struct wim_info *new_wim_info;
 943         struct image_info *image_info;
 944         int ret;
 945
 946         DEBUG("Copying XML data between WIM files for source image %d.", image);
 947
 948         wimlib_assert(old_wim_info != NULL);
 949         wimlib_assert(image >= 1 && image <= old_wim_info->num_images);
 950
 951         if (*new_wim_info_p) {
 952                 new_wim_info = *new_wim_info_p;
 953         } else {
 954                 new_wim_info = CALLOC(1, sizeof(struct wim_info));
 955                 if (!new_wim_info)
 956                         goto err;
 957         }
 958
 959         image_info = add_image_info_struct(new_wim_info);
 960         if (!image_info)
 961                 goto err;
 962
 963         ret = clone_image_info(&old_wim_info->images[image - 1], image_info);
 964         if (ret != 0)
 965                 goto err_destroy_image_info;
 966
 967         image_info->index = new_wim_info->num_images;
 968
 969         if (dest_image_name) {
 970                 FREE(image_info->name);
 971                 image_info->name = TSTRDUP(dest_image_name);
 972                 if (!image_info->name)
 973                         goto err_destroy_image_info;
 974         }
 975         if (dest_image_description) {
 976                 FREE(image_info->description);
 977                 image_info->description = TSTRDUP(dest_image_description);
 978                 if (!image_info->description)
 979                         goto err_destroy_image_info;
 980         }
 981         *new_wim_info_p = new_wim_info;
 982         return 0;
 983 err_destroy_image_info:
 984         destroy_image_info(image_info);
 985 err:
 986         if (new_wim_info != *new_wim_info_p)
 987                 free_wim_info(new_wim_info);
 988         return WIMLIB_ERR_NOMEM;
 989 }
 990
 991 /* Removes an image from the XML information. */
 992 void
 993 xml_delete_image(struct wim_info **wim_info_p, int image)
 994 {
 995         struct wim_info *wim_info;
 996
 997         wim_info = *wim_info_p;
 998         wimlib_assert(image >= 1 && image <= wim_info->num_images);
 999         DEBUG("Deleting image %d from the XML data.", image);
1000
1001         destroy_image_info(&wim_info->images[image - 1]);
1002
1003         memmove(&wim_info->images[image - 1],
1004                 &wim_info->images[image],
1005                 (wim_info->num_images - image) * sizeof(struct image_info));
1006
1007         if (--wim_info->num_images == 0) {
1008                 free_wim_info(wim_info);
1009                 *wim_info_p = NULL;
1010         } else {
1011                 for (int i = image - 1; i < wim_info->num_images; i++)
1012                         wim_info->images[i].index--;
1013         }
1014 }
1015
1016 size_t
1017 xml_get_max_image_name_len(const WIMStruct *w)
1018 {
1019         size_t max_len = 0;
1020         if (w->wim_info) {
1021                 for (int i = 0; i < w->wim_info->num_images; i++) {
1022                         size_t len = tstrlen(w->wim_info->images[i].name);
1023                         if (len > max_len)
1024                                 max_len = len;
1025                 }
1026         }
1027         return max_len;
1028 }
1029
1030 #ifdef ENABLE_CUSTOM_MEMORY_ALLOCATOR
1031 void
1032 xml_set_memory_allocator(void *(*malloc_func)(size_t),
1033                          void (*free_func)(void *),
1034                          void *(*realloc_func)(void *, size_t))
1035 {
1036         xmlMemSetup(free_func, malloc_func, realloc_func, STRDUP);
1037 }
1038 #endif
1039
1040 static int
1041 calculate_dentry_statistics(struct wim_dentry *dentry, void *arg)
1042 {
1043         struct image_info *info = arg;
1044         const struct wim_inode *inode = dentry->d_inode;
1045         struct wim_lookup_table_entry *lte;
1046
1047         /* Update directory count and file count.
1048          *
1049          * Each dentry counts as either a file or a directory, but not both.
1050          * The root directory is an exception: it is not counted at all.
1051          *
1052          * Symbolic links and junction points (and presumably other reparse
1053          * points) count as regular files.  This is despite the fact that
1054          * junction points have FILE_ATTRIBUTE_DIRECTORY set.
1055          */
1056         if (dentry_is_root(dentry))
1057                 return 0;
1058
1059         if (inode_is_directory(inode))
1060                 info->dir_count++;
1061         else
1062                 info->file_count++;
1063
1064         /*
1065          * Update total bytes and hard link bytes.
1066          *
1067          * Unfortunately there are some inconsistencies/bugs in the way this is
1068          * done.
1069          *
1070          * If there are no alternate data streams in the image, the "total
1071          * bytes" is the sum of the size of the un-named data stream of each
1072          * inode times the link count of that inode.  In other words, it would
1073          * be the total number of bytes of regular files you would have if you
1074          * extracted the full image without any hard-links.  The "hard link
1075          * bytes" is equal to the "total bytes" minus the size of the un-named
1076          * data stream of each inode.  In other words, the "hard link bytes"
1077          * counts the size of the un-named data stream for all the links to each
1078          * inode except the first one.
1079          *
1080          * Reparse points and directories don't seem to be counted in either the
1081          * total bytes or the hard link bytes.
1082          *
1083          * And now we get to the most confusing part, the alternate data
1084          * streams.  They are not counted in the "total bytes".  However, if the
1085          * link count of an inode with alternate data streams is 2 or greater,
1086          * the size of all the alternate data streams is included in the "hard
1087          * link bytes", and this size is multiplied by the link count (NOT one
1088          * less than the link count).
1089          */
1090         lte = inode_unnamed_lte(inode, info->lookup_table);
1091         if (lte) {
1092                 info->total_bytes += wim_resource_size(lte);
1093                 if (!dentry_is_first_in_inode(dentry))
1094                         info->hard_link_bytes += wim_resource_size(lte);
1095         }
1096
1097         if (inode->i_nlink >= 2 && dentry_is_first_in_inode(dentry)) {
1098                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
1099                         if (inode->i_ads_entries[i].stream_name_nbytes) {
1100                                 lte = inode_stream_lte(inode, i + 1, info->lookup_table);
1101                                 if (lte) {
1102                                         info->hard_link_bytes += inode->i_nlink *
1103                                                                  wim_resource_size(lte);
1104                                 }
1105                         }
1106                 }
1107         }
1108         return 0;
1109 }
1110
1111 /*
1112  * Calculate what to put in the <FILECOUNT>, <DIRCOUNT>, <TOTALBYTES>, and
1113  * <HARDLINKBYTES> elements of each <IMAGE>.
1114  *
1115  * Please note there is no official documentation for exactly how this is done.
1116  * But, see calculate_dentry_statistics().
1117  */
1118 void
1119 xml_update_image_info(WIMStruct *w, int image)
1120 {
1121         struct image_info *image_info;
1122
1123         DEBUG("Updating the image info for image %d", image);
1124
1125         image_info = &w->wim_info->images[image - 1];
1126
1127         image_info->file_count      = 0;
1128         image_info->dir_count       = 0;
1129         image_info->total_bytes     = 0;
1130         image_info->hard_link_bytes = 0;
1131         image_info->lookup_table = w->lookup_table;
1132
1133         for_dentry_in_tree(w->image_metadata[image - 1]->root_dentry,
1134                            calculate_dentry_statistics,
1135                            image_info);
1136         image_info->last_modification_time = get_wim_timestamp();
1137 }
1138
1139 /* Adds an image to the XML information. */
1140 int
1141 xml_add_image(WIMStruct *w, const tchar *name)
1142 {
1143         struct wim_info *wim_info;
1144         struct image_info *image_info;
1145
1146         wimlib_assert(name != NULL);
1147
1148         /* If this is the first image, allocate the struct wim_info.  Otherwise
1149          * use the existing struct wim_info. */
1150         if (w->wim_info) {
1151                 wim_info = w->wim_info;
1152         } else {
1153                 wim_info = CALLOC(1, sizeof(struct wim_info));
1154                 if (!wim_info)
1155                         return WIMLIB_ERR_NOMEM;
1156         }
1157
1158         image_info = add_image_info_struct(wim_info);
1159         if (!image_info)
1160                 goto out_free_wim_info;
1161
1162         if (!(image_info->name = TSTRDUP(name)))
1163                 goto out_destroy_image_info;
1164
1165         w->wim_info = wim_info;
1166         image_info->index = wim_info->num_images;
1167         image_info->creation_time = get_wim_timestamp();
1168         xml_update_image_info(w, image_info->index);
1169         return 0;
1170
1171 out_destroy_image_info:
1172         destroy_image_info(image_info);
1173         wim_info->num_images--;
1174 out_free_wim_info:
1175         if (wim_info != w->wim_info)
1176                 FREE(wim_info);
1177         return WIMLIB_ERR_NOMEM;
1178 }
1179
1180 /* Prints information about the specified image from struct wim_info structure.
1181  * */
1182 void
1183 print_image_info(const struct wim_info *wim_info, int image)
1184 {
1185         const struct image_info *image_info;
1186         const tchar *desc;
1187         tchar buf[50];
1188
1189         wimlib_assert(image >= 1 && image <= wim_info->num_images);
1190
1191         image_info = &wim_info->images[image - 1];
1192
1193         tprintf(T("Index:                  %d\n"), image_info->index);
1194         tprintf(T("Name:                   %"TS"\n"), image_info->name);
1195
1196         /* Always print the Description: part even if there is no
1197          * description. */
1198         if (image_info->description)
1199                 desc = image_info->description;
1200         else
1201                 desc = T("");
1202         tprintf(T("Description:            %"TS"\n"), desc);
1203
1204         if (image_info->display_name) {
1205                 tprintf(T("Display Name:           %"TS"\n"),
1206                         image_info->display_name);
1207         }
1208
1209         if (image_info->display_description) {
1210                 tprintf(T("Display Description:    %"TS"\n"),
1211                         image_info->display_description);
1212         }
1213
1214         tprintf(T("Directory Count:        %"PRIu64"\n"), image_info->dir_count);
1215         tprintf(T("File Count:             %"PRIu64"\n"), image_info->file_count);
1216         tprintf(T("Total Bytes:            %"PRIu64"\n"), image_info->total_bytes);
1217         tprintf(T("Hard Link Bytes:        %"PRIu64"\n"), image_info->hard_link_bytes);
1218
1219         wim_timestamp_to_str(image_info->creation_time, buf, sizeof(buf));
1220         tprintf(T("Creation Time:          %"TS"\n"), buf);
1221
1222         wim_timestamp_to_str(image_info->last_modification_time, buf, sizeof(buf));
1223         tprintf(T("Last Modification Time: %"TS"\n"), buf);
1224         if (image_info->windows_info_exists)
1225                 print_windows_info(&image_info->windows_info);
1226         if (image_info->flags)
1227                 tprintf(T("Flags:                  %"TS"\n"), image_info->flags);
1228         tputchar('\n');
1229 }
1230
1231 void
1232 libxml_global_init(void)
1233 {
1234         xmlInitParser();
1235         xmlInitCharEncodingHandlers();
1236 }
1237
1238 void
1239 libxml_global_cleanup(void)
1240 {
1241         xmlCleanupParser();
1242         xmlCleanupCharEncodingHandlers();
1243 }
1244
1245 /*
1246  * Reads the XML data from a WIM file.
1247  */
1248 int
1249 read_xml_data(int in_fd,
1250               const struct resource_entry *res_entry,
1251               struct wim_info **info_ret)
1252 {
1253         utf16lechar *xml_data;
1254         xmlDoc *doc;
1255         xmlNode *root;
1256         int ret;
1257
1258         DEBUG("XML data is %"PRIu64" bytes at offset %"PRIu64"",
1259               (u64)res_entry->size, res_entry->offset);
1260
1261         if (resource_is_compressed(res_entry)) {
1262                 ERROR("XML data is supposed to be uncompressed");
1263                 ret = WIMLIB_ERR_XML;
1264                 goto out;
1265         }
1266
1267         if (res_entry->size < 2) {
1268                 ERROR("XML data must be at least 2 bytes long");
1269                 ret = WIMLIB_ERR_XML;
1270                 goto out;
1271         }
1272
1273         xml_data = MALLOC(res_entry->size + 3);
1274         if (!xml_data) {
1275                 ret = WIMLIB_ERR_NOMEM;
1276                 goto out;
1277         }
1278
1279         if (full_pread(in_fd, xml_data,
1280                        res_entry->size, res_entry->offset) != res_entry->size)
1281         {
1282                 ERROR_WITH_ERRNO("Error reading XML data");
1283                 ret = WIMLIB_ERR_READ;
1284                 goto out_free_xml_data;
1285         }
1286
1287         /* Null-terminate just in case */
1288         ((u8*)xml_data)[res_entry->size] = 0;
1289         ((u8*)xml_data)[res_entry->size + 1] = 0;
1290         ((u8*)xml_data)[res_entry->size + 2] = 0;
1291
1292         DEBUG("Parsing XML using libxml2 to create XML tree");
1293
1294         doc = xmlReadMemory((const char *)xml_data,
1295                             res_entry->size, "noname.xml", "UTF-16", 0);
1296
1297         if (!doc) {
1298                 ERROR("Failed to parse XML data");
1299                 ret = WIMLIB_ERR_XML;
1300                 goto out_free_xml_data;
1301         }
1302
1303         DEBUG("Constructing WIM information structure from XML tree.");
1304
1305         root = xmlDocGetRootElement(doc);
1306         if (!root) {
1307                 ERROR("WIM XML data is an empty XML document");
1308                 ret = WIMLIB_ERR_XML;
1309                 goto out_free_doc;
1310         }
1311
1312         if (!node_is_element(root) || !node_name_is(root, "WIM")) {
1313                 ERROR("Expected <WIM> for the root XML element");
1314                 ret = WIMLIB_ERR_XML;
1315                 goto out_free_doc;
1316         }
1317         ret = xml_read_wim_info(root, info_ret);
1318 out_free_doc:
1319         DEBUG("Freeing XML tree.");
1320         xmlFreeDoc(doc);
1321 out_free_xml_data:
1322         FREE(xml_data);
1323 out:
1324         return ret;
1325 }
1326
1327 /*
1328  * Writes XML data to a WIM file.
1329  *
1330  * If @total_bytes is non-zero, it specifies what to write to the TOTALBYTES
1331  * element in the XML data.  If zero, TOTALBYTES is given the default value of
1332  * the offset of the XML data.
1333  */
1334 int
1335 write_xml_data(const struct wim_info *wim_info, int image, int out_fd,
1336                u64 total_bytes, struct resource_entry *out_res_entry)
1337 {
1338         xmlCharEncodingHandler *encoding_handler;
1339         xmlOutputBuffer *out_buffer;
1340         xmlTextWriter *writer;
1341         int ret;
1342         off_t start_offset;
1343         off_t end_offset;
1344
1345         wimlib_assert(image == WIMLIB_ALL_IMAGES ||
1346                         (wim_info != NULL && image >= 1 &&
1347                          image <= wim_info->num_images));
1348
1349         start_offset = filedes_offset(out_fd);
1350         if (start_offset == -1)
1351                 return WIMLIB_ERR_WRITE;
1352
1353         DEBUG("Writing XML data for image %d at offset %"PRIu64,
1354               image, start_offset);
1355
1356         /* 2 bytes endianness marker for UTF-16LE.  This is _required_ for WIM
1357          * XML data. */
1358         static u8 bom[2] = {0xff, 0xfe};
1359         if (full_write(out_fd, bom, 2) != 2) {
1360                 ERROR_WITH_ERRNO("Error writing XML data");
1361                 return WIMLIB_ERR_WRITE;
1362         }
1363
1364         /* The contents of the <TOTALBYTES> element in the XML data, under the
1365          * <WIM> element (not the <IMAGE> element), is for non-split WIMs the
1366          * size of the WIM file excluding the XML data and integrity table.
1367          * This should be equal to the current position in the output stream,
1368          * since the XML data and integrity table are the last elements of the
1369          * WIM.
1370          *
1371          * For split WIMs, <TOTALBYTES> takes into account the entire WIM, not
1372          * just the current part.  In that case, @total_bytes should be passed
1373          * in to this function. */
1374         if (total_bytes == 0)
1375                 total_bytes = start_offset;
1376
1377         /* The encoding of the XML data must be UTF-16LE. */
1378         encoding_handler = xmlGetCharEncodingHandler(XML_CHAR_ENCODING_UTF16LE);
1379         if (!encoding_handler) {
1380                 ERROR("Failed to get XML character encoding handler for UTF-16LE");
1381                 ret = WIMLIB_ERR_LIBXML_UTF16_HANDLER_NOT_AVAILABLE;
1382                 goto out;
1383         }
1384
1385         out_buffer = xmlOutputBufferCreateFd(out_fd, encoding_handler);
1386         if (!out_buffer) {
1387                 ERROR("Failed to allocate xmlOutputBuffer");
1388                 ret = WIMLIB_ERR_NOMEM;
1389                 goto out;
1390         }
1391
1392         writer = xmlNewTextWriter(out_buffer);
1393         if (!writer) {
1394                 ERROR("Failed to allocate xmlTextWriter");
1395                 ret = WIMLIB_ERR_NOMEM;
1396                 goto out_output_buffer_close;
1397         }
1398
1399         DEBUG("Writing <WIM> element");
1400
1401         ret = xmlTextWriterStartElement(writer, "WIM");
1402         if (ret < 0)
1403                 goto out_write_error;
1404
1405         ret = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
1406                                               total_bytes);
1407         if (ret < 0)
1408                 goto out_write_error;
1409
1410         if (wim_info != NULL) {
1411                 int first, last;
1412                 if (image == WIMLIB_ALL_IMAGES) {
1413                         first = 1;
1414                         last = wim_info->num_images;
1415                 } else {
1416                         first = image;
1417                         last = image;
1418                 }
1419                 DEBUG("Writing %d <IMAGE> elements", last - first + 1);
1420                 for (int i = first; i <= last; i++) {
1421                         ret = xml_write_image_info(writer, &wim_info->images[i - 1]);
1422                         if (ret) {
1423                                 if (ret < 0)
1424                                         goto out_write_error;
1425                                 goto out_free_text_writer;
1426                         }
1427                 }
1428         }
1429
1430         ret = xmlTextWriterEndElement(writer);
1431         if (ret < 0)
1432                 goto out_write_error;
1433
1434         ret = xmlTextWriterEndDocument(writer);
1435         if (ret < 0)
1436                 goto out_write_error;
1437
1438         ret = xmlTextWriterFlush(writer);
1439         if (ret < 0)
1440                 goto out_write_error;
1441
1442         DEBUG("Ended XML document");
1443
1444         end_offset = filedes_offset(out_fd);
1445         if (end_offset == -1) {
1446                 ret = WIMLIB_ERR_WRITE;
1447         } else {
1448                 ret = 0;
1449                 out_res_entry->offset        = start_offset;
1450                 out_res_entry->size          = end_offset - start_offset;
1451                 out_res_entry->original_size = end_offset - start_offset;
1452                 out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
1453         }
1454 out_free_text_writer:
1455         /* xmlFreeTextWriter will free the attached xmlOutputBuffer. */
1456         xmlFreeTextWriter(writer);
1457         goto out;
1458 out_output_buffer_close:
1459         xmlOutputBufferClose(out_buffer);
1460 out:
1461         if (ret == 0)
1462                 DEBUG("Successfully wrote XML data");
1463         return ret;
1464 out_write_error:
1465         ERROR("Error writing XML data");
1466         ret = WIMLIB_ERR_WRITE;
1467         goto out_free_text_writer;
1468 }
1469
1470 /* Returns the name of the specified image. */
1471 WIMLIBAPI const tchar *
1472 wimlib_get_image_name(const WIMStruct *w, int image)
1473 {
1474         if (image < 1 || image > w->hdr.image_count)
1475                 return NULL;
1476         return w->wim_info->images[image - 1].name;
1477 }
1478
1479 /* Returns the description of the specified image. */
1480 WIMLIBAPI const tchar *
1481 wimlib_get_image_description(const WIMStruct *w, int image)
1482 {
1483         if (image < 1 || image > w->hdr.image_count)
1484                 return NULL;
1485         return w->wim_info->images[image - 1].description;
1486 }
1487
1488 /* Determines if an image name is already used by some image in the WIM. */
1489 WIMLIBAPI bool
1490 wimlib_image_name_in_use(const WIMStruct *w, const tchar *name)
1491 {
1492         if (!name || !*name)
1493                 return false;
1494         for (int i = 1; i <= w->hdr.image_count; i++)
1495                 if (!tstrcmp(w->wim_info->images[i - 1].name, name))
1496                         return true;
1497         return false;
1498 }
1499
1500
1501 /* Extracts the raw XML data to a file stream. */
1502 WIMLIBAPI int
1503 wimlib_extract_xml_data(WIMStruct *w, FILE *fp)
1504 {
1505         size_t size;
1506         void *buf;
1507         int ret;
1508
1509         size = w->hdr.xml_res_entry.size;
1510         if (sizeof(size_t) < sizeof(u64))
1511                 if (size != w->hdr.xml_res_entry.size)
1512                         return WIMLIB_ERR_INVALID_PARAM;
1513
1514         buf = MALLOC(size);
1515         if (!buf)
1516                 return WIMLIB_ERR_NOMEM;
1517
1518         if (full_pread(w->in_fd,
1519                        buf,
1520                        w->hdr.xml_res_entry.size,
1521                        w->hdr.xml_res_entry.offset) != w->hdr.xml_res_entry.size)
1522         {
1523                 ERROR_WITH_ERRNO("Error reading XML data");
1524                 ret = WIMLIB_ERR_READ;
1525                 goto out_free_buf;
1526         }
1527
1528         if (fwrite(buf, 1, size, fp) != size) {
1529                 ERROR_WITH_ERRNO("Failed to extract XML data");
1530                 ret = WIMLIB_ERR_WRITE;
1531         } else {
1532                 ret = 0;
1533         }
1534 out_free_buf:
1535         FREE(buf);
1536         return ret;
1537 }
1538
1539 /* Sets the name of an image in the WIM. */
1540 WIMLIBAPI int
1541 wimlib_set_image_name(WIMStruct *w, int image, const tchar *name)
1542 {
1543         tchar *p;
1544         int i;
1545
1546         DEBUG("Setting the name of image %d to %"TS, image, name);
1547
1548         if (!name || !*name) {
1549                 ERROR("Must specify a non-empty string for the image name");
1550                 return WIMLIB_ERR_INVALID_PARAM;
1551         }
1552
1553         if (image < 1 || image > w->hdr.image_count) {
1554                 ERROR("%d is not a valid image", image);
1555                 return WIMLIB_ERR_INVALID_IMAGE;
1556         }
1557
1558         for (i = 1; i <= w->hdr.image_count; i++) {
1559                 if (i == image)
1560                         continue;
1561                 if (tstrcmp(w->wim_info->images[i - 1].name, name) == 0) {
1562                         ERROR("The name \"%"TS"\" is already in use in the WIM!",
1563                               name);
1564                         return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1565                 }
1566         }
1567
1568         p = TSTRDUP(name);
1569         if (!p)
1570                 return WIMLIB_ERR_NOMEM;
1571
1572         FREE(w->wim_info->images[image - 1].name);
1573         w->wim_info->images[image - 1].name = p;
1574         return 0;
1575 }
1576
1577 static int
1578 do_set_image_info_str(WIMStruct *w, int image, const tchar *tstr,
1579                       size_t offset)
1580 {
1581         tchar *tstr_copy;
1582         tchar **dest_tstr_p;
1583
1584         if (image < 1 || image > w->hdr.image_count) {
1585                 ERROR("%d is not a valid image", image);
1586                 return WIMLIB_ERR_INVALID_IMAGE;
1587         }
1588         if (tstr) {
1589                 tstr_copy = TSTRDUP(tstr);
1590                 if (!tstr_copy)
1591                         return WIMLIB_ERR_NOMEM;
1592         } else {
1593                 tstr_copy = NULL;
1594         }
1595         dest_tstr_p = (tchar**)((void*)&w->wim_info->images[image - 1] + offset);
1596
1597         FREE(*dest_tstr_p);
1598         *dest_tstr_p = tstr_copy;
1599         return 0;
1600 }
1601
1602 /* Sets the description of an image in the WIM. */
1603 WIMLIBAPI int
1604 wimlib_set_image_descripton(WIMStruct *w, int image,
1605                             const tchar *description)
1606 {
1607         return do_set_image_info_str(w, image, description,
1608                                      offsetof(struct image_info, description));
1609 }
1610
1611 /* Set the <FLAGS> element of a WIM image */
1612 WIMLIBAPI int
1613 wimlib_set_image_flags(WIMStruct *w, int image, const tchar *flags)
1614 {
1615         return do_set_image_info_str(w, image, flags,
1616                                      offsetof(struct image_info, flags));
1617 }