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