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 #include "wimlib/write.h"
  40
  41 #include <libxml/encoding.h>
  42 #include <libxml/parser.h>
  43 #include <libxml/tree.h>
  44 #include <libxml/xmlwriter.h>
  45 #include <limits.h>
  46 #include <string.h>
  47 #include <unistd.h>
  48
  49 /* Structures used to form an in-memory representation of the XML data (other
  50  * than the raw parse tree from libxml). */
  51
  52 struct windows_version {
  53         u64 major;
  54         u64 minor;
  55         u64 build;
  56         u64 sp_build;
  57         u64 sp_level;
  58 };
  59
  60 struct windows_info {
  61         u64      arch;
  62         tchar   *product_name;
  63         tchar   *edition_id;
  64         tchar   *installation_type;
  65         tchar   *pkeyconfigversion;
  66         tchar   *hal;
  67         tchar   *product_type;
  68         tchar   *product_suite;
  69         tchar  **languages;
  70         tchar   *default_language;
  71         size_t   num_languages;
  72         tchar   *system_root;
  73         bool     windows_version_exists;
  74         struct   windows_version windows_version;
  75 };
  76
  77 struct image_info {
  78         int index;
  79         bool windows_info_exists;
  80         u64 dir_count;
  81         u64 file_count;
  82         u64 total_bytes;
  83         u64 hard_link_bytes;
  84         u64 creation_time;
  85         u64 last_modification_time;
  86         struct windows_info windows_info;
  87         tchar *name;
  88         tchar *description;
  89         tchar *display_name;
  90         tchar *display_description;
  91         tchar *flags;
  92         bool wimboot;
  93
  94         /* Note: must update clone_image_info() if adding new fields here  */
  95
  96         struct wim_lookup_table *lookup_table; /* temporary field */
  97 };
  98
  99 /* A struct wim_info structure corresponds to the entire XML data for a WIM file. */
 100 struct wim_info {
 101         u64 total_bytes;
 102         int num_images;
 103         /* Array of `struct image_info's, one for each image in the WIM that is
 104          * mentioned in the XML data. */
 105         struct image_info *images;
 106 };
 107
 108 struct xml_string_spec {
 109         const char *name;
 110         size_t offset;
 111 };
 112
 113 #define ELEM(STRING_NAME, MEMBER_NAME) \
 114         {STRING_NAME, offsetof(struct image_info, MEMBER_NAME)}
 115 static const struct xml_string_spec
 116 image_info_xml_string_specs[] = {
 117         ELEM("NAME", name),
 118         ELEM("DESCRIPTION", description),
 119         ELEM("DISPLAYNAME", display_name),
 120         ELEM("DISPLAYDESCRIPTION", display_description),
 121         ELEM("FLAGS", flags),
 122 };
 123 #undef ELEM
 124
 125 #define ELEM(STRING_NAME, MEMBER_NAME) \
 126         {STRING_NAME, offsetof(struct windows_info, MEMBER_NAME)}
 127 static const struct xml_string_spec
 128 windows_info_xml_string_specs[] = {
 129         ELEM("PRODUCTNAME", product_name),
 130         ELEM("EDITIONID", edition_id),
 131         ELEM("INSTALLATIONTYPE", installation_type),
 132         ELEM("HAL", hal),
 133         ELEM("PRODUCTTYPE", product_type),
 134         ELEM("PRODUCTSUITE", product_suite),
 135 };
 136 #undef ELEM
 137
 138 u64
 139 wim_info_get_total_bytes(const struct wim_info *info)
 140 {
 141         if (info)
 142                 return info->total_bytes;
 143         else
 144                 return 0;
 145 }
 146
 147 u64
 148 wim_info_get_image_hard_link_bytes(const struct wim_info *info, int image)
 149 {
 150         if (info)
 151                 return info->images[image - 1].hard_link_bytes;
 152         else
 153                 return 0;
 154 }
 155
 156 u64
 157 wim_info_get_image_total_bytes(const struct wim_info *info, int image)
 158 {
 159         if (info)
 160                 return info->images[image - 1].total_bytes;
 161         else
 162                 return 0;
 163 }
 164
 165 unsigned
 166 wim_info_get_num_images(const struct wim_info *info)
 167 {
 168         if (info)
 169                 return info->num_images;
 170         else
 171                 return 0;
 172 }
 173
 174 void
 175 wim_info_set_wimboot(struct wim_info *info, int image, bool value)
 176 {
 177         info->images[image - 1].wimboot = value;
 178 }
 179
 180 bool
 181 wim_info_get_wimboot(const struct wim_info *info, int image)
 182 {
 183         return info->images[image - 1].wimboot;
 184 }
 185
 186 /* Architecture constants are from w64 mingw winnt.h  */
 187 #define PROCESSOR_ARCHITECTURE_INTEL 0
 188 #define PROCESSOR_ARCHITECTURE_MIPS 1
 189 #define PROCESSOR_ARCHITECTURE_ALPHA 2
 190 #define PROCESSOR_ARCHITECTURE_PPC 3
 191 #define PROCESSOR_ARCHITECTURE_SHX 4
 192 #define PROCESSOR_ARCHITECTURE_ARM 5
 193 #define PROCESSOR_ARCHITECTURE_IA64 6
 194 #define PROCESSOR_ARCHITECTURE_ALPHA64 7
 195 #define PROCESSOR_ARCHITECTURE_MSIL 8
 196 #define PROCESSOR_ARCHITECTURE_AMD64 9
 197 #define PROCESSOR_ARCHITECTURE_IA32_ON_WIN64 10
 198
 199 /* Returns a statically allocated string that is a string representation of the
 200  * architecture number. */
 201 static const tchar *
 202 get_arch(int arch)
 203 {
 204         switch (arch) {
 205         case PROCESSOR_ARCHITECTURE_INTEL:
 206                 return T("x86");
 207         case PROCESSOR_ARCHITECTURE_MIPS:
 208                 return T("MIPS");
 209         case PROCESSOR_ARCHITECTURE_ARM:
 210                 return T("ARM");
 211         case PROCESSOR_ARCHITECTURE_IA64:
 212                 return T("ia64");
 213         case PROCESSOR_ARCHITECTURE_AMD64:
 214                 return T("x86_64");
 215         default:
 216                 return T("unknown");
 217         }
 218 }
 219
 220
 221 /* Iterate through the children of an xmlNode. */
 222 #define for_node_child(parent, child)   \
 223         for (child = parent->children; child != NULL; child = child->next)
 224
 225 /* Utility functions for xmlNodes */
 226 static inline bool
 227 node_is_element(xmlNode *node)
 228 {
 229         return node->type == XML_ELEMENT_NODE;
 230 }
 231
 232 static inline bool
 233 node_is_text(xmlNode *node)
 234 {
 235         return node->type == XML_TEXT_NODE;
 236 }
 237
 238 static inline bool
 239 node_name_is(xmlNode *node, const char *name)
 240 {
 241         /* For now, both upper case and lower case element names are accepted. */
 242         return strcasecmp((const char *)node->name, name) == 0;
 243 }
 244
 245 static u64
 246 node_get_number(const xmlNode *u64_node, int base)
 247 {
 248         xmlNode *child;
 249         for_node_child(u64_node, child)
 250                 if (node_is_text(child))
 251                         return strtoull(child->content, NULL, base);
 252         return 0;
 253 }
 254
 255 /* Finds the text node that is a child of an element node and returns its
 256  * content converted to a 64-bit unsigned integer.  Returns 0 if no text node is
 257  * found. */
 258 static u64
 259 node_get_u64(const xmlNode *u64_node)
 260 {
 261         return node_get_number(u64_node, 10);
 262 }
 263
 264 /* Like node_get_u64(), but expects a number in base 16. */
 265 static u64
 266 node_get_hex_u64(const xmlNode *u64_node)
 267 {
 268         return node_get_number(u64_node, 16);
 269 }
 270
 271 static int
 272 node_get_string(const xmlNode *string_node, tchar **tstr_ret)
 273 {
 274         xmlNode *child;
 275         tchar *tstr = NULL;
 276         int ret;
 277
 278         if (*tstr_ret)
 279                 return 0;
 280
 281         for_node_child(string_node, child) {
 282                 if (node_is_text(child) && child->content) {
 283                         ret = utf8_to_tstr_simple(child->content, &tstr);
 284                         if (ret)
 285                                 return ret;
 286                         break;
 287                 }
 288         }
 289         *tstr_ret = tstr;
 290         return 0;
 291 }
 292
 293 /* Returns the timestamp from a time node.  It has child elements <HIGHPART> and
 294  * <LOWPART> that are then used to construct a 64-bit timestamp. */
 295 static u64
 296 node_get_timestamp(const xmlNode *time_node)
 297 {
 298         u32 high_part = 0;
 299         u32 low_part = 0;
 300         xmlNode *child;
 301         for_node_child(time_node, child) {
 302                 if (!node_is_element(child))
 303                         continue;
 304                 if (node_name_is(child, "HIGHPART"))
 305                         high_part = node_get_hex_u64(child);
 306                 else if (node_name_is(child, "LOWPART"))
 307                         low_part = node_get_hex_u64(child);
 308         }
 309         return (u64)low_part | ((u64)high_part << 32);
 310 }
 311
 312 /* Used to sort an array of struct image_infos by their image indices. */
 313 static int
 314 sort_by_index(const void *p1, const void *p2)
 315 {
 316         int index_1 = ((const struct image_info*)p1)->index;
 317         int index_2 = ((const struct image_info*)p2)->index;
 318         if (index_1 < index_2)
 319                 return -1;
 320         else if (index_1 > index_2)
 321                 return 1;
 322         else
 323                 return 0;
 324 }
 325
 326
 327 /* Frees memory allocated inside a struct windows_info structure. */
 328 static void
 329 destroy_windows_info(struct windows_info *windows_info)
 330 {
 331         FREE(windows_info->product_name);
 332         FREE(windows_info->edition_id);
 333         FREE(windows_info->installation_type);
 334         FREE(windows_info->hal);
 335         FREE(windows_info->product_type);
 336         FREE(windows_info->product_suite);
 337         FREE(windows_info->pkeyconfigversion);
 338         for (size_t i = 0; i < windows_info->num_languages; i++)
 339                 FREE(windows_info->languages[i]);
 340         FREE(windows_info->languages);
 341         FREE(windows_info->default_language);
 342         FREE(windows_info->system_root);
 343 }
 344
 345 /* Frees memory allocated inside a struct image_info structure. */
 346 static void
 347 destroy_image_info(struct image_info *image_info)
 348 {
 349         FREE(image_info->name);
 350         FREE(image_info->description);
 351         FREE(image_info->flags);
 352         FREE(image_info->display_name);
 353         FREE(image_info->display_description);
 354         destroy_windows_info(&image_info->windows_info);
 355         memset(image_info, 0, sizeof(struct image_info));
 356 }
 357
 358 void
 359 free_wim_info(struct wim_info *info)
 360 {
 361         if (info) {
 362                 if (info->images) {
 363                         for (int i = 0; i < info->num_images; i++)
 364                                 destroy_image_info(&info->images[i]);
 365                         FREE(info->images);
 366                 }
 367                 FREE(info);
 368         }
 369 }
 370
 371 /* Reads the information from a <VERSION> element inside the <WINDOWS> element.
 372  * */
 373 static void
 374 xml_read_windows_version(const xmlNode *version_node,
 375                          struct windows_version* windows_version)
 376 {
 377         xmlNode *child;
 378         for_node_child(version_node, child) {
 379                 if (!node_is_element(child))
 380                         continue;
 381                 if (node_name_is(child, "MAJOR"))
 382                         windows_version->major = node_get_u64(child);
 383                 else if (node_name_is(child, "MINOR"))
 384                         windows_version->minor = node_get_u64(child);
 385                 else if (node_name_is(child, "BUILD"))
 386                         windows_version->build = node_get_u64(child);
 387                 else if (node_name_is(child, "SPBUILD"))
 388                         windows_version->sp_build = node_get_u64(child);
 389                 else if (node_name_is(child, "SPLEVEL"))
 390                         windows_version->sp_level = node_get_u64(child);
 391         }
 392 }
 393
 394 /* Reads the information from a <LANGUAGE> element inside a <WINDOWS> element.
 395  * */
 396 static int
 397 xml_read_languages(const xmlNode *languages_node,
 398                    tchar ***languages_ret,
 399                    size_t *num_languages_ret,
 400                    tchar **default_language_ret)
 401 {
 402         xmlNode *child;
 403         size_t num_languages = 0;
 404         tchar **languages;
 405         int ret;
 406
 407         for_node_child(languages_node, child)
 408                 if (node_is_element(child) && node_name_is(child, "LANGUAGE"))
 409                         num_languages++;
 410
 411         languages = CALLOC(num_languages, sizeof(languages[0]));
 412         if (!languages)
 413                 return WIMLIB_ERR_NOMEM;
 414
 415         *languages_ret = languages;
 416         *num_languages_ret = num_languages;
 417
 418         ret = 0;
 419         for_node_child(languages_node, child) {
 420                 if (!node_is_element(child))
 421                         continue;
 422                 if (node_name_is(child, "LANGUAGE"))
 423                         ret = node_get_string(child, languages++);
 424                 else if (node_name_is(child, "DEFAULT"))
 425                         ret = node_get_string(child, default_language_ret);
 426                 if (ret != 0)
 427                         break;
 428         }
 429         return ret;
 430 }
 431
 432 /* Reads the information from a <WINDOWS> element inside an <IMAGE> element. */
 433 static int
 434 xml_read_windows_info(const xmlNode *windows_node,
 435                       struct windows_info *windows_info)
 436 {
 437         xmlNode *child;
 438         int ret = 0;
 439
 440         for_node_child(windows_node, child) {
 441                 if (!node_is_element(child))
 442                         continue;
 443                 if (node_name_is(child, "ARCH")) {
 444                         windows_info->arch = node_get_u64(child);
 445                 } else if (node_name_is(child, "PRODUCTNAME")) {
 446                         ret = node_get_string(child,
 447                                               &windows_info->product_name);
 448                 } else if (node_name_is(child, "EDITIONID")) {
 449                         ret = node_get_string(child,
 450                                               &windows_info->edition_id);
 451                 } else if (node_name_is(child, "INSTALLATIONTYPE")) {
 452                         ret = node_get_string(child,
 453                                               &windows_info->installation_type);
 454                 } else if (node_name_is(child, "PRODUCTTYPE")) {
 455                         ret = node_get_string(child,
 456                                               &windows_info->product_type);
 457                 } else if (node_name_is(child, "PRODUCTSUITE")) {
 458                         ret = node_get_string(child,
 459                                               &windows_info->product_suite);
 460                 } else if (node_name_is(child, "LANGUAGES")) {
 461                         ret = xml_read_languages(child,
 462                                                  &windows_info->languages,
 463                                                  &windows_info->num_languages,
 464                                                  &windows_info->default_language);
 465                 } else if (node_name_is(child, "VERSION")) {
 466                         xml_read_windows_version(child,
 467                                                 &windows_info->windows_version);
 468                         windows_info->windows_version_exists = true;
 469                 } else if (node_name_is(child, "SYSTEMROOT")) {
 470                         ret = node_get_string(child, &windows_info->system_root);
 471                 } else if (node_name_is(child, "HAL")) {
 472                         ret = node_get_string(child, &windows_info->hal);
 473                 } else if (node_name_is(child, "SERVICINGDATA")) {
 474                         xmlNode *grandchild;
 475
 476                         for_node_child(child, grandchild) {
 477                                 if (node_is_element(grandchild) &&
 478                                     node_name_is(grandchild, "PKEYCONFIGVERSION"))
 479                                 {
 480                                         ret = node_get_string(grandchild,
 481                                                               &windows_info->pkeyconfigversion);
 482                                 }
 483                         }
 484                 }
 485
 486                 if (ret != 0)
 487                         return ret;
 488         }
 489         return ret;
 490 }
 491
 492 /* Reads the information from an <IMAGE> element. */
 493 static int
 494 xml_read_image_info(xmlNode *image_node, struct image_info *image_info)
 495 {
 496         xmlNode *child;
 497         xmlChar *index_prop;
 498         int ret;
 499
 500         index_prop = xmlGetProp(image_node, "INDEX");
 501         if (index_prop) {
 502                 image_info->index = atoi(index_prop);
 503                 FREE(index_prop);
 504         } else {
 505                 image_info->index = 1;
 506         }
 507
 508         ret = 0;
 509         for_node_child(image_node, child) {
 510                 if (!node_is_element(child))
 511                         continue;
 512                 if (node_name_is(child, "DIRCOUNT"))
 513                         image_info->dir_count = node_get_u64(child);
 514                 else if (node_name_is(child, "FILECOUNT"))
 515                         image_info->file_count = node_get_u64(child);
 516                 else if (node_name_is(child, "TOTALBYTES"))
 517                         image_info->total_bytes = node_get_u64(child);
 518                 else if (node_name_is(child, "HARDLINKBYTES"))
 519                         image_info->hard_link_bytes = node_get_u64(child);
 520                 else if (node_name_is(child, "CREATIONTIME"))
 521                         image_info->creation_time = node_get_timestamp(child);
 522                 else if (node_name_is(child, "LASTMODIFICATIONTIME"))
 523                         image_info->last_modification_time = node_get_timestamp(child);
 524                 else if (node_name_is(child, "WINDOWS")) {
 525                         DEBUG("Found <WINDOWS> tag");
 526                         ret = xml_read_windows_info(child,
 527                                                     &image_info->windows_info);
 528                         image_info->windows_info_exists = true;
 529                 } else if (node_name_is(child, "NAME")) {
 530                         ret = node_get_string(child, &image_info->name);
 531                 } else if (node_name_is(child, "DESCRIPTION")) {
 532                         ret = node_get_string(child, &image_info->description);
 533                 } else if (node_name_is(child, "FLAGS")) {
 534                         ret = node_get_string(child, &image_info->flags);
 535                 } else if (node_name_is(child, "DISPLAYNAME")) {
 536                         ret = node_get_string(child, &image_info->display_name);
 537                 } else if (node_name_is(child, "DISPLAYDESCRIPTION")) {
 538                         ret = node_get_string(child, &image_info->display_description);
 539                 } else if (node_name_is(child, "WIMBOOT")) {
 540                         if (node_get_u64(child) == 1) {
 541                                 image_info->wimboot = true;
 542                         }
 543                 }
 544                 if (ret != 0)
 545                         return ret;
 546         }
 547         if (!image_info->name) {
 548                 tchar *empty_name;
 549                 /*WARNING("Image with index %d has no name", image_info->index);*/
 550                 empty_name = MALLOC(sizeof(tchar));
 551                 if (!empty_name)
 552                         return WIMLIB_ERR_NOMEM;
 553                 *empty_name = T('\0');
 554                 image_info->name = empty_name;
 555         }
 556         return ret;
 557 }
 558
 559 /* Reads the information from a <WIM> element, which should be the root element
 560  * of the XML tree. */
 561 static int
 562 xml_read_wim_info(const xmlNode *wim_node, struct wim_info **wim_info_ret)
 563 {
 564         struct wim_info *wim_info;
 565         xmlNode *child;
 566         int ret;
 567         int num_images;
 568         int i;
 569
 570         wim_info = CALLOC(1, sizeof(struct wim_info));
 571         if (!wim_info)
 572                 return WIMLIB_ERR_NOMEM;
 573
 574         /* Count how many images there are. */
 575         num_images = 0;
 576         for_node_child(wim_node, child) {
 577                 if (node_is_element(child) && node_name_is(child, "IMAGE")) {
 578                         if (unlikely(num_images == MAX_IMAGES)) {
 579                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 580                                 goto err;
 581                         }
 582                         num_images++;
 583                 }
 584         }
 585
 586         if (num_images > 0) {
 587                 /* Allocate the array of struct image_infos and fill them in. */
 588                 wim_info->images = CALLOC(num_images, sizeof(wim_info->images[0]));
 589                 if (!wim_info->images) {
 590                         ret = WIMLIB_ERR_NOMEM;
 591                         goto err;
 592                 }
 593                 wim_info->num_images = num_images;
 594                 i = 0;
 595                 for_node_child(wim_node, child) {
 596                         if (!node_is_element(child))
 597                                 continue;
 598                         if (node_name_is(child, "IMAGE")) {
 599                                 DEBUG("Found <IMAGE> tag");
 600                                 ret = xml_read_image_info(child,
 601                                                           &wim_info->images[i]);
 602                                 if (ret != 0)
 603                                         goto err;
 604                                 i++;
 605                         } else if (node_name_is(child, "TOTALBYTES")) {
 606                                 wim_info->total_bytes = node_get_u64(child);
 607                         } else if (node_name_is(child, "ESD")) {
 608                                 xmlNode *esdchild;
 609                                 for_node_child(child, esdchild) {
 610                                         if (node_is_element(esdchild) &&
 611                                             node_name_is(esdchild, "ENCRYPTED"))
 612                                         {
 613                                                 ret = WIMLIB_ERR_WIM_IS_ENCRYPTED;
 614                                                 goto err;
 615                                         }
 616                                 }
 617                         }
 618                 }
 619
 620                 /* Sort the array of image info by image index. */
 621                 qsort(wim_info->images, num_images,
 622                       sizeof(struct image_info), sort_by_index);
 623
 624                 /* Make sure the image indices make sense */
 625                 for (i = 0; i < num_images; i++) {
 626                         if (wim_info->images[i].index != i + 1) {
 627                                 ERROR("WIM images are not indexed [1...%d] "
 628                                       "in XML data as expected",
 629                                       num_images);
 630                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 631                                 goto err;
 632                         }
 633                 }
 634
 635         }
 636         *wim_info_ret = wim_info;
 637         return 0;
 638 err:
 639         free_wim_info(wim_info);
 640         return ret;
 641 }
 642
 643 /* Prints the information contained in a `struct windows_info'.
 644  *
 645  * Warning: any strings printed here are in UTF-8 encoding.  If the locale
 646  * character encoding is not UTF-8, the printed strings may be garbled. */
 647 static void
 648 print_windows_info(const struct windows_info *windows_info)
 649 {
 650         const struct windows_version *windows_version;
 651
 652         tprintf(T("Architecture:           %"TS"\n"),
 653                 get_arch(windows_info->arch));
 654
 655         if (windows_info->product_name) {
 656                 tprintf(T("Product Name:           %"TS"\n"),
 657                         windows_info->product_name);
 658         }
 659
 660         if (windows_info->edition_id) {
 661                 tprintf(T("Edition ID:             %"TS"\n"),
 662                         windows_info->edition_id);
 663         }
 664
 665         if (windows_info->installation_type) {
 666                 tprintf(T("Installation Type:      %"TS"\n"),
 667                         windows_info->installation_type);
 668         }
 669
 670         if (windows_info->hal) {
 671                 tprintf(T("HAL:                    %"TS"\n"),
 672                               windows_info->hal);
 673         }
 674
 675         if (windows_info->product_type) {
 676                 tprintf(T("Product Type:           %"TS"\n"),
 677                         windows_info->product_type);
 678         }
 679
 680         if (windows_info->product_suite) {
 681                 tprintf(T("Product Suite:          %"TS"\n"),
 682                         windows_info->product_suite);
 683         }
 684
 685         tprintf(T("Languages:              "));
 686         for (size_t i = 0; i < windows_info->num_languages; i++) {
 687
 688                 tfputs(windows_info->languages[i], stdout);
 689                 tputchar(T(' '));
 690         }
 691         tputchar(T('\n'));
 692         if (windows_info->default_language) {
 693                 tprintf(T("Default Language:       %"TS"\n"),
 694                         windows_info->default_language);
 695         }
 696         if (windows_info->system_root) {
 697                 tprintf(T("System Root:            %"TS"\n"),
 698                               windows_info->system_root);
 699         }
 700
 701         if (windows_info->windows_version_exists) {
 702                 windows_version = &windows_info->windows_version;
 703                 tprintf(T("Major Version:          %"PRIu64"\n"),
 704                         windows_version->major);
 705                 tprintf(T("Minor Version:          %"PRIu64"\n"),
 706                         windows_version->minor);
 707                 tprintf(T("Build:                  %"PRIu64"\n"),
 708                         windows_version->build);
 709                 tprintf(T("Service Pack Build:     %"PRIu64"\n"),
 710                         windows_version->sp_build);
 711                 tprintf(T("Service Pack Level:     %"PRIu64"\n"),
 712                         windows_version->sp_level);
 713         }
 714 }
 715
 716 static int
 717 xml_write_string(xmlTextWriter *writer, const char *name,
 718                  const tchar *tstr)
 719 {
 720         if (tstr) {
 721                 char *utf8_str;
 722                 int rc = tstr_to_utf8_simple(tstr, &utf8_str);
 723                 if (rc)
 724                         return rc;
 725                 rc = xmlTextWriterWriteElement(writer, name, utf8_str);
 726                 FREE(utf8_str);
 727                 if (rc < 0)
 728                         return rc;
 729         }
 730         return 0;
 731 }
 732
 733 static int
 734 xml_write_strings_from_specs(xmlTextWriter *writer,
 735                              const void *struct_with_strings,
 736                              const struct xml_string_spec specs[],
 737                              size_t num_specs)
 738 {
 739         for (size_t i = 0; i < num_specs; i++) {
 740                 int rc = xml_write_string(writer, specs[i].name,
 741                                       *(const tchar * const *)
 742                                         (struct_with_strings + specs[i].offset));
 743                 if (rc)
 744                         return rc;
 745         }
 746         return 0;
 747 }
 748
 749 static int
 750 dup_strings_from_specs(const void *old_struct_with_strings,
 751                        void *new_struct_with_strings,
 752                        const struct xml_string_spec specs[],
 753                        size_t num_specs)
 754 {
 755         for (size_t i = 0; i < num_specs; i++) {
 756                 const tchar *old_str = *(const tchar * const *)
 757                                         ((const void*)old_struct_with_strings + specs[i].offset);
 758                 tchar **new_str_p = (tchar **)((void*)new_struct_with_strings + specs[i].offset);
 759                 if (old_str) {
 760                         *new_str_p = TSTRDUP(old_str);
 761                         if (!*new_str_p)
 762                                 return WIMLIB_ERR_NOMEM;
 763                 }
 764         }
 765         return 0;
 766 }
 767
 768 /* Writes the information contained in a `struct windows_version' to the XML
 769  * document being written.  This is the <VERSION> element inside the <WINDOWS>
 770  * element. */
 771 static int
 772 xml_write_windows_version(xmlTextWriter *writer,
 773                           const struct windows_version *version)
 774 {
 775         int rc;
 776         rc = xmlTextWriterStartElement(writer, "VERSION");
 777         if (rc < 0)
 778                 return rc;
 779
 780         rc = xmlTextWriterWriteFormatElement(writer, "MAJOR", "%"PRIu64,
 781                                              version->major);
 782         if (rc < 0)
 783                 return rc;
 784
 785         rc = xmlTextWriterWriteFormatElement(writer, "MINOR", "%"PRIu64,
 786                                              version->minor);
 787         if (rc < 0)
 788                 return rc;
 789
 790         rc = xmlTextWriterWriteFormatElement(writer, "BUILD", "%"PRIu64,
 791                                              version->build);
 792         if (rc < 0)
 793                 return rc;
 794
 795         rc = xmlTextWriterWriteFormatElement(writer, "SPBUILD", "%"PRIu64,
 796                                              version->sp_build);
 797         if (rc < 0)
 798                 return rc;
 799
 800         rc = xmlTextWriterWriteFormatElement(writer, "SPLEVEL", "%"PRIu64,
 801                                              version->sp_level);
 802         if (rc < 0)
 803                 return rc;
 804
 805         rc = xmlTextWriterEndElement(writer); /* </VERSION> */
 806         if (rc < 0)
 807                 return rc;
 808
 809         return 0;
 810 }
 811
 812 /* Writes the information contained in a `struct windows_info' to the XML
 813  * document being written. This is the <WINDOWS> element. */
 814 static int
 815 xml_write_windows_info(xmlTextWriter *writer,
 816                        const struct windows_info *windows_info)
 817 {
 818         int rc;
 819         rc = xmlTextWriterStartElement(writer, "WINDOWS");
 820         if (rc < 0)
 821                 return rc;
 822
 823         rc = xmlTextWriterWriteFormatElement(writer, "ARCH", "%"PRIu64,
 824                                              windows_info->arch);
 825         if (rc < 0)
 826                 return rc;
 827
 828         rc = xml_write_strings_from_specs(writer,
 829                                           windows_info,
 830                                           windows_info_xml_string_specs,
 831                                           ARRAY_LEN(windows_info_xml_string_specs));
 832         if (rc)
 833                 return rc;
 834
 835         if (windows_info->num_languages) {
 836                 rc = xmlTextWriterStartElement(writer, "LANGUAGES");
 837                 if (rc < 0)
 838                         return rc;
 839
 840                 for (size_t i = 0; i < windows_info->num_languages; i++) {
 841                         rc = xml_write_string(writer, "LANGUAGE",
 842                                               windows_info->languages[i]);
 843                         if (rc)
 844                                 return rc;
 845                 }
 846
 847                 rc = xml_write_string(writer, "DEFAULT",
 848                                       windows_info->default_language);
 849                 if (rc)
 850                         return rc;
 851
 852                 rc = xmlTextWriterEndElement(writer); /* </LANGUAGES> */
 853                 if (rc < 0)
 854                         return rc;
 855         }
 856
 857         if (windows_info->pkeyconfigversion) {
 858                 rc = xmlTextWriterStartElement(writer, "SERVICINGDATA");
 859                 if (rc < 0)
 860                         return rc;
 861
 862                 rc = xml_write_string(writer, "PKEYCONFIGVERSION",
 863                                       windows_info->pkeyconfigversion);
 864                 if (rc)
 865                         return rc;
 866
 867                 rc = xmlTextWriterEndElement(writer);
 868                 if (rc < 0)
 869                         return rc;
 870         }
 871
 872         if (windows_info->windows_version_exists) {
 873                 rc = xml_write_windows_version(writer, &windows_info->windows_version);
 874                 if (rc)
 875                         return rc;
 876         }
 877
 878         rc = xml_write_string(writer, "SYSTEMROOT", windows_info->system_root);
 879         if (rc)
 880                 return rc;
 881
 882         rc = xmlTextWriterEndElement(writer); /* </WINDOWS> */
 883         if (rc < 0)
 884                 return rc;
 885
 886         return 0;
 887 }
 888
 889 /* Writes a time element to the XML document being constructed in memory. */
 890 static int
 891 xml_write_time(xmlTextWriter *writer, const char *element_name, u64 time)
 892 {
 893         int rc;
 894         rc = xmlTextWriterStartElement(writer, element_name);
 895         if (rc < 0)
 896                 return rc;
 897
 898         rc = xmlTextWriterWriteFormatElement(writer, "HIGHPART",
 899                                              "0x%08"PRIX32, (u32)(time >> 32));
 900         if (rc < 0)
 901                 return rc;
 902
 903         rc = xmlTextWriterWriteFormatElement(writer, "LOWPART",
 904                                              "0x%08"PRIX32, (u32)time);
 905         if (rc < 0)
 906                 return rc;
 907
 908         rc = xmlTextWriterEndElement(writer); /* </@element_name> */
 909         if (rc < 0)
 910                 return rc;
 911         return 0;
 912 }
 913
 914 /* Writes an <IMAGE> element to the XML document. */
 915 static int
 916 xml_write_image_info(xmlTextWriter *writer, const struct image_info *image_info)
 917 {
 918         int rc;
 919         rc = xmlTextWriterStartElement(writer, "IMAGE");
 920         if (rc < 0)
 921                 return rc;
 922
 923         rc = xmlTextWriterWriteFormatAttribute(writer, "INDEX", "%d",
 924                                                image_info->index);
 925         if (rc < 0)
 926                 return rc;
 927
 928         rc = xmlTextWriterWriteFormatElement(writer, "DIRCOUNT", "%"PRIu64,
 929                                              image_info->dir_count);
 930         if (rc < 0)
 931                 return rc;
 932
 933         rc = xmlTextWriterWriteFormatElement(writer, "FILECOUNT", "%"PRIu64,
 934                                              image_info->file_count);
 935         if (rc < 0)
 936                 return rc;
 937
 938         rc = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
 939                                              image_info->total_bytes);
 940         if (rc < 0)
 941                 return rc;
 942
 943         rc = xmlTextWriterWriteFormatElement(writer, "HARDLINKBYTES", "%"PRIu64,
 944                                              image_info->hard_link_bytes);
 945         if (rc < 0)
 946                 return rc;
 947
 948         rc = xml_write_time(writer, "CREATIONTIME", image_info->creation_time);
 949         if (rc < 0)
 950                 return rc;
 951
 952         rc = xml_write_time(writer, "LASTMODIFICATIONTIME",
 953                             image_info->last_modification_time);
 954         if (rc < 0)
 955                 return rc;
 956
 957         if (image_info->windows_info_exists) {
 958                 rc = xml_write_windows_info(writer, &image_info->windows_info);
 959                 if (rc)
 960                         return rc;
 961         }
 962
 963         rc = xml_write_strings_from_specs(writer, image_info,
 964                                           image_info_xml_string_specs,
 965                                           ARRAY_LEN(image_info_xml_string_specs));
 966         if (rc)
 967                 return rc;
 968
 969         if (image_info->wimboot) {
 970                 rc = xmlTextWriterWriteFormatElement(writer, "WIMBOOT", "%d", 1);
 971                 if (rc < 0)
 972                         return rc;
 973         }
 974
 975         rc = xmlTextWriterEndElement(writer); /* </IMAGE> */
 976         if (rc < 0)
 977                 return rc;
 978
 979         return 0;
 980 }
 981
 982
 983
 984 /* Makes space for another image in the XML information and return a pointer to
 985  * it.*/
 986 static struct image_info *
 987 add_image_info_struct(struct wim_info *wim_info)
 988 {
 989         struct image_info *images;
 990
 991         images = CALLOC(wim_info->num_images + 1, sizeof(struct image_info));
 992         if (!images)
 993                 return NULL;
 994         memcpy(images, wim_info->images,
 995                wim_info->num_images * sizeof(struct image_info));
 996         FREE(wim_info->images);
 997         wim_info->images = images;
 998         wim_info->num_images++;
 999         return &images[wim_info->num_images - 1];
1000 }
1001
1002 static int
1003 clone_windows_info(const struct windows_info *old, struct windows_info *new)
1004 {
1005         int ret;
1006
1007         new->arch = old->arch;
1008
1009         ret = dup_strings_from_specs(old, new, windows_info_xml_string_specs,
1010                                      ARRAY_LEN(windows_info_xml_string_specs));
1011         if (ret)
1012                 return ret;
1013
1014         if (old->pkeyconfigversion) {
1015                 new->pkeyconfigversion = TSTRDUP(old->pkeyconfigversion);
1016                 if (new->pkeyconfigversion == NULL)
1017                         return WIMLIB_ERR_NOMEM;
1018         }
1019
1020         if (old->languages) {
1021                 new->languages = CALLOC(old->num_languages, sizeof(new->languages[0]));
1022                 if (!new->languages)
1023                         return WIMLIB_ERR_NOMEM;
1024                 new->num_languages = old->num_languages;
1025                 for (size_t i = 0; i < new->num_languages; i++) {
1026                         if (!old->languages[i])
1027                                 continue;
1028                         new->languages[i] = TSTRDUP(old->languages[i]);
1029                         if (!new->languages[i])
1030                                 return WIMLIB_ERR_NOMEM;
1031                 }
1032         }
1033         if (old->default_language &&
1034                         !(new->default_language = TSTRDUP(old->default_language)))
1035                 return WIMLIB_ERR_NOMEM;
1036         if (old->system_root && !(new->system_root = TSTRDUP(old->system_root)))
1037                 return WIMLIB_ERR_NOMEM;
1038         if (old->windows_version_exists) {
1039                 new->windows_version_exists = true;
1040                 memcpy(&new->windows_version, &old->windows_version,
1041                        sizeof(old->windows_version));
1042         }
1043         return 0;
1044 }
1045
1046 static int
1047 clone_image_info(const struct image_info *old, struct image_info *new)
1048 {
1049         int ret;
1050
1051         new->dir_count              = old->dir_count;
1052         new->file_count             = old->file_count;
1053         new->total_bytes            = old->total_bytes;
1054         new->hard_link_bytes        = old->hard_link_bytes;
1055         new->creation_time          = old->creation_time;
1056         new->last_modification_time = old->last_modification_time;
1057
1058         ret = dup_strings_from_specs(old, new,
1059                                      image_info_xml_string_specs,
1060                                      ARRAY_LEN(image_info_xml_string_specs));
1061         if (ret)
1062                 return ret;
1063
1064         if (old->windows_info_exists) {
1065                 new->windows_info_exists = true;
1066                 ret = clone_windows_info(&old->windows_info,
1067                                          &new->windows_info);
1068                 if (ret)
1069                         return ret;
1070         }
1071         new->wimboot = old->wimboot;
1072         return 0;
1073 }
1074
1075 /* Copies the XML information for an image between WIM files.
1076  *
1077  * @dest_image_name and @dest_image_description are ignored if they are NULL;
1078  * otherwise, they are used to override the image name and/or image description
1079  * from the XML data in the source WIM file.
1080  *
1081  * On failure, WIMLIB_ERR_NOMEM is returned and no changes are made.  Otherwise,
1082  * 0 is returned and the WIM information at *new_wim_info_p is modified.
1083  */
1084 int
1085 xml_export_image(const struct wim_info *old_wim_info,
1086                  int image,
1087                  struct wim_info **new_wim_info_p,
1088                  const tchar *dest_image_name,
1089                  const tchar *dest_image_description)
1090 {
1091         struct wim_info *new_wim_info;
1092         struct image_info *image_info;
1093         int ret;
1094
1095         DEBUG("Copying XML data between WIM files for source image %d.", image);
1096
1097         wimlib_assert(old_wim_info != NULL);
1098         wimlib_assert(image >= 1 && image <= old_wim_info->num_images);
1099
1100         if (*new_wim_info_p) {
1101                 new_wim_info = *new_wim_info_p;
1102         } else {
1103                 new_wim_info = CALLOC(1, sizeof(struct wim_info));
1104                 if (!new_wim_info)
1105                         goto err;
1106         }
1107
1108         image_info = add_image_info_struct(new_wim_info);
1109         if (!image_info)
1110                 goto err;
1111
1112         ret = clone_image_info(&old_wim_info->images[image - 1], image_info);
1113         if (ret != 0)
1114                 goto err_destroy_image_info;
1115
1116         image_info->index = new_wim_info->num_images;
1117
1118         if (dest_image_name) {
1119                 FREE(image_info->name);
1120                 image_info->name = TSTRDUP(dest_image_name);
1121                 if (!image_info->name)
1122                         goto err_destroy_image_info;
1123         }
1124         if (dest_image_description) {
1125                 FREE(image_info->description);
1126                 image_info->description = TSTRDUP(dest_image_description);
1127                 if (!image_info->description)
1128                         goto err_destroy_image_info;
1129         }
1130         *new_wim_info_p = new_wim_info;
1131         return 0;
1132 err_destroy_image_info:
1133         destroy_image_info(image_info);
1134 err:
1135         if (new_wim_info != *new_wim_info_p)
1136                 free_wim_info(new_wim_info);
1137         return WIMLIB_ERR_NOMEM;
1138 }
1139
1140 /* Removes an image from the XML information. */
1141 void
1142 xml_delete_image(struct wim_info **wim_info_p, int image)
1143 {
1144         struct wim_info *wim_info;
1145
1146         wim_info = *wim_info_p;
1147         wimlib_assert(image >= 1 && image <= wim_info->num_images);
1148         DEBUG("Deleting image %d from the XML data.", image);
1149
1150         destroy_image_info(&wim_info->images[image - 1]);
1151
1152         memmove(&wim_info->images[image - 1],
1153                 &wim_info->images[image],
1154                 (wim_info->num_images - image) * sizeof(struct image_info));
1155
1156         if (--wim_info->num_images == 0) {
1157                 free_wim_info(wim_info);
1158                 *wim_info_p = NULL;
1159         } else {
1160                 for (int i = image - 1; i < wim_info->num_images; i++)
1161                         wim_info->images[i].index--;
1162         }
1163 }
1164
1165 size_t
1166 xml_get_max_image_name_len(const WIMStruct *wim)
1167 {
1168         size_t max_len = 0;
1169         for (u32 i = 0; i < wim->hdr.image_count; i++)
1170                 max_len = max(max_len, tstrlen(wim->wim_info->images[i].name));
1171         return max_len;
1172 }
1173
1174 void
1175 xml_set_memory_allocator(void *(*malloc_func)(size_t),
1176                          void (*free_func)(void *),
1177                          void *(*realloc_func)(void *, size_t))
1178 {
1179         xmlMemSetup(free_func, malloc_func, realloc_func, STRDUP);
1180 }
1181
1182 static int
1183 calculate_dentry_statistics(struct wim_dentry *dentry, void *arg)
1184 {
1185         struct image_info *info = arg;
1186         const struct wim_inode *inode = dentry->d_inode;
1187         struct wim_lookup_table_entry *lte;
1188
1189         /* Update directory count and file count.
1190          *
1191          * Each dentry counts as either a file or a directory, but not both.
1192          * The root directory is an exception: it is not counted at all.
1193          *
1194          * Symbolic links and junction points (and presumably other reparse
1195          * points) count as regular files.  This is despite the fact that
1196          * junction points have FILE_ATTRIBUTE_DIRECTORY set.
1197          */
1198         if (dentry_is_root(dentry))
1199                 return 0;
1200
1201         if (inode_is_directory(inode))
1202                 info->dir_count++;
1203         else
1204                 info->file_count++;
1205
1206         /*
1207          * Update total bytes and hard link bytes.
1208          *
1209          * Unfortunately there are some inconsistencies/bugs in the way this is
1210          * done.
1211          *
1212          * If there are no alternate data streams in the image, the "total
1213          * bytes" is the sum of the size of the un-named data stream of each
1214          * inode times the link count of that inode.  In other words, it would
1215          * be the total number of bytes of regular files you would have if you
1216          * extracted the full image without any hard-links.  The "hard link
1217          * bytes" is equal to the "total bytes" minus the size of the un-named
1218          * data stream of each inode.  In other words, the "hard link bytes"
1219          * counts the size of the un-named data stream for all the links to each
1220          * inode except the first one.
1221          *
1222          * Reparse points and directories don't seem to be counted in either the
1223          * total bytes or the hard link bytes.
1224          *
1225          * And now we get to the most confusing part, the alternate data
1226          * streams.  They are not counted in the "total bytes".  However, if the
1227          * link count of an inode with alternate data streams is 2 or greater,
1228          * the size of all the alternate data streams is included in the "hard
1229          * link bytes", and this size is multiplied by the link count (NOT one
1230          * less than the link count).
1231          */
1232         lte = inode_unnamed_lte(inode, info->lookup_table);
1233         if (lte) {
1234                 info->total_bytes += lte->size;
1235                 if (!dentry_is_first_in_inode(dentry))
1236                         info->hard_link_bytes += lte->size;
1237         }
1238
1239         if (inode->i_nlink >= 2 && dentry_is_first_in_inode(dentry)) {
1240                 for (unsigned i = 0; i < inode->i_num_ads; i++) {
1241                         if (inode->i_ads_entries[i].stream_name_nbytes) {
1242                                 lte = inode_stream_lte(inode, i + 1, info->lookup_table);
1243                                 if (lte) {
1244                                         info->hard_link_bytes += inode->i_nlink *
1245                                                                  lte->size;
1246                                 }
1247                         }
1248                 }
1249         }
1250         return 0;
1251 }
1252
1253 /*
1254  * Calculate what to put in the <FILECOUNT>, <DIRCOUNT>, <TOTALBYTES>, and
1255  * <HARDLINKBYTES> elements of each <IMAGE>.
1256  *
1257  * Please note there is no official documentation for exactly how this is done.
1258  * But, see calculate_dentry_statistics().
1259  */
1260 void
1261 xml_update_image_info(WIMStruct *wim, int image)
1262 {
1263         struct image_info *image_info;
1264
1265         DEBUG("Updating the image info for image %d", image);
1266
1267         image_info = &wim->wim_info->images[image - 1];
1268
1269         image_info->file_count      = 0;
1270         image_info->dir_count       = 0;
1271         image_info->total_bytes     = 0;
1272         image_info->hard_link_bytes = 0;
1273         image_info->lookup_table = wim->lookup_table;
1274
1275         for_dentry_in_tree(wim->image_metadata[image - 1]->root_dentry,
1276                            calculate_dentry_statistics,
1277                            image_info);
1278         image_info->last_modification_time = get_wim_timestamp();
1279 }
1280
1281 /* Adds an image to the XML information. */
1282 int
1283 xml_add_image(WIMStruct *wim, const tchar *name)
1284 {
1285         struct wim_info *wim_info;
1286         struct image_info *image_info;
1287
1288         wimlib_assert(name != NULL);
1289
1290         /* If this is the first image, allocate the struct wim_info.  Otherwise
1291          * use the existing struct wim_info. */
1292         if (wim->wim_info) {
1293                 wim_info = wim->wim_info;
1294         } else {
1295                 wim_info = CALLOC(1, sizeof(struct wim_info));
1296                 if (!wim_info)
1297                         return WIMLIB_ERR_NOMEM;
1298         }
1299
1300         image_info = add_image_info_struct(wim_info);
1301         if (!image_info)
1302                 goto out_free_wim_info;
1303
1304         if (!(image_info->name = TSTRDUP(name)))
1305                 goto out_destroy_image_info;
1306
1307         wim->wim_info = wim_info;
1308         image_info->index = wim_info->num_images;
1309         image_info->creation_time = get_wim_timestamp();
1310         xml_update_image_info(wim, image_info->index);
1311         return 0;
1312
1313 out_destroy_image_info:
1314         destroy_image_info(image_info);
1315         wim_info->num_images--;
1316 out_free_wim_info:
1317         if (wim_info != wim->wim_info)
1318                 FREE(wim_info);
1319         return WIMLIB_ERR_NOMEM;
1320 }
1321
1322 /* Prints information about the specified image from struct wim_info structure.
1323  * */
1324 void
1325 print_image_info(const struct wim_info *wim_info, int image)
1326 {
1327         const struct image_info *image_info;
1328         const tchar *desc;
1329         tchar buf[50];
1330
1331         wimlib_assert(image >= 1 && image <= wim_info->num_images);
1332
1333         image_info = &wim_info->images[image - 1];
1334
1335         tprintf(T("Index:                  %d\n"), image_info->index);
1336         tprintf(T("Name:                   %"TS"\n"), image_info->name);
1337
1338         /* Always print the Description: part even if there is no
1339          * description. */
1340         if (image_info->description)
1341                 desc = image_info->description;
1342         else
1343                 desc = T("");
1344         tprintf(T("Description:            %"TS"\n"), desc);
1345
1346         if (image_info->display_name) {
1347                 tprintf(T("Display Name:           %"TS"\n"),
1348                         image_info->display_name);
1349         }
1350
1351         if (image_info->display_description) {
1352                 tprintf(T("Display Description:    %"TS"\n"),
1353                         image_info->display_description);
1354         }
1355
1356         tprintf(T("Directory Count:        %"PRIu64"\n"), image_info->dir_count);
1357         tprintf(T("File Count:             %"PRIu64"\n"), image_info->file_count);
1358         tprintf(T("Total Bytes:            %"PRIu64"\n"), image_info->total_bytes);
1359         tprintf(T("Hard Link Bytes:        %"PRIu64"\n"), image_info->hard_link_bytes);
1360
1361         wim_timestamp_to_str(image_info->creation_time, buf, sizeof(buf));
1362         tprintf(T("Creation Time:          %"TS"\n"), buf);
1363
1364         wim_timestamp_to_str(image_info->last_modification_time, buf, sizeof(buf));
1365         tprintf(T("Last Modification Time: %"TS"\n"), buf);
1366         if (image_info->windows_info_exists)
1367                 print_windows_info(&image_info->windows_info);
1368         if (image_info->flags)
1369                 tprintf(T("Flags:                  %"TS"\n"), image_info->flags);
1370         tprintf(T("WIMBoot compatible:     %"TS"\n"),
1371                 image_info->wimboot ? T("yes") : T("no"));
1372         tputchar('\n');
1373 }
1374
1375 void
1376 libxml_global_init(void)
1377 {
1378         xmlInitParser();
1379         xmlInitCharEncodingHandlers();
1380 }
1381
1382 void
1383 libxml_global_cleanup(void)
1384 {
1385         xmlCleanupParser();
1386         xmlCleanupCharEncodingHandlers();
1387 }
1388
1389 /* Reads the XML data from a WIM file.  */
1390 int
1391 read_wim_xml_data(WIMStruct *wim)
1392 {
1393         void *buf;
1394         size_t bufsize;
1395         u8 *xml_data;
1396         xmlDoc *doc;
1397         xmlNode *root;
1398         int ret;
1399
1400         ret = wimlib_get_xml_data(wim, &buf, &bufsize);
1401         if (ret)
1402                 goto out;
1403         xml_data = buf;
1404
1405         doc = xmlReadMemory((const char *)xml_data, bufsize,
1406                             NULL, "UTF-16LE", 0);
1407         if (!doc) {
1408                 ERROR("Failed to parse XML data");
1409                 ret = WIMLIB_ERR_XML;
1410                 goto out_free_xml_data;
1411         }
1412
1413         root = xmlDocGetRootElement(doc);
1414         if (!root || !node_is_element(root) || !node_name_is(root, "WIM")) {
1415                 ERROR("WIM XML data is invalid");
1416                 ret = WIMLIB_ERR_XML;
1417                 goto out_free_doc;
1418         }
1419
1420         ret = xml_read_wim_info(root, &wim->wim_info);
1421 out_free_doc:
1422         xmlFreeDoc(doc);
1423 out_free_xml_data:
1424         FREE(xml_data);
1425 out:
1426         return ret;
1427 }
1428
1429 /* Prepares an in-memory buffer containing the UTF-16LE XML data for a WIM file.
1430  *
1431  * total_bytes is the number to write in <TOTALBYTES>, or
1432  * WIM_TOTALBYTES_USE_EXISTING to use the existing value in memory, or
1433  * WIM_TOTALBYTES_OMIT to omit <TOTALBYTES> entirely.
1434  */
1435 static int
1436 prepare_wim_xml_data(WIMStruct *wim, int image, u64 total_bytes,
1437                      u8 **xml_data_ret, size_t *xml_len_ret)
1438 {
1439         xmlCharEncodingHandler *encoding_handler;
1440         xmlBuffer *buf;
1441         xmlOutputBuffer *outbuf;
1442         xmlTextWriter *writer;
1443         int ret;
1444         int first, last;
1445         const xmlChar *content;
1446         int len;
1447         u8 *xml_data;
1448         size_t xml_len;
1449
1450         /* Open an xmlTextWriter that writes to an in-memory buffer using
1451          * UTF-16LE encoding.  */
1452
1453         encoding_handler = xmlGetCharEncodingHandler(XML_CHAR_ENCODING_UTF16LE);
1454         if (!encoding_handler) {
1455                 ERROR("Failed to get XML character encoding handler for UTF-16LE");
1456                 ret = WIMLIB_ERR_LIBXML_UTF16_HANDLER_NOT_AVAILABLE;
1457                 goto out;
1458         }
1459
1460         buf = xmlBufferCreate();
1461         if (!buf) {
1462                 ERROR("Failed to create xmlBuffer");
1463                 ret = WIMLIB_ERR_NOMEM;
1464                 goto out;
1465         }
1466
1467         outbuf = xmlOutputBufferCreateBuffer(buf, encoding_handler);
1468         if (!outbuf) {
1469                 ERROR("Failed to allocate xmlOutputBuffer");
1470                 ret = WIMLIB_ERR_NOMEM;
1471                 goto out_buffer_free;
1472         }
1473
1474         writer = xmlNewTextWriter(outbuf);
1475         if (!writer) {
1476                 ERROR("Failed to allocate xmlTextWriter");
1477                 ret = WIMLIB_ERR_NOMEM;
1478                 goto out_output_buffer_close;
1479         }
1480
1481         /* Write the XML document.  */
1482
1483         ret = xmlTextWriterStartElement(writer, "WIM");
1484         if (ret < 0)
1485                 goto out_write_error;
1486
1487         /* The contents of the <TOTALBYTES> element in the XML data, under the
1488          * <WIM> element (not the <IMAGE> element), is for non-split WIMs the
1489          * size of the WIM file excluding the XML data and integrity table.
1490          * For split WIMs, <TOTALBYTES> takes into account the entire WIM, not
1491          * just the current part.  */
1492         if (total_bytes != WIM_TOTALBYTES_OMIT) {
1493                 if (total_bytes == WIM_TOTALBYTES_USE_EXISTING) {
1494                         if (wim->wim_info)
1495                                 total_bytes = wim->wim_info->total_bytes;
1496                         else
1497                                 total_bytes = 0;
1498                 }
1499                 ret = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES",
1500                                                       "%"PRIu64, total_bytes);
1501                 if (ret < 0)
1502                         goto out_write_error;
1503         }
1504
1505         if (image == WIMLIB_ALL_IMAGES) {
1506                 first = 1;
1507                 last = wim->hdr.image_count;
1508         } else {
1509                 first = image;
1510                 last = image;
1511         }
1512
1513         for (int i = first; i <= last; i++) {
1514                 ret = xml_write_image_info(writer, &wim->wim_info->images[i - 1]);
1515                 if (ret) {
1516                         if (ret < 0)
1517                                 goto out_write_error;
1518                         goto out_free_text_writer;
1519                 }
1520         }
1521
1522         ret = xmlTextWriterEndElement(writer);
1523         if (ret < 0)
1524                 goto out_write_error;
1525
1526         ret = xmlTextWriterEndDocument(writer);
1527         if (ret < 0)
1528                 goto out_write_error;
1529
1530         ret = xmlTextWriterFlush(writer);
1531         if (ret < 0)
1532                 goto out_write_error;
1533
1534         /* Retrieve the buffer into which the document was written.  */
1535
1536         content = xmlBufferContent(buf);
1537         len = xmlBufferLength(buf);
1538
1539         /* Copy the data into a new buffer, and prefix it with the UTF-16LE BOM
1540          * (byte order mark), which is required by MS's software to understand
1541          * the data.  */
1542
1543         xml_len = len + 2;
1544         xml_data = MALLOC(xml_len);
1545         if (!xml_data) {
1546                 ret = WIMLIB_ERR_NOMEM;
1547                 goto out_free_text_writer;
1548         }
1549         xml_data[0] = 0xff;
1550         xml_data[1] = 0xfe;
1551         memcpy(&xml_data[2], content, len);
1552
1553         /* Clean up libxml objects and return success.  */
1554         *xml_data_ret = xml_data;
1555         *xml_len_ret = xml_len;
1556         ret = 0;
1557 out_free_text_writer:
1558         /* xmlFreeTextWriter will free the attached xmlOutputBuffer.  */
1559         xmlFreeTextWriter(writer);
1560         goto out_buffer_free;
1561 out_output_buffer_close:
1562         xmlOutputBufferClose(outbuf);
1563 out_buffer_free:
1564         xmlBufferFree(buf);
1565 out:
1566         DEBUG("ret=%d", ret);
1567         return ret;
1568
1569 out_write_error:
1570         ERROR("Error writing XML data");
1571         ret = WIMLIB_ERR_WRITE;
1572         goto out_free_text_writer;
1573 }
1574
1575 /* Writes the XML data to a WIM file.  */
1576 int
1577 write_wim_xml_data(WIMStruct *wim, int image, u64 total_bytes,
1578                    struct wim_reshdr *out_reshdr,
1579                    int write_resource_flags)
1580 {
1581         int ret;
1582         u8 *xml_data;
1583         size_t xml_len;
1584
1585         DEBUG("Writing WIM XML data (image=%d, offset=%"PRIu64")",
1586               image, wim->out_fd.offset);
1587
1588         ret = prepare_wim_xml_data(wim, image, total_bytes,
1589                                    &xml_data, &xml_len);
1590         if (ret)
1591                 return ret;
1592
1593         /* Write the XML data uncompressed.  Although wimlib can handle
1594          * compressed XML data, MS software cannot.  */
1595         ret = write_wim_resource_from_buffer(xml_data,
1596                                              xml_len,
1597                                              WIM_RESHDR_FLAG_METADATA,
1598                                              &wim->out_fd,
1599                                              WIMLIB_COMPRESSION_TYPE_NONE,
1600                                              0,
1601                                              out_reshdr,
1602                                              NULL,
1603                                              write_resource_flags);
1604         FREE(xml_data);
1605         DEBUG("ret=%d", ret);
1606         return ret;
1607 }
1608
1609 /* API function documented in wimlib.h  */
1610 WIMLIBAPI const tchar *
1611 wimlib_get_image_name(const WIMStruct *wim, int image)
1612 {
1613         if (image < 1 || image > wim->hdr.image_count)
1614                 return NULL;
1615         return wim->wim_info->images[image - 1].name;
1616 }
1617
1618 /* API function documented in wimlib.h  */
1619 WIMLIBAPI const tchar *
1620 wimlib_get_image_description(const WIMStruct *wim, int image)
1621 {
1622         if (image < 1 || image > wim->hdr.image_count)
1623                 return NULL;
1624         return wim->wim_info->images[image - 1].description;
1625 }
1626
1627 /* API function documented in wimlib.h  */
1628 WIMLIBAPI bool
1629 wimlib_image_name_in_use(const WIMStruct *wim, const tchar *name)
1630 {
1631         if (!name || !*name)
1632                 return false;
1633         for (int i = 1; i <= wim->hdr.image_count; i++)
1634                 if (!tstrcmp(wim->wim_info->images[i - 1].name, name))
1635                         return true;
1636         return false;
1637 }
1638
1639
1640 /* API function documented in wimlib.h  */
1641 WIMLIBAPI int
1642 wimlib_get_xml_data(WIMStruct *wim, void **buf_ret, size_t *bufsize_ret)
1643 {
1644         const struct wim_reshdr *xml_reshdr;
1645
1646         if (wim->filename == NULL && filedes_is_seekable(&wim->in_fd))
1647                 return WIMLIB_ERR_INVALID_PARAM;
1648
1649         if (buf_ret == NULL || bufsize_ret == NULL)
1650                 return WIMLIB_ERR_INVALID_PARAM;
1651
1652         xml_reshdr = &wim->hdr.xml_data_reshdr;
1653
1654         DEBUG("Reading XML data.");
1655         *bufsize_ret = xml_reshdr->uncompressed_size;
1656         return wim_reshdr_to_data(xml_reshdr, wim, buf_ret);
1657 }
1658
1659 WIMLIBAPI int
1660 wimlib_extract_xml_data(WIMStruct *wim, FILE *fp)
1661 {
1662         int ret;
1663         void *buf;
1664         size_t bufsize;
1665
1666         ret = wimlib_get_xml_data(wim, &buf, &bufsize);
1667         if (ret)
1668                 return ret;
1669
1670         if (fwrite(buf, 1, bufsize, fp) != bufsize) {
1671                 ERROR_WITH_ERRNO("Failed to extract XML data");
1672                 ret = WIMLIB_ERR_WRITE;
1673         }
1674         FREE(buf);
1675         return ret;
1676 }
1677
1678 /* API function documented in wimlib.h  */
1679 WIMLIBAPI int
1680 wimlib_set_image_name(WIMStruct *wim, int image, const tchar *name)
1681 {
1682         tchar *p;
1683         int i;
1684         int ret;
1685
1686         DEBUG("Setting the name of image %d to %"TS, image, name);
1687
1688         ret = can_modify_wim(wim);
1689         if (ret)
1690                 return ret;
1691
1692         if (name == NULL)
1693                 name = T("");
1694
1695         if (image < 1 || image > wim->hdr.image_count) {
1696                 ERROR("%d is not a valid image", image);
1697                 return WIMLIB_ERR_INVALID_IMAGE;
1698         }
1699
1700         for (i = 1; i <= wim->hdr.image_count; i++) {
1701                 if (i == image)
1702                         continue;
1703                 if (!tstrcmp(wim->wim_info->images[i - 1].name, name)) {
1704                         ERROR("The name \"%"TS"\" is already in use in the WIM!",
1705                               name);
1706                         return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1707                 }
1708         }
1709
1710         p = TSTRDUP(name);
1711         if (!p)
1712                 return WIMLIB_ERR_NOMEM;
1713
1714         FREE(wim->wim_info->images[image - 1].name);
1715         wim->wim_info->images[image - 1].name = p;
1716         return 0;
1717 }
1718
1719 static int
1720 do_set_image_info_str(WIMStruct *wim, int image, const tchar *tstr,
1721                       size_t offset)
1722 {
1723         tchar *tstr_copy;
1724         tchar **dest_tstr_p;
1725         int ret;
1726
1727         ret = can_modify_wim(wim);
1728         if (ret)
1729                 return ret;
1730
1731         if (image < 1 || image > wim->hdr.image_count) {
1732                 ERROR("%d is not a valid image", image);
1733                 return WIMLIB_ERR_INVALID_IMAGE;
1734         }
1735         if (tstr) {
1736                 tstr_copy = TSTRDUP(tstr);
1737                 if (!tstr_copy)
1738                         return WIMLIB_ERR_NOMEM;
1739         } else {
1740                 tstr_copy = NULL;
1741         }
1742         dest_tstr_p = (tchar**)((void*)&wim->wim_info->images[image - 1] + offset);
1743
1744         FREE(*dest_tstr_p);
1745         *dest_tstr_p = tstr_copy;
1746         return 0;
1747 }
1748
1749 /* API function documented in wimlib.h  */
1750 WIMLIBAPI int
1751 wimlib_set_image_descripton(WIMStruct *wim, int image,
1752                             const tchar *description)
1753 {
1754         return do_set_image_info_str(wim, image, description,
1755                                      offsetof(struct image_info, description));
1756 }
1757
1758 /* API function documented in wimlib.h  */
1759 WIMLIBAPI int
1760 wimlib_set_image_flags(WIMStruct *wim, int image, const tchar *flags)
1761 {
1762         return do_set_image_info_str(wim, image, flags,
1763                                      offsetof(struct image_info, flags));
1764 }