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