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