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                         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                 empty_name = MALLOC(sizeof(tchar));
 541                 if (!empty_name)
 542                         return WIMLIB_ERR_NOMEM;
 543                 *empty_name = T('\0');
 544                 image_info->name = empty_name;
 545         }
 546         return ret;
 547 }
 548
 549 /* Reads the information from a <WIM> element, which should be the root element
 550  * of the XML tree. */
 551 static int
 552 xml_read_wim_info(const xmlNode *wim_node, struct wim_info **wim_info_ret)
 553 {
 554         struct wim_info *wim_info;
 555         xmlNode *child;
 556         int ret;
 557         int num_images;
 558         int i;
 559
 560         wim_info = CALLOC(1, sizeof(struct wim_info));
 561         if (!wim_info)
 562                 return WIMLIB_ERR_NOMEM;
 563
 564         /* Count how many images there are. */
 565         num_images = 0;
 566         for_node_child(wim_node, child) {
 567                 if (node_is_element(child) && node_name_is(child, "IMAGE")) {
 568                         if (unlikely(num_images == MAX_IMAGES)) {
 569                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 570                                 goto err;
 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                                 ret = xml_read_image_info(child,
 590                                                           &wim_info->images[i]);
 591                                 if (ret != 0)
 592                                         goto err;
 593                                 i++;
 594                         } else if (node_name_is(child, "TOTALBYTES")) {
 595                                 wim_info->total_bytes = node_get_u64(child);
 596                         } else if (node_name_is(child, "ESD")) {
 597                                 xmlNode *esdchild;
 598                                 for_node_child(child, esdchild) {
 599                                         if (node_is_element(esdchild) &&
 600                                             node_name_is(esdchild, "ENCRYPTED"))
 601                                         {
 602                                                 ret = WIMLIB_ERR_WIM_IS_ENCRYPTED;
 603                                                 goto err;
 604                                         }
 605                                 }
 606                         }
 607                 }
 608
 609                 /* Sort the array of image info by image index. */
 610                 qsort(wim_info->images, num_images,
 611                       sizeof(struct image_info), sort_by_index);
 612
 613                 /* Make sure the image indices make sense */
 614                 for (i = 0; i < num_images; i++) {
 615                         if (wim_info->images[i].index != i + 1) {
 616                                 ERROR("WIM images are not indexed [1...%d] "
 617                                       "in XML data as expected",
 618                                       num_images);
 619                                 ret = WIMLIB_ERR_IMAGE_COUNT;
 620                                 goto err;
 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 static void
 634 print_windows_info(const struct windows_info *windows_info)
 635 {
 636         const struct windows_version *windows_version;
 637
 638         tprintf(T("Architecture:           %"TS"\n"),
 639                 get_arch(windows_info->arch));
 640
 641         if (windows_info->product_name) {
 642                 tprintf(T("Product Name:           %"TS"\n"),
 643                         windows_info->product_name);
 644         }
 645
 646         if (windows_info->edition_id) {
 647                 tprintf(T("Edition ID:             %"TS"\n"),
 648                         windows_info->edition_id);
 649         }
 650
 651         if (windows_info->installation_type) {
 652                 tprintf(T("Installation Type:      %"TS"\n"),
 653                         windows_info->installation_type);
 654         }
 655
 656         if (windows_info->hal) {
 657                 tprintf(T("HAL:                    %"TS"\n"),
 658                               windows_info->hal);
 659         }
 660
 661         if (windows_info->product_type) {
 662                 tprintf(T("Product Type:           %"TS"\n"),
 663                         windows_info->product_type);
 664         }
 665
 666         if (windows_info->product_suite) {
 667                 tprintf(T("Product Suite:          %"TS"\n"),
 668                         windows_info->product_suite);
 669         }
 670
 671         tprintf(T("Languages:              "));
 672         for (size_t i = 0; i < windows_info->num_languages; i++) {
 673
 674                 tfputs(windows_info->languages[i], stdout);
 675                 tputchar(T(' '));
 676         }
 677         tputchar(T('\n'));
 678         if (windows_info->default_language) {
 679                 tprintf(T("Default Language:       %"TS"\n"),
 680                         windows_info->default_language);
 681         }
 682         if (windows_info->system_root) {
 683                 tprintf(T("System Root:            %"TS"\n"),
 684                               windows_info->system_root);
 685         }
 686
 687         if (windows_info->windows_version_exists) {
 688                 windows_version = &windows_info->windows_version;
 689                 tprintf(T("Major Version:          %"PRIu64"\n"),
 690                         windows_version->major);
 691                 tprintf(T("Minor Version:          %"PRIu64"\n"),
 692                         windows_version->minor);
 693                 tprintf(T("Build:                  %"PRIu64"\n"),
 694                         windows_version->build);
 695                 tprintf(T("Service Pack Build:     %"PRIu64"\n"),
 696                         windows_version->sp_build);
 697                 tprintf(T("Service Pack Level:     %"PRIu64"\n"),
 698                         windows_version->sp_level);
 699         }
 700 }
 701
 702 static int
 703 xml_write_string(xmlTextWriter *writer, const char *name,
 704                  const tchar *tstr)
 705 {
 706         if (tstr) {
 707                 char *utf8_str;
 708                 int rc = tstr_to_utf8_simple(tstr, &utf8_str);
 709                 if (rc)
 710                         return rc;
 711                 rc = xmlTextWriterWriteElement(writer, name, utf8_str);
 712                 FREE(utf8_str);
 713                 if (rc < 0)
 714                         return rc;
 715         }
 716         return 0;
 717 }
 718
 719 static int
 720 xml_write_strings_from_specs(xmlTextWriter *writer,
 721                              const void *struct_with_strings,
 722                              const struct xml_string_spec specs[],
 723                              size_t num_specs)
 724 {
 725         for (size_t i = 0; i < num_specs; i++) {
 726                 int rc = xml_write_string(writer, specs[i].name,
 727                                       *(const tchar * const *)
 728                                         (struct_with_strings + specs[i].offset));
 729                 if (rc)
 730                         return rc;
 731         }
 732         return 0;
 733 }
 734
 735 static int
 736 dup_strings_from_specs(const void *old_struct_with_strings,
 737                        void *new_struct_with_strings,
 738                        const struct xml_string_spec specs[],
 739                        size_t num_specs)
 740 {
 741         for (size_t i = 0; i < num_specs; i++) {
 742                 const tchar *old_str = *(const tchar * const *)
 743                                         ((const void*)old_struct_with_strings + specs[i].offset);
 744                 tchar **new_str_p = (tchar **)((void*)new_struct_with_strings + specs[i].offset);
 745                 if (old_str) {
 746                         *new_str_p = TSTRDUP(old_str);
 747                         if (!*new_str_p)
 748                                 return WIMLIB_ERR_NOMEM;
 749                 }
 750         }
 751         return 0;
 752 }
 753
 754 /* Writes the information contained in a `struct windows_version' to the XML
 755  * document being written.  This is the <VERSION> element inside the <WINDOWS>
 756  * element. */
 757 static int
 758 xml_write_windows_version(xmlTextWriter *writer,
 759                           const struct windows_version *version)
 760 {
 761         int rc;
 762         rc = xmlTextWriterStartElement(writer, "VERSION");
 763         if (rc < 0)
 764                 return rc;
 765
 766         rc = xmlTextWriterWriteFormatElement(writer, "MAJOR", "%"PRIu64,
 767                                              version->major);
 768         if (rc < 0)
 769                 return rc;
 770
 771         rc = xmlTextWriterWriteFormatElement(writer, "MINOR", "%"PRIu64,
 772                                              version->minor);
 773         if (rc < 0)
 774                 return rc;
 775
 776         rc = xmlTextWriterWriteFormatElement(writer, "BUILD", "%"PRIu64,
 777                                              version->build);
 778         if (rc < 0)
 779                 return rc;
 780
 781         rc = xmlTextWriterWriteFormatElement(writer, "SPBUILD", "%"PRIu64,
 782                                              version->sp_build);
 783         if (rc < 0)
 784                 return rc;
 785
 786         rc = xmlTextWriterWriteFormatElement(writer, "SPLEVEL", "%"PRIu64,
 787                                              version->sp_level);
 788         if (rc < 0)
 789                 return rc;
 790
 791         rc = xmlTextWriterEndElement(writer); /* </VERSION> */
 792         if (rc < 0)
 793                 return rc;
 794
 795         return 0;
 796 }
 797
 798 /* Writes the information contained in a `struct windows_info' to the XML
 799  * document being written. This is the <WINDOWS> element. */
 800 static int
 801 xml_write_windows_info(xmlTextWriter *writer,
 802                        const struct windows_info *windows_info)
 803 {
 804         int rc;
 805         rc = xmlTextWriterStartElement(writer, "WINDOWS");
 806         if (rc < 0)
 807                 return rc;
 808
 809         rc = xmlTextWriterWriteFormatElement(writer, "ARCH", "%"PRIu64,
 810                                              windows_info->arch);
 811         if (rc < 0)
 812                 return rc;
 813
 814         rc = xml_write_strings_from_specs(writer,
 815                                           windows_info,
 816                                           windows_info_xml_string_specs,
 817                                           ARRAY_LEN(windows_info_xml_string_specs));
 818         if (rc)
 819                 return rc;
 820
 821         if (windows_info->num_languages) {
 822                 rc = xmlTextWriterStartElement(writer, "LANGUAGES");
 823                 if (rc < 0)
 824                         return rc;
 825
 826                 for (size_t i = 0; i < windows_info->num_languages; i++) {
 827                         rc = xml_write_string(writer, "LANGUAGE",
 828                                               windows_info->languages[i]);
 829                         if (rc)
 830                                 return rc;
 831                 }
 832
 833                 rc = xml_write_string(writer, "DEFAULT",
 834                                       windows_info->default_language);
 835                 if (rc)
 836                         return rc;
 837
 838                 rc = xmlTextWriterEndElement(writer); /* </LANGUAGES> */
 839                 if (rc < 0)
 840                         return rc;
 841         }
 842
 843         if (windows_info->pkeyconfigversion) {
 844                 rc = xmlTextWriterStartElement(writer, "SERVICINGDATA");
 845                 if (rc < 0)
 846                         return rc;
 847
 848                 rc = xml_write_string(writer, "PKEYCONFIGVERSION",
 849                                       windows_info->pkeyconfigversion);
 850                 if (rc)
 851                         return rc;
 852
 853                 rc = xmlTextWriterEndElement(writer);
 854                 if (rc < 0)
 855                         return rc;
 856         }
 857
 858         if (windows_info->windows_version_exists) {
 859                 rc = xml_write_windows_version(writer, &windows_info->windows_version);
 860                 if (rc)
 861                         return rc;
 862         }
 863
 864         rc = xml_write_string(writer, "SYSTEMROOT", windows_info->system_root);
 865         if (rc)
 866                 return rc;
 867
 868         rc = xmlTextWriterEndElement(writer); /* </WINDOWS> */
 869         if (rc < 0)
 870                 return rc;
 871
 872         return 0;
 873 }
 874
 875 /* Writes a time element to the XML document being constructed in memory. */
 876 static int
 877 xml_write_time(xmlTextWriter *writer, const char *element_name, u64 time)
 878 {
 879         int rc;
 880         rc = xmlTextWriterStartElement(writer, element_name);
 881         if (rc < 0)
 882                 return rc;
 883
 884         rc = xmlTextWriterWriteFormatElement(writer, "HIGHPART",
 885                                              "0x%08"PRIX32, (u32)(time >> 32));
 886         if (rc < 0)
 887                 return rc;
 888
 889         rc = xmlTextWriterWriteFormatElement(writer, "LOWPART",
 890                                              "0x%08"PRIX32, (u32)time);
 891         if (rc < 0)
 892                 return rc;
 893
 894         rc = xmlTextWriterEndElement(writer); /* </@element_name> */
 895         if (rc < 0)
 896                 return rc;
 897         return 0;
 898 }
 899
 900 /* Writes an <IMAGE> element to the XML document. */
 901 static int
 902 xml_write_image_info(xmlTextWriter *writer, const struct image_info *image_info)
 903 {
 904         int rc;
 905         rc = xmlTextWriterStartElement(writer, "IMAGE");
 906         if (rc < 0)
 907                 return rc;
 908
 909         rc = xmlTextWriterWriteFormatAttribute(writer, "INDEX", "%d",
 910                                                image_info->index);
 911         if (rc < 0)
 912                 return rc;
 913
 914         rc = xmlTextWriterWriteFormatElement(writer, "DIRCOUNT", "%"PRIu64,
 915                                              image_info->dir_count);
 916         if (rc < 0)
 917                 return rc;
 918
 919         rc = xmlTextWriterWriteFormatElement(writer, "FILECOUNT", "%"PRIu64,
 920                                              image_info->file_count);
 921         if (rc < 0)
 922                 return rc;
 923
 924         rc = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES", "%"PRIu64,
 925                                              image_info->total_bytes);
 926         if (rc < 0)
 927                 return rc;
 928
 929         rc = xmlTextWriterWriteFormatElement(writer, "HARDLINKBYTES", "%"PRIu64,
 930                                              image_info->hard_link_bytes);
 931         if (rc < 0)
 932                 return rc;
 933
 934         rc = xml_write_time(writer, "CREATIONTIME", image_info->creation_time);
 935         if (rc < 0)
 936                 return rc;
 937
 938         rc = xml_write_time(writer, "LASTMODIFICATIONTIME",
 939                             image_info->last_modification_time);
 940         if (rc < 0)
 941                 return rc;
 942
 943         if (image_info->windows_info_exists) {
 944                 rc = xml_write_windows_info(writer, &image_info->windows_info);
 945                 if (rc)
 946                         return rc;
 947         }
 948
 949         rc = xml_write_strings_from_specs(writer, image_info,
 950                                           image_info_xml_string_specs,
 951                                           ARRAY_LEN(image_info_xml_string_specs));
 952         if (rc)
 953                 return rc;
 954
 955         if (image_info->wimboot) {
 956                 rc = xmlTextWriterWriteFormatElement(writer, "WIMBOOT", "%d", 1);
 957                 if (rc < 0)
 958                         return rc;
 959         }
 960
 961         rc = xmlTextWriterEndElement(writer); /* </IMAGE> */
 962         if (rc < 0)
 963                 return rc;
 964
 965         return 0;
 966 }
 967
 968
 969
 970 /* Makes space for another image in the XML information and return a pointer to
 971  * it.*/
 972 static struct image_info *
 973 add_image_info_struct(struct wim_info *wim_info)
 974 {
 975         struct image_info *images;
 976
 977         images = CALLOC(wim_info->num_images + 1, sizeof(struct image_info));
 978         if (!images)
 979                 return NULL;
 980         memcpy(images, wim_info->images,
 981                wim_info->num_images * sizeof(struct image_info));
 982         FREE(wim_info->images);
 983         wim_info->images = images;
 984         wim_info->num_images++;
 985         return &images[wim_info->num_images - 1];
 986 }
 987
 988 static int
 989 clone_windows_info(const struct windows_info *old, struct windows_info *new)
 990 {
 991         int ret;
 992
 993         new->arch = old->arch;
 994
 995         ret = dup_strings_from_specs(old, new, windows_info_xml_string_specs,
 996                                      ARRAY_LEN(windows_info_xml_string_specs));
 997         if (ret)
 998                 return ret;
 999
1000         if (old->pkeyconfigversion) {
1001                 new->pkeyconfigversion = TSTRDUP(old->pkeyconfigversion);
1002                 if (new->pkeyconfigversion == NULL)
1003                         return WIMLIB_ERR_NOMEM;
1004         }
1005
1006         if (old->languages) {
1007                 new->languages = CALLOC(old->num_languages, sizeof(new->languages[0]));
1008                 if (!new->languages)
1009                         return WIMLIB_ERR_NOMEM;
1010                 new->num_languages = old->num_languages;
1011                 for (size_t i = 0; i < new->num_languages; i++) {
1012                         if (!old->languages[i])
1013                                 continue;
1014                         new->languages[i] = TSTRDUP(old->languages[i]);
1015                         if (!new->languages[i])
1016                                 return WIMLIB_ERR_NOMEM;
1017                 }
1018         }
1019         if (old->default_language &&
1020                         !(new->default_language = TSTRDUP(old->default_language)))
1021                 return WIMLIB_ERR_NOMEM;
1022         if (old->system_root && !(new->system_root = TSTRDUP(old->system_root)))
1023                 return WIMLIB_ERR_NOMEM;
1024         if (old->windows_version_exists) {
1025                 new->windows_version_exists = true;
1026                 memcpy(&new->windows_version, &old->windows_version,
1027                        sizeof(old->windows_version));
1028         }
1029         return 0;
1030 }
1031
1032 static int
1033 clone_image_info(const struct image_info *old, struct image_info *new)
1034 {
1035         int ret;
1036
1037         new->dir_count              = old->dir_count;
1038         new->file_count             = old->file_count;
1039         new->total_bytes            = old->total_bytes;
1040         new->hard_link_bytes        = old->hard_link_bytes;
1041         new->creation_time          = old->creation_time;
1042         new->last_modification_time = old->last_modification_time;
1043
1044         ret = dup_strings_from_specs(old, new,
1045                                      image_info_xml_string_specs,
1046                                      ARRAY_LEN(image_info_xml_string_specs));
1047         if (ret)
1048                 return ret;
1049
1050         if (old->windows_info_exists) {
1051                 new->windows_info_exists = true;
1052                 ret = clone_windows_info(&old->windows_info,
1053                                          &new->windows_info);
1054                 if (ret)
1055                         return ret;
1056         }
1057         new->wimboot = old->wimboot;
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
1174         /* Update directory count and file count.
1175          *
1176          * Each dentry counts as either a file or a directory, but not both.
1177          * The root directory is an exception: it is not counted at all.
1178          *
1179          * Symbolic links and junction points (and presumably other reparse
1180          * points) count as regular files.  This is despite the fact that
1181          * junction points have FILE_ATTRIBUTE_DIRECTORY set.
1182          */
1183
1184         if (!dentry_is_root(dentry)) {
1185                 if (inode_is_directory(inode))
1186                         info->dir_count++;
1187                 else
1188                         info->file_count++;
1189         }
1190
1191         /*
1192          * Update total bytes and hard link bytes.
1193          *
1194          * We try to act the same as the MS implementation, even though there
1195          * are some inconsistencies/bugs in the way it operates.
1196          *
1197          * If there are no alternate data streams in the image, the "total
1198          * bytes" is the sum of the size of the un-named data stream of each
1199          * inode times the link count of that inode.  In other words, it would
1200          * be the total number of bytes of regular files you would have if you
1201          * extracted the full image without any hard-links.  The "hard link
1202          * bytes" is equal to the "total bytes" minus the size of the un-named
1203          * data stream of each inode.  In other words, the "hard link bytes"
1204          * counts the size of the un-named data stream for all the links to each
1205          * inode except the first one.
1206          *
1207          * Reparse points and directories don't seem to be counted in either the
1208          * total bytes or the hard link bytes.
1209          *
1210          * And now we get to the most confusing part, the alternate data
1211          * streams.  They are not counted in the "total bytes".  However, if the
1212          * link count of an inode with alternate data streams is 2 or greater,
1213          * the size of all the alternate data streams is included in the "hard
1214          * link bytes", and this size is multiplied by the link count (NOT one
1215          * less than the link count).
1216          */
1217         if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
1218                                      FILE_ATTRIBUTE_REPARSE_POINT)))
1219         {
1220                 struct wim_lookup_table_entry *lte;
1221
1222                 lte = inode_unnamed_lte(inode, info->lookup_table);
1223                 if (lte) {
1224                         info->total_bytes += lte->size;
1225                         if (!dentry_is_first_in_inode(dentry))
1226                                 info->hard_link_bytes += lte->size;
1227                 }
1228
1229                 if (inode->i_nlink >= 2 && dentry_is_first_in_inode(dentry)) {
1230                         for (unsigned i = 0; i < inode->i_num_ads; i++) {
1231                                 if (inode->i_ads_entries[i].stream_name_nbytes) {
1232                                         lte = inode_stream_lte(inode, i + 1, info->lookup_table);
1233                                         if (lte) {
1234                                                 info->hard_link_bytes += inode->i_nlink *
1235                                                                          lte->size;
1236                                         }
1237                                 }
1238                         }
1239                 }
1240         }
1241         return 0;
1242 }
1243
1244 /*
1245  * Calculate what to put in the <FILECOUNT>, <DIRCOUNT>, <TOTALBYTES>, and
1246  * <HARDLINKBYTES> elements of each <IMAGE>.
1247  */
1248 void
1249 xml_update_image_info(WIMStruct *wim, int image)
1250 {
1251         struct image_info *image_info;
1252
1253         DEBUG("Updating the image info for image %d", image);
1254
1255         image_info = &wim->wim_info->images[image - 1];
1256
1257         image_info->file_count      = 0;
1258         image_info->dir_count       = 0;
1259         image_info->total_bytes     = 0;
1260         image_info->hard_link_bytes = 0;
1261         image_info->lookup_table = wim->lookup_table;
1262
1263         for_dentry_in_tree(wim->image_metadata[image - 1]->root_dentry,
1264                            calculate_dentry_statistics,
1265                            image_info);
1266         image_info->last_modification_time = get_wim_timestamp();
1267 }
1268
1269 /* Adds an image to the XML information. */
1270 int
1271 xml_add_image(WIMStruct *wim, const tchar *name)
1272 {
1273         struct wim_info *wim_info;
1274         struct image_info *image_info;
1275
1276         wimlib_assert(name != NULL);
1277
1278         /* If this is the first image, allocate the struct wim_info.  Otherwise
1279          * use the existing struct wim_info. */
1280         if (wim->wim_info) {
1281                 wim_info = wim->wim_info;
1282         } else {
1283                 wim_info = CALLOC(1, sizeof(struct wim_info));
1284                 if (!wim_info)
1285                         return WIMLIB_ERR_NOMEM;
1286         }
1287
1288         image_info = add_image_info_struct(wim_info);
1289         if (!image_info)
1290                 goto out_free_wim_info;
1291
1292         if (!(image_info->name = TSTRDUP(name)))
1293                 goto out_destroy_image_info;
1294
1295         wim->wim_info = wim_info;
1296         image_info->index = wim_info->num_images;
1297         image_info->creation_time = get_wim_timestamp();
1298         xml_update_image_info(wim, image_info->index);
1299         return 0;
1300
1301 out_destroy_image_info:
1302         destroy_image_info(image_info);
1303         wim_info->num_images--;
1304 out_free_wim_info:
1305         if (wim_info != wim->wim_info)
1306                 FREE(wim_info);
1307         return WIMLIB_ERR_NOMEM;
1308 }
1309
1310 /* Prints information about the specified image from struct wim_info structure.
1311  * */
1312 void
1313 print_image_info(const struct wim_info *wim_info, int image)
1314 {
1315         const struct image_info *image_info;
1316         const tchar *desc;
1317         tchar buf[50];
1318
1319         wimlib_assert(image >= 1 && image <= wim_info->num_images);
1320
1321         image_info = &wim_info->images[image - 1];
1322
1323         tprintf(T("Index:                  %d\n"), image_info->index);
1324         tprintf(T("Name:                   %"TS"\n"), image_info->name);
1325
1326         /* Always print the Description: part even if there is no
1327          * description. */
1328         if (image_info->description)
1329                 desc = image_info->description;
1330         else
1331                 desc = T("");
1332         tprintf(T("Description:            %"TS"\n"), desc);
1333
1334         if (image_info->display_name) {
1335                 tprintf(T("Display Name:           %"TS"\n"),
1336                         image_info->display_name);
1337         }
1338
1339         if (image_info->display_description) {
1340                 tprintf(T("Display Description:    %"TS"\n"),
1341                         image_info->display_description);
1342         }
1343
1344         tprintf(T("Directory Count:        %"PRIu64"\n"), image_info->dir_count);
1345         tprintf(T("File Count:             %"PRIu64"\n"), image_info->file_count);
1346         tprintf(T("Total Bytes:            %"PRIu64"\n"), image_info->total_bytes);
1347         tprintf(T("Hard Link Bytes:        %"PRIu64"\n"), image_info->hard_link_bytes);
1348
1349         wim_timestamp_to_str(image_info->creation_time, buf, sizeof(buf));
1350         tprintf(T("Creation Time:          %"TS"\n"), buf);
1351
1352         wim_timestamp_to_str(image_info->last_modification_time, buf, sizeof(buf));
1353         tprintf(T("Last Modification Time: %"TS"\n"), buf);
1354         if (image_info->windows_info_exists)
1355                 print_windows_info(&image_info->windows_info);
1356         if (image_info->flags)
1357                 tprintf(T("Flags:                  %"TS"\n"), image_info->flags);
1358         tprintf(T("WIMBoot compatible:     %"TS"\n"),
1359                 image_info->wimboot ? T("yes") : T("no"));
1360         tputchar('\n');
1361 }
1362
1363 void
1364 libxml_global_init(void)
1365 {
1366         xmlInitParser();
1367         xmlInitCharEncodingHandlers();
1368 }
1369
1370 void
1371 libxml_global_cleanup(void)
1372 {
1373         xmlCleanupParser();
1374         xmlCleanupCharEncodingHandlers();
1375 }
1376
1377 /* Reads the XML data from a WIM file.  */
1378 int
1379 read_wim_xml_data(WIMStruct *wim)
1380 {
1381         void *buf;
1382         size_t bufsize;
1383         u8 *xml_data;
1384         xmlDoc *doc;
1385         xmlNode *root;
1386         int ret;
1387
1388         ret = wimlib_get_xml_data(wim, &buf, &bufsize);
1389         if (ret)
1390                 goto out;
1391         xml_data = buf;
1392
1393         doc = xmlReadMemory((const char *)xml_data, bufsize,
1394                             NULL, "UTF-16LE", 0);
1395         if (!doc) {
1396                 ERROR("Failed to parse XML data");
1397                 ret = WIMLIB_ERR_XML;
1398                 goto out_free_xml_data;
1399         }
1400
1401         root = xmlDocGetRootElement(doc);
1402         if (!root || !node_is_element(root) || !node_name_is(root, "WIM")) {
1403                 ERROR("WIM XML data is invalid");
1404                 ret = WIMLIB_ERR_XML;
1405                 goto out_free_doc;
1406         }
1407
1408         ret = xml_read_wim_info(root, &wim->wim_info);
1409 out_free_doc:
1410         xmlFreeDoc(doc);
1411 out_free_xml_data:
1412         FREE(xml_data);
1413 out:
1414         return ret;
1415 }
1416
1417 /* Prepares an in-memory buffer containing the UTF-16LE XML data for a WIM file.
1418  *
1419  * total_bytes is the number to write in <TOTALBYTES>, or
1420  * WIM_TOTALBYTES_USE_EXISTING to use the existing value in memory, or
1421  * WIM_TOTALBYTES_OMIT to omit <TOTALBYTES> entirely.
1422  */
1423 static int
1424 prepare_wim_xml_data(WIMStruct *wim, int image, u64 total_bytes,
1425                      u8 **xml_data_ret, size_t *xml_len_ret)
1426 {
1427         xmlCharEncodingHandler *encoding_handler;
1428         xmlBuffer *buf;
1429         xmlOutputBuffer *outbuf;
1430         xmlTextWriter *writer;
1431         int ret;
1432         int first, last;
1433         const xmlChar *content;
1434         int len;
1435         u8 *xml_data;
1436         size_t xml_len;
1437
1438         /* Open an xmlTextWriter that writes to an in-memory buffer using
1439          * UTF-16LE encoding.  */
1440
1441         encoding_handler = xmlGetCharEncodingHandler(XML_CHAR_ENCODING_UTF16LE);
1442         if (!encoding_handler) {
1443                 ERROR("Failed to get XML character encoding handler for UTF-16LE");
1444                 ret = WIMLIB_ERR_LIBXML_UTF16_HANDLER_NOT_AVAILABLE;
1445                 goto out;
1446         }
1447
1448         buf = xmlBufferCreate();
1449         if (!buf) {
1450                 ERROR("Failed to create xmlBuffer");
1451                 ret = WIMLIB_ERR_NOMEM;
1452                 goto out;
1453         }
1454
1455         outbuf = xmlOutputBufferCreateBuffer(buf, encoding_handler);
1456         if (!outbuf) {
1457                 ERROR("Failed to allocate xmlOutputBuffer");
1458                 ret = WIMLIB_ERR_NOMEM;
1459                 goto out_buffer_free;
1460         }
1461
1462         writer = xmlNewTextWriter(outbuf);
1463         if (!writer) {
1464                 ERROR("Failed to allocate xmlTextWriter");
1465                 ret = WIMLIB_ERR_NOMEM;
1466                 goto out_output_buffer_close;
1467         }
1468
1469         /* Write the XML document.  */
1470
1471         ret = xmlTextWriterStartElement(writer, "WIM");
1472         if (ret < 0)
1473                 goto out_write_error;
1474
1475         /* The contents of the <TOTALBYTES> element in the XML data, under the
1476          * <WIM> element (not the <IMAGE> element), is for non-split WIMs the
1477          * size of the WIM file excluding the XML data and integrity table.
1478          * For split WIMs, <TOTALBYTES> takes into account the entire WIM, not
1479          * just the current part.  */
1480         if (total_bytes != WIM_TOTALBYTES_OMIT) {
1481                 if (total_bytes == WIM_TOTALBYTES_USE_EXISTING) {
1482                         if (wim->wim_info)
1483                                 total_bytes = wim->wim_info->total_bytes;
1484                         else
1485                                 total_bytes = 0;
1486                 }
1487                 ret = xmlTextWriterWriteFormatElement(writer, "TOTALBYTES",
1488                                                       "%"PRIu64, total_bytes);
1489                 if (ret < 0)
1490                         goto out_write_error;
1491         }
1492
1493         if (image == WIMLIB_ALL_IMAGES) {
1494                 first = 1;
1495                 last = wim->hdr.image_count;
1496         } else {
1497                 first = image;
1498                 last = image;
1499         }
1500
1501         for (int i = first; i <= last; i++) {
1502                 ret = xml_write_image_info(writer, &wim->wim_info->images[i - 1]);
1503                 if (ret) {
1504                         if (ret < 0)
1505                                 goto out_write_error;
1506                         goto out_free_text_writer;
1507                 }
1508         }
1509
1510         ret = xmlTextWriterEndElement(writer);
1511         if (ret < 0)
1512                 goto out_write_error;
1513
1514         ret = xmlTextWriterEndDocument(writer);
1515         if (ret < 0)
1516                 goto out_write_error;
1517
1518         ret = xmlTextWriterFlush(writer);
1519         if (ret < 0)
1520                 goto out_write_error;
1521
1522         /* Retrieve the buffer into which the document was written.  */
1523
1524         content = xmlBufferContent(buf);
1525         len = xmlBufferLength(buf);
1526
1527         /* Copy the data into a new buffer, and prefix it with the UTF-16LE BOM
1528          * (byte order mark), which is required by MS's software to understand
1529          * the data.  */
1530
1531         xml_len = len + 2;
1532         xml_data = MALLOC(xml_len);
1533         if (!xml_data) {
1534                 ret = WIMLIB_ERR_NOMEM;
1535                 goto out_free_text_writer;
1536         }
1537         xml_data[0] = 0xff;
1538         xml_data[1] = 0xfe;
1539         memcpy(&xml_data[2], content, len);
1540
1541         /* Clean up libxml objects and return success.  */
1542         *xml_data_ret = xml_data;
1543         *xml_len_ret = xml_len;
1544         ret = 0;
1545 out_free_text_writer:
1546         /* xmlFreeTextWriter will free the attached xmlOutputBuffer.  */
1547         xmlFreeTextWriter(writer);
1548         goto out_buffer_free;
1549 out_output_buffer_close:
1550         xmlOutputBufferClose(outbuf);
1551 out_buffer_free:
1552         xmlBufferFree(buf);
1553 out:
1554         DEBUG("ret=%d", ret);
1555         return ret;
1556
1557 out_write_error:
1558         ERROR("Error writing XML data");
1559         ret = WIMLIB_ERR_WRITE;
1560         goto out_free_text_writer;
1561 }
1562
1563 /* Writes the XML data to a WIM file.  */
1564 int
1565 write_wim_xml_data(WIMStruct *wim, int image, u64 total_bytes,
1566                    struct wim_reshdr *out_reshdr,
1567                    int write_resource_flags)
1568 {
1569         int ret;
1570         u8 *xml_data;
1571         size_t xml_len;
1572
1573         DEBUG("Writing WIM XML data (image=%d, offset=%"PRIu64")",
1574               image, wim->out_fd.offset);
1575
1576         ret = prepare_wim_xml_data(wim, image, total_bytes,
1577                                    &xml_data, &xml_len);
1578         if (ret)
1579                 return ret;
1580
1581         /* Write the XML data uncompressed.  Although wimlib can handle
1582          * compressed XML data, MS software cannot.  */
1583         ret = write_wim_resource_from_buffer(xml_data,
1584                                              xml_len,
1585                                              WIM_RESHDR_FLAG_METADATA,
1586                                              &wim->out_fd,
1587                                              WIMLIB_COMPRESSION_TYPE_NONE,
1588                                              0,
1589                                              out_reshdr,
1590                                              NULL,
1591                                              write_resource_flags);
1592         FREE(xml_data);
1593         DEBUG("ret=%d", ret);
1594         return ret;
1595 }
1596
1597 /* API function documented in wimlib.h  */
1598 WIMLIBAPI const tchar *
1599 wimlib_get_image_name(const WIMStruct *wim, int image)
1600 {
1601         if (image < 1 || image > wim->hdr.image_count)
1602                 return NULL;
1603         return wim->wim_info->images[image - 1].name;
1604 }
1605
1606 /* API function documented in wimlib.h  */
1607 WIMLIBAPI const tchar *
1608 wimlib_get_image_description(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].description;
1613 }
1614
1615 /* API function documented in wimlib.h  */
1616 WIMLIBAPI bool
1617 wimlib_image_name_in_use(const WIMStruct *wim, const tchar *name)
1618 {
1619         if (!name || !*name)
1620                 return false;
1621         for (int i = 1; i <= wim->hdr.image_count; i++)
1622                 if (!tstrcmp(wim->wim_info->images[i - 1].name, name))
1623                         return true;
1624         return false;
1625 }
1626
1627
1628 /* API function documented in wimlib.h  */
1629 WIMLIBAPI int
1630 wimlib_get_xml_data(WIMStruct *wim, void **buf_ret, size_t *bufsize_ret)
1631 {
1632         const struct wim_reshdr *xml_reshdr;
1633
1634         if (wim->filename == NULL && filedes_is_seekable(&wim->in_fd))
1635                 return WIMLIB_ERR_INVALID_PARAM;
1636
1637         if (buf_ret == NULL || bufsize_ret == NULL)
1638                 return WIMLIB_ERR_INVALID_PARAM;
1639
1640         xml_reshdr = &wim->hdr.xml_data_reshdr;
1641
1642         DEBUG("Reading XML data.");
1643         *bufsize_ret = xml_reshdr->uncompressed_size;
1644         return wim_reshdr_to_data(xml_reshdr, wim, buf_ret);
1645 }
1646
1647 WIMLIBAPI int
1648 wimlib_extract_xml_data(WIMStruct *wim, FILE *fp)
1649 {
1650         int ret;
1651         void *buf;
1652         size_t bufsize;
1653
1654         ret = wimlib_get_xml_data(wim, &buf, &bufsize);
1655         if (ret)
1656                 return ret;
1657
1658         if (fwrite(buf, 1, bufsize, fp) != bufsize) {
1659                 ERROR_WITH_ERRNO("Failed to extract XML data");
1660                 ret = WIMLIB_ERR_WRITE;
1661         }
1662         FREE(buf);
1663         return ret;
1664 }
1665
1666 /* API function documented in wimlib.h  */
1667 WIMLIBAPI int
1668 wimlib_set_image_name(WIMStruct *wim, int image, const tchar *name)
1669 {
1670         tchar *p;
1671         int i;
1672         int ret;
1673
1674         DEBUG("Setting the name of image %d to %"TS, image, name);
1675
1676         ret = can_modify_wim(wim);
1677         if (ret)
1678                 return ret;
1679
1680         if (name == NULL)
1681                 name = T("");
1682
1683         if (image < 1 || image > wim->hdr.image_count) {
1684                 ERROR("%d is not a valid image", image);
1685                 return WIMLIB_ERR_INVALID_IMAGE;
1686         }
1687
1688         for (i = 1; i <= wim->hdr.image_count; i++) {
1689                 if (i == image)
1690                         continue;
1691                 if (!tstrcmp(wim->wim_info->images[i - 1].name, name)) {
1692                         ERROR("The name \"%"TS"\" is already in use in the WIM!",
1693                               name);
1694                         return WIMLIB_ERR_IMAGE_NAME_COLLISION;
1695                 }
1696         }
1697
1698         p = TSTRDUP(name);
1699         if (!p)
1700                 return WIMLIB_ERR_NOMEM;
1701
1702         FREE(wim->wim_info->images[image - 1].name);
1703         wim->wim_info->images[image - 1].name = p;
1704         return 0;
1705 }
1706
1707 static int
1708 do_set_image_info_str(WIMStruct *wim, int image, const tchar *tstr,
1709                       size_t offset)
1710 {
1711         tchar *tstr_copy;
1712         tchar **dest_tstr_p;
1713         int ret;
1714
1715         ret = can_modify_wim(wim);
1716         if (ret)
1717                 return ret;
1718
1719         if (image < 1 || image > wim->hdr.image_count) {
1720                 ERROR("%d is not a valid image", image);
1721                 return WIMLIB_ERR_INVALID_IMAGE;
1722         }
1723         if (tstr) {
1724                 tstr_copy = TSTRDUP(tstr);
1725                 if (!tstr_copy)
1726                         return WIMLIB_ERR_NOMEM;
1727         } else {
1728                 tstr_copy = NULL;
1729         }
1730         dest_tstr_p = (tchar**)((void*)&wim->wim_info->images[image - 1] + offset);
1731
1732         FREE(*dest_tstr_p);
1733         *dest_tstr_p = tstr_copy;
1734         return 0;
1735 }
1736
1737 /* API function documented in wimlib.h  */
1738 WIMLIBAPI int
1739 wimlib_set_image_descripton(WIMStruct *wim, int image,
1740                             const tchar *description)
1741 {
1742         return do_set_image_info_str(wim, image, description,
1743                                      offsetof(struct image_info, description));
1744 }
1745
1746 /* API function documented in wimlib.h  */
1747 WIMLIBAPI int
1748 wimlib_set_image_flags(WIMStruct *wim, int image, const tchar *flags)
1749 {
1750         return do_set_image_info_str(wim, image, flags,
1751                                      offsetof(struct image_info, flags));
1752 }