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