wimlib.net Git - wimlib/blob - src/xml_windows.c

   1 /*
   2  * xml_windows.c
   3  *
   4  * Set Windows-specific metadata in a WIM file's XML document based on the image
   5  * contents.
   6  */
   7
   8 /*
   9  * Copyright 2016-2023 Eric Biggers
  10  *
  11  * This file is free software; you can redistribute it and/or modify it under
  12  * the terms of the GNU Lesser General Public License as published by the Free
  13  * Software Foundation; either version 3 of the License, or (at your option) any
  14  * later version.
  15  *
  16  * This file is distributed in the hope that it will be useful, but WITHOUT
  17  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
  18  * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
  19  * details.
  20  *
  21  * You should have received a copy of the GNU Lesser General Public License
  22  * along with this file; if not, see https://www.gnu.org/licenses/.
  23  */
  24
  25 #ifdef HAVE_CONFIG_H
  26 #  include "config.h"
  27 #endif
  28
  29 #include <stdlib.h>
  30
  31 #include "wimlib.h"
  32 #include "wimlib/blob_table.h"
  33 #include "wimlib/dentry.h"
  34 #include "wimlib/encoding.h"
  35 #include "wimlib/endianness.h"
  36 #include "wimlib/error.h"
  37 #include "wimlib/metadata.h"
  38 #include "wimlib/registry.h"
  39 #include "wimlib/wim.h"
  40 #include "wimlib/xml_windows.h"
  41
  42 /* Context for a call to set_windows_specific_info()  */
  43 struct windows_info_ctx {
  44         WIMStruct *wim;
  45         int image;
  46         bool oom_encountered;
  47         bool debug_enabled;
  48 };
  49
  50 /* For debugging purposes, the environmental variable WIMLIB_DEBUG_XML_INFO can
  51  * be set to enable messages about certain things not being as expected in the
  52  * registry or other files used as information sources.  */
  53
  54 #define XML_WARN(format, ...)                   \
  55         if (ctx->debug_enabled)                 \
  56                 WARNING(format, ##__VA_ARGS__)
  57
  58 /* Set a property in the XML document, with error checking.  */
  59 static void
  60 set_string_property(struct windows_info_ctx *ctx,
  61                     const tchar *name, const tchar *value)
  62 {
  63         int ret = wimlib_set_image_property(ctx->wim, ctx->image, name, value);
  64         if (likely(!ret))
  65                 return;
  66
  67         ctx->oom_encountered |= (ret == WIMLIB_ERR_NOMEM);
  68         WARNING("Failed to set image property \"%"TS"\" to value "
  69                 "\"%"TS"\": %"TS, name, value, wimlib_get_error_string(ret));
  70 }
  71
  72 /* Set a property in the XML document, with error checking.  */
  73 static void
  74 set_number_property(struct windows_info_ctx *ctx, const tchar *name, s64 value)
  75 {
  76         tchar buffer[32];
  77         tsprintf(buffer, T("%"PRIi64""), value);
  78         set_string_property(ctx, name, buffer);
  79 }
  80
  81 /* Check the result of a registry hive operation.  If unsuccessful, possibly
  82  * print debugging information.  Return true iff successful.  */
  83 static bool
  84 check_hive_status(struct windows_info_ctx *ctx, enum hive_status status,
  85                   const tchar *key, const tchar *value)
  86 {
  87         if (likely(status == HIVE_OK))
  88                 return true;
  89
  90         ctx->oom_encountered |= (status == HIVE_OUT_OF_MEMORY);
  91         XML_WARN("%s; key=%"TS" value=%"TS, hive_status_to_string(status),
  92                  (key ? key : T("(null)")), (value ? value : T("(null)")));
  93         return false;
  94 }
  95
  96 static bool
  97 is_registry_valid(struct windows_info_ctx *ctx, const void *hive_mem,
  98                   size_t hive_size)
  99 {
 100         enum hive_status status;
 101
 102         status = hive_validate(hive_mem, hive_size);
 103         return check_hive_status(ctx, status, NULL, NULL);
 104 }
 105
 106 static bool
 107 get_string_from_registry(struct windows_info_ctx *ctx, const struct regf *regf,
 108                          const tchar *key_name, const tchar *value_name,
 109                          tchar **value_ret)
 110 {
 111         enum hive_status status;
 112
 113         status = hive_get_string(regf, key_name, value_name, value_ret);
 114         return check_hive_status(ctx, status, key_name, value_name);
 115 }
 116
 117 static bool
 118 get_number_from_registry(struct windows_info_ctx *ctx, const struct regf *regf,
 119                          const tchar *key_name, const tchar *value_name,
 120                          s64 *value_ret)
 121 {
 122         enum hive_status status;
 123
 124         status = hive_get_number(regf, key_name, value_name, value_ret);
 125         return check_hive_status(ctx, status, key_name, value_name);
 126 }
 127
 128 static bool
 129 list_subkeys_in_registry(struct windows_info_ctx *ctx, const struct regf *regf,
 130                          const tchar *key_name, tchar ***subkeys_ret)
 131 {
 132         enum hive_status status;
 133
 134         status = hive_list_subkeys(regf, key_name, subkeys_ret);
 135         return check_hive_status(ctx, status, key_name, NULL);
 136 }
 137
 138 /* Copy a string value from a registry hive to the XML document.  */
 139 static void
 140 copy_registry_string(struct windows_info_ctx *ctx, const struct regf *regf,
 141                      const tchar *key_name, const tchar *value_name,
 142                      const tchar *property_name)
 143 {
 144         tchar *string;
 145
 146         if (get_string_from_registry(ctx, regf, key_name, value_name, &string)) {
 147                 set_string_property(ctx, property_name, string);
 148                 FREE(string);
 149         }
 150 }
 151
 152 /* A table that map Windows language IDs, sorted numerically, to their language
 153  * names.  It was generated by tools/generate_language_id_map.c.  */
 154 static const struct {
 155         u16 id;
 156         u16 name_start_offset;
 157 } language_id_map[453] = {
 158         {0x0000,    0}, {0x0001,    6}, {0x0002,   12}, {0x0003,   18},
 159         {0x0004,   24}, {0x0005,   30}, {0x0006,   36}, {0x0007,   42},
 160         {0x0008,   48}, {0x0009,   54}, {0x000a,   60}, {0x000b,   66},
 161         {0x000c,   72}, {0x000d,   78}, {0x000e,   84}, {0x000f,   90},
 162         {0x0010,   96}, {0x0011,  102}, {0x0012,  108}, {0x0013,  114},
 163         {0x0014,  120}, {0x0015,  126}, {0x0016,  132}, {0x0017,  138},
 164         {0x0018,  144}, {0x0019,  150}, {0x001a,  156}, {0x001b,  162},
 165         {0x001c,  168}, {0x001d,  174}, {0x001e,  180}, {0x001f,  186},
 166         {0x0020,  192}, {0x0021,  198}, {0x0022,  204}, {0x0023,  210},
 167         {0x0024,  216}, {0x0025,  222}, {0x0026,  228}, {0x0027,  234},
 168         {0x0028,  240}, {0x0029,  251}, {0x002a,  257}, {0x002b,  263},
 169         {0x002c,  269}, {0x002d,  280}, {0x002e,  286}, {0x002f,  293},
 170         {0x0030,  299}, {0x0031,  305}, {0x0032,  311}, {0x0033,  317},
 171         {0x0034,  323}, {0x0035,  329}, {0x0036,  335}, {0x0037,  341},
 172         {0x0038,  347}, {0x0039,  353}, {0x003a,  359}, {0x003b,  365},
 173         {0x003c,  371}, {0x003d,  377}, {0x003e,  384}, {0x003f,  390},
 174         {0x0040,  396}, {0x0041,  402}, {0x0042,  408}, {0x0043,  414},
 175         {0x0044,  425}, {0x0045,  431}, {0x0046,  437}, {0x0047,  443},
 176         {0x0048,  449}, {0x0049,  455}, {0x004a,  461}, {0x004b,  467},
 177         {0x004c,  473}, {0x004d,  479}, {0x004e,  485}, {0x004f,  491},
 178         {0x0050,  497}, {0x0051,  503}, {0x0052,  509}, {0x0053,  515},
 179         {0x0054,  521}, {0x0055,  527}, {0x0056,  533}, {0x0057,  539},
 180         {0x0058,  546}, {0x0059,  553}, {0x005a,  564}, {0x005b,  571},
 181         {0x005c,  577}, {0x005d,  589}, {0x005e,  600}, {0x005f,  606},
 182         {0x0060,  618}, {0x0061,  629}, {0x0062,  635}, {0x0063,  641},
 183         {0x0064,  647}, {0x0065,  654}, {0x0066,  660}, {0x0067,  667},
 184         {0x0068,  678}, {0x0069,  689}, {0x006a,  696}, {0x006b,  702},
 185         {0x006c,  709}, {0x006d,  716}, {0x006e,  722}, {0x006f,  728},
 186         {0x0070,  734}, {0x0071,  740}, {0x0072,  751}, {0x0073,  757},
 187         {0x0074,  763}, {0x0075,  769}, {0x0076,  776}, {0x0077,  783},
 188         {0x0078,  789}, {0x0079,  795}, {0x007a,  803}, {0x007c,  810},
 189         {0x007e,  817}, {0x007f,  823}, {0x0080,  824}, {0x0081,  830},
 190         {0x0082,  836}, {0x0083,  842}, {0x0084,  848}, {0x0085,  855},
 191         {0x0086,  862}, {0x0087,  874}, {0x0088,  880}, {0x008c,  886},
 192         {0x0091,  893}, {0x0092,  899}, {0x0400,  910}, {0x0401,  916},
 193         {0x0402,  922}, {0x0403,  928}, {0x0404,  934}, {0x0405,  940},
 194         {0x0406,  946}, {0x0407,  952}, {0x0408,  958}, {0x0409,  964},
 195         {0x040a,  970}, {0x040b,  983}, {0x040c,  989}, {0x040d,  995},
 196         {0x040e, 1001}, {0x040f, 1007}, {0x0410, 1013}, {0x0411, 1019},
 197         {0x0412, 1025}, {0x0413, 1031}, {0x0414, 1037}, {0x0415, 1043},
 198         {0x0416, 1049}, {0x0417, 1055}, {0x0418, 1061}, {0x0419, 1067},
 199         {0x041a, 1073}, {0x041b, 1079}, {0x041c, 1085}, {0x041d, 1091},
 200         {0x041e, 1097}, {0x041f, 1103}, {0x0420, 1109}, {0x0421, 1115},
 201         {0x0422, 1121}, {0x0423, 1127}, {0x0424, 1133}, {0x0425, 1139},
 202         {0x0426, 1145}, {0x0427, 1151}, {0x0428, 1157}, {0x0429, 1168},
 203         {0x042a, 1174}, {0x042b, 1180}, {0x042c, 1186}, {0x042d, 1197},
 204         {0x042e, 1203}, {0x042f, 1210}, {0x0430, 1216}, {0x0431, 1222},
 205         {0x0432, 1228}, {0x0433, 1234}, {0x0434, 1240}, {0x0435, 1246},
 206         {0x0436, 1252}, {0x0437, 1258}, {0x0438, 1264}, {0x0439, 1270},
 207         {0x043a, 1276}, {0x043b, 1282}, {0x043d, 1288}, {0x043e, 1295},
 208         {0x043f, 1301}, {0x0440, 1307}, {0x0441, 1313}, {0x0442, 1319},
 209         {0x0443, 1325}, {0x0444, 1336}, {0x0445, 1342}, {0x0446, 1348},
 210         {0x0447, 1354}, {0x0448, 1360}, {0x0449, 1366}, {0x044a, 1372},
 211         {0x044b, 1378}, {0x044c, 1384}, {0x044d, 1390}, {0x044e, 1396},
 212         {0x044f, 1402}, {0x0450, 1408}, {0x0451, 1414}, {0x0452, 1420},
 213         {0x0453, 1426}, {0x0454, 1432}, {0x0455, 1438}, {0x0456, 1444},
 214         {0x0457, 1450}, {0x0458, 1457}, {0x0459, 1464}, {0x045a, 1475},
 215         {0x045b, 1482}, {0x045c, 1488}, {0x045d, 1500}, {0x045e, 1511},
 216         {0x045f, 1517}, {0x0460, 1529}, {0x0461, 1540}, {0x0462, 1546},
 217         {0x0463, 1552}, {0x0464, 1558}, {0x0465, 1565}, {0x0466, 1571},
 218         {0x0467, 1578}, {0x0468, 1589}, {0x0469, 1600}, {0x046a, 1607},
 219         {0x046b, 1613}, {0x046c, 1620}, {0x046d, 1627}, {0x046e, 1633},
 220         {0x046f, 1639}, {0x0470, 1645}, {0x0471, 1651}, {0x0472, 1662},
 221         {0x0473, 1668}, {0x0474, 1674}, {0x0475, 1680}, {0x0476, 1687},
 222         {0x0477, 1694}, {0x0478, 1700}, {0x0479, 1706}, {0x047a, 1714},
 223         {0x047c, 1721}, {0x047e, 1728}, {0x0480, 1734}, {0x0481, 1740},
 224         {0x0482, 1746}, {0x0483, 1752}, {0x0484, 1758}, {0x0485, 1765},
 225         {0x0486, 1772}, {0x0487, 1784}, {0x0488, 1790}, {0x048c, 1796},
 226         {0x0491, 1803}, {0x0492, 1809}, {0x0501, 1820}, {0x05fe, 1829},
 227         {0x0800, 1839}, {0x0801, 1845}, {0x0803, 1851}, {0x0804, 1866},
 228         {0x0807, 1872}, {0x0809, 1878}, {0x080a, 1884}, {0x080c, 1890},
 229         {0x0810, 1896}, {0x0813, 1902}, {0x0814, 1908}, {0x0816, 1914},
 230         {0x0818, 1920}, {0x0819, 1926}, {0x081a, 1932}, {0x081d, 1943},
 231         {0x0820, 1949}, {0x082c, 1955}, {0x082e, 1966}, {0x0832, 1973},
 232         {0x083b, 1979}, {0x083c, 1985}, {0x083e, 1991}, {0x0843, 1997},
 233         {0x0845, 2008}, {0x0846, 2014}, {0x0849, 2025}, {0x0850, 2031},
 234         {0x0859, 2042}, {0x085d, 2053}, {0x085f, 2064}, {0x0860, 2076},
 235         {0x0861, 2087}, {0x0867, 2093}, {0x086b, 2104}, {0x0873, 2111},
 236         {0x0901, 2117}, {0x09ff, 2131}, {0x0c00, 2141}, {0x0c01, 2147},
 237         {0x0c04, 2153}, {0x0c07, 2159}, {0x0c09, 2165}, {0x0c0a, 2171},
 238         {0x0c0c, 2177}, {0x0c1a, 2183}, {0x0c3b, 2194}, {0x0c50, 2200},
 239         {0x0c51, 2211}, {0x0c6b, 2217}, {0x1000, 2224}, {0x1001, 2235},
 240         {0x1004, 2241}, {0x1007, 2247}, {0x1009, 2253}, {0x100a, 2259},
 241         {0x100c, 2265}, {0x101a, 2271}, {0x103b, 2277}, {0x105f, 2284},
 242         {0x1401, 2296}, {0x1404, 2302}, {0x1407, 2308}, {0x1409, 2314},
 243         {0x140a, 2320}, {0x140c, 2326}, {0x141a, 2332}, {0x143b, 2343},
 244         {0x1801, 2350}, {0x1809, 2356}, {0x180a, 2362}, {0x180c, 2368},
 245         {0x181a, 2374}, {0x183b, 2385}, {0x1c01, 2392}, {0x1c09, 2398},
 246         {0x1c0a, 2404}, {0x1c0c, 2410}, {0x1c1a, 2417}, {0x1c3b, 2428},
 247         {0x2000, 2435}, {0x2001, 2441}, {0x2009, 2447}, {0x200a, 2453},
 248         {0x200c, 2459}, {0x201a, 2465}, {0x203b, 2476}, {0x2400, 2483},
 249         {0x2401, 2489}, {0x2409, 2495}, {0x240a, 2502}, {0x240c, 2508},
 250         {0x241a, 2514}, {0x243b, 2525}, {0x2800, 2532}, {0x2801, 2538},
 251         {0x2809, 2544}, {0x280a, 2550}, {0x280c, 2556}, {0x281a, 2562},
 252         {0x2c00, 2573}, {0x2c01, 2579}, {0x2c09, 2585}, {0x2c0a, 2591},
 253         {0x2c0c, 2597}, {0x2c1a, 2603}, {0x3000, 2614}, {0x3001, 2620},
 254         {0x3009, 2626}, {0x300a, 2632}, {0x300c, 2638}, {0x301a, 2644},
 255         {0x3400, 2655}, {0x3401, 2661}, {0x3409, 2667}, {0x340a, 2673},
 256         {0x340c, 2679}, {0x3800, 2685}, {0x3801, 2691}, {0x3809, 2697},
 257         {0x380a, 2703}, {0x380c, 2709}, {0x3c00, 2715}, {0x3c01, 2721},
 258         {0x3c09, 2727}, {0x3c0a, 2733}, {0x3c0c, 2739}, {0x4000, 2745},
 259         {0x4001, 2751}, {0x4009, 2757}, {0x400a, 2763}, {0x4400, 2769},
 260         {0x4409, 2775}, {0x440a, 2781}, {0x4800, 2787}, {0x4809, 2793},
 261         {0x480a, 2799}, {0x4c00, 2805}, {0x4c09, 2811}, {0x4c0a, 2817},
 262         {0x500a, 2823}, {0x540a, 2829}, {0x580a, 2835}, {0x5c0a, 2842},
 263         {0x641a, 2848}, {0x681a, 2859}, {0x6c1a, 2870}, {0x701a, 2881},
 264         {0x703b, 2892}, {0x742c, 2899}, {0x743b, 2910}, {0x7804, 2917},
 265         {0x7814, 2923}, {0x781a, 2929}, {0x782c, 2940}, {0x783b, 2951},
 266         {0x7843, 2958}, {0x7850, 2969}, {0x785d, 2975}, {0x785f, 2986},
 267         {0x7c04, 2998}, {0x7c14, 3004}, {0x7c1a, 3010}, {0x7c28, 3021},
 268         {0x7c2e, 3032}, {0x7c3b, 3039}, {0x7c43, 3046}, {0x7c46, 3057},
 269         {0x7c50, 3068}, {0x7c59, 3079}, {0x7c5c, 3090}, {0x7c5d, 3102},
 270         {0x7c5f, 3113}, {0x7c67, 3125}, {0x7c68, 3136}, {0x7c86, 3147},
 271         {0x7c92, 3159},
 272 };
 273
 274 /* All the language names; generated by tools/generate_language_id_map.c.
 275  * For compactness, this is a 'char' string rather than a 'tchar' string.  */
 276 static const char language_names[3170] =
 277         "en-US\0ar-SA\0bg-BG\0ca-ES\0zh-CN\0cs-CZ\0da-DK\0de-DE\0el-GR\0en-US\0"
 278         "es-ES\0fi-FI\0fr-FR\0he-IL\0hu-HU\0is-IS\0it-IT\0ja-JP\0ko-KR\0nl-NL\0"
 279         "nb-NO\0pl-PL\0pt-BR\0rm-CH\0ro-RO\0ru-RU\0hr-HR\0sk-SK\0sq-AL\0sv-SE\0"
 280         "th-TH\0tr-TR\0ur-PK\0id-ID\0uk-UA\0be-BY\0sl-SI\0et-EE\0lv-LV\0lt-LT\0"
 281         "tg-Cyrl-TJ\0fa-IR\0vi-VN\0hy-AM\0az-Latn-AZ\0eu-ES\0hsb-DE\0mk-MK\0"
 282         "st-ZA\0ts-ZA\0tn-ZA\0ve-ZA\0xh-ZA\0zu-ZA\0af-ZA\0ka-GE\0fo-FO\0hi-IN\0"
 283         "mt-MT\0se-NO\0ga-IE\0yi-001\0ms-MY\0kk-KZ\0ky-KG\0sw-KE\0tk-TM\0"
 284         "uz-Latn-UZ\0tt-RU\0bn-BD\0pa-IN\0gu-IN\0or-IN\0ta-IN\0te-IN\0kn-IN\0"
 285         "ml-IN\0as-IN\0mr-IN\0sa-IN\0mn-MN\0bo-CN\0cy-GB\0km-KH\0lo-LA\0my-MM\0"
 286         "gl-ES\0kok-IN\0mni-IN\0sd-Arab-PK\0syr-SY\0si-LK\0chr-Cher-US\0"
 287         "iu-Latn-CA\0am-ET\0tzm-Latn-DZ\0ks-Arab-IN\0ne-NP\0fy-NL\0ps-AF\0"
 288         "fil-PH\0dv-MV\0bin-NG\0ff-Latn-SN\0ha-Latn-NG\0ibb-NG\0yo-NG\0quz-BO\0"
 289         "nso-ZA\0ba-RU\0lb-LU\0kl-GL\0ig-NG\0kr-Latn-NG\0om-ET\0ti-ER\0gn-PY\0"
 290         "haw-US\0la-001\0so-SO\0ii-CN\0pap-029\0arn-CL\0moh-CA\0br-FR\0\0"
 291         "ug-CN\0mi-NZ\0oc-FR\0co-FR\0gsw-CH\0sah-RU\0quc-Latn-GT\0rw-RW\0"
 292         "wo-SN\0prs-AF\0gd-GB\0ku-Arab-IQ\0en-US\0ar-SA\0bg-BG\0ca-ES\0zh-TW\0"
 293         "cs-CZ\0da-DK\0de-DE\0el-GR\0en-US\0es-ES_tradnl\0fi-FI\0fr-FR\0he-IL\0"
 294         "hu-HU\0is-IS\0it-IT\0ja-JP\0ko-KR\0nl-NL\0nb-NO\0pl-PL\0pt-BR\0rm-CH\0"
 295         "ro-RO\0ru-RU\0hr-HR\0sk-SK\0sq-AL\0sv-SE\0th-TH\0tr-TR\0ur-PK\0id-ID\0"
 296         "uk-UA\0be-BY\0sl-SI\0et-EE\0lv-LV\0lt-LT\0tg-Cyrl-TJ\0fa-IR\0vi-VN\0"
 297         "hy-AM\0az-Latn-AZ\0eu-ES\0hsb-DE\0mk-MK\0st-ZA\0ts-ZA\0tn-ZA\0ve-ZA\0"
 298         "xh-ZA\0zu-ZA\0af-ZA\0ka-GE\0fo-FO\0hi-IN\0mt-MT\0se-NO\0yi-001\0"
 299         "ms-MY\0kk-KZ\0ky-KG\0sw-KE\0tk-TM\0uz-Latn-UZ\0tt-RU\0bn-IN\0pa-IN\0"
 300         "gu-IN\0or-IN\0ta-IN\0te-IN\0kn-IN\0ml-IN\0as-IN\0mr-IN\0sa-IN\0mn-MN\0"
 301         "bo-CN\0cy-GB\0km-KH\0lo-LA\0my-MM\0gl-ES\0kok-IN\0mni-IN\0sd-Deva-IN\0"
 302         "syr-SY\0si-LK\0chr-Cher-US\0iu-Cans-CA\0am-ET\0tzm-Arab-MA\0"
 303         "ks-Arab-IN\0ne-NP\0fy-NL\0ps-AF\0fil-PH\0dv-MV\0bin-NG\0ff-Latn-NG\0"
 304         "ha-Latn-NG\0ibb-NG\0yo-NG\0quz-BO\0nso-ZA\0ba-RU\0lb-LU\0kl-GL\0"
 305         "ig-NG\0kr-Latn-NG\0om-ET\0ti-ET\0gn-PY\0haw-US\0la-001\0so-SO\0ii-CN\0"
 306         "pap-029\0arn-CL\0moh-CA\0br-FR\0ug-CN\0mi-NZ\0oc-FR\0co-FR\0gsw-FR\0"
 307         "sah-RU\0quc-Latn-GT\0rw-RW\0wo-SN\0prs-AF\0gd-GB\0ku-Arab-IQ\0"
 308         "qps-ploc\0qps-ploca\0en-US\0ar-IQ\0ca-ES-valencia\0zh-CN\0de-CH\0"
 309         "en-GB\0es-MX\0fr-BE\0it-CH\0nl-BE\0nn-NO\0pt-PT\0ro-MD\0ru-MD\0"
 310         "sr-Latn-CS\0sv-FI\0ur-IN\0az-Cyrl-AZ\0dsb-DE\0tn-BW\0se-SE\0ga-IE\0"
 311         "ms-BN\0uz-Cyrl-UZ\0bn-BD\0pa-Arab-PK\0ta-LK\0mn-Mong-CN\0sd-Arab-PK\0"
 312         "iu-Latn-CA\0tzm-Latn-DZ\0ks-Deva-IN\0ne-IN\0ff-Latn-SN\0quz-EC\0"
 313         "ti-ER\0qps-Latn-x-sh\0qps-plocm\0en-US\0ar-EG\0zh-HK\0de-AT\0en-AU\0"
 314         "es-ES\0fr-CA\0sr-Cyrl-CS\0se-FI\0mn-Mong-MN\0dz-BT\0quz-PE\0"
 315         "ks-Arab-IN\0ar-LY\0zh-SG\0de-LU\0en-CA\0es-GT\0fr-CH\0hr-BA\0smj-NO\0"
 316         "tzm-Tfng-MA\0ar-DZ\0zh-MO\0de-LI\0en-NZ\0es-CR\0fr-LU\0bs-Latn-BA\0"
 317         "smj-SE\0ar-MA\0en-IE\0es-PA\0fr-MC\0sr-Latn-BA\0sma-NO\0ar-TN\0en-ZA\0"
 318         "es-DO\0fr-029\0sr-Cyrl-BA\0sma-SE\0en-US\0ar-OM\0en-JM\0es-VE\0fr-RE\0"
 319         "bs-Cyrl-BA\0sms-FI\0en-US\0ar-YE\0en-029\0es-CO\0fr-CD\0sr-Latn-RS\0"
 320         "smn-FI\0en-US\0ar-SY\0en-BZ\0es-PE\0fr-SN\0sr-Cyrl-RS\0en-US\0ar-JO\0"
 321         "en-TT\0es-AR\0fr-CM\0sr-Latn-ME\0en-US\0ar-LB\0en-ZW\0es-EC\0fr-CI\0"
 322         "sr-Cyrl-ME\0en-US\0ar-KW\0en-PH\0es-CL\0fr-ML\0en-US\0ar-AE\0en-ID\0"
 323         "es-UY\0fr-MA\0en-US\0ar-BH\0en-HK\0es-PY\0fr-HT\0en-US\0ar-QA\0en-IN\0"
 324         "es-BO\0en-US\0en-MY\0es-SV\0en-US\0en-SG\0es-HN\0en-US\0en-AE\0es-NI\0"
 325         "es-PR\0es-US\0es-419\0es-CU\0bs-Cyrl-BA\0bs-Latn-BA\0sr-Cyrl-RS\0"
 326         "sr-Latn-RS\0smn-FI\0az-Cyrl-AZ\0sms-FI\0zh-CN\0nn-NO\0bs-Latn-BA\0"
 327         "az-Latn-AZ\0sma-SE\0uz-Cyrl-UZ\0mn-MN\0iu-Cans-CA\0tzm-Tfng-MA\0"
 328         "zh-HK\0nb-NO\0sr-Latn-RS\0tg-Cyrl-TJ\0dsb-DE\0smj-SE\0uz-Latn-UZ\0"
 329         "pa-Arab-PK\0mn-Mong-CN\0sd-Arab-PK\0chr-Cher-US\0iu-Latn-CA\0"
 330         "tzm-Latn-DZ\0ff-Latn-SN\0ha-Latn-NG\0quc-Latn-GT\0ku-Arab-IQ\0";
 331
 332 /* Translate a Windows language ID to its name.  Returns NULL if the ID is not
 333  * recognized.  */
 334 static const char *
 335 language_id_to_name(u16 id)
 336 {
 337         int l = 0;
 338         int r = ARRAY_LEN(language_id_map) - 1;
 339         do {
 340                 int m = (l + r) / 2;
 341                 if (id < language_id_map[m].id)
 342                         r = m - 1;
 343                 else if (id > language_id_map[m].id)
 344                         l = m + 1;
 345                 else
 346                         return &language_names[language_id_map[m].name_start_offset];
 347         } while (l <= r);
 348         return NULL;
 349 }
 350
 351 /* PE binary processor architecture codes (common ones only)  */
 352 #define IMAGE_FILE_MACHINE_I386         0x014C
 353 #define IMAGE_FILE_MACHINE_ARM          0x01C0
 354 #define IMAGE_FILE_MACHINE_ARMV7        0x01C4
 355 #define IMAGE_FILE_MACHINE_THUMB        0x01C2
 356 #define IMAGE_FILE_MACHINE_IA64         0x0200
 357 #define IMAGE_FILE_MACHINE_AMD64        0x8664
 358 #define IMAGE_FILE_MACHINE_ARM64        0xAA64
 359
 360 /* Windows API processor architecture codes (common ones only)  */
 361 #define PROCESSOR_ARCHITECTURE_INTEL    0
 362 #define PROCESSOR_ARCHITECTURE_ARM      5
 363 #define PROCESSOR_ARCHITECTURE_IA64     6
 364 #define PROCESSOR_ARCHITECTURE_AMD64    9
 365 #define PROCESSOR_ARCHITECTURE_ARM64    12
 366
 367 /* Translate a processor architecture code as given in a PE binary to the code
 368  * used by the Windows API.  Returns -1 if the code is not recognized.  */
 369 static int
 370 pe_arch_to_windows_arch(unsigned pe_arch)
 371 {
 372         switch (pe_arch) {
 373         case IMAGE_FILE_MACHINE_I386:
 374                 return PROCESSOR_ARCHITECTURE_INTEL;
 375         case IMAGE_FILE_MACHINE_ARM:
 376         case IMAGE_FILE_MACHINE_ARMV7:
 377         case IMAGE_FILE_MACHINE_THUMB:
 378                 return PROCESSOR_ARCHITECTURE_ARM;
 379         case IMAGE_FILE_MACHINE_IA64:
 380                 return PROCESSOR_ARCHITECTURE_IA64;
 381         case IMAGE_FILE_MACHINE_AMD64:
 382                 return PROCESSOR_ARCHITECTURE_AMD64;
 383         case IMAGE_FILE_MACHINE_ARM64:
 384                 return PROCESSOR_ARCHITECTURE_ARM64;
 385         }
 386         return -1;
 387 }
 388
 389 /* Gather information from kernel32.dll.  */
 390 static void
 391 set_info_from_kernel32(struct windows_info_ctx *ctx,
 392                        const void *contents, size_t size)
 393 {
 394         u32 e_lfanew;
 395         const u8 *pe_hdr;
 396         unsigned pe_arch;
 397         int arch;
 398
 399         /* Read the processor architecture from the executable header.  */
 400
 401         if (size < 0x40)
 402                 goto invalid;
 403
 404         e_lfanew = le32_to_cpu(*(le32 *)((u8 *)contents + 0x3C));
 405         if (e_lfanew > size || size - e_lfanew < 6 || (e_lfanew & 3))
 406                 goto invalid;
 407
 408         pe_hdr = (u8 *)contents + e_lfanew;
 409         if (*(le32 *)pe_hdr != cpu_to_le32(0x00004550)) /* "PE\0\0"  */
 410                 goto invalid;
 411
 412         pe_arch = le16_to_cpu(*(le16 *)(pe_hdr + 4));
 413         arch = pe_arch_to_windows_arch(pe_arch);
 414         if (arch >= 0) {
 415                 /* Save the processor architecture in the XML document.  */
 416                 set_number_property(ctx, T("WINDOWS/ARCH"), arch);
 417         } else {
 418                 XML_WARN("Architecture value %x from kernel32.dll "
 419                          "header not recognized", pe_arch);
 420         }
 421         return;
 422
 423 invalid:
 424         XML_WARN("kernel32.dll is not a valid PE binary.");
 425 }
 426
 427 /* Gather information from the SOFTWARE registry hive.  */
 428 static void
 429 set_info_from_software_hive(struct windows_info_ctx *ctx,
 430                             const struct regf *regf)
 431 {
 432         const tchar *version_key = T("Microsoft\\Windows NT\\CurrentVersion");
 433         s64 major_version = -1;
 434         s64 minor_version = -1;
 435         tchar *version_string;
 436         tchar *build_string;
 437
 438         /* Image flags  */
 439         copy_registry_string(ctx, regf, version_key, T("EditionID"),
 440                              T("FLAGS"));
 441
 442         /* Image display name  */
 443         copy_registry_string(ctx, regf, version_key, T("ProductName"),
 444                              T("DISPLAYNAME"));
 445
 446         /* Image display description  */
 447         copy_registry_string(ctx, regf, version_key, T("ProductName"),
 448                              T("DISPLAYDESCRIPTION"));
 449
 450         /* Edition ID  */
 451         copy_registry_string(ctx, regf, version_key, T("EditionID"),
 452                              T("WINDOWS/EDITIONID"));
 453
 454         /* Installation type  */
 455         copy_registry_string(ctx, regf, version_key, T("InstallationType"),
 456                              T("WINDOWS/INSTALLATIONTYPE"));
 457
 458         /* Product name  */
 459         copy_registry_string(ctx, regf, version_key, T("ProductName"),
 460                              T("WINDOWS/PRODUCTNAME"));
 461
 462         /* Major and minor version number  */
 463
 464         /* Note: in Windows 10, CurrentVersion was apparently fixed at 6.3.
 465          * Instead, the new values CurrentMajorVersionNumber and
 466          * CurrentMinorVersionNumber should be used.  */
 467
 468         get_number_from_registry(ctx, regf, version_key,
 469                                  T("CurrentMajorVersionNumber"), &major_version);
 470
 471         get_number_from_registry(ctx, regf, version_key,
 472                                  T("CurrentMinorVersionNumber"), &minor_version);
 473
 474         if (major_version < 0 || minor_version < 0) {
 475                 if (get_string_from_registry(ctx, regf, version_key,
 476                                              T("CurrentVersion"),
 477                                              &version_string))
 478                 {
 479                         if (2 != tscanf(version_string, T("%"PRIi64".%"PRIi64),
 480                                         &major_version, &minor_version))
 481                         {
 482                                 XML_WARN("Unrecognized CurrentVersion: %"TS,
 483                                          version_string);
 484                         }
 485                         FREE(version_string);
 486                 }
 487         }
 488
 489         if (major_version >= 0) {
 490                 set_number_property(ctx, T("WINDOWS/VERSION/MAJOR"),
 491                                     major_version);
 492                 if (minor_version >= 0) {
 493                         set_number_property(ctx, T("WINDOWS/VERSION/MINOR"),
 494                                             minor_version);
 495                 }
 496         }
 497
 498         /* Build number  */
 499
 500         /* Note: "CurrentBuild" is marked as obsolete in Windows XP registries
 501          * (example value:  "1.511.1 () (Obsolete data - do not use)"), and
 502          * "CurrentBuildNumber" contains the correct value.  But oddly enough,
 503          * it is "CurrentBuild" that contains the correct value on *later*
 504          * versions of Windows.  */
 505         if (get_string_from_registry(ctx, regf, version_key, T("CurrentBuild"),
 506                                      &build_string))
 507         {
 508                 if (tstrchr(build_string, T('.'))) {
 509                         FREE(build_string);
 510                         build_string = NULL;
 511                         get_string_from_registry(ctx, regf, version_key,
 512                                                  T("CurrentBuildNumber"),
 513                                                  &build_string);
 514                 }
 515                 if (build_string) {
 516                         set_string_property(ctx, T("WINDOWS/VERSION/BUILD"),
 517                                             build_string);
 518                         FREE(build_string);
 519                 }
 520         }
 521 }
 522
 523 /* Gather the default language from the SYSTEM registry hive.  */
 524 static void
 525 set_default_language(struct windows_info_ctx *ctx, const struct regf *regf)
 526 {
 527         tchar *string;
 528         unsigned language_id;
 529
 530         if (!get_string_from_registry(ctx, regf,
 531                                       T("ControlSet001\\Control\\Nls\\Language"),
 532                                       T("InstallLanguage"), &string))
 533                 return;
 534
 535         if (1 == tscanf(string, T("%x"), &language_id)) {
 536                 const char *language_name = language_id_to_name(language_id);
 537                 if (language_name) {
 538                         size_t len = strlen(language_name);
 539                         tchar tstr[len + 1];
 540                         for (size_t i = 0; i <= len; i++)
 541                                 tstr[i] = language_name[i];
 542                         set_string_property(ctx, T("WINDOWS/LANGUAGES/DEFAULT"),
 543                                             tstr);
 544                         FREE(string);
 545                         return;
 546                 }
 547         }
 548         XML_WARN("Unrecognized InstallLanguage: %"TS, string);
 549         FREE(string);
 550 }
 551
 552 /* Gather information from the SYSTEM registry hive.  */
 553 static void
 554 set_info_from_system_hive(struct windows_info_ctx *ctx, const struct regf *regf)
 555 {
 556         const tchar *windows_key = T("ControlSet001\\Control\\Windows");
 557         const tchar *uilanguages_key = T("ControlSet001\\Control\\MUI\\UILanguages");
 558         const tchar *productoptions_key = T("ControlSet001\\Control\\ProductOptions");
 559         s64 spbuild;
 560         s64 splevel;
 561         tchar **subkeys;
 562
 563         /* Service pack build  */
 564         if (get_number_from_registry(ctx, regf, windows_key,
 565                                      T("CSDBuildNumber"), &spbuild))
 566                 set_number_property(ctx, T("WINDOWS/VERSION/SPBUILD"), spbuild);
 567
 568         /* Service pack level  */
 569         if (get_number_from_registry(ctx, regf, windows_key,
 570                                      T("CSDVersion"), &splevel))
 571                 set_number_property(ctx, T("WINDOWS/VERSION/SPLEVEL"), splevel >> 8);
 572
 573         /* Product type  */
 574         copy_registry_string(ctx, regf, productoptions_key, T("ProductType"),
 575                              T("WINDOWS/PRODUCTTYPE"));
 576
 577         /* Product suite  */
 578         copy_registry_string(ctx, regf, productoptions_key, T("ProductSuite"),
 579                              T("WINDOWS/PRODUCTSUITE"));
 580
 581         /* Hardware abstraction layer  */
 582         copy_registry_string(ctx, regf,
 583                              T("ControlSet001\\Control\\Class\\{4D36E966-E325-11CE-BFC1-08002BE10318}\\0000"),
 584                              T("MatchingDeviceId"),
 585                              T("WINDOWS/HAL"));
 586
 587         /* Languages  */
 588         if (list_subkeys_in_registry(ctx, regf, uilanguages_key, &subkeys)) {
 589                 tchar property_name[64];
 590                 for (tchar **p = subkeys; *p; p++) {
 591                         tsprintf(property_name,
 592                                  T("WINDOWS/LANGUAGES/LANGUAGE[%zu]"), p - subkeys + 1);
 593                         set_string_property(ctx, property_name, *p);
 594                 }
 595                 hive_free_subkeys_list(subkeys);
 596         }
 597
 598         /* Default language  */
 599         set_default_language(ctx, regf);
 600 }
 601
 602 /* Load the contents of a file in the currently selected WIM image into memory.
 603  */
 604 static void *
 605 load_file_contents(struct windows_info_ctx *ctx,
 606                    const struct wim_dentry *dentry, const char *filename,
 607                    size_t *size_ret)
 608 {
 609         const struct blob_descriptor *blob;
 610         void *contents;
 611         int ret;
 612
 613         if (!dentry) {
 614                 XML_WARN("%s does not exist", filename);
 615                 return NULL;
 616         }
 617
 618         blob = inode_get_blob_for_unnamed_data_stream(dentry->d_inode,
 619                                                       ctx->wim->blob_table);
 620         if (!blob) {
 621                 XML_WARN("%s has no contents", filename);
 622                 return NULL;
 623         }
 624
 625         ret = read_blob_into_alloc_buf(blob, &contents);
 626         if (ret) {
 627                 XML_WARN("Error loading %s (size=%"PRIu64"): %"TS,
 628                          filename, blob->size, wimlib_get_error_string(ret));
 629                 ctx->oom_encountered |= (ret == WIMLIB_ERR_NOMEM &&
 630                                          blob->size < 100000000);
 631                 return NULL;
 632         }
 633
 634         *size_ret = blob->size;
 635         return contents;
 636 }
 637
 638 /* Load and validate a registry hive file.  */
 639 static void *
 640 load_hive(struct windows_info_ctx *ctx, const struct wim_dentry *dentry,
 641           const char *filename)
 642 {
 643         void *hive_mem;
 644         size_t hive_size;
 645
 646         hive_mem = load_file_contents(ctx, dentry, filename, &hive_size);
 647         if (hive_mem && !is_registry_valid(ctx, hive_mem, hive_size)) {
 648                 XML_WARN("%s is not a valid registry hive!", filename);
 649                 FREE(hive_mem);
 650                 hive_mem = NULL;
 651         }
 652         return hive_mem;
 653 }
 654
 655 /* Set the WINDOWS/SYSTEMROOT property to the name of the directory specified by
 656  * 'systemroot'.  */
 657 static void
 658 set_systemroot_property(struct windows_info_ctx *ctx,
 659                         const struct wim_dentry *systemroot)
 660 {
 661         utf16lechar *uname;
 662         const tchar *name;
 663         size_t name_nbytes;
 664         int ret;
 665
 666         /* to uppercase ...  */
 667         uname = utf16le_dupz(systemroot->d_name, systemroot->d_name_nbytes);
 668         if (!uname) {
 669                 ctx->oom_encountered = true;
 670                 goto out;
 671         }
 672         for (size_t i = 0; i < systemroot->d_name_nbytes / 2; i++)
 673                 uname[i] = cpu_to_le16(upcase[le16_to_cpu(uname[i])]);
 674
 675         /* to tstring ...  */
 676         ret = utf16le_get_tstr(uname, systemroot->d_name_nbytes,
 677                                &name, &name_nbytes);
 678         if (ret) {
 679                 ctx->oom_encountered |= (ret == WIMLIB_ERR_NOMEM);
 680                 XML_WARN("Failed to get systemroot name: %"TS,
 681                          wimlib_get_error_string(ret));
 682                 goto out;
 683         }
 684         set_string_property(ctx, T("WINDOWS/SYSTEMROOT"), name);
 685         utf16le_put_tstr(name);
 686 out:
 687         FREE(uname);
 688 }
 689
 690 static int
 691 do_set_windows_specific_info(WIMStruct *wim,
 692                              const struct wim_dentry *systemroot,
 693                              const struct wim_dentry *kernel32,
 694                              const struct wim_dentry *software,
 695                              const struct wim_dentry *system)
 696 {
 697         void *contents;
 698         size_t size;
 699         struct windows_info_ctx _ctx = {
 700                 .wim = wim,
 701                 .image = wim->current_image,
 702                 .oom_encountered = false,
 703                 .debug_enabled = (tgetenv(T("WIMLIB_DEBUG_XML_INFO")) != NULL),
 704         }, *ctx = &_ctx;
 705
 706         set_systemroot_property(ctx, systemroot);
 707
 708         if ((contents = load_file_contents(ctx, kernel32, "kernel32.dll", &size))) {
 709                 set_info_from_kernel32(ctx, contents, size);
 710                 FREE(contents);
 711         }
 712
 713         if ((contents = load_hive(ctx, software, "SOFTWARE"))) {
 714                 set_info_from_software_hive(ctx, contents);
 715                 FREE(contents);
 716         }
 717
 718         if ((contents = load_hive(ctx, system, "SYSTEM"))) {
 719                 set_info_from_system_hive(ctx, contents);
 720                 FREE(contents);
 721         }
 722
 723         if (ctx->oom_encountered) {
 724                 ERROR("Ran out of memory while setting Windows-specific "
 725                       "metadata in the WIM file's XML document.");
 726                 return WIMLIB_ERR_NOMEM;
 727         }
 728
 729         return 0;
 730 }
 731
 732 /* Windows */
 733 static const utf16lechar windows_name[] = {
 734         cpu_to_le16('W'), cpu_to_le16('i'), cpu_to_le16('n'),
 735         cpu_to_le16('d'), cpu_to_le16('o'), cpu_to_le16('w'),
 736         cpu_to_le16('s'),
 737 };
 738
 739 /* System32 */
 740 static const utf16lechar system32_name[] = {
 741         cpu_to_le16('S'), cpu_to_le16('y'), cpu_to_le16('s'),
 742         cpu_to_le16('t'), cpu_to_le16('e'), cpu_to_le16('m'),
 743         cpu_to_le16('3'), cpu_to_le16('2'),
 744 };
 745
 746 /* kernel32.dll */
 747 static const utf16lechar kernel32_name[] = {
 748         cpu_to_le16('k'), cpu_to_le16('e'), cpu_to_le16('r'),
 749         cpu_to_le16('n'), cpu_to_le16('e'), cpu_to_le16('l'),
 750         cpu_to_le16('3'), cpu_to_le16('2'), cpu_to_le16('.'),
 751         cpu_to_le16('d'), cpu_to_le16('l'), cpu_to_le16('l'),
 752 };
 753
 754 /* config */
 755 static const utf16lechar config_name[] = {
 756         cpu_to_le16('c'), cpu_to_le16('o'), cpu_to_le16('n'),
 757         cpu_to_le16('f'), cpu_to_le16('i'), cpu_to_le16('g'),
 758 };
 759
 760 /* SOFTWARE */
 761 static const utf16lechar software_name[] = {
 762         cpu_to_le16('S'), cpu_to_le16('O'), cpu_to_le16('F'),
 763         cpu_to_le16('T'), cpu_to_le16('W'), cpu_to_le16('A'),
 764         cpu_to_le16('R'), cpu_to_le16('E'),
 765 };
 766
 767 /* SYSTEM */
 768 static const utf16lechar system_name[] = {
 769         cpu_to_le16('S'), cpu_to_le16('Y'), cpu_to_le16('S'),
 770         cpu_to_le16('T'), cpu_to_le16('E'), cpu_to_le16('M'),
 771 };
 772
 773 #define GET_CHILD(parent, child_name)                           \
 774         get_dentry_child_with_utf16le_name(parent,              \
 775                                            child_name,          \
 776                                            sizeof(child_name),  \
 777                                            WIMLIB_CASE_INSENSITIVE)
 778
 779 static bool
 780 is_default_systemroot(const struct wim_dentry *potential_systemroot)
 781 {
 782         return !cmp_utf16le_strings(potential_systemroot->d_name,
 783                                     potential_systemroot->d_name_nbytes / 2,
 784                                     windows_name,
 785                                     ARRAY_LEN(windows_name),
 786                                     true);
 787 }
 788
 789 /*
 790  * Set Windows-specific XML information for the currently selected WIM image.
 791  *
 792  * This process is heavily based on heuristics and hard-coded logic related to
 793  * where Windows stores certain types of information.  Therefore, it simply
 794  * tries to set as much information as possible.  If there's a problem, it skips
 795  * the affected information and proceeds to the next part.  It only returns an
 796  * error code if there was a severe problem such as out-of-memory.
 797  */
 798 int
 799 set_windows_specific_info(WIMStruct *wim)
 800 {
 801         const struct wim_dentry *root, *potential_systemroot,
 802                                 *best_systemroot = NULL,
 803                                 *best_kernel32 = NULL,
 804                                 *best_software = NULL,
 805                                 *best_system = NULL;
 806         int best_score = 0;
 807
 808         root = wim_get_current_root_dentry(wim);
 809         if (!root)
 810                 return 0;
 811
 812         /* Find the system root.  This is usually the toplevel directory
 813          * "Windows", but it might be a different toplevel directory.  Choose
 814          * the directory that contains the greatest number of the files we want:
 815          * System32/kernel32.dll, System32/config/SOFTWARE, and
 816          * System32/config/SYSTEM.  Compare all names case insensitively.  */
 817         for_dentry_child(potential_systemroot, root) {
 818                 const struct wim_dentry *system32, *kernel32, *config,
 819                                         *software = NULL, *system = NULL;
 820                 int score;
 821
 822                 if (!dentry_is_directory(potential_systemroot))
 823                         continue;
 824                 system32 = GET_CHILD(potential_systemroot, system32_name);
 825                 if (!system32)
 826                         continue;
 827                 kernel32 = GET_CHILD(system32, kernel32_name);
 828                 config = GET_CHILD(system32, config_name);
 829                 if (config) {
 830                         software = GET_CHILD(config, software_name);
 831                         system = GET_CHILD(config, system_name);
 832                 }
 833
 834                 score = !!kernel32 + !!software + !!system;
 835                 if (score >= best_score) {
 836                         /* If there's a tie, prefer the "Windows" directory.  */
 837                         if (score > best_score ||
 838                             is_default_systemroot(potential_systemroot))
 839                         {
 840                                 best_score = score;
 841                                 best_systemroot = potential_systemroot;
 842                                 best_kernel32 = kernel32;
 843                                 best_software = software;
 844                                 best_system = system;
 845                         }
 846                 }
 847         }
 848
 849         if (likely(best_score == 0))
 850                 return 0;  /* No Windows system root found.  */
 851
 852         /* Found the Windows system root.  */
 853         return do_set_windows_specific_info(wim, best_systemroot, best_kernel32,
 854                                             best_software, best_system);
 855 }