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

   1 /*
   2  * join.c
   3  *
   4  * Join split WIMs (sometimes named as .swm files) together into one WIM.
   5  */
   6
   7 /*
   8  * Copyright (C) 2012 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 #include "wimlib_internal.h"
  27 #include "lookup_table.h"
  28 #include "xml.h"
  29 #include <stdlib.h>
  30
  31 static int copy_lte_to_table(struct lookup_table_entry *lte, void *table)
  32 {
  33         struct lookup_table_entry *copy;
  34         copy = MALLOC(sizeof(struct lookup_table_entry));
  35         if (!copy)
  36                 return WIMLIB_ERR_NOMEM;
  37         memcpy(copy, lte, sizeof(struct lookup_table_entry));
  38         lookup_table_insert(table, copy);
  39         return 0;
  40 }
  41
  42 static int lookup_table_join(struct lookup_table *table,
  43                              struct lookup_table *new)
  44 {
  45         return for_lookup_table_entry(new, copy_lte_to_table, table);
  46 }
  47
  48
  49 static int cmp_swms_by_part_number(const void *swm1, const void *swm2)
  50 {
  51         u16 partno_1 = (*(WIMStruct**)swm1)->hdr.part_number;
  52         u16 partno_2 = (*(WIMStruct**)swm2)->hdr.part_number;
  53         return (int)partno_1 - (int)partno_2;
  54 }
  55
  56 int verify_swm_set(WIMStruct *w, WIMStruct **additional_swms,
  57                    unsigned num_additional_swms)
  58 {
  59         unsigned total_parts = w->hdr.total_parts;
  60         int ctype;
  61         const u8 *guid;
  62
  63         if (total_parts != num_additional_swms + 1) {
  64                 ERROR("`%s' says there are %u parts in the spanned set, "
  65                       "but %s%u part%s provided",
  66                       w->filename, w->hdr.total_parts,
  67                       (num_additional_swms + 1 < w->hdr.total_parts) ? "only " : "",
  68                       num_additional_swms + 1,
  69                       (num_additional_swms) ? "s were" : " was");
  70                 return WIMLIB_ERR_SPLIT_INVALID;
  71         }
  72         if (w->hdr.part_number != 1) {
  73                 ERROR("WIM `%s' is not the first part of the split WIM.",
  74                       w->filename);
  75                 return WIMLIB_ERR_SPLIT_INVALID;
  76         }
  77         for (unsigned i = 0; i < num_additional_swms; i++) {
  78                 if (additional_swms[i]->hdr.total_parts != total_parts) {
  79                         ERROR("WIM `%s' says there are %u parts in the spanned set, "
  80                               "but %u parts were provided",
  81                               additional_swms[i]->filename,
  82                               additional_swms[i]->hdr.total_parts,
  83                               total_parts);
  84                         return WIMLIB_ERR_SPLIT_INVALID;
  85                 }
  86         }
  87
  88         /* keep track of ctype and guid just to make sure they are the same for
  89          * all the WIMs. */
  90         ctype = wimlib_get_compression_type(w);
  91         guid = w->hdr.guid;
  92
  93         WIMStruct *parts_to_swms[num_additional_swms];
  94         ZERO_ARRAY(parts_to_swms);
  95         for (unsigned i = 0; i < num_additional_swms; i++) {
  96
  97                 WIMStruct *swm = additional_swms[i];
  98
  99                 if (wimlib_get_compression_type(swm) != ctype) {
 100                         ERROR("The split WIMs do not all have the same "
 101                               "compression type");
 102                         return WIMLIB_ERR_SPLIT_INVALID;
 103                 }
 104                 if (memcmp(guid, swm->hdr.guid, WIM_GID_LEN) != 0) {
 105                         ERROR("The split WIMs do not all have the same "
 106                               "GUID");
 107                         return WIMLIB_ERR_SPLIT_INVALID;
 108                 }
 109                 if (swm->hdr.part_number == 1) {
 110                         ERROR("WIMs `%s' and `%s' both are marked as the "
 111                               "first WIM in the spanned set",
 112                               w->filename, swm->filename);
 113                         return WIMLIB_ERR_SPLIT_INVALID;
 114                 }
 115                 if (swm->hdr.part_number == 0 ||
 116                     swm->hdr.part_number > total_parts)
 117                 {
 118                         ERROR("WIM `%s' says it is part %u in the spanned set, "
 119                               "but the part number must be in the range "
 120                               "[1, %u]",
 121                               swm->filename, swm->hdr.part_number, total_parts);
 122                         return WIMLIB_ERR_SPLIT_INVALID;
 123                 }
 124                 if (parts_to_swms[swm->hdr.part_number - 2])
 125                 {
 126                         ERROR("`%s' and `%s' are both marked as part %u of %u "
 127                               "in the spanned set",
 128                               parts_to_swms[swm->hdr.part_number - 2]->filename,
 129                               swm->filename,
 130                               swm->hdr.part_number,
 131                               total_parts);
 132                         return WIMLIB_ERR_SPLIT_INVALID;
 133                 } else {
 134                         parts_to_swms[swm->hdr.part_number - 2] = swm;
 135                 }
 136         }
 137         return 0;
 138 }
 139
 140 /*
 141  * Joins lookup tables from the parts of a split WIM.
 142  *
 143  * @w specifies the first part, while @additional_swms and @num_additional_swms
 144  * specify an array of points to the WIMStruct's for additional split WIM parts.
 145  *
 146  * On success, 0 is returned on a pointer to the joined lookup table is returned
 147  * in @table_ret.
 148  *
 149  * The reason we join the lookup tables is so:
 150  *      - We only have to search one lookup table to find the location of a
 151  *      resource in the entire split WIM.
 152  *      - Each lookup table entry will have a pointer to its split WIM part (and
 153  *      a part number field, although we don't really use it).
 154  */
 155 int new_joined_lookup_table(WIMStruct *w,
 156                             WIMStruct **additional_swms,
 157                             unsigned num_additional_swms,
 158                             struct lookup_table **table_ret)
 159 {
 160         struct lookup_table *table;
 161         int ret;
 162         unsigned i;
 163
 164
 165         table = new_lookup_table(9001);
 166         if (!table)
 167                 return WIMLIB_ERR_NOMEM;
 168         ret = lookup_table_join(table, w->lookup_table);
 169         if (ret != 0)
 170                 goto out_free_table;
 171         for (i = 0; i < num_additional_swms; i++) {
 172                 ret = lookup_table_join(table, additional_swms[i]->lookup_table);
 173                 if (ret != 0)
 174                         goto out_free_table;
 175         }
 176         *table_ret = table;
 177         return 0;
 178 out_free_table:
 179         free_lookup_table(table);
 180         return ret;
 181 }
 182
 183
 184 static int join_wims(WIMStruct **swms, uint num_swms, WIMStruct *joined_wim,
 185                      int write_flags)
 186 {
 187         uint i;
 188         int ret;
 189         FILE *out_fp = joined_wim->out_fp;
 190         u64 total_bytes = wim_info_get_total_bytes(swms[0]->wim_info);
 191
 192         swms[0]->write_metadata = false;
 193         for (i = 0; i < num_swms; i++) {
 194                 if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS) {
 195                         off_t cur_offset = ftello(out_fp);
 196                         printf("Writing resources from part %u of %u "
 197                                "(%"PRIu64" of %"PRIu64" bytes, %.0f%% done)\n",
 198                                i + 1, num_swms, cur_offset, total_bytes,
 199                                (double)cur_offset / total_bytes * 100.0);
 200                 }
 201                 swms[i]->fp = fopen(swms[i]->filename, "rb");
 202                 if (!swms[i]->fp) {
 203                         ERROR_WITH_ERRNO("Failed to reopen `%s'",
 204                                          swms[i]->filename);
 205                         return WIMLIB_ERR_OPEN;
 206                 }
 207                 swms[i]->out_fp = out_fp;
 208                 swms[i]->hdr.part_number = 1;
 209                 ret = for_lookup_table_entry(swms[i]->lookup_table,
 210                                              copy_resource, swms[i]);
 211                 if (ret != 0)
 212                         return ret;
 213                 if (i != 0) {
 214                         fclose(swms[i]->fp);
 215                         swms[i]->fp = NULL;
 216                 }
 217         }
 218         swms[0]->write_metadata = true;
 219         if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS)
 220                 printf("Writing %d metadata resources\n",
 221                         swms[0]->hdr.image_count);
 222
 223         for (i = 0; i < swms[0]->hdr.image_count; i++) {
 224                 ret = copy_resource(swms[0]->image_metadata[i].metadata_lte,
 225                                     swms[0]);
 226                 if (ret != 0)
 227                         return ret;
 228         }
 229
 230         off_t lookup_table_offset = ftello(out_fp);
 231
 232         if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS)
 233                 printf("Writing lookup tables, XML data, and header\n");
 234         /* Now write the lookup table for the joined wim.  Since the lookup
 235          * table has no header, we can just concatenate the lookup tables of all
 236          * the SWM parts. */
 237         for (i = 0; i < num_swms; i++) {
 238                 ret = write_lookup_table(swms[i]->lookup_table, out_fp);
 239                 if (ret != 0)
 240                         return ret;
 241         }
 242         off_t xml_data_offset = ftello(out_fp);
 243
 244         if (lookup_table_offset == -1 || xml_data_offset == -1) {
 245                 ERROR_WITH_ERRNO("Failed to get file offset");
 246                 return WIMLIB_ERR_WRITE;
 247         }
 248         swms[0]->hdr.lookup_table_res_entry.offset = lookup_table_offset;
 249         swms[0]->hdr.lookup_table_res_entry.size =
 250                                         xml_data_offset - lookup_table_offset;
 251
 252
 253         /* finish_write is called on the first swm, not the joined_wim, because
 254          * the first swm is the one that has the image metadata and XML data
 255          * attached to it.  */
 256         swms[0]->hdr.flags &= ~WIM_HDR_FLAG_SPANNED;
 257         swms[0]->hdr.total_parts = 1;
 258         return finish_write(swms[0], WIM_ALL_IMAGES, write_flags, 0);
 259 }
 260
 261
 262 WIMLIBAPI int wimlib_join(const char **swm_names, unsigned num_swms,
 263                           const char *output_path, int flags)
 264 {
 265         int i;
 266         int ret;
 267         int part_idx;
 268         int write_flags = 0;
 269         WIMStruct *joined_wim = NULL;
 270         WIMStruct *swms[num_swms];
 271
 272         int ctype;
 273         u8 *guid;
 274
 275         if (num_swms < 1)
 276                 return WIMLIB_ERR_INVALID_PARAM;
 277
 278         ZERO_ARRAY(swms);
 279
 280         for (i = 0; i < num_swms; i++) {
 281                 ret = wimlib_open_wim(swm_names[i],
 282                                       flags | WIMLIB_OPEN_FLAG_SPLIT_OK, &swms[i]);
 283                 if (ret != 0)
 284                         goto out;
 285
 286                 /* don't open all the parts at the same time, in case there are
 287                  * a lot of them */
 288                 fclose(swms[i]->fp);
 289                 swms[i]->fp = NULL;
 290         }
 291
 292         qsort(swms, num_swms, sizeof(swms[0]), cmp_swms_by_part_number);
 293
 294         ret = verify_swm_set(swms[0], &swms[1], num_swms - 1);
 295         if (ret != 0)
 296                 goto out;
 297
 298         joined_wim = new_wim_struct();
 299         if (!joined_wim) {
 300                 ret = WIMLIB_ERR_NOMEM;
 301                 goto out;
 302         }
 303
 304         if (flags & WIMLIB_OPEN_FLAG_CHECK_INTEGRITY)
 305                 write_flags |= WIMLIB_WRITE_FLAG_CHECK_INTEGRITY;
 306         if (flags & WIMLIB_OPEN_FLAG_SHOW_PROGRESS)
 307                 write_flags |= WIMLIB_WRITE_FLAG_SHOW_PROGRESS;
 308
 309         ret = begin_write(joined_wim, output_path, write_flags);
 310         if (ret != 0)
 311                 goto out;
 312         ret = join_wims(swms, num_swms, joined_wim, write_flags);
 313 out:
 314         for (i = 0; i < num_swms; i++) {
 315                 /* out_fp is the same in all the swms and joined_wim; only close
 316                  * it one time, when freeing joined_wim. */
 317                 if (swms[i]) {
 318                         swms[i]->out_fp = NULL;
 319                         wimlib_free(swms[i]);
 320                 }
 321         }
 322         wimlib_free(joined_wim);
 323         return ret;
 324 }