]> wimlib.net Git - wimlib/blobdiff - src/join.c
Cache compression type in WIMStruct and wim_lookup_table_entry
[wimlib] / src / join.c
index 3710eec85ec72ec02fee1bae2d857cf328fd79b2..2c459d30aa1d4826a80835cf3ee16021d823eabe 100644 (file)
@@ -5,7 +5,7 @@
  */
 
 /*
- * Copyright (C) 2012 Eric Biggers
+ * Copyright (C) 2012, 2013 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  */
 
-#include "wimlib_internal.h"
-#include "lookup_table.h"
-#include "xml.h"
-#include <stdlib.h>
+#ifdef HAVE_CONFIG_H
+#  include "config.h"
+#endif
 
-static int copy_lte_to_table(struct lookup_table_entry *lte, void *table)
-{
-       struct lookup_table_entry *copy;
-       copy = new_lookup_table_entry();
-       if (!copy)
-               return WIMLIB_ERR_NOMEM;
-       memcpy(copy, lte, sizeof(struct lookup_table_entry));
-       lookup_table_insert(table, copy);
-       return 0;
-}
-
-static int lookup_table_join(struct lookup_table *table,
-                            struct lookup_table *new)
-{
-       return for_lookup_table_entry(new, copy_lte_to_table, table);
-}
+#include "wimlib.h"
+#include "wimlib/lookup_table.h"
+#include "wimlib/metadata.h"
+#include "wimlib/resource.h"
+#include "wimlib/swm.h"
+#include "wimlib/write.h"
+#include "wimlib/xml.h"
 
+#include <stdlib.h> /* for qsort() */
 
-static int cmp_swms_by_part_number(const void *swm1, const void *swm2)
-{
-       u16 partno_1 = (*(WIMStruct**)swm1)->hdr.part_number;
-       u16 partno_2 = (*(WIMStruct**)swm2)->hdr.part_number;
-       return (int)partno_1 - (int)partno_2;
-}
-
-int verify_swm_set(WIMStruct *w, WIMStruct **additional_swms,
-                  unsigned num_additional_swms)
+static int
+join_wims(WIMStruct **swms, unsigned num_swms,
+         WIMStruct *joined_wim, int write_flags,
+         wimlib_progress_func_t progress_func)
 {
-       unsigned total_parts = w->hdr.total_parts;
-       int ctype;
-       const u8 *guid;
-
-       if (total_parts != num_additional_swms + 1) {
-               ERROR("`%s' says there are %u parts in the spanned set, "
-                     "but %s%u part%s provided",
-                     w->filename, w->hdr.total_parts,
-                     (num_additional_swms + 1 < w->hdr.total_parts) ? "only " : "",
-                     num_additional_swms + 1,
-                     (num_additional_swms) ? "s were" : " was");
-               return WIMLIB_ERR_SPLIT_INVALID;
-       }
-       if (w->hdr.part_number != 1) {
-               ERROR("WIM `%s' is not the first part of the split WIM.",
-                     w->filename);
-               return WIMLIB_ERR_SPLIT_INVALID;
-       }
-       for (unsigned i = 0; i < num_additional_swms; i++) {
-               if (additional_swms[i]->hdr.total_parts != total_parts) {
-                       ERROR("WIM `%s' says there are %u parts in the spanned set, "
-                             "but %u parts were provided", 
-                             additional_swms[i]->filename,
-                             additional_swms[i]->hdr.total_parts,
-                             total_parts);
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               }
-       }
-
-       /* keep track of ctype and guid just to make sure they are the same for
-        * all the WIMs. */
-       ctype = wimlib_get_compression_type(w);
-       guid = w->hdr.guid;
-
-       WIMStruct *parts_to_swms[num_additional_swms];
-       ZERO_ARRAY(parts_to_swms);
-       for (unsigned i = 0; i < num_additional_swms; i++) {
-
-               WIMStruct *swm = additional_swms[i];
-
-               if (wimlib_get_compression_type(swm) != ctype) {
-                       ERROR("The split WIMs do not all have the same "
-                             "compression type");
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               }
-               if (memcmp(guid, swm->hdr.guid, WIM_GID_LEN) != 0) {
-                       ERROR("The split WIMs do not all have the same "
-                             "GUID");
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               }
-               if (swm->hdr.part_number == 1) {
-                       ERROR("WIMs `%s' and `%s' both are marked as the "
-                             "first WIM in the spanned set", 
-                             w->filename, swm->filename);
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               }
-               if (swm->hdr.part_number == 0 ||
-                   swm->hdr.part_number > total_parts)
-               {
-                       ERROR("WIM `%s' says it is part %u in the spanned set, "
-                             "but the part number must be in the range "
-                             "[1, %u]",
-                             swm->filename, swm->hdr.part_number, total_parts);
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               }
-               if (parts_to_swms[swm->hdr.part_number - 2])
-               {
-                       ERROR("`%s' and `%s' are both marked as part %u of %u "
-                             "in the spanned set",
-                             parts_to_swms[swm->hdr.part_number - 2]->filename,
-                             swm->filename,
-                             swm->hdr.part_number,
-                             total_parts);
-                       return WIMLIB_ERR_SPLIT_INVALID;
-               } else {
-                       parts_to_swms[swm->hdr.part_number - 2] = swm;
-               }
-       }
-       return 0;
-}
-
-int new_joined_lookup_table(WIMStruct *w,
-                           WIMStruct **additional_swms,
-                           unsigned num_additional_swms,
-                           struct lookup_table **table_ret)
-{
-       struct lookup_table *table;
        int ret;
        unsigned i;
+       union wimlib_progress_info progress;
+       u64 total_bytes = 0;
+       u64 part_bytes;
+       u64 swm_part_sizes[num_swms];
 
-
-       table = new_lookup_table(9001);
-       if (!table)
-               return WIMLIB_ERR_NOMEM;
-       ret = lookup_table_join(table, w->lookup_table);
-       if (ret != 0)
-               goto out_free_table;
-       for (i = 0; i < num_additional_swms; i++) {
-               ret = lookup_table_join(table, additional_swms[i]->lookup_table);
-               if (ret != 0)
-                       goto out_free_table;
+       /* Calculate total size of the streams in the split WIM parts. */
+       for (i = 0; i < num_swms; i++) {
+               part_bytes = lookup_table_total_stream_size(swms[i]->lookup_table);
+               swm_part_sizes[i] = part_bytes;
+               total_bytes += part_bytes;
        }
-       *table_ret = table;
-       return 0;
-out_free_table:
-       free_lookup_table(table);
-       return ret;
-}
 
-
-static int join_wims(WIMStruct **swms, uint num_swms, WIMStruct *joined_wim,
-                    int write_flags)
-{
-       uint i;
-       int ret;
-       FILE *out_fp = joined_wim->out_fp;
-       u64 total_bytes = wim_info_get_total_bytes(swms[0]->wim_info);
-
-       swms[0]->write_metadata = false;
-       for (i = 0; i < num_swms; i++) {
-               if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS) {
-                       off_t cur_offset = ftello(out_fp);
-                       printf("Writing resources from part %u of %u "
-                              "(%"PRIu64" of %"PRIu64" bytes, %.0f%% done)\n",
-                              i + 1, num_swms, cur_offset, total_bytes,
-                              (double)cur_offset / total_bytes * 100.0);
-               }
-               swms[i]->fp = fopen(swms[i]->filename, "rb");
-               if (!swms[i]->fp) {
-                       ERROR_WITH_ERRNO("Failed to reopen `%s'",
-                                        swms[i]->filename);
-                       return WIMLIB_ERR_OPEN;
-               }
-               swms[i]->out_fp = out_fp;
-               swms[i]->hdr.part_number = 1;
-               ret = for_lookup_table_entry(swms[i]->lookup_table, 
-                                            copy_resource, swms[i]);
-               if (ret != 0)
-                       return ret;
-               if (i != 0) {
-                       fclose(swms[i]->fp);
-                       swms[i]->fp = NULL;
-               }
+       if (progress_func) {
+               progress.join.total_bytes     = total_bytes;
+               progress.join.total_parts     = swms[0]->hdr.total_parts;
+               progress.join.completed_bytes = 0;
+               progress.join.completed_parts = 0;
+               progress_func(WIMLIB_PROGRESS_MSG_JOIN_STREAMS, &progress);
        }
-       swms[0]->write_metadata = true;
-       if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS)
-               printf("Writing %d metadata resources\n", 
-                       swms[0]->hdr.image_count);
 
-       for (i = 0; i < swms[0]->hdr.image_count; i++) {
-               ret = copy_resource(swms[0]->image_metadata[i].metadata_lte, 
-                                   swms[0]);
-               if (ret != 0)
+       /* Write the non-metadata resources from each SWM part */
+       for (i = 0; i < num_swms; i++) {
+               ret = reopen_wim(swms[i]);
+               if (ret)
                        return ret;
-       }
+               swms[i]->out_fd = joined_wim->out_fd;
+               swms[i]->hdr.part_number = 1;
 
-       off_t lookup_table_offset = ftello(out_fp);
+               ret = for_lookup_table_entry_pos_sorted(swms[i]->lookup_table,
+                                                       copy_resource,
+                                                       swms[i]);
+               swms[i]->out_fd = -1;
+               if (i != 0)
+                       close_wim(swms[i]);
 
-       if (write_flags & WIMLIB_WRITE_FLAG_SHOW_PROGRESS)
-               printf("Writing lookup tables, XML data, and header\n");
-       /* Now write the lookup table for the joined wim.  Since the lookup
-        * table has no header, we can just concatenate the lookup tables of all
-        * the SWM parts. */
-       for (i = 0; i < num_swms; i++) {
-               ret = write_lookup_table(swms[i]->lookup_table, out_fp);
-               if (ret != 0)
+               if (ret)
                        return ret;
-       }
-       off_t xml_data_offset = ftello(out_fp);
 
-       if (lookup_table_offset == -1 || xml_data_offset == -1) {
-               ERROR_WITH_ERRNO("Failed to get file offset");
-               return WIMLIB_ERR_WRITE;
+               if (progress_func) {
+                       progress.join.completed_bytes += swm_part_sizes[i];
+                       progress.join.completed_parts++;
+                       progress_func(WIMLIB_PROGRESS_MSG_JOIN_STREAMS, &progress);
+               }
        }
-       swms[0]->hdr.lookup_table_res_entry.offset = lookup_table_offset;
-       swms[0]->hdr.lookup_table_res_entry.size = 
-                                       xml_data_offset - lookup_table_offset;
 
+       /* Copy the metadata resources from the first SWM part */
+       joined_wim->hdr.image_count = swms[0]->hdr.image_count;
+       for (i = 0; i < joined_wim->hdr.image_count; i++) {
+               ret = copy_resource(swms[0]->image_metadata[i]->metadata_lte,
+                                   joined_wim);
+               if (ret)
+                       return ret;
+       }
 
-       /* finish_write is called on the first swm, not the joined_wim, because
-        * the first swm is the one that has the image metadata and XML data
-        * attached to it.  */
-       swms[0]->hdr.flags &= ~WIM_HDR_FLAG_SPANNED;
-       swms[0]->hdr.total_parts = 1;
-       return finish_write(swms[0], WIM_ALL_IMAGES, write_flags, 0);
+       /* Write lookup table, XML data, and optional integrity table */
+       merge_lookup_tables(joined_wim, swms, num_swms);
+       free_wim_info(joined_wim->wim_info);
+       joined_wim->wim_info = swms[0]->wim_info;
+       joined_wim->image_metadata = swms[0]->image_metadata;
+       ret = finish_write(joined_wim, WIMLIB_ALL_IMAGES, write_flags, progress_func);
+       joined_wim->wim_info = NULL;
+       joined_wim->image_metadata = NULL;
+       return ret;
 }
 
+static int
+cmp_swms_by_part_number(const void *swm1, const void *swm2)
+{
+       u16 partno_1 = (*(const WIMStruct**)swm1)->hdr.part_number;
+       u16 partno_2 = (*(const WIMStruct**)swm2)->hdr.part_number;
+       return (int)partno_1 - (int)partno_2;
+}
 
-WIMLIBAPI int wimlib_join(const char **swm_names, unsigned num_swms, 
-                         const char *output_path, int flags)
+/*
+ * Join a set of split WIMs into a stand-alone WIM.
+ */
+WIMLIBAPI int
+wimlib_join(const tchar * const *swm_names,
+           unsigned num_swms,
+           const tchar *output_path,
+           int swm_open_flags,
+           int wim_write_flags,
+           wimlib_progress_func_t progress_func)
 {
-       int i;
        int ret;
-       int part_idx;
-       int write_flags = 0;
        WIMStruct *joined_wim = NULL;
-       WIMStruct *swms[num_swms];
+       unsigned i;
 
-       int ctype;
-       u8 *guid;
+       swm_open_flags |= WIMLIB_OPEN_FLAG_SPLIT_OK;
+       wim_write_flags &= WIMLIB_WRITE_MASK_PUBLIC;
 
-       if (num_swms < 1)
+       if (num_swms < 1 || num_swms > 0xffff)
                return WIMLIB_ERR_INVALID_PARAM;
 
+       WIMStruct *swms[num_swms];
        ZERO_ARRAY(swms);
 
        for (i = 0; i < num_swms; i++) {
-               ret = wimlib_open_wim(swm_names[i], 
-                                     flags | WIMLIB_OPEN_FLAG_SPLIT_OK, &swms[i]);
-               if (ret != 0)
-                       goto out;
+               ret = wimlib_open_wim(swm_names[i], swm_open_flags, &swms[i],
+                                     progress_func);
+               if (ret)
+                       goto out_free_wims;
 
-               /* don't open all the parts at the same time, in case there are
+               /* Don't open all the parts at the same time, in case there are
                 * a lot of them */
-               fclose(swms[i]->fp);
-               swms[i]->fp = NULL;
+               close_wim(swms[i]);
        }
 
        qsort(swms, num_swms, sizeof(swms[0]), cmp_swms_by_part_number);
 
        ret = verify_swm_set(swms[0], &swms[1], num_swms - 1);
-       if (ret != 0)
-               goto out;
-
-       joined_wim = new_wim_struct();
-       if (!joined_wim) {
-               ret = WIMLIB_ERR_NOMEM;
-               goto out;
-       }
-
-       if (flags & WIMLIB_OPEN_FLAG_CHECK_INTEGRITY)
-               write_flags |= WIMLIB_WRITE_FLAG_CHECK_INTEGRITY;
-       if (flags & WIMLIB_OPEN_FLAG_SHOW_PROGRESS)
-               write_flags |= WIMLIB_WRITE_FLAG_SHOW_PROGRESS;
-
-       ret = begin_write(joined_wim, output_path, write_flags);
-       if (ret != 0)
-               goto out;
-       ret = join_wims(swms, num_swms, joined_wim, write_flags);
-out:
-       for (i = 0; i < num_swms; i++) {
-               /* out_fp is the same in all the swms and joined_wim; only close
-                * it one time, when freeing joined_wim. */
-               if (swms[i]) {
-                       swms[i]->out_fp = NULL;
-                       wimlib_free(swms[i]);
-               }
-       }
+       if (ret)
+               goto out_free_wims;
+
+       ret = wimlib_create_new_wim(swms[0]->compression_type, &joined_wim);
+       if (ret)
+               goto out_free_wims;
+
+       ret = begin_write(joined_wim, output_path, wim_write_flags);
+       if (ret)
+               goto out_free_wims;
+       ret = join_wims(swms, num_swms, joined_wim, wim_write_flags,
+                       progress_func);
+out_free_wims:
+       for (i = 0; i < num_swms; i++)
+               wimlib_free(swms[i]);
        wimlib_free(joined_wim);
        return ret;
 }