X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fjoin.c;h=b2125d13d5e03f5911ddbcc2b8ce052a632f5077;hp=08b360bad3dd704fbfb40b1e0cd7451086dc5cd8;hb=f02f8a0274377459bf90d5a70cb7b72480a3df60;hpb=10a87017a0a82d34ed3981e1f5e586b5b8613e3f

diff --git a/src/join.c b/src/join.c
index 08b360ba..b2125d13 100644
--- a/src/join.c
+++ b/src/join.c
@@ -5,7 +5,7 @@
  */
 
 /*
- * Copyright (C) 2012, 2013 Biggers
+ * Copyright (C) 2012, 2013 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
@@ -23,192 +23,196 @@
  * 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 move_lte_to_table(struct wim_lookup_table_entry *lte,
-			     void *other_tab)
-{
-	hlist_del(&lte->hash_list);
-	lookup_table_insert((struct wim_lookup_table*)other_tab, lte);
-	return 0;
-}
-
-static int lookup_table_join(struct wim_lookup_table *table,
-			     struct wim_lookup_table *new)
-{
-	return for_lookup_table_entry(new, move_lte_to_table, table);
-}
+#include "wimlib.h"
+#include "wimlib/error.h"
+#include "wimlib/types.h"
+#include "wimlib/util.h"
+#include "wimlib/wim.h"
 
 /*
- * new_joined_lookup_table: - Join lookup tables from the parts of a split WIM.
+ * verify_swm_set: - Sanity checks to make sure a set of WIMs correctly
+ *		     correspond to a spanned set.
  *
- * @w specifies the first part, while @additional_swms and @num_additional_swms
- * specify an array of pointers to the WIMStruct's for additional split WIM parts.
+ * @wim:
+ *	Part 1 of the set.
  *
- * The lookup table entries are *moved* to the new table.
+ * @additional_swms:
+ *	All parts of the set other than part 1.
  *
- * On success, 0 is returned on a pointer to the joined lookup table is returned
- * in @table_ret.
+ * @num_additional_swms:
+ *	Number of WIMStructs in @additional_swms.  Or, the total number of parts
+ *	in the set minus 1.
  *
- * The reason we join the lookup tables is so we only have to search one lookup
- * table to find the location of a resource in the entire WIM.
+ * @return:
+ *	0 on success; WIMLIB_ERR_SPLIT_INVALID if the set is not valid.
  */
-int new_joined_lookup_table(WIMStruct *w,
-			    WIMStruct **additional_swms,
-			    unsigned num_additional_swms,
-			    struct wim_lookup_table **table_ret)
+static int
+verify_swm_set(WIMStruct *wim, WIMStruct **additional_swms,
+	       unsigned num_additional_swms)
 {
-	struct wim_lookup_table *table;
-	int ret;
-	unsigned i;
-
-	table = new_lookup_table(9001);
-	if (!table)
-		return WIMLIB_ERR_NOMEM;
-
-	if (w)
-		lookup_table_join(table, w->lookup_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;
+	unsigned total_parts = wim->hdr.total_parts;
+	int ctype;
+	u32 chunk_size;
+	const u8 *guid;
+
+	if (total_parts != num_additional_swms + 1) {
+		ERROR("`%"TS"' says there are %u parts in the spanned set, "
+		      "but %"TS"%u part%"TS" provided",
+		      wim->filename, total_parts,
+		      (num_additional_swms + 1 < total_parts) ? T("only ") : T(""),
+		      num_additional_swms + 1,
+		      (num_additional_swms) ? T("s were") : T(" was"));
+		return WIMLIB_ERR_SPLIT_INVALID;
 	}
-	*table_ret = table;
-	return 0;
-out_free_table:
-	free_lookup_table(table);
-	return ret;
-}
-
-
-static int join_wims(WIMStruct **swms, unsigned num_swms,
-		     WIMStruct *joined_wim, int write_flags,
-		     wimlib_progress_func_t progress_func)
-{
-	int ret;
-	unsigned i;
-	union wimlib_progress_info progress;
-	u64 total_bytes = 0;
-	u64 part_bytes;
-	u64 swm_part_sizes[num_swms];
-
-	/* 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;
+	if (wim->hdr.part_number != 1) {
+		ERROR("WIM `%"TS"' is not the first part of the split WIM.",
+		      wim->filename);
+		return WIMLIB_ERR_SPLIT_INVALID;
 	}
-
-	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);
+	for (unsigned i = 0; i < num_additional_swms; i++) {
+		if (additional_swms[i]->hdr.total_parts != total_parts) {
+			ERROR("WIM `%"TS"' 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;
+		}
 	}
 
-	/* Write the resources (streams and metadata resources) from each SWM
-	 * part */
-	swms[0]->write_metadata = true;
-	for (i = 0; i < num_swms; i++) {
-		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 = joined_wim->out_fp;
-		swms[i]->hdr.part_number = 1;
-		ret = for_lookup_table_entry(swms[i]->lookup_table,
-					     copy_resource, swms[i]);
-		swms[i]->out_fp = NULL;
-		fclose(swms[i]->fp);
-		swms[i]->fp = NULL;
-
-		if (ret != 0)
-			return ret;
-
-		if (progress_func) {
-			progress.join.completed_bytes += swm_part_sizes[i];
-			progress.join.completed_parts++;
-			progress_func(WIMLIB_PROGRESS_MSG_JOIN_STREAMS, &progress);
+	/* Keep track of the compression type, chunk size, and GUID to make sure
+	 * they are the same for all the WIMs.  */
+	ctype = wim->compression_type;
+	chunk_size = wim->chunk_size;
+	guid = wim->hdr.guid;
+
+	{
+		/* parts_to_swms is not allocated at function scope because it
+		 * should only be allocated after num_additional_swms was
+		 * checked to be the same as wim->hdr.total_parts.  Otherwise, it
+		 * could be unexpectedly high and cause a stack overflow. */
+		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 (swm->compression_type != ctype) {
+				ERROR("The split WIMs do not all have the same "
+				      "compression type");
+				return WIMLIB_ERR_SPLIT_INVALID;
+			}
+			if (swm->chunk_size != chunk_size &&
+			    ctype != WIMLIB_COMPRESSION_TYPE_NONE) {
+				ERROR("The split WIMs do not all have the same "
+				      "chunk size");
+				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 `%"TS"' and `%"TS"' both are marked "
+				      "as the first WIM in the spanned set",
+				      wim->filename, swm->filename);
+				return WIMLIB_ERR_SPLIT_INVALID;
+			}
+			if (swm->hdr.part_number == 0 ||
+			    swm->hdr.part_number > total_parts)
+			{
+				ERROR("WIM `%"TS"' 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("`%"TS"' and `%"TS"' 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;
+			}
 		}
 	}
-
-	joined_wim->hdr.image_count = swms[0]->hdr.image_count;
-	for (i = 0; i < num_swms; i++)
-		lookup_table_join(joined_wim->lookup_table, swms[i]->lookup_table);
-
-	free_wim_info(joined_wim->wim_info);
-	joined_wim->wim_info = swms[0]->wim_info;
-	ret = finish_write(joined_wim, WIMLIB_ALL_IMAGES, write_flags, progress_func);
-	joined_wim->wim_info = 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;
+	return 0;
 }
 
-/*
- * Join a set of split WIMs into a stand-alone WIM.
- */
-WIMLIBAPI int wimlib_join(const char **swm_names, unsigned num_swms,
-			  const char *output_path, int swm_open_flags,
-			  int wim_write_flags,
-			  wimlib_progress_func_t progress_func)
+/* API function documented in wimlib.h  */
+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 ret;
-	WIMStruct *joined_wim = NULL;
 	unsigned i;
-
-	swm_open_flags |= WIMLIB_OPEN_FLAG_SPLIT_OK;
-	wim_write_flags &= WIMLIB_WRITE_MASK_PUBLIC;
+	unsigned j;
+	WIMStruct *swm0;
+	WIMStruct **additional_swms;
+	unsigned num_additional_swms;
 
 	if (num_swms < 1 || num_swms > 0xffff)
 		return WIMLIB_ERR_INVALID_PARAM;
+	num_additional_swms = num_swms - 1;
+
+	additional_swms = CALLOC(num_additional_swms, sizeof(additional_swms[0]));
+	if (!additional_swms)
+		return WIMLIB_ERR_NOMEM;
 
-	WIMStruct *swms[num_swms];
-	ZERO_ARRAY(swms);
+	swm0 = NULL;
+	for (i = 0, j = 0; i < num_swms; i++) {
+		WIMStruct *swm;
 
-	for (i = 0; i < num_swms; i++) {
-		ret = wimlib_open_wim(swm_names[i], swm_open_flags, &swms[i],
+		ret = wimlib_open_wim(swm_names[i], swm_open_flags, &swm,
 				      progress_func);
-		if (ret != 0)
-			goto out;
+		if (ret)
+			goto out_free_swms;
+		if (swm->hdr.part_number == 1 && swm0 == NULL)
+			swm0 = swm;
+		else
+			additional_swms[j++] = swm;
+	}
 
-		/* 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;
+	if (!swm0) {
+		ret = WIMLIB_ERR_SPLIT_INVALID;
+		goto out_free_swms;
 	}
 
-	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;
-
-	ret = wimlib_create_new_wim(wimlib_get_compression_type(swms[0]),
-				    &joined_wim);
-	if (ret != 0)
-		goto out;
-
-	ret = begin_write(joined_wim, output_path, wim_write_flags);
-	if (ret != 0)
-		goto out;
-	ret = join_wims(swms, num_swms, joined_wim, wim_write_flags,
-			progress_func);
-out:
-	for (i = 0; i < num_swms; i++)
-		wimlib_free(swms[i]);
-	wimlib_free(joined_wim);
+	ret = verify_swm_set(swm0, additional_swms, num_additional_swms);
+	if (ret)
+		goto out_free_swms;
+
+	ret = wimlib_reference_resources(swm0, additional_swms,
+					 num_additional_swms, 0);
+	if (ret)
+		goto out_free_swms;
+
+	swm0->guid_set_explicitly = 1;
+
+	/* It is reasonably safe to provide, WIMLIB_WRITE_FLAG_STREAMS_OK, as we
+	 * have verified that the specified split WIM parts form a spanned set.
+	 */
+	ret = wimlib_write(swm0, output_path, WIMLIB_ALL_IMAGES,
+			   wim_write_flags | WIMLIB_WRITE_FLAG_STREAMS_OK,
+			   1, progress_func);
+out_free_swms:
+	for (i = 0; i < num_additional_swms; i++)
+		wimlib_free(additional_swms[i]);
+	FREE(additional_swms);
+	wimlib_free(swm0);
 	return ret;
 }