X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fjoin.c;h=8a0abe5d663e912c0a28d6058c820cd368eebedc;hp=07cbfa971c92622b08eb72df6b39e9d26e18524a;hb=f303b46312f8d8be4210fba66082d5a7572dbd70;hpb=512d3f87a1e7b59ca19ae6d6965dbcf7f4a17c15

diff --git a/src/join.c b/src/join.c
index 07cbfa97..8a0abe5d 100644
--- a/src/join.c
+++ b/src/join.c
@@ -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.
  *
@@ -23,62 +23,60 @@
  * 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 = MALLOC(sizeof(struct lookup_table_entry));
-	if (!copy)
-		return WIMLIB_ERR_NOMEM;
-	memcpy(copy, lte, sizeof(struct lookup_table_entry));
-	INIT_LIST_HEAD(&copy->lte_group_list);
-	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);
-}
-
-
-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;
-}
+#include "wimlib.h"
+#include "wimlib/error.h"
+#include "wimlib/types.h"
+#include "wimlib/util.h"
+#include "wimlib/wim.h"
 
-int verify_swm_set(WIMStruct *w, WIMStruct **additional_swms,
-		   unsigned num_additional_swms)
+/*
+ * verify_swm_set: - Sanity checks to make sure a set of WIMs correctly
+ *		     correspond to a spanned set.
+ *
+ * @wim:
+ *	Part 1 of the set.
+ *
+ * @additional_swms:
+ *	All parts of the set other than part 1.
+ *
+ * @num_additional_swms:
+ *	Number of WIMStructs in @additional_swms.  Or, the total number of parts
+ *	in the set minus 1.
+ *
+ * @return:
+ *	0 on success; WIMLIB_ERR_SPLIT_INVALID if the set is not valid.
+ */
+static int
+verify_swm_set(WIMStruct *wim, WIMStruct **additional_swms,
+	       unsigned num_additional_swms)
 {
-	unsigned total_parts = w->hdr.total_parts;
+	unsigned total_parts = wim->hdr.total_parts;
 	int ctype;
+	u32 chunk_size;
 	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 " : "",
+		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) ? "s were" : " was");
+		      (num_additional_swms) ? T("s were") : T(" 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);
+	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;
 	}
 	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", 
+			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);
@@ -86,225 +84,137 @@ int verify_swm_set(WIMStruct *w, WIMStruct **additional_swms,
 		}
 	}
 
-	/* 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;
+	/* 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;
+			}
 		}
 	}
 	return 0;
 }
 
-int new_joined_lookup_table(WIMStruct *w,
-			    WIMStruct **additional_swms,
-			    unsigned num_additional_swms,
-			    struct lookup_table **table_ret)
+/* 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)
 {
-	struct lookup_table *table;
 	int ret;
 	unsigned i;
+	unsigned j;
+	WIMStruct *swm0;
+	WIMStruct **additional_swms;
+	unsigned num_additional_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;
-	}
-	*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;
-		}
-	}
-	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)
-			return ret;
-	}
-
-	off_t lookup_table_offset = ftello(out_fp);
-
-	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)
-			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;
-	}
-	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;
-
-
-	/* 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);
-}
-
-
-WIMLIBAPI int wimlib_join(const char **swm_names, unsigned num_swms, 
-			  const char *output_path, int flags)
-{
-	int i;
-	int ret;
-	int part_idx;
-	int write_flags = 0;
-	WIMStruct *joined_wim = NULL;
-	WIMStruct *swms[num_swms];
-
-	int ctype;
-	u8 *guid;
-
-	if (num_swms < 1)
+	if (num_swms < 1 || num_swms > 0xffff)
 		return WIMLIB_ERR_INVALID_PARAM;
+	num_additional_swms = num_swms - 1;
 
-	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;
+	additional_swms = CALLOC((num_additional_swms + 1),
+				 sizeof(additional_swms[0]));
+	if (!additional_swms)
+		return WIMLIB_ERR_NOMEM;
 
-		/* 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;
+	swm0 = NULL;
+	for (i = 0, j = 0; i < num_swms; i++) {
+		WIMStruct *swm;
+
+		ret = wimlib_open_wim(swm_names[i], swm_open_flags, &swm,
+				      progress_func);
+		if (ret)
+			goto out_free_swms;
+		if (swm->hdr.part_number == 1 && swm0 == NULL)
+			swm0 = swm;
+		else
+			additional_swms[j++] = swm;
 	}
 
-	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 (!swm0) {
+		ERROR("Part 1 of the split WIM was not specified!");
+		ret = WIMLIB_ERR_SPLIT_INVALID;
+		goto out_free_swms;
 	}
 
-	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]);
-		}
-	}
-	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 + 1; i++)
+		wimlib_free(additional_swms[i]);
+	FREE(additional_swms);
+	wimlib_free(swm0);
 	return ret;
 }