X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fsplit.c;h=2d855a6f0e10df18514f4cb3dd13367dd37782e0;hp=f9811d97dd69ee57fc879e7d4d93ce724d9b1116;hb=48fbbd68e82d879e2cb6f384c1440e81373338c7;hpb=60b8f54df8fed44136bdc8ec615ee62703d87b69

diff --git a/src/split.c b/src/split.c
index f9811d97..2d855a6f 100644
--- a/src/split.c
+++ b/src/split.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.
  *
@@ -23,243 +23,208 @@
  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  */
 
-#include "wimlib_internal.h"
-#include "lookup_table.h"
-#include "xml.h"
-#include "io.h"
-
-struct args {
-	WIMStruct *w;
-	char *swm_base_name;
-	size_t swm_base_name_len;
-	const char *swm_suffix;
-	struct lookup_table_entry *lte_chain_head;
-	struct lookup_table_entry *lte_chain_tail;
-	int    part_number;
-	int    write_flags;
-	long   size_remaining;
-	size_t part_size;
-	u64    total_bytes;
-	u64    total_bytes_written;
+#ifdef HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#include "wimlib.h"
+#include "wimlib/error.h"
+#include "wimlib/list.h"
+#include "wimlib/lookup_table.h"
+#include "wimlib/metadata.h"
+#include "wimlib/resource.h"
+#include "wimlib/wim.h"
+#include "wimlib/write.h"
+
+#ifdef HAVE_ALLOCA_H
+#  include <alloca.h>
+#else
+#  include <stdlib.h>
+#endif
+
+struct swm_part_info {
+	struct list_head stream_list;
+	u64 size;
 };
 
-static int finish_swm(WIMStruct *w, struct lookup_table_entry *lte_chain_head,
-		      int write_flags)
+static void
+copy_part_info(struct swm_part_info *dst, struct swm_part_info *src)
 {
-	off_t lookup_table_offset = ftello(w->out_fp);
-	int ret;
-
-	DEBUG("Writing lookup table for SWM (offset %"PRIu64")", 
-			lookup_table_offset);
-
-	while (lte_chain_head != NULL) {
-		print_lookup_table_entry(lte_chain_head);
-
-		ret = write_lookup_table_entry(lte_chain_head, w->out_fp);
-		if (ret != 0)
-			return ret;
-		struct lookup_table_entry *prev = lte_chain_head;
-		lte_chain_head = lte_chain_head->next_lte_in_swm;
-		prev->next_lte_in_swm = NULL;
-	}
-	off_t xml_data_offset = ftello(w->out_fp);
-
-	if (lookup_table_offset == -1 || xml_data_offset == -1)
-		return WIMLIB_ERR_WRITE;
-	w->hdr.lookup_table_res_entry.offset = lookup_table_offset;
-	w->hdr.lookup_table_res_entry.size = 
-				xml_data_offset - lookup_table_offset;
-	ret = finish_write(w, WIM_ALL_IMAGES, write_flags, 0);
-	if (ret != 0)
-		return ret;
-
-	ret = fclose(w->out_fp);
-	if (ret != 0)
-		ret = WIMLIB_ERR_WRITE;
-	w->out_fp = NULL;
-	return ret;
+	list_transfer(&src->stream_list, &dst->stream_list);
+	dst->size = src->size;
 }
 
-static int copy_resource_to_swm(struct lookup_table_entry *lte, void *__args)
-{
-	struct args *args = (struct args*)__args;
-	WIMStruct *w = args->w;
-	int ret;
-
-	/* metadata resources were already written. */
-	if (lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
-		return 0;
+struct swm_info {
+	struct swm_part_info *parts;
+	unsigned num_parts;
+	unsigned num_alloc_parts;
+	u64 total_bytes;
+	u64 max_part_size;
+};
 
-	if (args->size_remaining < 0 || 
-			(u64)args->size_remaining < lte->resource_entry.size) {
+static int
+write_split_wim(WIMStruct *orig_wim, const tchar *swm_name,
+		struct swm_info *swm_info, int write_flags,
+		wimlib_progress_func_t progress_func)
+{
+	size_t swm_name_len;
+	tchar *swm_name_buf;
+	const tchar *dot;
+	tchar *swm_suffix;
+	size_t swm_base_name_len;
 
-		/* No space for this resource.  Finish the previous swm and
-		 * start a new one. */
+	union wimlib_progress_info progress;
+	unsigned part_number;
+	int ret;
+	u8 guid[WIMLIB_GUID_LEN];
+
+	swm_name_len = tstrlen(swm_name);
+	swm_name_buf = alloca((swm_name_len + 20) * sizeof(tchar));
+	tstrcpy(swm_name_buf, swm_name);
+	dot = tstrchr(swm_name_buf, T('.'));
+	if (dot) {
+		swm_base_name_len = dot - swm_name_buf;
+		swm_suffix = alloca((tstrlen(dot) + 1) * sizeof(tchar));
+		tstrcpy(swm_suffix, dot);
+	} else {
+		swm_base_name_len = swm_name_len;
+		swm_suffix = alloca(1 * sizeof(tchar));
+		swm_suffix[0] = T('\0');
+	}
 
-		ret = finish_swm(w, args->lte_chain_head, args->write_flags);
+	progress.split.completed_bytes = 0;
+	progress.split.total_bytes = 0;
+	for (part_number = 1; part_number <= swm_info->num_parts; part_number++)
+		progress.split.total_bytes += swm_info->parts[part_number - 1].size;
+	progress.split.total_parts = swm_info->num_parts;
+	progress.split.part_name = swm_name_buf;
 
-		args->lte_chain_tail = NULL;
-		args->lte_chain_head = NULL;
+	randomize_byte_array(guid, WIMLIB_GUID_LEN);
 
-		sprintf(args->swm_base_name + args->swm_base_name_len, "%d", 
-			++args->part_number);
-		strcat(args->swm_base_name, args->swm_suffix);
+	for (part_number = 1; part_number <= swm_info->num_parts; part_number++) {
+		int part_write_flags;
 
-		w->hdr.part_number = args->part_number;
+		if (part_number != 1) {
+			tsprintf(swm_name_buf + swm_base_name_len,
+				 T("%u%"TS), part_number, swm_suffix);
+		}
 
-		if (args->write_flags & WIMLIB_OPEN_FLAG_SHOW_PROGRESS)
-			printf("Writing `%s' (%"PRIu64" of %"PRIu64" bytes, "
-			       "%.0f%% done)\n", 
-			       args->swm_base_name, 
-			       args->total_bytes_written,
-			       args->total_bytes,
-			       (double)args->total_bytes_written /
-			           (double)args->total_bytes * 100.0);
+		progress.split.cur_part_number = part_number;
+		if (progress_func) {
+			progress_func(WIMLIB_PROGRESS_MSG_SPLIT_BEGIN_PART,
+				      &progress);
+		}
 
-		ret = begin_write(w, args->swm_base_name, args->write_flags);
-		if (ret != 0)
+		part_write_flags = write_flags & WIMLIB_WRITE_MASK_PUBLIC;
+		part_write_flags |= WIMLIB_WRITE_FLAG_USE_EXISTING_TOTALBYTES;
+		if (part_number != 1)
+			part_write_flags |= WIMLIB_WRITE_FLAG_NO_METADATA;
+
+		ret = write_wim_part(orig_wim,
+				     swm_name_buf,
+				     WIMLIB_ALL_IMAGES,
+				     part_write_flags,
+				     1,
+				     NULL,
+				     part_number,
+				     swm_info->num_parts,
+				     &swm_info->parts[part_number - 1].stream_list,
+				     guid);
+		if (ret)
 			return ret;
-		args->size_remaining = args->part_size;
+
+		progress.split.completed_bytes += swm_info->parts[part_number - 1].size;
+		if (progress_func) {
+			progress_func(WIMLIB_PROGRESS_MSG_SPLIT_END_PART,
+				      &progress);
+		}
 	}
-	args->size_remaining -= lte->resource_entry.size;
-	args->total_bytes_written += lte->resource_entry.size;
-	if (args->lte_chain_tail)
-		args->lte_chain_tail->next_lte_in_swm = lte;
-	else
-		args->lte_chain_head = lte;
-	args->lte_chain_tail = lte;
-	return copy_resource(lte, w);
+	return 0;
 }
 
-/* Splits the WIM file @wimfile into multiple parts prefixed by @swm_name with
- * size at most @part_size. */
-WIMLIBAPI int wimlib_split(const char *wimfile, const char *swm_name, 
-			   size_t part_size, int flags)
+static int
+add_stream_to_swm(struct wim_lookup_table_entry *lte, void *_swm_info)
 {
-	int ret;
-	WIMStruct *w;
-	int write_flags = 0;
-	size_t swm_name_len = strlen(swm_name);
-	size_t swm_base_name_len;
-	char name[swm_name_len + 20];
-	char *swm_suffix;
-
-	struct lookup_table_entry *lte_chain_head = NULL;
-	struct lookup_table_entry *lte_chain_tail = NULL;
-	long size_remaining = part_size;
-	u64 total_bytes_written = 0;
-	u64 total_bytes;
-
-	ret = wimlib_open_wim(wimfile, flags, &w);
-	if (ret != 0)
-		return ret;
-
-	total_bytes = wim_info_get_total_bytes(w->wim_info);
-
-	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;
-
-	w->hdr.flags |= WIM_HDR_FLAG_SPANNED;
-	w->hdr.boot_idx = 0;
-	randomize_byte_array(w->hdr.guid, WIM_GID_LEN);
-	ret = begin_write(w, swm_name, write_flags);
-	if (ret != 0)
-		return ret;
-
-	swm_suffix = strchr(swm_name, '.');
-	memcpy(name, swm_name, swm_name_len + 1);
-	if (swm_suffix) {
-		swm_base_name_len = swm_suffix - swm_name;
-	} else {
-		swm_base_name_len = swm_name_len;
-		name[sizeof(name) - 1] = '\0';
-		swm_suffix = &name[sizeof(name) - 1];
+	struct swm_info *swm_info = _swm_info;
+	u64 stream_size;
+
+	stream_size = lte->resource_entry.size;
+
+	/* - Start first part if no parts have been started so far;
+	 * - Start next part if adding this stream exceeds maximum part size,
+	 *   UNLESS the stream is metadata or if no streams at all have been
+	 *   added to the current part.
+	 */
+	if (swm_info->num_parts == 0 ||
+	    ((swm_info->parts[swm_info->num_parts - 1].size +
+			stream_size >= swm_info->max_part_size)
+	     && !((lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) ||
+		   swm_info->parts[swm_info->num_parts - 1].size == 0)))
+	{
+		if (swm_info->num_parts == swm_info->num_alloc_parts) {
+			struct swm_part_info *parts;
+			size_t num_alloc_parts = swm_info->num_alloc_parts;
+
+			num_alloc_parts += 8;
+			parts = MALLOC(num_alloc_parts * sizeof(parts[0]));
+			if (!parts)
+				return WIMLIB_ERR_NOMEM;
+
+			for (unsigned i = 0; i < swm_info->num_parts; i++)
+				copy_part_info(&parts[i], &swm_info->parts[i]);
+
+			FREE(swm_info->parts);
+			swm_info->parts = parts;
+			swm_info->num_alloc_parts = num_alloc_parts;
+		}
+		swm_info->num_parts++;
+		INIT_LIST_HEAD(&swm_info->parts[swm_info->num_parts - 1].stream_list);
+		swm_info->parts[swm_info->num_parts - 1].size = 0;
 	}
-
-	if (write_flags & WIMLIB_OPEN_FLAG_SHOW_PROGRESS)
-		printf("Writing `%s' (%.2f %% done)\n", 
-			swm_name, 
-			(double)total_bytes_written /
-				(double)total_bytes * 100.0);
-
-	w->write_metadata = true;
-	for (int i = 0; i < w->hdr.image_count; i++) {
-		struct lookup_table_entry *metadata_lte;
-
-		DEBUG("Writing metadata resource %d", i);
-
-		metadata_lte = w->image_metadata[i].metadata_lte;
-		ret = copy_resource(metadata_lte, w);
-		if (ret != 0)
-			return ret;
-		size_remaining -= metadata_lte->resource_entry.size;
-		total_bytes_written += metadata_lte->resource_entry.size;
-		if (lte_chain_tail)
-			lte_chain_tail->next_lte_in_swm = metadata_lte;
-		else
-			lte_chain_head = metadata_lte;
-		lte_chain_tail = metadata_lte;
+	swm_info->parts[swm_info->num_parts - 1].size += stream_size;
+	if (!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)) {
+		list_add_tail(&lte->write_streams_list,
+			      &swm_info->parts[swm_info->num_parts - 1].stream_list);
 	}
-	w->write_metadata = false;
+	swm_info->total_bytes += stream_size;
+	return 0;
+}
 
-	struct args args = {
-		.w                 = w,
-		.swm_base_name     = name,
-		.swm_base_name_len = swm_base_name_len,
-		.swm_suffix        = swm_suffix,
-		.lte_chain_head    = lte_chain_head,
-		.lte_chain_tail    = lte_chain_tail,
-		.part_number       = 1,
-		.write_flags       = write_flags,
-		.size_remaining    = size_remaining,
-		.part_size         = part_size,
-		.total_bytes        = total_bytes,
-		.total_bytes_written = total_bytes_written,
-	};
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_split(WIMStruct *wim, const tchar *swm_name,
+	     u64 part_size, int write_flags,
+	     wimlib_progress_func_t progress_func)
+{
+	struct swm_info swm_info;
+	unsigned i;
+	int ret;
 
-	ret = for_lookup_table_entry(w->lookup_table, copy_resource_to_swm, &args);
-	if (ret != 0)
-		return ret;
+	if (swm_name == NULL || swm_name[0] == T('\0') || part_size == 0)
+		return WIMLIB_ERR_INVALID_PARAM;
 
-	ret = finish_swm(w, args.lte_chain_head, write_flags);
-	if (ret != 0)
-		return ret;
+	if (!wim_has_metadata(wim))
+		return WIMLIB_ERR_INVALID_PARAM;
 
+	memset(&swm_info, 0, sizeof(swm_info));
+	swm_info.max_part_size = part_size;
 
-	/* The swms are all ready now, except the total_parts and part_number
-	 * fields in their headers are wrong (we don't know the total parts
-	 * until they are all written).  Fix them. */
-	int total_parts = args.part_number;
-	for (int i = 1; i <= total_parts; i++) {
-		const char *p;
-		if (i == 1) {
-			p = swm_name;
-		} else {
-			sprintf(name + swm_base_name_len, "%d", i);
-			p = strcat(name, swm_suffix);
-		}
+	for (i = 0; i < wim->hdr.image_count; i++) {
+		ret = add_stream_to_swm(wim->image_metadata[i]->metadata_lte,
+					&swm_info);
+		if (ret)
+			goto out_free_swm_info;
+	}
 
-		FILE *fp = fopen(p, "r+b");
-		if (!fp) {
-			ERROR_WITH_ERRNO("Failed to open `%s'", p);
-			return WIMLIB_ERR_OPEN;
-		}
-		u8 buf[4];
-		put_u16(&buf[0], i);
-		put_u16(&buf[2], total_parts);
+	ret = for_lookup_table_entry_pos_sorted(wim->lookup_table,
+						add_stream_to_swm,
+						&swm_info);
+	if (ret)
+		goto out_free_swm_info;
 
-		if (fseek(fp, 40, SEEK_SET) != 0 || 
-				fwrite(buf, 1, sizeof(buf), fp) != sizeof(buf)
-				|| fclose(fp) != 0) {
-			ERROR_WITH_ERRNO("Error overwriting header of `%s'",
-					 name);
-			return WIMLIB_ERR_WRITE;
-		}
-	}
-	if (write_flags & WIMLIB_OPEN_FLAG_SHOW_PROGRESS)
-		puts("Done!");
-	wimlib_free(w);
-	return 0;
+	ret = write_split_wim(wim, swm_name, &swm_info, write_flags,
+			      progress_func);
+out_free_swm_info:
+	FREE(swm_info.parts);
+	return ret;
 }