X-Git-Url: https://mattmccutchen.net/rsync/rsync.git/blobdiff_plain/b5c6a6aeeb7d526e023454ea681b7381945bced8..a1cc199b34b4bc65cba2dce0cbb04c18f17271ac:/flist.c

diff --git a/flist.c b/flist.c
index 763d826f..06b75ba9 100644
--- a/flist.c
+++ b/flist.c
@@ -53,6 +53,7 @@ extern int preserve_uid;
 extern int preserve_gid;
 extern int relative_paths;
 extern int implied_dirs;
+extern int prune_empty_dirs;
 extern int copy_links;
 extern int copy_unsafe_links;
 extern int protocol_version;
@@ -75,6 +76,7 @@ unsigned int file_struct_len;
 
 static char empty_sum[MD4_SUM_LENGTH];
 static int flist_count_offset;
+static int max_dir_depth = 0;
 
 static void clean_flist(struct file_list *flist, int strip_root, int no_dups);
 static void output_flist(struct file_list *flist);
@@ -633,6 +635,8 @@ static struct file_struct *receive_file_entry(struct file_list *flist,
 		bp[-1] = '\0';
 		lastdir_depth = count_dir_elements(lastdir);
 		file->dir.depth = lastdir_depth + 1;
+		if (lastdir_depth >= max_dir_depth)
+			max_dir_depth = lastdir_depth + 1;
 	} else if (dirname) {
 		file->dirname = dirname; /* we're reusing lastname */
 		file->dir.depth = lastdir_depth + 1;
@@ -1439,11 +1443,33 @@ int flist_find(struct file_list *flist, struct file_struct *f)
 
 	while (low <= high) {
 		mid = (low + high) / 2;
-		for (mid_up = mid; !flist->files[mid_up]->basename; mid_up++) {}
-		if (mid_up <= high)
-			ret = f_name_cmp(flist->files[mid_up], f);
-		else
-			ret = 1;
+		if (flist->files[mid]->basename)
+			mid_up = mid;
+		else {
+			/* Scan for the next non-empty entry using the cached
+			 * distance values.  If the value isn't fully up-to-
+			 * date, update it. */
+			mid_up = mid + flist->files[mid]->dir.depth;
+			if (!flist->files[mid_up]->basename) {
+				do {
+				    mid_up += flist->files[mid_up]->dir.depth;
+				} while (!flist->files[mid_up]->basename);
+				flist->files[mid]->dir.depth = mid_up - mid;
+			}
+			if (mid_up > high) {
+				/* If there's nothing left above us, set high to
+				 * a non-empty entry below us and continue. */
+				high = mid - flist->files[mid]->length;
+				if (!flist->files[high]->basename) {
+					do {
+					    high -= flist->files[high]->length;
+					} while (!flist->files[high]->basename);
+					flist->files[mid]->length = mid - high;
+				}
+				continue;
+			}
+		}
+		ret = f_name_cmp(flist->files[mid_up], f);
 		if (ret == 0) {
 			if (protocol_version < 29
 			    && S_ISDIR(flist->files[mid_up]->mode)
@@ -1465,9 +1491,15 @@ int flist_find(struct file_list *flist, struct file_struct *f)
  */
 void clear_file(int i, struct file_list *flist)
 {
-	if (flist->hlink_pool && flist->files[i]->link_u.idev)
-		pool_free(flist->hlink_pool, 0, flist->files[i]->link_u.idev);
-	memset(flist->files[i], 0, file_struct_len);
+	struct file_struct *file = flist->files[i];
+	if (flist->hlink_pool && file->link_u.idev)
+		pool_free(flist->hlink_pool, 0, file->link_u.idev);
+	memset(file, 0, file_struct_len);
+	/* In an empty entry, dir.depth is an offset to the next non-empty
+	 * entry.  Likewise for length in the opposite direction.  We assume
+	 * that we're alone for now since flist_find() will collate adjacent
+	 * items for any entries that are encountered during the find. */
+	file->length = file->dir.depth = 1;
 }
 
 /*
@@ -1515,6 +1547,7 @@ void flist_free(struct file_list *flist)
  */
 static void clean_flist(struct file_list *flist, int strip_root, int no_dups)
 {
+	char fbuf[MAXPATHLEN];
 	int i, prev_i = 0;
 
 	if (!flist)
@@ -1568,10 +1601,10 @@ static void clean_flist(struct file_list *flist, int strip_root, int no_dups)
 			if (verbose > 1 && !am_server) {
 				rprintf(FINFO,
 					"removing duplicate name %s from file list (%d)\n",
-					f_name(file, NULL), drop);
+					f_name(file, fbuf), drop);
 			}
-			/* Make sure that if we unduplicate '.', that we don't
-			 * lose track of a user-specified top directory. */
+			/* Make sure we don't lose track of a user-specified
+			 * top directory. */
 			flist->files[keep]->flags |= flist->files[drop]->flags
 						   & (FLAG_TOP_DIR|FLAG_DEL_HERE);
 
@@ -1599,16 +1632,58 @@ static void clean_flist(struct file_list *flist, int strip_root, int no_dups)
 
 			if (!file->dirname)
 				continue;
-			if (*file->dirname == '/') {
-				char *s = file->dirname + 1;
-				while (*s == '/') s++;
-				memmove(file->dirname, s, strlen(s) + 1);
-			}
-
+			while (*file->dirname == '/')
+				file->dirname++;
 			if (!*file->dirname)
 				file->dirname = NULL;
 		}
 	}
+
+	if (prune_empty_dirs && no_dups && max_dir_depth) {
+		int j, cur_depth = 0;
+		int *maybe_dirs = new_array(int, max_dir_depth);
+
+		maybe_dirs[0] = -1;
+
+		for (i = flist->low; i <= flist->high; i++) {
+			struct file_struct *file = flist->files[i];
+
+			if (S_ISDIR(file->mode) && file->dir.depth) {
+				j = cur_depth;
+				cur_depth = file->dir.depth - 1;
+				for ( ; j >= cur_depth; j--) {
+					if (maybe_dirs[j] < 0)
+						continue;
+					clear_file(maybe_dirs[j], flist);
+				}
+				if (is_excluded(f_name(file, fbuf), 1,
+						       ALL_FILTERS)) {
+					for (j = 0; j <= cur_depth; j++)
+						maybe_dirs[j] = -1;
+				} else
+					maybe_dirs[cur_depth] = i;
+			} else if (maybe_dirs[cur_depth] >= 0) {
+				for (j = 0; j <= cur_depth; j++)
+					maybe_dirs[j] = -1;
+			}
+		}
+		for (j = cur_depth; j >= 0; j--) {
+			if (maybe_dirs[j] < 0)
+				continue;
+			clear_file(maybe_dirs[j], flist);
+		}
+
+		for (i = flist->low; i <= flist->high; i++) {
+			if (flist->files[i]->basename)
+				break;
+		}
+		flist->low = i;
+		for (i = flist->high; i >= flist->low; i--) {
+			if (flist->files[i]->basename)
+				break;
+		}
+		flist->high = i;
+	}
 }
 
 static void output_flist(struct file_list *flist)