Move the pattern matching from gensquashfs to dir_tree_iterator_t

A simple unit test is added that mainly checks for the behavior of
recursing into a sub-tree and only matching the children at the end,
but not reporting the parents that don't match. The behavior is
inteded to immitate the `find` command.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

8 files changed, 146 insertions, 74 deletions

diff --git a/bin/gensquashfs/src/fstree_from_dir.c b/bin/gensquashfs/src/fstree_from_dir.c
index c960430..6709368 100644
--- a/bin/gensquashfs/src/fstree_from_dir.c
+++ b/bin/gensquashfs/src/fstree_from_dir.c
@@ -21,8 +21,7 @@ static sqfs_u32 clamp_timestamp(sqfs_s64 ts)
 	return ts;
 }
 
-int fstree_from_dir(fstree_t *fs, dir_iterator_t *dir,
-		    scan_node_callback cb, void *user)
+int fstree_from_dir(fstree_t *fs, dir_iterator_t *dir)
 {
 	for (;;) {
 		dir_entry_t *ent = NULL;
@@ -40,18 +39,7 @@ int fstree_from_dir(fstree_t *fs, dir_iterator_t *dir,
 
 		n = fstree_get_node_by_path(fs, fs->root, ent->name,
 					    false, true);
-		if (n == NULL)
-			ret = 1;
-
-		if (ret == 0 && cb != NULL)
-			ret = cb(user, ent);
-
-		if (ret < 0) {
-			free(ent);
-			return -1;
-		}
-
-		if (ret > 0) {
+		if (n == NULL) {
 			if (S_ISDIR(ent->mode))
 				dir_tree_iterator_skip(dir);
 			free(ent);
diff --git a/bin/gensquashfs/src/fstree_from_file.c b/bin/gensquashfs/src/fstree_from_file.c
index 4cdfe66..318c129 100644
--- a/bin/gensquashfs/src/fstree_from_file.c
+++ b/bin/gensquashfs/src/fstree_from_file.c
@@ -17,15 +17,6 @@
 #include <errno.h>
 #include <ctype.h>
 
-struct glob_context {
-	unsigned int glob_flags;
-	char *name_pattern;
-};
-
-enum {
-	GLOB_FLAG_PATH = 0x00010000,
-};
-
 static const struct {
 	const char *name;
 	unsigned int clear_flag;
@@ -118,29 +109,6 @@ static int add_hard_link(fstree_t *fs, const char *filename, size_t line_num,
 	return 0;
 }
 
-static int glob_node_callback(void *user, dir_entry_t *ent)
-{
-	struct glob_context *ctx = user;
-	int ret;
-
-	if (ctx->name_pattern != NULL) {
-		if (ctx->glob_flags & GLOB_FLAG_PATH) {
-			ret = fnmatch(ctx->name_pattern,
-				      ent->name, FNM_PATHNAME);
-		} else {
-			const char *name = strrchr(ent->name, '/');
-			name = (name == NULL) ? ent->name : (name + 1);
-
-			ret = fnmatch(ctx->name_pattern, name, 0);
-		}
-
-		if (ret != 0)
-			return 1;
-	}
-
-	return 0;
-}
-
 static size_t name_string_length(const char *str)
 {
 	size_t len = 0;
@@ -182,19 +150,15 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 		      const char *basepath, unsigned int glob_flags,
 		      const char *extra)
 {
+	char *name_pattern = NULL, *prefix = NULL;
 	unsigned int scan_flags = 0, all_flags;
 	dir_iterator_t *dir = NULL;
-	struct glob_context ctx;
 	bool first_clear_flag;
 	size_t i, count, len;
 	dir_tree_cfg_t cfg;
 	tree_node_t *root;
-	char *prefix;
 	int ret;
 
-	memset(&ctx, 0, sizeof(ctx));
-	ctx.glob_flags = glob_flags;
-
 	/* fetch the actual target node */
 	root = fstree_get_node_by_path(fs, fs->root, path, true, false);
 	if (root == NULL) {
@@ -258,14 +222,14 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 
 			len = name_string_length(extra);
 
-			free(ctx.name_pattern);
-			ctx.name_pattern = strndup(extra, len);
+			free(name_pattern);
+			name_pattern = strndup(extra, len);
 			extra += len;
 
 			while (isspace(*extra))
 				++extra;
 
-			quote_remove(ctx.name_pattern);
+			quote_remove(name_pattern);
 			continue;
 		}
 
@@ -275,15 +239,15 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 
 			len = name_string_length(extra);
 
-			free(ctx.name_pattern);
-			ctx.name_pattern = strndup(extra, len);
+			free(name_pattern);
+			name_pattern = strndup(extra, len);
 			extra += len;
 
 			while (isspace(*extra))
 				++extra;
 
-			quote_remove(ctx.name_pattern);
-			ctx.glob_flags |= GLOB_FLAG_PATH;
+			quote_remove(name_pattern);
+			glob_flags |= DIR_SCAN_MATCH_FULL_PATH;
 			continue;
 		}
 
@@ -297,7 +261,7 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 
 			fprintf(stderr, "%s: " PRI_SZ ": unknown option.\n",
 				filename, line_num);
-			free(ctx.name_pattern);
+			free(name_pattern);
 			free(prefix);
 			return -1;
 		} else {
@@ -316,6 +280,7 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 	cfg.def_gid = basic->st_gid;
 	cfg.def_mode = basic->st_mode;
 	cfg.prefix = prefix;
+	cfg.name_pattern = name_pattern;
 
 	if (basepath == NULL) {
 		dir = dir_tree_iterator_create(extra == NULL ? "." : extra,
@@ -341,13 +306,13 @@ static int glob_files(fstree_t *fs, const char *filename, size_t line_num,
 	}
 
 	if (dir != NULL) {
-		ret = fstree_from_dir(fs, dir, glob_node_callback, &ctx);
+		ret = fstree_from_dir(fs, dir);
 		sqfs_drop(dir);
 	} else {
 		ret = -1;
 	}
 
-	free(ctx.name_pattern);
+	free(name_pattern);
 	free(prefix);
 	return ret;
 }
diff --git a/bin/gensquashfs/src/mkfs.c b/bin/gensquashfs/src/mkfs.c
index 8ad394b..b64a217 100644
--- a/bin/gensquashfs/src/mkfs.c
+++ b/bin/gensquashfs/src/mkfs.c
@@ -179,7 +179,7 @@ int main(int argc, char **argv)
 		if (dir == NULL)
 			goto out;
 
-		ret = fstree_from_dir(&sqfs.fs, dir, NULL, NULL);
+		ret = fstree_from_dir(&sqfs.fs, dir);
 		sqfs_drop(dir);
 		if (ret != 0)
 			goto out;
diff --git a/bin/gensquashfs/src/mkfs.h b/bin/gensquashfs/src/mkfs.h
index 3157694..81ad038 100644
--- a/bin/gensquashfs/src/mkfs.h
+++ b/bin/gensquashfs/src/mkfs.h
@@ -42,15 +42,6 @@
 #include <errno.h>
 #include <ctype.h>
 
-/*
-  Optionally used by fstree_from_dir to execute custom actions for
-  each discovered node.
-
-  If it returns a value > 0, the new node is discarded, if < 0, scanning is
-  aborted and returns a failure status.
- */
-typedef int (*scan_node_callback)(void *user, dir_entry_t *ent);
-
 typedef struct {
 	sqfs_writer_cfg_t cfg;
 	unsigned int dirscan_flags;
@@ -129,8 +120,7 @@ int fstree_from_file_stream(fstree_t *fs, istream_t *file,
 
   Returns 0 on success, prints to stderr on failure.
  */
-int fstree_from_dir(fstree_t *fs, dir_iterator_t *dir,
-		    scan_node_callback cb, void *user);
+int fstree_from_dir(fstree_t *fs, dir_iterator_t *dir);
 
 int fstree_sort_files(fstree_t *fs, istream_t *sortfile);
 
diff --git a/include/util/dir_tree_iterator.h b/include/util/dir_tree_iterator.h
index b5ebe90..204f609 100644
--- a/include/util/dir_tree_iterator.h
+++ b/include/util/dir_tree_iterator.h
@@ -25,6 +25,7 @@ enum {
 
 	DIR_SCAN_ONE_FILESYSTEM = 0x1000,
 	DIR_SCAN_NO_RECURSION = 0x2000,
+	DIR_SCAN_MATCH_FULL_PATH = 0x4000,
 };
 
 typedef struct {
@@ -34,6 +35,7 @@ typedef struct {
 	sqfs_u32 def_mode;
 	sqfs_s64 def_mtime;
 	const char *prefix;
+	const char *name_pattern;
 } dir_tree_cfg_t;
 
 dir_iterator_t *dir_tree_iterator_create(const char *path,
diff --git a/lib/util/Makemodule.am b/lib/util/Makemodule.am
index 37eac93..15c5b88 100644
--- a/lib/util/Makemodule.am
+++ b/lib/util/Makemodule.am
@@ -96,11 +96,16 @@ test_dir_tree_iterator2_LDADD = libutil.a libcompat.a
 test_dir_tree_iterator2_CPPFLAGS = $(AM_CPPFLAGS)
 test_dir_tree_iterator2_CPPFLAGS += -DTESTPATH=$(top_srcdir)/lib/util/test/testdir
 
+test_dir_tree_iterator3_SOURCES = lib/util/test/dir_tree_iterator3.c
+test_dir_tree_iterator3_LDADD = libutil.a libcompat.a
+test_dir_tree_iterator3_CPPFLAGS = $(AM_CPPFLAGS)
+test_dir_tree_iterator3_CPPFLAGS += -DTESTPATH=$(top_srcdir)/lib/util/test/testdir
+
 LIBUTIL_TESTS = \
 	test_str_table test_rbtree test_xxhash test_threadpool test_ismemzero \
 	test_canonicalize_name test_filename_sane test_filename_sane_w32 \
 	test_sdate_epoch test_hex_decode test_base64_decode test_dir_iterator \
-	test_dir_tree_iterator test_dir_tree_iterator2
+	test_dir_tree_iterator test_dir_tree_iterator2 test_dir_tree_iterator3
 
 check_PROGRAMS += $(LIBUTIL_TESTS)
 TESTS += $(LIBUTIL_TESTS)
diff --git a/lib/util/src/dir_tree_iterator.c b/lib/util/src/dir_tree_iterator.c
index 55fbabb..a63209b 100644
--- a/lib/util/src/dir_tree_iterator.c
+++ b/lib/util/src/dir_tree_iterator.c
@@ -216,6 +216,23 @@ retry:
 		}
 	}
 
+	if (it->cfg.name_pattern != NULL) {
+		if (it->cfg.flags & DIR_SCAN_MATCH_FULL_PATH) {
+			ret = fnmatch(it->cfg.name_pattern,
+				      ent->name, FNM_PATHNAME);
+		} else {
+			const char *name = strrchr(ent->name, '/');
+			name = (name == NULL) ? ent->name : (name + 1);
+
+			ret = fnmatch(it->cfg.name_pattern, name, 0);
+		}
+
+		if (ret != 0) {
+			free(ent);
+			goto retry;
+		}
+	}
+
 	*out = ent;
 	return it->state;
 fail:
diff --git a/lib/util/test/dir_tree_iterator3.c b/lib/util/test/dir_tree_iterator3.c
new file mode 100644
index 0000000..7b14190
--- /dev/null
+++ b/lib/util/test/dir_tree_iterator3.c
@@ -0,0 +1,105 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * dir_tree_iterator2.c
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+
+#include "util/dir_tree_iterator.h"
+#include "sqfs/error.h"
+#include "util/test.h"
+#include "compat.h"
+
+static int compare_entries(const void *a, const void *b)
+{
+	const dir_entry_t *const *lhs = a;
+	const dir_entry_t *const *rhs = b;
+
+	return strcmp((*lhs)->name, (*rhs)->name);
+}
+
+int main(int argc, char **argv)
+{
+	dir_entry_t *ent[17];
+	dir_iterator_t *dir;
+	dir_tree_cfg_t cfg;
+	size_t i;
+	int ret;
+	(void)argc; (void)argv;
+
+	/********** match name **********/
+	memset(&cfg, 0, sizeof(cfg));
+	cfg.name_pattern = "file_x*";
+
+	dir = dir_tree_iterator_create(TEST_PATH, &cfg);
+	TEST_NOT_NULL(dir);
+
+	for (i = 0; i < 3; ++i) {
+		ret = dir->next(dir, &ent[i]);
+		TEST_NOT_NULL(ent[i]);
+		TEST_EQUAL_I(ret, 0);
+		printf("READ %s\n", ent[i]->name);
+	}
+
+	ret = dir->next(dir, &ent[3]);
+	TEST_NULL(ent[3]);
+	TEST_ASSERT(ret > 0);
+
+	dir = sqfs_drop(dir);
+
+	qsort(ent, 3, sizeof(ent[0]), compare_entries);
+
+	printf("After sort:\n");
+	for (i = 0; i < 3; ++i)
+		printf("%s\n", ent[i]->name);
+
+	TEST_STR_EQUAL(ent[0]->name, "dirb/dirx/file_x0");
+	TEST_ASSERT(S_ISREG(ent[0]->mode));
+	TEST_STR_EQUAL(ent[1]->name, "dirb/dirx/file_x1");
+	TEST_ASSERT(S_ISREG(ent[1]->mode));
+	TEST_STR_EQUAL(ent[2]->name, "dirb/dirx/file_x2");
+	TEST_ASSERT(S_ISREG(ent[2]->mode));
+
+	for (i = 0; i < 3; ++i)
+		free(ent[i]);
+
+	/********** match path **********/
+	memset(&cfg, 0, sizeof(cfg));
+	cfg.name_pattern = "dir*/file_*0";
+	cfg.flags |= DIR_SCAN_MATCH_FULL_PATH;
+
+	dir = dir_tree_iterator_create(TEST_PATH, &cfg);
+	TEST_NOT_NULL(dir);
+
+	for (i = 0; i < 3; ++i) {
+		ret = dir->next(dir, &ent[i]);
+		TEST_NOT_NULL(ent[i]);
+		TEST_EQUAL_I(ret, 0);
+		printf("READ %s\n", ent[i]->name);
+	}
+
+	ret = dir->next(dir, &ent[3]);
+	TEST_NULL(ent[3]);
+	TEST_ASSERT(ret > 0);
+
+	dir = sqfs_drop(dir);
+
+	qsort(ent, 3, sizeof(ent[0]), compare_entries);
+
+	printf("After sort:\n");
+	for (i = 0; i < 3; ++i)
+		printf("%s\n", ent[i]->name);
+
+	TEST_STR_EQUAL(ent[0]->name, "dira/file_a0");
+	TEST_ASSERT(S_ISREG(ent[0]->mode));
+	TEST_STR_EQUAL(ent[1]->name, "dirb/file_b0");
+	TEST_ASSERT(S_ISREG(ent[1]->mode));
+	TEST_STR_EQUAL(ent[2]->name, "dirc/file_c0");
+	TEST_ASSERT(S_ISREG(ent[2]->mode));
+
+	for (i = 0; i < 3; ++i)
+		free(ent[i]);
+
+	return EXIT_SUCCESS;
+}