Split out Windows directory iteration code to a dir_iterator_t type

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

6 files changed, 269 insertions, 131 deletions

diff --git a/bin/gensquashfs/Makemodule.am b/bin/gensquashfs/Makemodule.am
index de9ea4c..ff426ae 100644
--- a/bin/gensquashfs/Makemodule.am
+++ b/bin/gensquashfs/Makemodule.am
@@ -67,7 +67,7 @@ fstree_fuzz_SOURCES = bin/gensquashfs/test/fstree_fuzz.c \
 			bin/gensquashfs/src/fstree_from_dir.c \
 			bin/gensquashfs/src/mkfs.h
 fstree_fuzz_CPPFLAGS = $(AM_CPPFLAGS) -I$(top_srcdir)/bin/gensquashfs/src
-fstree_fuzz_LDADD = libfstree.a libio.a libutil.a libcompat.a
+fstree_fuzz_LDADD = libcommon.a libfstree.a libio.a libutil.a libcompat.a
 
 GENSQUASHFS_TESTS = \
 	test_filemap_xattr test_fstree_from_file test_fstree_from_dir \
diff --git a/bin/gensquashfs/src/fstree_from_dir.c b/bin/gensquashfs/src/fstree_from_dir.c
index d117435..2ece45a 100644
--- a/bin/gensquashfs/src/fstree_from_dir.c
+++ b/bin/gensquashfs/src/fstree_from_dir.c
@@ -13,48 +13,28 @@
 #include <errno.h>
 
 #if defined(_WIN32) || defined(__WINDOWS__)
-#define UNIX_EPOCH_ON_W32 11644473600UL
-#define W32_TICS_PER_SEC 10000000UL
-
-static sqfs_u32 w32time_to_sqfs_time(const FILETIME *ft)
-{
-	sqfs_u64 w32ts;
-
-	w32ts = ft->dwHighDateTime;
-	w32ts <<= 32UL;
-	w32ts |= ft->dwLowDateTime;
-
-	w32ts /= W32_TICS_PER_SEC;
-
-	if (w32ts <= UNIX_EPOCH_ON_W32)
-		return 0;
-
-	w32ts -= UNIX_EPOCH_ON_W32;
-
-	return (w32ts < 0x0FFFFFFFFUL) ? w32ts : 0xFFFFFFFF;
-}
-
 static int add_node(fstree_t *fs, tree_node_t *root,
 		    scan_node_callback cb, void *user,
 		    unsigned int flags,
-		    const LPWIN32_FIND_DATAW entry)
+		    const dir_entry_t *entry)
 {
+	size_t length = strlen(entry->name);
 	tree_node_t *n;
-	DWORD length;
 
-	if (entry->cFileName[0] == '.') {
-		if (entry->cFileName[1] == '\0')
+	if (entry->name[0] == '.') {
+		if (length == 1)
 			return 0;
 
-		if (entry->cFileName[1] == '.' && entry->cFileName[2] == '\0')
+		if (entry->name[1] == '.' && length == 2)
 			return 0;
 	}
 
-	length = WideCharToMultiByte(CP_UTF8, 0, entry->cFileName,
-				     -1, NULL, 0, NULL, NULL);
-	if (length <= 0) {
-		w32_perror("converting path to UTF-8");
-		return -1;
+	if (S_ISDIR(entry->mode)) {
+		if (flags & DIR_SCAN_NO_DIR)
+			return 0;
+	} else {
+		if (flags & DIR_SCAN_NO_FILE)
+			return 0;
 	}
 
 	n = calloc(1, sizeof(*n) + length + 1);
@@ -63,25 +43,9 @@ static int add_node(fstree_t *fs, tree_node_t *root,
 		return -1;
 	}
 
+	n->mode = entry->mode;
 	n->name = (char *)n->payload;
-	WideCharToMultiByte(CP_UTF8, 0, entry->cFileName, -1,
-			    n->name, length + 1, NULL, NULL);
-
-	if (entry->dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
-		if (flags & DIR_SCAN_NO_DIR) {
-			free(n);
-			return 0;
-		}
-
-		n->mode = S_IFDIR | 0755;
-	} else {
-		if (flags & DIR_SCAN_NO_FILE) {
-			free(n);
-			return 0;
-		}
-
-		n->mode = S_IFREG | 0644;
-	}
+	memcpy(n->name, entry->name, length);
 
 	if (cb != NULL) {
 		int ret = cb(user, fs, n);
@@ -93,7 +57,13 @@ static int add_node(fstree_t *fs, tree_node_t *root,
 	}
 
 	if (flags & DIR_SCAN_KEEP_TIME) {
-		n->mod_time = w32time_to_sqfs_time(&(entry->ftLastWriteTime));
+		if (entry->mtime < 0) {
+			n->mod_time = 0;
+		} else if (entry->mtime > 0x0FFFFFFFFLL) {
+			n->mod_time = 0xFFFFFFFF;
+		} else {
+			n->mod_time = entry->mtime;
+		}
 	} else {
 		n->mod_time = fs->defaults.mtime;
 	}
@@ -103,81 +73,49 @@ static int add_node(fstree_t *fs, tree_node_t *root,
 }
 
 static int scan_dir(fstree_t *fs, tree_node_t *root,
-		    const char *path, const WCHAR *wpath,
+		    const char *path,
 		    scan_node_callback cb, void *user,
 		    unsigned int flags)
 {
-	WIN32_FIND_DATAW entry;
-	HANDLE dirhnd;
+	dir_iterator_t *it = dir_iterator_create(path);
 
-	dirhnd = FindFirstFileW(wpath, &entry);
+	if (it == NULL)
+		return -1;
 
-	if (dirhnd == INVALID_HANDLE_VALUE)
-		goto fail_perror;
+	for (;;) {
+		dir_entry_t *ent;
+		int ret;
 
-	do {
-		if (add_node(fs, root, cb, user, flags, &entry))
-			goto fail;
-	} while (FindNextFileW(dirhnd, &entry));
+		ret = it->next(it, &ent);
+		if (ret > 0)
+			break;
+		if (ret < 0) {
+			sqfs_perror(path, "reading directory entry", ret);
+			sqfs_drop(it);
+			return -1;
+		}
 
-	if (GetLastError() != ERROR_NO_MORE_FILES)
-		goto fail_perror;
+		ret = add_node(fs, root, cb, user, flags, ent);
+		free(ent);
+		if (ret != 0) {
+			sqfs_drop(it);
+			return -1;
+		}
+	}
 
-	FindClose(dirhnd);
+	sqfs_drop(it);
 	return 0;
-fail_perror:
-	w32_perror(path);
-fail:
-	if (dirhnd != INVALID_HANDLE_VALUE)
-		FindClose(dirhnd);
-	return -1;
 }
 
 int fstree_from_dir(fstree_t *fs, tree_node_t *root,
 		    const char *path, scan_node_callback cb,
 		    void *user, unsigned int flags)
 {
-	WCHAR *wpath = NULL, *new = NULL;
-	size_t len, newlen;
 	tree_node_t *n;
 
-	/* path -> to_wchar(path) + L"\*" */
-	wpath = path_to_windows(path);
-	if (wpath == NULL) {
-		fprintf(stderr, "%s: allocation failure.\n", path);
+	if (scan_dir(fs, root, path, cb, user, flags))
 		return -1;
-	}
-
-	for (len = 0; wpath[len] != '\0'; ++len)
-		;
-
-	newlen = len + 1;
-
-	if (len > 0 && wpath[len - 1] != '\\')
-		newlen += 1;
-
-	new = realloc(wpath, sizeof(wpath[0]) * (newlen + 1));
-	if (new == NULL) {
-		fprintf(stderr, "%s: allocation failure.\n", path);
-		goto fail;
-	}
-
-	wpath = new;
 
-	if (len > 0 && wpath[len - 1] != '\\')
-		wpath[len++] = '\\';
-
-	wpath[len++] = '*';
-	wpath[len++] = '\0';
-
-	/* scan directory contents */
-	if (scan_dir(fs, root, path, wpath, cb, user, flags))
-		goto fail;
-
-	free(wpath);
-	wpath = NULL;
-
-	/* recursion step */
 	if (flags & DIR_SCAN_NO_RECURSION)
 		return 0;
 
@@ -190,9 +128,6 @@ int fstree_from_dir(fstree_t *fs, tree_node_t *root,
 	}
 
 	return 0;
-fail:
-	free(wpath);
-	return -1;
 }
 
 int fstree_from_subdir(fstree_t *fs, tree_node_t *root,
@@ -200,8 +135,7 @@ int fstree_from_subdir(fstree_t *fs, tree_node_t *root,
 		       scan_node_callback cb, void *user,
 		       unsigned int flags)
 {
-	size_t len, plen, slen;
-	WCHAR *wpath = NULL;
+	size_t plen, slen;
 	char *temp = NULL;
 	tree_node_t *n;
 
@@ -211,9 +145,7 @@ int fstree_from_subdir(fstree_t *fs, tree_node_t *root,
 	if (slen == 0)
 		return fstree_from_dir(fs, root, path, cb, user, flags);
 
-	len = plen + 1 + slen + 2;
-
-	temp = calloc(1, len + 1);
+	temp = calloc(1, plen + 1 + slen + 1);
 	if (temp == NULL) {
 		fprintf(stderr, "%s/%s: allocation failure.\n", path, subdir);
 		return -1;
@@ -222,29 +154,16 @@ int fstree_from_subdir(fstree_t *fs, tree_node_t *root,
 	memcpy(temp, path, plen);
 	temp[plen] = '/';
 	memcpy(temp + plen + 1, subdir, slen);
-	temp[plen + 1 + slen    ] = '/';
-	temp[plen + 1 + slen + 1] = '*';
-	temp[plen + 1 + slen + 2] = '\0';
-
-	wpath = path_to_windows(temp);
-	if (wpath == NULL) {
-		fprintf(stderr, "%s: allocation failure.\n", temp);
-		goto fail;
-	}
+	temp[plen + 1 + slen] = '\0';
 
-	if (scan_dir(fs, root, temp, wpath, cb, user, flags))
+	if (scan_dir(fs, root, temp, cb, user, flags))
 		goto fail;
 
-	free(wpath);
-	wpath = NULL;
-
 	if (flags & DIR_SCAN_NO_RECURSION) {
 		free(temp);
 		return 0;
 	}
 
-	temp[plen + 1 + slen] = '\0';
-
 	for (n = root->data.dir.children; n != NULL; n = n->next) {
 		if (!S_ISDIR(n->mode))
 			continue;
@@ -257,7 +176,6 @@ int fstree_from_subdir(fstree_t *fs, tree_node_t *root,
 	return 0;
 fail:
 	free(temp);
-	free(wpath);
 	return -1;
 
 }
diff --git a/bin/gensquashfs/src/mkfs.h b/bin/gensquashfs/src/mkfs.h
index 516ab87..6b69465 100644
--- a/bin/gensquashfs/src/mkfs.h
+++ b/bin/gensquashfs/src/mkfs.h
@@ -12,6 +12,7 @@
 
 #include "common.h"
 #include "fstree.h"
+#include "util/dir_iterator.h"
 #include "util/util.h"
 #include "io/file.h"
 
diff --git a/include/util/dir_iterator.h b/include/util/dir_iterator.h
new file mode 100644
index 0000000..66c13f3
--- /dev/null
+++ b/include/util/dir_iterator.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * dir_iterator.h
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#ifndef UTIL_DIR_ITERATOR_H
+#define UTIL_DIR_ITERATOR_H
+
+#include "sqfs/predef.h"
+
+typedef struct {
+	sqfs_s64 mtime;
+	sqfs_u64 dev;
+	sqfs_u64 rdev;
+	sqfs_u32 uid;
+	sqfs_u32 gid;
+	sqfs_u16 mode;
+
+	char name[];
+} dir_entry_t;
+
+typedef struct dir_iterator_t {
+	sqfs_object_t obj;
+
+	int (*next)(struct dir_iterator_t *it,
+		    dir_entry_t **out);
+} dir_iterator_t;
+
+dir_iterator_t *dir_iterator_create(const char *path);
+
+#endif /* UTIL_DIR_ITERATOR_H */
diff --git a/lib/util/Makemodule.am b/lib/util/Makemodule.am
index 0a0e50e..8e9b01d 100644
--- a/lib/util/Makemodule.am
+++ b/lib/util/Makemodule.am
@@ -2,6 +2,7 @@ libutil_a_SOURCES = include/util/util.h include/util/str_table.h \
 	include/util/hash_table.h include/util/test.h include/util/rbtree.h \
 	include/util/array.h include/util/threadpool.h \
 	include/util/w32threadwrap.h include/util/mempool.h \
+	include/util/dir_iterator.h \
 	lib/util/src/str_table.c lib/util/src/alloc.c lib/util/src/rbtree.c \
 	lib/util/src/array.c lib/util/src/xxhash.c lib/util/src/hash_table.c \
 	lib/util/src/fast_urem_by_const.h lib/util/src/threadpool_serial.c \
@@ -14,6 +15,7 @@ libutil_a_CPPFLAGS = $(AM_CPPFLAGS)
 
 if WINDOWS
 libutil_a_CFLAGS += -DWINVER=0x0600 -D_WIN32_WINNT=0x0600
+libutil_a_SOURCES += lib/util/src/w32_dir_iterator.c
 endif
 
 if HAVE_PTHREAD
diff --git a/lib/util/src/w32_dir_iterator.c b/lib/util/src/w32_dir_iterator.c
new file mode 100644
index 0000000..1a330ae
--- /dev/null
+++ b/lib/util/src/w32_dir_iterator.c
@@ -0,0 +1,185 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * w32_dir_iterator.c
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "util/dir_iterator.h"
+#include "util/util.h"
+#include "sqfs/error.h"
+
+#include <windows.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+#define UNIX_EPOCH_ON_W32 11644473600UL
+#define W32_TICS_PER_SEC 10000000UL
+
+enum {
+	STATE_HAVE_ENTRY = 0,
+	STATE_NEED_ENTRY = 1,
+	STATE_ERROR = 2,
+	STATE_EOF = 3,
+};
+
+typedef struct {
+	dir_iterator_t base;
+
+	WIN32_FIND_DATAW ent;
+	HANDLE dirhnd;
+	int state;
+} dir_iterator_win32_t;
+
+static sqfs_s64 w32time_to_unix(const FILETIME *ft)
+{
+	sqfs_u64 w32ts;
+
+	w32ts = ft->dwHighDateTime;
+	w32ts <<= 32UL;
+	w32ts |= ft->dwLowDateTime;
+
+	w32ts /= W32_TICS_PER_SEC;
+
+	if (w32ts <= UNIX_EPOCH_ON_W32)
+		return -((sqfs_s64)(UNIX_EPOCH_ON_W32 - w32ts));
+
+	return w32ts - UNIX_EPOCH_ON_W32;
+}
+
+static int dir_iterator_next(dir_iterator_t *it, dir_entry_t **out)
+{
+	dir_iterator_win32_t *w32 = (dir_iterator_win32_t *)it;
+	dir_entry_t *ent = NULL;
+	DWORD length;
+
+	*out = NULL;
+
+	switch (w32->state) {
+	case STATE_HAVE_ENTRY:
+		break;
+	case STATE_NEED_ENTRY:
+		if (!FindNextFileW(w32->dirhnd, &w32->ent)) {
+			if (GetLastError() == ERROR_NO_MORE_FILES) {
+				w32->state = STATE_EOF;
+				return 1;
+			}
+			w32->state = STATE_ERROR;
+			return SQFS_ERROR_INTERNAL;
+		}
+
+		w32->state = STATE_HAVE_ENTRY;
+		break;
+	case STATE_EOF:
+		return 1;
+	default:
+		return SQFS_ERROR_INTERNAL;
+	}
+
+	/* allocate */
+	length = WideCharToMultiByte(CP_UTF8, 0, w32->ent.cFileName,
+				     -1, NULL, 0, NULL, NULL);
+	if (length <= 0) {
+		w32->state = STATE_ERROR;
+		return SQFS_ERROR_ALLOC;
+	}
+
+	ent = alloc_flex(sizeof(*ent), 1, length + 1);
+	if (ent == NULL) {
+		w32->state = STATE_ERROR;
+		return SQFS_ERROR_ALLOC;
+	}
+
+	/* convert and fill entry fields */
+	WideCharToMultiByte(CP_UTF8, 0, w32->ent.cFileName, -1,
+			    ent->name, length + 1, NULL, NULL);
+
+	if (w32->ent.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
+		ent->mode = S_IFDIR | 0755;
+	} else {
+		ent->mode = S_IFREG | 0644;
+	}
+
+	ent->mtime = w32time_to_unix(&(w32->ent.ftLastWriteTime));
+
+	/* return and update state */
+	*out = ent;
+	w32->state = STATE_NEED_ENTRY;
+	return 0;
+}
+
+static void dir_iterator_destroy(sqfs_object_t *obj)
+{
+	dir_iterator_win32_t *dir = (dir_iterator_win32_t *)obj;
+
+	FindClose(dir->dirhnd);
+	free(dir);
+}
+
+dir_iterator_t *dir_iterator_create(const char *path)
+{
+	WCHAR *wpath = NULL, *new = NULL;
+	WIN32_FIND_DATAW first_entry;
+	dir_iterator_win32_t *it;
+	size_t len, newlen;
+	HANDLE dirhnd;
+
+	/* convert path to UTF-16, append "\\*" */
+	wpath = path_to_windows(path);
+	if (wpath == NULL) {
+		fprintf(stderr, "%s: allocation failure.\n", path);
+		return NULL;
+	}
+
+	for (len = 0; wpath[len] != '\0'; ++len)
+		;
+
+	newlen = len + 1;
+
+	if (len > 0 && wpath[len - 1] != '\\')
+		newlen += 1;
+
+	new = realloc(wpath, sizeof(wpath[0]) * (newlen + 1));
+	if (new == NULL) {
+		fprintf(stderr, "%s: allocation failure.\n", path);
+		free(wpath);
+		return NULL;
+	}
+
+	wpath = new;
+
+	if (len > 0 && wpath[len - 1] != '\\')
+		wpath[len++] = '\\';
+
+	wpath[len++] = '*';
+	wpath[len++] = '\0';
+
+	/* get the directory handle AND the first entry */
+	dirhnd = FindFirstFileW(wpath, &first_entry);
+
+	if (dirhnd == INVALID_HANDLE_VALUE) {
+		w32_perror(path);
+		free(wpath);
+		return NULL;
+	}
+
+	free(wpath);
+	wpath = NULL;
+
+	/* wrap it up */
+	it = calloc(1, sizeof(*it));
+	if (it == NULL) {
+		fprintf(stderr, "%s: allocation failure.\n", path);
+		FindClose(dirhnd);
+		return NULL;
+	}
+
+	sqfs_object_init(it, dir_iterator_destroy, NULL);
+
+	((dir_iterator_t *)it)->next = dir_iterator_next;
+
+	it->state = STATE_HAVE_ENTRY;
+	it->dirhnd = dirhnd;
+	it->ent = first_entry;
+	return (dir_iterator_t *)it;
+}