3 files changed, 732 insertions, 0 deletions

diff --git a/lib/io/src/dir_tree_iterator.c b/lib/io/src/dir_tree_iterator.c
new file mode 100644
index 0000000..b05eeb3
--- /dev/null
+++ b/lib/io/src/dir_tree_iterator.c
@@ -0,0 +1,310 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * dir_tree_iterator.c
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/dir_tree_iterator.h"
+#include "util/util.h"
+#include "sqfs/error.h"
+
+#include <stdlib.h>
+#include <string.h>
+
+typedef struct dir_stack_t {
+	struct dir_stack_t *next;
+	dir_iterator_t *dir;
+	char name[];
+} dir_stack_t;
+
+typedef struct {
+	dir_iterator_t base;
+
+	dir_tree_cfg_t cfg;
+	int state;
+	dir_stack_t *top;
+} dir_tree_iterator_t;
+
+static void pop(dir_tree_iterator_t *it)
+{
+	if (it->top != NULL) {
+		dir_stack_t *ent = it->top;
+		it->top = it->top->next;
+
+		sqfs_drop(ent->dir);
+		free(ent);
+	}
+}
+
+static int push(dir_tree_iterator_t *it, const char *name, dir_iterator_t *dir)
+{
+	dir_stack_t *ent = alloc_flex(sizeof(*ent), 1, strlen(name) + 1);
+
+	if (ent == NULL)
+		return SQFS_ERROR_ALLOC;
+
+	strcpy(ent->name, name);
+	ent->dir = sqfs_grab(dir);
+	ent->next = it->top;
+	it->top = ent;
+	return 0;
+}
+
+static bool should_skip(const dir_tree_iterator_t *dir, const dir_entry_t *ent)
+{
+	unsigned int type_mask;
+
+	if (!strcmp(ent->name, ".") || !strcmp(ent->name, ".."))
+		return true;
+
+	if ((dir->cfg.flags & DIR_SCAN_ONE_FILESYSTEM)) {
+		if (ent->dev != ((const dir_iterator_t *)dir)->dev)
+			return true;
+	}
+
+	switch (ent->mode & S_IFMT) {
+	case S_IFSOCK: type_mask = DIR_SCAN_NO_SOCK;  break;
+	case S_IFLNK:  type_mask = DIR_SCAN_NO_SLINK; break;
+	case S_IFREG:  type_mask = DIR_SCAN_NO_FILE;  break;
+	case S_IFBLK:  type_mask = DIR_SCAN_NO_BLK;   break;
+	case S_IFCHR:  type_mask = DIR_SCAN_NO_CHR;   break;
+	case S_IFIFO:  type_mask = DIR_SCAN_NO_FIFO;  break;
+	default:       type_mask = 0;                 break;
+	}
+
+	return (dir->cfg.flags & type_mask) != 0;
+}
+
+static dir_entry_t *expand_path(const dir_tree_iterator_t *it, dir_entry_t *ent)
+{
+	size_t slen = strlen(ent->name) + 1, plen = 0;
+	dir_stack_t *sit;
+	char *dst;
+
+	for (sit = it->top; sit != NULL; sit = sit->next) {
+		if (sit->name[0] != '\0')
+			plen += strlen(sit->name) + 1;
+	}
+
+	if (it->cfg.prefix != NULL && it->cfg.prefix[0] != '\0')
+		plen += strlen(it->cfg.prefix) + 1;
+
+	if (plen > 0) {
+		void *new = realloc(ent, sizeof(*ent) + plen + slen);
+		if (new == NULL) {
+			free(ent);
+			return NULL;
+		}
+
+		ent = new;
+		memmove(ent->name + plen, ent->name, slen);
+		dst = ent->name + plen;
+
+		for (sit = it->top; sit != NULL; sit = sit->next) {
+			size_t len = strlen(sit->name);
+			if (len > 0) {
+				*(--dst) = '/';
+				dst -= len;
+				memcpy(dst, sit->name, len);
+			}
+		}
+
+		if (it->cfg.prefix != NULL && it->cfg.prefix[0] != '\0') {
+			size_t len = strlen(it->cfg.prefix);
+			memcpy(ent->name, it->cfg.prefix, len);
+			ent->name[len] = '/';
+		}
+	}
+
+	return ent;
+}
+
+static void apply_changes(const dir_tree_iterator_t *it, dir_entry_t *ent)
+{
+	if (!(it->cfg.flags & DIR_SCAN_KEEP_TIME))
+		ent->mtime = it->cfg.def_mtime;
+
+	if (!(it->cfg.flags & DIR_SCAN_KEEP_UID))
+		ent->uid = it->cfg.def_uid;
+
+	if (!(it->cfg.flags & DIR_SCAN_KEEP_GID))
+		ent->gid = it->cfg.def_gid;
+
+	if (!(it->cfg.flags & DIR_SCAN_KEEP_MODE)) {
+		ent->mode &= ~(07777);
+		ent->mode |= it->cfg.def_mode & 07777;
+	}
+}
+
+/*****************************************************************************/
+
+static void destroy(sqfs_object_t *obj)
+{
+	dir_tree_iterator_t *it = (dir_tree_iterator_t *)obj;
+
+	while (it->top != NULL)
+		pop(it);
+
+	free(it);
+}
+
+static int next(dir_iterator_t *base, dir_entry_t **out)
+{
+	dir_tree_iterator_t *it = (dir_tree_iterator_t *)base;
+	dir_iterator_t *sub;
+	dir_entry_t *ent;
+	int ret;
+retry:
+	*out = NULL;
+	sub = NULL;
+	ent = NULL;
+
+	if (it->state != 0)
+		return it->state;
+
+	for (;;) {
+		if (it->top == NULL) {
+			ret = 1;
+			goto fail;
+		}
+
+		ret = it->top->dir->next(it->top->dir, &ent);
+		if (ret < 0)
+			goto fail;
+
+		if (ret > 0) {
+			pop(it);
+			continue;
+		}
+
+		if (should_skip(it, ent)) {
+			free(ent);
+			ent = NULL;
+			continue;
+		}
+
+		break;
+	}
+
+	ent = expand_path(it, ent);
+	if (ent == NULL) {
+		it->state = SQFS_ERROR_ALLOC;
+		return it->state;
+	}
+
+	apply_changes(it, ent);
+
+	if (S_ISDIR(ent->mode)) {
+		if (!(it->cfg.flags & DIR_SCAN_NO_RECURSION)) {
+			const char *name = strrchr(ent->name, '/');
+			name = (name == NULL) ? ent->name : (name + 1);
+
+			ret = it->top->dir->open_subdir(it->top->dir, &sub);
+			if (ret != 0)
+				goto fail;
+
+			ret = push(it, name, sub);
+			sqfs_drop(sub);
+			if (ret != 0)
+				goto fail;
+		}
+
+		if (it->cfg.flags & DIR_SCAN_NO_DIR) {
+			free(ent);
+			goto 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:
+	free(ent);
+	it->state = ret;
+	return it->state;
+}
+
+static int read_link(dir_iterator_t *base, char **out)
+{
+	dir_tree_iterator_t *it = (dir_tree_iterator_t *)base;
+
+	if (it->top == NULL) {
+		*out = NULL;
+		return SQFS_ERROR_NO_ENTRY;
+	}
+
+	return it->top->dir->read_link(it->top->dir, out);
+}
+
+static int open_subdir(dir_iterator_t *base, dir_iterator_t **out)
+{
+	dir_tree_iterator_t *it = (dir_tree_iterator_t *)base;
+
+	if (it->top == NULL) {
+		*out = NULL;
+		return SQFS_ERROR_NO_ENTRY;
+	}
+
+	return it->top->dir->open_subdir(it->top->dir, out);
+}
+
+dir_iterator_t *dir_tree_iterator_create(const char *path,
+					 const dir_tree_cfg_t *cfg)
+{
+	dir_tree_iterator_t *it = calloc(1, sizeof(*it));
+	dir_iterator_t *dir;
+	int ret;
+
+	if (it == NULL) {
+		perror(path);
+		return NULL;
+	}
+
+	it->cfg = *cfg;
+
+	dir = dir_iterator_create(path);
+	if (dir == NULL)
+		goto fail;
+
+	ret = push(it, "", dir);
+	dir = sqfs_drop(dir);
+	if (ret != 0) {
+		fprintf(stderr, "%s: out of memory\n", path);
+		goto fail;
+	}
+
+	sqfs_object_init(it, destroy, NULL);
+	((dir_iterator_t *)it)->dev = it->top->dir->dev;
+	((dir_iterator_t *)it)->next = next;
+	((dir_iterator_t *)it)->read_link = read_link;
+	((dir_iterator_t *)it)->open_subdir = open_subdir;
+
+	return (dir_iterator_t *)it;
+fail:
+	free(it);
+	return NULL;
+}
+
+void dir_tree_iterator_skip(dir_iterator_t *base)
+{
+	dir_tree_iterator_t *it = (dir_tree_iterator_t *)base;
+
+	pop(it);
+}
diff --git a/lib/io/src/unix/dir_iterator.c b/lib/io/src/unix/dir_iterator.c
new file mode 100644
index 0000000..d67fe83
--- /dev/null
+++ b/lib/io/src/unix/dir_iterator.c
@@ -0,0 +1,204 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * dir_iterator.c
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/dir_iterator.h"
+#include "util/util.h"
+#include "sqfs/error.h"
+
+#include <sys/stat.h>
+#include <stdlib.h>
+#include <dirent.h>
+#include <string.h>
+#include <errno.h>
+
+typedef struct {
+	dir_iterator_t base;
+
+	struct dirent *ent;
+	struct stat sb;
+	int state;
+	DIR *dir;
+} unix_dir_iterator_t;
+
+static void dir_destroy(sqfs_object_t *obj)
+{
+	unix_dir_iterator_t *it = (unix_dir_iterator_t *)obj;
+
+	closedir(it->dir);
+	free(it);
+}
+
+static int dir_read_link(dir_iterator_t *base, char **out)
+{
+	unix_dir_iterator_t *it = (unix_dir_iterator_t *)base;
+	ssize_t ret;
+	size_t size;
+	char *str;
+
+	*out = NULL;
+
+	if (it->state < 0)
+		return it->state;
+
+	if (it->state > 0 || it->ent == NULL)
+		return SQFS_ERROR_NO_ENTRY;
+
+	if ((sizeof(it->sb.st_size) > sizeof(size_t)) &&
+	    it->sb.st_size > SIZE_MAX) {
+		return SQFS_ERROR_ALLOC;
+	}
+
+	if (SZ_ADD_OV((size_t)it->sb.st_size, 1, &size))
+		return SQFS_ERROR_ALLOC;
+
+	str = calloc(1, size);
+	if (str == NULL)
+		return SQFS_ERROR_ALLOC;
+
+	ret = readlinkat(dirfd(it->dir), it->ent->d_name,
+			 str, (size_t)it->sb.st_size);
+	if (ret < 0) {
+		free(str);
+		return SQFS_ERROR_IO;
+	}
+
+	str[ret] = '\0';
+
+	*out = str;
+	return 0;
+}
+
+static int dir_next(dir_iterator_t *base, dir_entry_t **out)
+{
+	unix_dir_iterator_t *it = (unix_dir_iterator_t *)base;
+	dir_entry_t *decoded;
+	size_t len;
+
+	*out = NULL;
+	if (it->state != 0)
+		return it->state;
+
+	errno = 0;
+	it->ent = readdir(it->dir);
+
+	if (it->ent == NULL) {
+		if (errno != 0) {
+			it->state = SQFS_ERROR_IO;
+		} else {
+			it->state = 1;
+		}
+
+		return it->state;
+	}
+
+	if (fstatat(dirfd(it->dir), it->ent->d_name,
+		    &it->sb, AT_SYMLINK_NOFOLLOW)) {
+		it->state = SQFS_ERROR_IO;
+		return it->state;
+	}
+
+	len = strlen(it->ent->d_name);
+
+	decoded = alloc_flex(sizeof(*decoded), 1, len + 1);
+	if (decoded == NULL) {
+		it->state = SQFS_ERROR_ALLOC;
+		return it->state;
+	}
+
+	memcpy(decoded->name, it->ent->d_name, len);
+	decoded->mtime = it->sb.st_mtime;
+	decoded->dev = it->sb.st_dev;
+	decoded->rdev = it->sb.st_rdev;
+	decoded->uid = it->sb.st_uid;
+	decoded->gid = it->sb.st_gid;
+	decoded->mode = it->sb.st_mode;
+
+	*out = decoded;
+	return it->state;
+}
+
+static int dir_open_subdir(dir_iterator_t *base, dir_iterator_t **out)
+{
+	const unix_dir_iterator_t *it = (const unix_dir_iterator_t *)base;
+	unix_dir_iterator_t *sub = NULL;
+	int fd;
+
+	*out = NULL;
+
+	if (it->state < 0)
+		return it->state;
+
+	if (it->state > 0 || it->ent == NULL)
+		return SQFS_ERROR_NO_ENTRY;
+
+	fd = openat(dirfd(it->dir), it->ent->d_name, O_RDONLY | O_DIRECTORY);
+	if (fd < 0) {
+		if (errno == ENOTDIR)
+			return SQFS_ERROR_NOT_DIR;
+		return SQFS_ERROR_IO;
+	}
+
+	sub = calloc(1, sizeof(*sub));
+	if (sub == NULL)
+		goto fail_alloc;
+
+	sub->dir = fdopendir(fd);
+	if (sub->dir == NULL)
+		goto fail_alloc;
+
+	if (fstat(dirfd(sub->dir), &sub->sb)) {
+		free(sub);
+		return SQFS_ERROR_IO;
+	}
+
+	sqfs_object_init(sub, dir_destroy, NULL);
+	((dir_iterator_t *)sub)->dev = sub->sb.st_dev;
+	((dir_iterator_t *)sub)->next = dir_next;
+	((dir_iterator_t *)sub)->read_link = dir_read_link;
+	((dir_iterator_t *)sub)->open_subdir = dir_open_subdir;
+
+	*out = (dir_iterator_t *)sub;
+	return 0;
+fail_alloc:
+	free(sub);
+	close(fd);
+	return SQFS_ERROR_ALLOC;
+}
+
+dir_iterator_t *dir_iterator_create(const char *path)
+{
+	unix_dir_iterator_t *it = calloc(1, sizeof(*it));
+
+	if (it == NULL) {
+		perror(path);
+		return NULL;
+	}
+
+	it->state = 0;
+	it->dir = opendir(path);
+
+	if (it->dir == NULL) {
+		perror(path);
+		free(it);
+		return NULL;
+	}
+
+	if (fstat(dirfd(it->dir), &it->sb)) {
+		perror(path);
+		closedir(it->dir);
+		free(it);
+		return NULL;
+	}
+
+	sqfs_object_init(it, dir_destroy, NULL);
+	((dir_iterator_t *)it)->dev = it->sb.st_dev;
+	((dir_iterator_t *)it)->next = dir_next;
+	((dir_iterator_t *)it)->read_link = dir_read_link;
+	((dir_iterator_t *)it)->open_subdir = dir_open_subdir;
+
+	return (dir_iterator_t *)it;
+}
diff --git a/lib/io/src/win32/dir_iterator.c b/lib/io/src/win32/dir_iterator.c
new file mode 100644
index 0000000..23d00f4
--- /dev/null
+++ b/lib/io/src/win32/dir_iterator.c
@@ -0,0 +1,218 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * dir_iterator.c
+ *
+ * Copyright (C) 2023 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "config.h"
+#include "io/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
+
+typedef struct {
+	dir_iterator_t base;
+
+	WIN32_FIND_DATAW ent;
+	HANDLE dirhnd;
+	int state;
+	bool is_first;
+
+	WCHAR path[];
+} 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_read_link(dir_iterator_t *it, char **out)
+{
+	(void)it;
+	*out = NULL;
+	return SQFS_ERROR_UNSUPPORTED;
+}
+
+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;
+
+	if (w32->state == 0 && !w32->is_first) {
+		if (!FindNextFileW(w32->dirhnd, &w32->ent)) {
+			if (GetLastError() == ERROR_NO_MORE_FILES) {
+				w32->state = 1;
+			} else {
+				w32->state = SQFS_ERROR_IO;
+			}
+		}
+	}
+
+	w32->is_first = false;
+
+	if (w32->state != 0)
+		goto out;
+
+	length = WideCharToMultiByte(CP_UTF8, 0, w32->ent.cFileName,
+				     -1, NULL, 0, NULL, NULL);
+	if (length <= 0) {
+		w32->state = SQFS_ERROR_ALLOC;
+		goto out;
+	}
+
+	ent = alloc_flex(sizeof(*ent), 1, length + 1);
+	if (ent == NULL) {
+		w32->state = SQFS_ERROR_ALLOC;
+		goto out;
+	}
+
+	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));
+out:
+	*out = ent;
+	return w32->state;
+}
+
+static void dir_iterator_destroy(sqfs_object_t *obj)
+{
+	dir_iterator_win32_t *dir = (dir_iterator_win32_t *)obj;
+
+	FindClose(dir->dirhnd);
+	free(dir);
+}
+
+static int dir_iterator_open_subdir(dir_iterator_t *it, dir_iterator_t **out)
+{
+	const dir_iterator_win32_t *dir = (const dir_iterator_win32_t *)it;
+	dir_iterator_win32_t *sub = NULL;
+	size_t plen, slen, total;
+
+	*out = NULL;
+
+	if (dir->state != 0)
+		return (dir->state > 0) ? SQFS_ERROR_NO_ENTRY : dir->state;
+
+	if (!(dir->ent.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY))
+		return SQFS_ERROR_NOT_DIR;
+
+	plen = wcslen(dir->path) - 1;
+	slen = wcslen(dir->ent.cFileName);
+	total = plen + slen + 3;
+
+	sub = alloc_flex(sizeof(*sub), sizeof(WCHAR), total);
+	if (sub == NULL)
+		return SQFS_ERROR_ALLOC;
+
+	memcpy(sub->path, dir->path, plen * sizeof(WCHAR));
+	memcpy(sub->path + plen, dir->ent.cFileName, slen * sizeof(WCHAR));
+	sub->path[plen + slen    ] = '\\';
+	sub->path[plen + slen + 1] = '*';
+	sub->path[plen + slen + 2] = '\0';
+
+	sqfs_object_init(sub, dir_iterator_destroy, NULL);
+	((dir_iterator_t *)sub)->next = dir_iterator_next;
+	((dir_iterator_t *)sub)->read_link = dir_iterator_read_link;
+	((dir_iterator_t *)sub)->open_subdir = dir_iterator_open_subdir;
+	sub->is_first = true;
+	sub->state = 0;
+
+	sub->dirhnd = FindFirstFileW(sub->path, &sub->ent);
+	if (sub->dirhnd == INVALID_HANDLE_VALUE) {
+		free(sub);
+		return SQFS_ERROR_IO;
+	}
+
+	*out = (dir_iterator_t *)sub;
+	return 0;
+}
+
+dir_iterator_t *dir_iterator_create(const char *path)
+{
+	dir_iterator_win32_t *it;
+	size_t len, newlen;
+	WCHAR *wpath = NULL;
+	void *new = NULL;
+
+	/* convert path to UTF-16, append "\\*" */
+	wpath = path_to_windows(path);
+	if (wpath == NULL)
+		goto fail_alloc;
+
+	len = wcslen(wpath);
+	newlen = len + 1;
+
+	if (len > 0 && wpath[len - 1] != '\\')
+		newlen += 1;
+
+	new = realloc(wpath, sizeof(wpath[0]) * (newlen + 1));
+	if (new == NULL)
+		goto fail_alloc;
+
+	wpath = new;
+
+	if (len > 0 && wpath[len - 1] != '\\')
+		wpath[len++] = '\\';
+
+	wpath[len++] = '*';
+	wpath[len++] = '\0';
+
+	/* create the sourrounding iterator structure */
+	new = realloc(wpath, sizeof(*it) + len * sizeof(wpath[0]));
+	if (new == NULL)
+		goto fail_alloc;
+
+	it = new;
+	wpath = NULL;
+	memmove(it->path, new, len * sizeof(wpath[0]));
+
+	/* initialize */
+	memset(it, 0, offsetof(dir_iterator_win32_t, path));
+	sqfs_object_init(it, dir_iterator_destroy, NULL);
+
+	((dir_iterator_t *)it)->next = dir_iterator_next;
+	((dir_iterator_t *)it)->read_link = dir_iterator_read_link;
+	((dir_iterator_t *)it)->open_subdir = dir_iterator_open_subdir;
+	it->is_first = true;
+	it->state = 0;
+
+	/* get the directory handle AND the first entry */
+	it->dirhnd = FindFirstFileW(it->path, &it->ent);
+
+	if (it->dirhnd == INVALID_HANDLE_VALUE) {
+		w32_perror(path);
+		free(it);
+		return NULL;
+	}
+
+	return (dir_iterator_t *)it;
+fail_alloc:
+	fprintf(stderr, "%s: allocation failure.\n", path);
+	free(wpath);
+	return NULL;
+}