Update (fixes support for a few edge cases, and leaves nothing laying around)

Signed-off-by: Mattias Andrée <m@maandree.se>

1 files changed, 316 insertions, 0 deletions

diff --git a/util/repodiff.c b/util/repodiff.c
new file mode 100644
index 0000000..b27aef6
--- /dev/null
+++ b/util/repodiff.c
@@ -0,0 +1,316 @@
+/* See LICENSE file for copyright and license details. */
+#include <libsimple.h>
+#include <libsimple-arg.h>
+
+NUSAGE(2, "directory-1 file-list-1 directory-2 file-list-2");
+
+
+LIBSIMPLE_PURE
+static int
+parse_fd(const char *s, int dash_fd)
+{
+	int fd, d;
+
+	if (streq(s, "/dev/stdin"))
+		return STDIN_FILENO;
+	if (streq(s, "/dev/stdout"))
+		return STDOUT_FILENO;
+	if (streq(s, "/dev/stderr"))
+		return STDERR_FILENO;
+	if (streq(s, "-"))
+		return dash_fd;
+
+	if (strstarts(s, "/proc/self/fd/"))
+		s = &s[sizeof("/proc/self/fd/") - 1u];
+	else if (strstarts(s, "/dev/fd/"))
+		s = &s[sizeof("/dev/fd/") - 1u];
+	else
+		return -1;
+
+	if ('0' > *s || *s > '9')
+		return -1;
+	fd = (int)(*s++ - '0');
+	while ('0' <= *s && *s <= '9') {
+		d = (int)(*s - '0');
+		if (fd > (INT_MAX - d) / 10)
+			return -1;
+		fd = fd * 10 + d;
+	}
+
+	return *s ? -1 : fd;
+}
+
+
+static char **
+get_file_list(const char *path, size_t *n_out)
+{
+	char *buf = NULL;
+	size_t bufsize = 0u;
+	size_t len = 0u;
+	size_t off, i, j, d;
+	int fd, do_close;
+	ssize_t r;
+	char **list = NULL;
+	size_t listsize = 0u;
+
+	*n_out = 0u;
+
+	do_close = 0;
+	fd = parse_fd(path, STDIN_FILENO);
+	if (fd < 0) {
+		do_close = 1;
+		fd = open(path, O_RDONLY);
+		if (fd < 0)
+			eprintf("open %s O_RDONLY:", path);
+	}
+
+	for (;;) {
+		if (len == bufsize)
+			buf = erealloc(buf, bufsize += (size_t)8 << 10);
+		r = read(fd, &buf[len], bufsize - len);
+		if (r <= 0) {
+			if (!r)
+				break;
+			if (errno == EINTR)
+				continue;
+			eprintf("read %s:", path);
+		}
+		len += (size_t)r;
+	}
+
+	if (do_close)
+		close(fd);
+
+	buf = erealloc(buf, len + 1u);
+	buf[len++] = '\0';
+
+	off = 0u;
+	for (i = 0u; i < len; i++) {
+		if (buf[i])
+			continue;
+		if (i == off)
+			goto next;
+		while (buf[off] == '/')
+			off += 1u;
+		if (buf[off + 0u] == '.' && buf[off + 1u] == '/')
+			off += 2u;
+		if (!buf[off])
+			goto next;
+		if (buf[off + 0u] == '.' && !buf[off + 1u])
+			goto next;
+
+		d = buf[i - 1u] == '/' ? 1u : 0u;
+		buf[i - d] = '\0';
+		if (!buf[off])
+			goto next;
+
+		if (*n_out == listsize)
+			list = ereallocarray(list, listsize += 128u, sizeof(*list));
+		list[(*n_out)++] = estrdup(&buf[off]);
+		for (j = i - 1u - d; j > off; j--) {
+			if (buf[j] != '/')
+				continue;
+			buf[j] = '\0';
+			if (*n_out == listsize)
+				list = ereallocarray(list, listsize += 128u, sizeof(*list));
+			list[(*n_out)++] = estrdup(&buf[off]);
+		}
+
+	next:
+		off = i + 1u;
+	}
+
+	free(buf);
+	return list;
+}
+
+
+static size_t
+uniq(char **list, size_t n)
+{
+	size_t r, w = 1u;
+	if (!n)
+		return 0u;
+	for (r = 1u; r < n; r++) {
+		if (strcmp(list[r], list[w - 1u]))
+			list[w++] = list[r];
+		else
+			free(list[r]);
+	}
+	return w;
+}
+
+
+static int
+files_equal(int fd1, const char *dir1, int fd2, const char *dir2, const char *path)
+{
+	static char buf1[8192];
+	static char buf2[sizeof(buf1)];
+	size_t len1, len2;
+	ssize_t r;
+
+	do {
+		len1 = 0u;
+		while (len1 < sizeof(buf1)) {
+			r = read(fd1, &buf1[len1], sizeof(buf1) - len1);
+			if (r <= 0) {
+				if (!r)
+					break;
+				if (errno == EINTR)
+					continue;
+				eprintf("read %s/%s:", dir1, path);
+			}
+			len1 += (size_t)r;
+		}
+
+		len2 = 0u;
+		while (len2 < sizeof(buf2)) {
+			r = read(fd2, &buf2[len2], sizeof(buf2));
+			if (r <= 0) {
+				if (!r)
+					break;
+				if (errno == EINTR)
+					continue;
+				eprintf("read %s/%s:", dir2, path);
+			}
+			len2 += (size_t)r;
+		}
+
+		if (len1 != len2 || memcmp(buf1, buf2, len1))
+			return 0;
+	} while (len1 && len2);
+
+	return !len1 && !len2;
+}
+
+
+int
+main(int argc, char *argv[])
+{
+	struct stat st1, st2;
+	const char *dir1, *dir2;
+	char **files1, **files2;
+	size_t nfiles1, nfiles2, i, j;
+	char *target1, *target2;
+	int dirfd1, dirfd2, cmp;
+	int fd1, fd2;
+	int ret = 0;
+
+	libsimple_default_failure_exit = 2;
+
+	ARGBEGIN {
+	default:
+		usage();
+	} ARGEND;
+
+	if (argc != 4)
+		usage();
+
+	dir1 = argv[0];
+	dir2 = argv[2];
+	dirfd1 = open(dir1, O_PATH);
+	if (dirfd1 < 0)
+		eprintf("open %s O_PATH", dir1);
+	dirfd2 = open(dir2, O_PATH);
+	if (dirfd2 < 0)
+		eprintf("open %s O_PATH", dir2);
+
+	files1 = get_file_list(argv[1], &nfiles1);
+	files2 = get_file_list(argv[3], &nfiles2);
+	libsimple_qsort_str((void *)files1, nfiles1);
+	libsimple_qsort_str((void *)files2, nfiles2);
+	nfiles1 = uniq(files1, nfiles1);
+	nfiles2 = uniq(files2, nfiles2);
+
+	i = j = 0u;
+	while (i < nfiles1 || j < nfiles2) {
+		if (i == nfiles1) {
+		exclusive_to_dir2:
+			weprintf("%s exists only in %s\n", files2[j], dir2);
+			j++;
+			ret = 1;
+			continue;
+		}
+		if (j == nfiles2) {
+		exclusive_to_dir1:
+			weprintf("%s exists only in %s\n", files1[i], dir1);
+			i++;
+			ret = 1;
+			continue;
+		}
+
+		cmp = strcmp(files1[i], files2[j]);
+		if (cmp < 0)
+			goto exclusive_to_dir1;
+		if (cmp > 1)
+			goto exclusive_to_dir2;
+
+		if (fstatat(dirfd1, files1[i], &st1, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT))
+			eprintf("fstatat %s %s AT_SYMLINK_NOFOLLOW|AT_NO_AUTOMOUNT:", dir1, files1[i]);
+		if (fstatat(dirfd2, files2[j], &st2, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT))
+			eprintf("fstatat %s %s AT_SYMLINK_NOFOLLOW|AT_NO_AUTOMOUNT:", dir2, files2[j]);
+
+		if ((st1.st_mode & S_IFMT) != (st2.st_mode & S_IFMT)) {
+			weprintf("%s has different file types in %s and %s\n", files1[i], dir1, dir2);
+			ret = 1;
+			goto next;
+		}
+
+		switch (st1.st_mode & S_IFMT) {
+		case S_IFREG:
+			if ((st1.st_mode ^ st2.st_mode) & S_IXUSR) {
+				weprintf("%s is executable in one of but in not both of %s and %s\n", files1[i], dir1, dir2);
+				ret = 1;
+			}
+			if (st1.st_size != st2.st_size) {
+				weprintf("%s has different sizes in %s and %s\n", files1[i], dir1, dir2);
+				ret = 1;
+				goto next;
+			}
+			fd1 = openat(dirfd1, files1[i], O_RDONLY | O_NOFOLLOW);
+			if (fd1 < 0)
+				eprintf("openat %s %s O_RDONLY|O_NOFOLLOW:", dir1, files1[i]);
+			fd2 = openat(dirfd2, files2[j], O_RDONLY | O_NOFOLLOW);
+			if (fd2 < 0)
+				eprintf("openat %s %s O_RDONLY|O_NOFOLLOW:", dir2, files2[j]);
+			if (!files_equal(fd1, dir1, fd2, dir2, files1[i])) {
+				weprintf("%s has different contents in %s and %s\n", files1[i], dir1, dir2);
+				ret = 1;
+			}
+			close(fd1);
+			close(fd2);
+			break;
+
+		case S_IFLNK:
+			target1 = libsimple_ereadlinkat(dirfd1, files1[i]);
+			target2 = libsimple_ereadlinkat(dirfd2, files2[j]);
+			if (strcmp(target1, target2)) {
+				weprintf("%s has different targets in %s and %s\n", files1[i], dir1, dir2);
+				ret = 1;
+			}
+			free(target1);
+			free(target2);
+			break;
+
+		case S_IFDIR:
+			break;
+
+		default:
+			eprintf("%s has unsupported file type\n", files1[i]);
+		}
+
+	next:
+		i++;
+		j++;
+	}
+
+	close(dirfd1);
+	close(dirfd2);
+	while (nfiles1--)
+		free(files1[nfiles1]);
+	while (nfiles2--)
+		free(files2[nfiles2]);
+	free(files1);
+	free(files2);
+	return ret;
+}