os: add existing_path function (#14536)

vlib/os/filepath.v

@@ -123,6 +123,52 @@ pub fn norm_path(path string) string {
 	return res
 }
 
+// existing_path returns the existing part of the given `path`.
+// An error is returned if there is no existing part of the given `path`.
+pub fn existing_path(path string) ?string {
+	err := error('path does not exist')
+	if path.len == 0 {
+		return err
+	}
+	if exists(path) {
+		return path
+	}
+	mut volume_len := 0
+	$if windows {
+		volume_len = win_volume_len(path)
+	}
+	if volume_len > 0 && is_slash(path[volume_len - 1]) {
+		volume_len++
+	}
+	mut sc := textscanner.new(path[volume_len..])
+	mut recent_path := path[..volume_len]
+	for sc.next() != -1 {
+		curr := u8(sc.current())
+		peek := sc.peek()
+		back := sc.peek_back()
+		if is_curr_dir_ref(back, curr, peek) {
+			continue
+		}
+		range := sc.ilen - sc.remaining() + volume_len
+		if is_slash(curr) && !is_slash(u8(peek)) {
+			recent_path = path[..range]
+			continue
+		}
+		if !is_slash(curr) && (peek == -1 || is_slash(u8(peek))) {
+			curr_path := path[..range]
+			if exists(curr_path) {
+				recent_path = curr_path
+				continue
+			}
+			if recent_path.len == 0 {
+				break
+			}
+			return recent_path
+		}
+	}
+	return err
+}
+
 // clean_path returns the "cleaned" version of the given `path`
 // by turning forward slashes into back slashes
 // on a Windows system and eliminating:
@@ -144,8 +190,7 @@ fn clean_path(path string) string {
 			continue
 		}
 		// skip reference to current dir (.)
-		if (back == -1 || is_slash(u8(back))) && curr == os.dot
+		if is_curr_dir_ref(back, curr, peek) {
-			&& (peek == -1 || is_slash(u8(peek))) {
 			// skip if the next byte is a path separator
 			if peek != -1 && is_slash(u8(peek)) {
 				sc.skip_n(1)
@@ -246,3 +291,13 @@ fn is_normal_path(path string) bool {
 	return (plen == 1 && is_slash(path[0])) || (plen >= 2 && is_slash(path[0])
 		&& !is_slash(path[1]))
 }
+
+// is_curr_dir_ref returns `true` if the 3 given integer construct
+// a reference to a current directory (.).
+// NOTE: a negative integer means that no byte is present
+fn is_curr_dir_ref(byte_one int, byte_two int, byte_three int) bool {
+	if u8(byte_two) != os.dot {
+		return false
+	}
+	return (byte_one < 0 || is_slash(u8(byte_one))) && (byte_three < 0 || is_slash(u8(byte_three)))
+}

vlib/os/filepath_test.v

@@ -36,11 +36,13 @@ fn test_clean_path() {
 		assert clean_path(r'\./path/dir\\file.exe') == r'\path\dir\file.exe'
 		assert clean_path(r'.') == ''
 		assert clean_path(r'./') == ''
+		assert clean_path('') == ''
 		assert clean_path(r'\./') == '\\'
 		assert clean_path(r'//\/\/////') == '\\'
 		return
 	}
 	assert clean_path('./../.././././//') == '../..'
+	assert clean_path('') == ''
 	assert clean_path('.') == ''
 	assert clean_path('./path/to/file.v//./') == 'path/to/file.v'
 	assert clean_path('./') == ''
@@ -127,3 +129,26 @@ fn test_abs_path() {
 	assert abs_path('path/../file.v/..') == wd
 	assert abs_path('///') == '/'
 }
+
+fn test_existing_path() {
+	wd := getwd()
+	$if windows {
+		assert existing_path('') or { '' } == ''
+		assert existing_path('..') or { '' } == '..'
+		assert existing_path('.') or { '' } == '.'
+		assert existing_path(wd) or { '' } == wd
+		assert existing_path('\\') or { '' } == '\\'
+		assert existing_path('$wd\\.\\\\does/not/exist\\.\\') or { '' } == '$wd\\.\\\\'
+		assert existing_path('$wd\\\\/\\.\\.\\/.') or { '' } == '$wd\\\\/\\.\\.\\/.'
+		assert existing_path('$wd\\././/\\/oh') or { '' } == '$wd\\././/\\/'
+		return
+	}
+	assert existing_path('') or { '' } == ''
+	assert existing_path('..') or { '' } == '..'
+	assert existing_path('.') or { '' } == '.'
+	assert existing_path(wd) or { '' } == wd
+	assert existing_path('/') or { '' } == '/'
+	assert existing_path('$wd/does/.///not/exist///.//') or { '' } == '$wd/'
+	assert existing_path('$wd//././/.//') or { '' } == '$wd//././/.//'
+	assert existing_path('$wd//././/.//oh') or { '' } == '$wd//././/.//'
+}