From 971c55cf3050e688728a7df97a6b60df1942c7c3 Mon Sep 17 00:00:00 2001
From: Ben <89769190+benwalksaway@users.noreply.github.com>
Date: Sat, 21 May 2022 00:16:29 +0200
Subject: [PATCH] os: add norm_path and abs_path function (#14435)

---
 .../workflows/v_apps_and_modules_compile.yml  |   2 +-
 vlib/os/filepath.v                            | 174 +++++++++++++++++-
 vlib/os/filepath_test.v                       | 100 ++++++++++
 vlib/v/builder/builder.v                      |   2 +-
 4 files changed, 270 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/v_apps_and_modules_compile.yml b/.github/workflows/v_apps_and_modules_compile.yml
index 0704cf79f9..a3b764f066 100644
--- a/.github/workflows/v_apps_and_modules_compile.yml
+++ b/.github/workflows/v_apps_and_modules_compile.yml
@@ -10,7 +10,7 @@ on:
       - "**.md"
 
 concurrency:
-  group: build-other-${{ github.event.pull_request.number || github.sha }}
+  group: build-v-apps-and-modules-${{ github.event.pull_request.number || github.sha }}
   cancel-in-progress: true
 
 jobs:
diff --git a/vlib/os/filepath.v b/vlib/os/filepath.v
index 2770d26e5b..c9d772a18a 100644
--- a/vlib/os/filepath.v
+++ b/vlib/os/filepath.v
@@ -1,13 +1,20 @@
 module os
 
+import strings
+import strings.textscanner
+
 // Collection of useful functions for manipulation, validation and analysis of system paths.
 // The following functions handle paths depending on the operating system,
 // therefore results may be different for certain operating systems.
 
 const (
-	fslash = `/`
-	bslash = `\\`
-	dot    = `.`
+	fslash  = `/`
+	bslash  = `\\`
+	dot     = `.`
+	qmark   = `?`
+	dot_dot = '..'
+	empty   = ''
+	dot_str = '.'
 )
 
 // is_abs_path returns `true` if the given `path` is absolute.
@@ -16,11 +23,152 @@ pub fn is_abs_path(path string) bool {
 		return false
 	}
 	$if windows {
-		return is_device_path(path) || is_drive_rooted(path) || is_normal_path(path)
+		return is_unc_path(path) || is_drive_rooted(path) || is_normal_path(path)
 	}
 	return path[0] == os.fslash
 }
 
+// abs_path joins the current working directory
+// with the given `path` (if the `path` is relative)
+// and returns the absolute path representation.
+pub fn abs_path(path string) string {
+	wd := getwd()
+	if path.len == 0 {
+		return wd
+	}
+	npath := norm_path(path)
+	if npath == os.dot_str {
+		return wd
+	}
+	if !is_abs_path(npath) {
+		mut sb := strings.new_builder(npath.len)
+		sb.write_string(wd)
+		sb.write_string(path_separator)
+		sb.write_string(npath)
+		return norm_path(sb.str())
+	}
+	return npath
+}
+
+// norm_path returns the normalized version of the given `path`
+// by resolving backlinks (..), turning forward slashes into
+// back slashes on a Windows system and eliminating:
+// - references to current directories (.)
+// - redundant path separators
+// - the last path separator
+[direct_array_access]
+pub fn norm_path(path string) string {
+	if path.len == 0 {
+		return '.'
+	}
+	rooted := is_abs_path(path)
+	volume := get_volume(path)
+	volume_len := volume.len
+	cpath := clean_path(path[volume_len..])
+	if cpath.len == 0 && volume_len == 0 {
+		return '.'
+	}
+	spath := cpath.split(path_separator)
+	if os.dot_dot !in spath {
+		return if volume_len != 0 { volume + cpath } else { cpath }
+	}
+	// resolve backlinks (..)
+	spath_len := spath.len
+	mut sb := strings.new_builder(cpath.len)
+	if rooted {
+		sb.write_string(path_separator)
+	}
+	mut new_path := []string{cap: spath_len}
+	mut backlink_count := 0
+	for i := spath_len - 1; i >= 0; i-- {
+		part := spath[i]
+		if part == os.empty {
+			continue
+		}
+		if part == os.dot_dot {
+			backlink_count++
+			continue
+		}
+		if backlink_count != 0 {
+			backlink_count--
+			continue
+		}
+		new_path.prepend(part)
+	}
+	// append backlink(s) to the path if backtracking
+	// is not possible and the given path is not rooted
+	if backlink_count != 0 && !rooted {
+		for i in 0 .. backlink_count {
+			sb.write_string(os.dot_dot)
+			if new_path.len == 0 && i == backlink_count - 1 {
+				break
+			}
+			sb.write_string(path_separator)
+		}
+	}
+	sb.write_string(new_path.join(path_separator))
+	res := sb.str()
+	if res.len == 0 {
+		if volume_len != 0 {
+			return volume
+		}
+		if !rooted {
+			return '.'
+		}
+		return path_separator
+	}
+	if volume_len != 0 {
+		return volume + res
+	}
+	return res
+}
+
+// clean_path returns the "cleaned" version of the given `path`
+// by turning forward slashes into back slashes
+// on a Windows system and eliminating:
+// - references to current directories (.)
+// - redundant separators
+// - the last path separator
+fn clean_path(path string) string {
+	if path.len == 0 {
+		return ''
+	}
+	mut sb := strings.new_builder(path.len)
+	mut sc := textscanner.new(path)
+	for sc.next() != -1 {
+		curr := u8(sc.current())
+		back := sc.peek_back()
+		peek := sc.peek()
+		// skip current path separator if last byte was a path separator
+		if back != -1 && is_slash(u8(back)) && is_slash(curr) {
+			continue
+		}
+		// skip reference to current dir (.)
+		if (back == -1 || is_slash(u8(back))) && curr == os.dot
+			&& (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)
+			}
+			continue
+		}
+		// turn foward slash into a back slash on a Windows system
+		$if windows {
+			if curr == os.fslash {
+				sb.write_u8(os.bslash)
+				continue
+			}
+		}
+		sb.write_u8(u8(sc.current()))
+	}
+	res := sb.str()
+	// eliminate the last path separator
+	if res.len > 1 && is_slash(res[res.len - 1]) {
+		return res[..res.len - 1]
+	}
+	return res
+}
+
 // win_volume_len returns the length of the
 // Windows volume/drive from the given `path`.
 fn win_volume_len(path string) int {
@@ -32,7 +180,7 @@ fn win_volume_len(path string) int {
 		return 2
 	}
 	// its UNC path / DOS device path?
-	if path.len >= 5 && starts_w_slash_slash(path) && !is_slash(path[2]) {
+	if plen >= 5 && starts_w_slash_slash(path) && !is_slash(path[2]) {
 		for i := 3; i < plen; i++ {
 			if is_slash(path[i]) {
 				if i + 1 >= plen || is_slash(path[i + 1]) {
@@ -51,6 +199,20 @@ fn win_volume_len(path string) int {
 	return 0
 }
 
+fn get_volume(path string) string {
+	$if !windows {
+		return ''
+	}
+	volume := path[..win_volume_len(path)]
+	if volume.len == 0 {
+		return ''
+	}
+	if volume[0] == os.fslash {
+		return volume.replace('/', '\\')
+	}
+	return volume
+}
+
 fn is_slash(b u8) bool {
 	$if windows {
 		return b == os.bslash || b == os.fslash
@@ -58,7 +220,7 @@ fn is_slash(b u8) bool {
 	return b == os.fslash
 }
 
-fn is_device_path(path string) bool {
+fn is_unc_path(path string) bool {
 	return win_volume_len(path) >= 5 && starts_w_slash_slash(path)
 }
 
diff --git a/vlib/os/filepath_test.v b/vlib/os/filepath_test.v
index 0c74b7633c..0d65d8738d 100644
--- a/vlib/os/filepath_test.v
+++ b/vlib/os/filepath_test.v
@@ -27,3 +27,103 @@ fn test_is_abs_path() {
 	assert !is_abs_path('./')
 	assert !is_abs_path('.')
 }
+
+fn test_clean_path() {
+	$if windows {
+		assert clean_path(r'\\path\to\files/file.v') == r'\path\to\files\file.v'
+		assert clean_path(r'\/\//\/') == '\\'
+		assert clean_path(r'./path\\dir/\\./\/\\/file.v\.\\\.') == r'path\dir\file.v'
+		assert clean_path(r'\./path/dir\\file.exe') == r'\path\dir\file.exe'
+		assert clean_path(r'.') == ''
+		assert clean_path(r'./') == ''
+		assert clean_path(r'\./') == '\\'
+		assert clean_path(r'//\/\/////') == '\\'
+		return
+	}
+	assert clean_path('./../.././././//') == '../..'
+	assert clean_path('.') == ''
+	assert clean_path('./path/to/file.v//./') == 'path/to/file.v'
+	assert clean_path('./') == ''
+	assert clean_path('/.') == '/'
+	assert clean_path('//path/./to/.///files/file.v///') == '/path/to/files/file.v'
+	assert clean_path('path/./to/.///files/.././file.v///') == 'path/to/files/../file.v'
+	assert clean_path('\\') == '\\'
+	assert clean_path('//////////') == '/'
+}
+
+fn test_norm_path() {
+	$if windows {
+		assert norm_path(r'C:/path/to//file.v\\') == r'C:\path\to\file.v'
+		assert norm_path(r'C:path\.\..\\\.\to//file.v') == r'C:to\file.v'
+		assert norm_path(r'D:path\.\..\..\\\\.\to//dir/..\') == r'D:..\to'
+		assert norm_path(r'D:/path\.\..\/..\file.v') == r'D:\file.v'
+		assert norm_path(r'') == '.'
+		assert norm_path(r'/') == '\\'
+		assert norm_path(r'\/') == '\\'
+		assert norm_path(r'path\../dir\..') == '.'
+		assert norm_path(r'.\.\') == '.'
+		assert norm_path(r'G:.\.\dir\././\.\.\\\\///to/././\file.v/./\\') == r'G:dir\to\file.v'
+		assert norm_path(r'G:\..\..\.\.\file.v\\\.') == r'G:\file.v'
+		assert norm_path(r'\\Server\share\\\dir/..\file.v\./.') == r'\\Server\share\file.v'
+		assert norm_path(r'\\.\device\\\dir/to/./file.v\.') == r'\\.\device\dir\to\file.v'
+		assert norm_path(r'C:dir/../dir2/../../../file.v') == r'C:..\..\file.v'
+		assert norm_path(r'\\.\C:\\\Users/\Documents//..') == r'\\.\C:\Users'
+		assert norm_path(r'\\.\C:\Users') == r'\\.\C:\Users'
+		assert norm_path(r'\\') == '\\'
+		assert norm_path(r'//') == '\\'
+		assert norm_path(r'\\\') == '\\'
+		assert norm_path(r'.') == '.'
+		assert norm_path(r'\\Server') == '\\Server'
+		assert norm_path(r'\\Server\') == '\\Server'
+		return
+	}
+	assert norm_path('/path/././../to/file//file.v/.') == '/to/file/file.v'
+	assert norm_path('path/././to/files/../../file.v/.') == 'path/file.v'
+	assert norm_path('path/././/../../to/file.v/.') == '../to/file.v'
+	assert norm_path('/path/././/../..///.././file.v/././') == '/file.v'
+	assert norm_path('path/././//../../../to/dir//.././file.v/././') == '../../to/file.v'
+	assert norm_path('path/../dir/..') == '.'
+	assert norm_path('../dir/..') == '..'
+	assert norm_path('/../dir/..') == '/'
+	assert norm_path('//././dir/../files/././/file.v') == '/files/file.v'
+	assert norm_path('/\\../dir/////////.') == '/\\../dir'
+	assert norm_path('/home/') == '/home'
+	assert norm_path('/home/////./.') == '/home'
+	assert norm_path('...') == '...'
+}
+
+fn test_abs_path() {
+	wd := getwd()
+	wd_w_sep := wd + path_separator
+	$if windows {
+		assert abs_path('path/to/file.v') == '${wd_w_sep}path\\to\\file.v'
+		assert abs_path('path/to/file.v') == '${wd_w_sep}path\\to\\file.v'
+		assert abs_path('/') == r'\'
+		assert abs_path(r'C:\path\to\files\file.v') == r'C:\path\to\files\file.v'
+		assert abs_path(r'C:\/\path\.\to\../files\file.v\.\\\.\') == r'C:\path\files\file.v'
+		assert abs_path(r'\\Host\share\files\..\..\.') == r'\\Host\share\'
+		assert abs_path(r'\\.\HardDiskvolume2\files\..\..\.') == r'\\.\HardDiskvolume2\'
+		assert abs_path(r'\\?\share') == r'\\?\share'
+		assert abs_path(r'\\.\') == r'\'
+		assert abs_path(r'G:/\..\\..\.\.\file.v\\.\.\\\\') == r'G:\file.v'
+		assert abs_path('files') == '${wd_w_sep}files'
+		assert abs_path('') == wd
+		assert abs_path('.') == wd
+		assert abs_path('files/../file.v') == '${wd_w_sep}file.v'
+		assert abs_path('///') == r'\'
+		assert abs_path('/path/to/file.v') == r'\path\to\file.v'
+		assert abs_path('D:/') == r'D:\'
+		assert abs_path(r'\\.\HardiskVolume6') == r'\\.\HardiskVolume6'
+		return
+	}
+	assert abs_path('/') == '/'
+	assert abs_path('.') == wd
+	assert abs_path('files') == '${wd_w_sep}files'
+	assert abs_path('') == wd
+	assert abs_path('files/../file.v') == '${wd_w_sep}file.v'
+	assert abs_path('///') == '/'
+	assert abs_path('/path/to/file.v') == '/path/to/file.v'
+	assert abs_path('/path/to/file.v/../..') == '/path'
+	assert abs_path('path/../file.v/..') == wd
+	assert abs_path('///') == '/'
+}
diff --git a/vlib/v/builder/builder.v b/vlib/v/builder/builder.v
index c635425043..ae6c23ae02 100644
--- a/vlib/v/builder/builder.v
+++ b/vlib/v/builder/builder.v
@@ -270,7 +270,7 @@ pub fn (b &Builder) import_graph() &depgraph.DepGraph {
 			deps << 'builtin'
 			if b.pref.backend == .c {
 				// TODO JavaScript backend doesn't handle os for now
-				if b.pref.is_vsh && p.mod.name !in ['os', 'dl'] {
+				if b.pref.is_vsh && p.mod.name !in ['os', 'dl', 'strings.textscanner'] {
 					deps << 'os'
 				}
 			}