Merge branch 'cygpath'

This topic branch adds the `cygpath` applet that closely imitates
MSYS2's tool of the same name.

Further, the `PATH` variable is special-cased to get the `cygpath`
treatment automagically: whenever a user sets it to a Unix-style path
list, we now try to convert that automatically to a Windows-style path
list (albeit with forward-slashes).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>

Author: dscho

configs/mingw32_defconfig

@@ -55,6 +55,7 @@ CONFIG_FEATURE_EURO=y
 CONFIG_SKIP_ANSI_EMULATION_DEFAULT=2
 CONFIG_FEATURE_IMPROVED_COLOUR_MAPPING=y
 CONFIG_FEATURE_EXTRA_FILE_DATA=y
+CONFIG_CYGPATH=y
 
 #
 # Build Options

configs/mingw64_defconfig

@@ -55,6 +55,7 @@ CONFIG_FEATURE_EURO=y
 CONFIG_SKIP_ANSI_EMULATION_DEFAULT=2
 CONFIG_FEATURE_IMPROVED_COLOUR_MAPPING=y
 CONFIG_FEATURE_EXTRA_FILE_DATA=y
+CONFIG_CYGPATH=y
 
 #
 # Build Options

libbb/appletlib.c

@@ -31,6 +31,7 @@
 static inline int *get_perrno(void) { return &errno; }
 
 #include "busybox.h"
+#include "path-convert.h"
 
 #if !(defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__) \
     || defined(__APPLE__) \
@@ -1225,6 +1226,22 @@ get_script_content(unsigned n UNUSED_PARAM)
 
 #endif /* defined(SINGLE_APPLET_MAIN) */
 
+#if ENABLE_PLATFORM_MINGW32
+static char *xwcstoutf(wchar_t *wcs)
+{
+	DWORD size = WideCharToMultiByte(CP_UTF8, 0, wcs, -1, NULL, 0, NULL, NULL) + 1;
+	char *buf;
+
+	if (!size)
+		bb_error_msg_and_die("could not convert '%ls' to UTF-8", wcs);
+	buf = xmalloc(size);
+	if (WideCharToMultiByte(CP_UTF8, 0, wcs, -1, buf, size, NULL, NULL))
+		return buf;
+	free(buf);
+	bb_error_msg_and_die("could not convert '%ls' to UTF-8", wcs);
+}
+#endif
+
 #if ENABLE_BUILD_LIBBUSYBOX
 int lbb_main(char **argv)
 #else
@@ -1333,19 +1350,29 @@ int main(int argc UNUSED_PARAM, char **argv)
 			/* Manually convert non-ASCII environment entries from UTF-16 */
 			for (i = 0; wenv[i]; i++) {
 				for (p = wenv + i; *p; p++)
-					if (*p & ~0x7f) {
+					if (!_wcsnicmp(wenv + i, L"PATH=", 5)) {
+						char *orig = xwcstoutf(wenv + i + 5);
+						char *converted = path_convert_path_list(orig, PATH_CONVERT_MIXED);
+						size_t len;
+
+						if (!converted)
+							bb_error_msg_and_die("could not convert path list '%s'", orig);
+
+						len = strlen(converted);
+						converted = xrealloc(converted, len + 6);
+						memmove(converted + 5, converted, len + 1);
+						memcpy(converted, "PATH=", 5);
+
+						putenv(converted);
+
+						free(converted);
+						free(orig);
+						break;
+					} else if (*p & ~0x7f) {
 						/* Non-ASCII name or value */
-						DWORD size = WideCharToMultiByte(CP_UTF8, 0,
-								wenv + i, -1, NULL, 0, NULL, NULL) + 1;
-						char *buf;
-
-						if (!size)
-							break;
-						buf = malloc(size);
-						if (!buf)
-							bb_error_msg_and_die("Out of memory");
-						if (WideCharToMultiByte(CP_UTF8, 0,
-								wenv + i, -1, buf, size, NULL, NULL))
+						char *buf = xwcstoutf(wenv + i);
+
+						if (buf)
 							/* if we could not convert, punt */
 							putenv(buf);
 						free(buf);

miscutils/cygpath.c

@@ -0,0 +1,86 @@
+/*
+ * Convert MSYS-style paths to Win32-style ones
+ *
+ * Copyright (C) 2022 Johannes Schindelin <johannes.schindelin@gmx.de>
+ *
+ * Licensed under GPLv2, see file LICENSE in this source tree.
+ */
+//config:config CYGPATH
+//config:	bool "cygpath"
+//config:	default y
+//config:	depends on PLATFORM_MINGW32
+//config:	help
+//config:	Convert Unix and Windows format paths
+
+//applet:IF_CYGPATH(APPLET_NOEXEC(cygpath, cygpath, BB_DIR_USR_BIN, BB_SUID_DROP, cygpath))
+
+//kbuild:lib-$(CONFIG_CYGPATH) += cygpath.o
+
+//usage:#define cygpath_trivial_usage
+//usage:       "ARG|ARGS"
+//usage:#define cygpath_full_usage "\n\n"
+//usage:       "Convert Unix and Windows format paths"
+
+#include "libbb.h"
+#include "path-convert.h"
+
+enum {
+    OPT_absolute  = (1 << 0),
+    OPT_mixed     = (1 << 1),
+    OPT_unix      = (1 << 2),
+    OPT_windows   = (1 << 3),
+    OPT_path_list = (1 << 4),
+};
+
+int cygpath_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
+int cygpath_main(int argc UNUSED_PARAM, char **argv)
+{
+    int i;
+    enum path_convert_flag flags = 0;
+#if ENABLE_LONG_OPTS
+    static const char cygpath_longopts[] ALIGN1 =
+        "absolute\0" No_argument "a"
+        "mixed\0" No_argument "m"
+        "unix\0" No_argument "u"
+        "windows\0" No_argument "w"
+        "path\0" No_argument "p"
+        ;
+#endif
+    int opt = getopt32long(argv, "amuwp", cygpath_longopts);
+    argv += optind;
+    argc -= optind;
+
+    if (opt & OPT_absolute)
+        flags |= PATH_CONVERT_ABSOLUTE;
+    if (opt & OPT_mixed)
+        flags |= PATH_CONVERT_MIXED;
+    if (opt & OPT_unix)
+        flags |= PATH_CONVERT_UNIX;
+    if (opt & OPT_windows)
+        flags |= PATH_CONVERT_WINDOWS;
+
+    for (i = 0; i < argc; i++) {
+        char *to_free = NULL;
+        const char *path = argv[i], *result;
+        char buffer[PATH_MAX_LONG];
+
+        if (!*argv[i]) {
+            bb_error_msg("can't convert empty path");
+            return EXIT_FAILURE;
+        }
+
+        if (opt & OPT_path_list)
+            result = to_free = path_convert_path_list(path, flags);
+        else
+            result = path_convert(path, buffer, sizeof(buffer), flags);
+
+        if (!result)
+            return EXIT_FAILURE;
+
+        printf("%s\n", result);
+
+        free(to_free);
+    }
+
+    return EXIT_SUCCESS;
+}

shell/ash.c

@@ -317,6 +317,10 @@ typedef long arith_t;
 # define CLEAR_RANDOM_T(rnd) ((void)0)
 #endif
 
+#if ENABLE_PLATFORM_MINGW32
+# include "path-convert.h"
+#endif
+
 #include "NUM_APPLETS.h"
 #if NUM_APPLETS == 1
 /* STANDALONE does not make sense, and won't compile */
@@ -2624,34 +2628,19 @@ bltinlookup(const char *name)
 static char *
 fix_pathvar(const char *path, int len)
 {
-	char *newpath = xstrdup(path);
-	char *p;
-	int modified = FALSE;
-
-	p = newpath + len;
-	while (*p) {
-		if (*p != ':' && *p != ';') {
-			/* skip drive */
-			if (isalpha(*p) && p[1] == ':')
-				p += 2;
-			/* skip through path component */
-			for (; *p != '\0' && *p != ':' && *p != ';'; ++p)
-				continue;
-		}
-		/* *p is ':', ';' or '\0' here */
-		if (*p == ':') {
-			*p++ = ';';
-			modified = TRUE;
-		}
-		else if (*p == ';') {
-			++p;
-		}
-	}
+	char *newpath = path_convert_path_list(path + len, PATH_CONVERT_MIXED);
+	size_t len2;
 
-	if (!modified) {
+	if (!strcmp(path + len, newpath)) {
 		free(newpath);
-		newpath = NULL;
+		return NULL;
 	}
+
+	len2 = strlen(newpath);
+	newpath = xrealloc(newpath, len + len2 + 1);
+	memmove(newpath + len, newpath, len2 + 1);
+	memcpy(newpath, path, len);
+
 	return newpath;
 }
 

testsuite/cygpath.tests

@@ -0,0 +1,54 @@
+#!/bin/sh
+set -x
+
+# Copyright 2022 by Johannes Schindelin <johannes.schindelin@gmx.de>
+# Licensed under GPLv2, see file LICENSE in this source tree.
+
+. ./testing.sh
+
+# testing "description" "command" "result" "infile" "stdin"
+pseudo_root="$(cygpath -am /)"
+res=$?
+case "$pseudo_root" in */) s="ends in a slash";; *) s=;; esac
+
+testing "pseudo root" 'echo $res,$s,"$pseudo_root"' "0,,$pseudo_root
+" "" ""
+
+testing "cygpath -aw" "cygpath -aw /c/ /c/123 /c/324" 'C:\\
+C:\\123
+C:\\324
+' "" ""
+
+testing "cygpath -am" "cygpath -am /c/ /c/123 /c/324 / /lib" "C:/
+C:/123
+C:/324
+$pseudo_root
+${pseudo_root}/lib
+" "" ""
+
+testing "cygpath -u" 'cygpath -u C:\\ abc' '/c/
+abc
+' "" ""
+
+testing "cygpath and dots" 'cygpath -am . ./. ../. ./..' "$PWD
+$PWD
+${PWD%/*}
+${PWD%/*}
+" "" ""
+
+testing "cygpath ''" '! cygpath "" abc 2>&1' "cygpath: can't convert empty path
+" "" ""
+
+testing "cygpath -p" 'cygpath -pam .:./.:/c/' "$PWD;$PWD;C:/
+" "" ""
+
+testing "cygpath -p" 'cygpath -pam .:' "$PWD;$PWD
+" "" ""
+
+testing "cygpath -u" "cygpath -pau \"$pseudo_root;${pseudo_root}/lib\"" "/:/lib
+" "" ""
+
+testing "cygpath -m <unix-style-path>" "cygpath -m /lib" "$pseudo_root/lib
+" "" ""
+
+exit $FAILCOUNT

win32/Kbuild

@@ -16,6 +16,7 @@ lib-$(CONFIG_PLATFORM_MINGW32) += process.o
 lib-$(CONFIG_PLATFORM_MINGW32) += match_class.o
 lib-$(CONFIG_PLATFORM_MINGW32) += mntent.o
 lib-$(CONFIG_PLATFORM_MINGW32) += net.o
+lib-$(CONFIG_PLATFORM_MINGW32) += path-convert.o
 lib-$(CONFIG_PLATFORM_MINGW32) += poll.o
 lib-$(CONFIG_PLATFORM_MINGW32) += popen.o
 lib-$(CONFIG_PLATFORM_MINGW32) += regex.o

win32/env.c

@@ -1,4 +1,5 @@
 #include "libbb.h"
+#include "path-convert.h"
 
 #undef getenv
 #undef putenv
@@ -18,16 +19,20 @@ char *mingw_getenv(const char *name)
 int setenv(const char *name, const char *value, int replace)
 {
 	int out;
-	char *envstr;
+	char *envstr, *to_free = NULL;
 
 	if (!name || !*name || strchr(name, '=') || !value) return -1;
 	if (!replace) {
 		if (getenv(name)) return 0;
 	}
 
+	if (!strcmp(name, "PATH"))
+		value = to_free = path_convert_path_list(value, PATH_CONVERT_WINDOWS);
+
 	envstr = xasprintf("%s=%s", name, value);
 	out = mingw_putenv(envstr);
 	free(envstr);
+	free(to_free);
 
 	return out;
 }

win32/mingw.c

@@ -303,7 +303,7 @@ wchar_t *mingw_pathconv(const char *path)
 			pseudo_root[pseudo_root_len++] = L'\\';
 	}
 
-	memcpy(result, pseudo_root, pseudo_root_len * sizeof(wchar_t));
+	memcpy(result, pseudo_root, (pseudo_root_len - !path[1]) * sizeof(wchar_t));
 	return pathconv_rest(result, pseudo_root_len, path + 1);
 }