CYG17 utf8 paths

Use cygwin_conv_path since cygwin-1.7 and support native utf-8 paths.

1 files changed, 218 insertions, 34 deletions

diff --git a/cygwin/cygwin.c b/cygwin/cygwin.c
index aa6938d279..0f5fe1d7fc 100644
--- a/cygwin/cygwin.c
+++ b/cygwin/cygwin.c
@@ -10,9 +10,13 @@
 #include <unistd.h>
 #include <process.h>
 #include <sys/cygwin.h>
+#include <cygwin/version.h>
 #include <mntent.h>
 #include <alloca.h>
 #include <dlfcn.h>
+#if (CYGWIN_VERSION_API_MINOR >= 181)
+#include <wchar.h>
+#endif
 
 /*
  * pp_system() implemented via spawn()
@@ -125,7 +129,7 @@ do_spawn (char *cmd)
 	    return do_spawnvp("sh",command);
 	}
 
-    Newx (PL_Argv,(s-cmd)/2+2,const char*);
+    Newx (PL_Argv,(s-cmd)/2+2,char*);
     PL_Cmd=savepvn (cmd,s-cmd);
     a=PL_Argv;
     for (s=PL_Cmd; *s;) {
@@ -143,6 +147,44 @@ do_spawn (char *cmd)
     return do_spawnvp(PL_Argv[0],(const char * const *)PL_Argv);
 }
 
+#if (CYGWIN_VERSION_API_MINOR >= 181)
+char*
+wide_to_utf8(const wchar_t *wbuf)
+{
+    char *buf;
+    int wlen = 0;
+    char *oldlocale = setlocale(LC_CTYPE, NULL);
+    setlocale(LC_CTYPE, "utf-8");
+
+    /* uvuni_to_utf8(buf, chr) or Encoding::_bytes_to_utf8(sv, "UCS-2BE"); */
+    wlen = wcsrtombs(NULL, (const wchar_t **)&wbuf, wlen, NULL);
+    buf = (char *) safemalloc(wlen+1);
+    wcsrtombs(buf, (const wchar_t **)&wbuf, wlen, NULL);
+
+    if (oldlocale) setlocale(LC_CTYPE, oldlocale);
+    else setlocale(LC_CTYPE, "C");
+    return buf;
+}
+
+wchar_t*
+utf8_to_wide(const char *buf)
+{
+    wchar_t *wbuf;
+    mbstate_t mbs;
+    char *oldlocale = setlocale(LC_CTYPE, NULL);
+    int wlen = sizeof(wchar_t)*strlen(buf);
+
+    setlocale(LC_CTYPE, "utf-8");
+    wbuf = (wchar_t *) safemalloc(wlen);
+    /* utf8_to_uvuni(pathname, wpath) or Encoding::_utf8_to_bytes(sv, "UCS-2BE"); */
+    wlen = mbsrtowcs(wbuf, (const char**)&buf, wlen, &mbs);
+
+    if (oldlocale) setlocale(LC_CTYPE, oldlocale);
+    else setlocale(LC_CTYPE, "C");
+    return wbuf;
+}
+#endif /* cygwin 1.7 */
+
 /* see also Cwd.pm */
 XS(Cygwin_cwd)
 {
@@ -194,7 +236,12 @@ XS(XS_Cygwin_winpid_to_pid)
 
     pid = (pid_t)SvIV(ST(0));
 
-    if ((RETVAL = cygwin32_winpid_to_pid(pid)) > 0) {
+#if (CYGWIN_VERSION_API_MINOR >= 181)
+    RETVAL = cygwin_winpid_to_pid(pid);
+#else
+    RETVAL = cygwin32_winpid_to_pid(pid);
+#endif
+    if (RETVAL > 0) {
         XSprePUSH; PUSHi((IV)RETVAL);
         XSRETURN(1);
     }
@@ -202,34 +249,91 @@ XS(XS_Cygwin_winpid_to_pid)
 }
 
 XS(XS_Cygwin_win_to_posix_path)
+
 {
     dXSARGS;
     int absolute_flag = 0;
     STRLEN len;
     int err;
-    char *pathname, *buf;
+    char *src_path;
+    char *posix_path;
+    int isutf8 = 0;
 
     if (items < 1 || items > 2)
         Perl_croak(aTHX_ "Usage: Cygwin::win_to_posix_path(pathname, [absolute])");
 
-    pathname = SvPV(ST(0), len);
+    src_path = SvPV(ST(0), len);
     if (items == 2)
 	absolute_flag = SvTRUE(ST(1));
 
     if (!len)
 	Perl_croak(aTHX_ "can't convert empty path");
-    buf = (char *) safemalloc (len + 260 + 1001);
+    isutf8 = SvUTF8(ST(0));
 
+#if (CYGWIN_VERSION_API_MINOR >= 181)
+    /* Check utf8 flag and use wide api then.
+       Size calculation: On overflow let cygwin_conv_path calculate the final size.
+     */
+    if (isutf8) {
+	int what = absolute_flag ? CCP_WIN_W_TO_POSIX : CCP_WIN_W_TO_POSIX | CCP_RELATIVE;
+	int wlen = sizeof(wchar_t)*(len + 260 + 1001);
+	wchar_t *wpath = (wchar_t *) safemalloc(sizeof(wchar_t)*len);
+	wchar_t *wbuf = (wchar_t *) safemalloc(wlen);
+	if (!IN_BYTES) {
+	    mbstate_t mbs;
+            char *oldlocale = setlocale(LC_CTYPE, NULL);
+            setlocale(LC_CTYPE, "utf-8");
+	    /* utf8_to_uvuni(src_path, wpath) or Encoding::_utf8_to_bytes(sv, "UCS-2BE"); */
+	    wlen = mbsrtowcs(wpath, (const char**)&src_path, wlen, &mbs);
+	    if (wlen > 0)
+		err = cygwin_conv_path(what, wpath, wbuf, wlen);
+            if (oldlocale) setlocale(LC_CTYPE, oldlocale);
+            else setlocale(LC_CTYPE, "C");
+	} else { /* use bytes; assume already ucs-2 encoded bytestream */
+	    err = cygwin_conv_path(what, src_path, wbuf, wlen);
+	}
+	if (err == ENOSPC) { /* our space assumption was wrong, not enough space */
+	    int newlen = cygwin_conv_path(what, wpath, wbuf, 0);
+	    wbuf = (wchar_t *) realloc(&wbuf, newlen);
+	    err = cygwin_conv_path(what, wpath, wbuf, newlen);
+	    wlen = newlen;
+	}
+	/* utf16_to_utf8(*p, *d, bytlen, *newlen) */
+	posix_path = (char *) safemalloc(wlen*3);
+	Perl_utf16_to_utf8(aTHX_ (U8*)&wpath, (U8*)posix_path, (I32)wlen*2, (I32*)&len);
+	/*
+	wlen = wcsrtombs(NULL, (const wchar_t **)&wbuf, wlen, NULL);
+	posix_path = (char *) safemalloc(wlen+1);
+	wcsrtombs(posix_path, (const wchar_t **)&wbuf, wlen, NULL);
+	*/
+    } else {
+	int what = absolute_flag ? CCP_WIN_A_TO_POSIX : CCP_WIN_A_TO_POSIX | CCP_RELATIVE;
+	posix_path = (char *) safemalloc (len + 260 + 1001);
+	err = cygwin_conv_path(what, src_path, posix_path, len + 260 + 1001);
+	if (err == ENOSPC) { /* our space assumption was wrong, not enough space */
+	    int newlen = cygwin_conv_path(what, src_path, posix_path, 0);
+	    posix_path = (char *) realloc(&posix_path, newlen);
+	    err = cygwin_conv_path(what, src_path, posix_path, newlen);
+	}
+    }
+#else
+    posix_path = (char *) safemalloc (len + 260 + 1001);
     if (absolute_flag)
-	err = cygwin_conv_to_full_posix_path(pathname, buf);
+	err = cygwin_conv_to_full_posix_path(src_path, posix_path);
     else
-	err = cygwin_conv_to_posix_path(pathname, buf);
+	err = cygwin_conv_to_posix_path(src_path, posix_path);
+#endif
     if (!err) {
-	ST(0) = sv_2mortal(newSVpv(buf, 0));
-	safefree(buf);
-       XSRETURN(1);
+	EXTEND(SP, 1);
+	ST(0) = sv_2mortal(newSVpv(posix_path, 0));
+	if (isutf8) { /* src was utf-8, so result should also */
+	    /* TODO: convert ANSI (local windows encoding) to utf-8 on cygwin-1.5 */
+	    SvUTF8_on(ST(0));
+	}
+	safefree(posix_path);
+        XSRETURN(1);
     } else {
-	safefree(buf);
+	safefree(posix_path);
 	XSRETURN_UNDEF;
     }
 }
@@ -240,29 +344,80 @@ XS(XS_Cygwin_posix_to_win_path)
     int absolute_flag = 0;
     STRLEN len;
     int err;
-    char *pathname, *buf;
+    char *src_path, *win_path;
+    int isutf8 = 0;
 
     if (items < 1 || items > 2)
         Perl_croak(aTHX_ "Usage: Cygwin::posix_to_win_path(pathname, [absolute])");
 
-    pathname = SvPV(ST(0), len);
+    src_path = SvPVx(ST(0), len);
     if (items == 2)
 	absolute_flag = SvTRUE(ST(1));
 
     if (!len)
 	Perl_croak(aTHX_ "can't convert empty path");
-    buf = (char *) safemalloc(len + 260 + 1001);
-
+    isutf8 = SvUTF8(ST(0));
+#if (CYGWIN_VERSION_API_MINOR >= 181)
+    /* Check utf8 flag and use wide api then.
+       Size calculation: On overflow let cygwin_conv_path calculate the final size.
+     */
+    if (isutf8) {
+	int what = absolute_flag ? CCP_POSIX_TO_WIN_W : CCP_POSIX_TO_WIN_W | CCP_RELATIVE;
+	int wlen = sizeof(wchar_t)*(len + 260 + 1001);
+	wchar_t *wpath = (wchar_t *) safemalloc(sizeof(wchar_t)*len);
+	wchar_t *wbuf = (wchar_t *) safemalloc(wlen);
+	char *oldlocale = setlocale(LC_CTYPE, NULL);
+	setlocale(LC_CTYPE, "utf-8");
+	if (!IN_BYTES) {
+	    mbstate_t mbs;
+	    /* utf8_to_uvuni(src_path, wpath) or Encoding::_utf8_to_bytes(sv, "UCS-2BE"); */
+	    wlen = mbsrtowcs(wpath, (const char**)&src_path, wlen, &mbs);
+	    if (wlen > 0)
+		err = cygwin_conv_path(what, wpath, wbuf, wlen);
+	} else { /* use bytes; assume already ucs-2 encoded bytestream */
+	    err = cygwin_conv_path(what, src_path, wbuf, wlen);
+	}
+	if (err == ENOSPC) { /* our space assumption was wrong, not enough space */
+	    int newlen = cygwin_conv_path(what, wpath, wbuf, 0);
+	    wbuf = (wchar_t *) realloc(&wbuf, newlen);
+	    err = cygwin_conv_path(what, wpath, wbuf, newlen);
+	    wlen = newlen;
+	}
+	/* also see utf8.c: Perl_utf16_to_utf8() or Encoding::_bytes_to_utf8(sv, "UCS-2BE"); */
+	wlen = wcsrtombs(NULL, (const wchar_t **)&wbuf, wlen, NULL);
+	win_path = (char *) safemalloc(wlen+1);
+	wcsrtombs(win_path, (const wchar_t **)&wbuf, wlen, NULL);
+	if (oldlocale) setlocale(LC_CTYPE, oldlocale);
+	else setlocale(LC_CTYPE, "C");
+    } else {
+	int what = absolute_flag ? CCP_POSIX_TO_WIN_A : CCP_POSIX_TO_WIN_A | CCP_RELATIVE;
+	win_path = (char *) safemalloc(len + 260 + 1001);
+	err = cygwin_conv_path(what, src_path, win_path, len + 260 + 1001);
+	if (err == ENOSPC) { /* our space assumption was wrong, not enough space */
+	    int newlen = cygwin_conv_path(what, src_path, win_path, 0);
+	    win_path = (char *) realloc(&win_path, newlen);
+	    err = cygwin_conv_path(what, src_path, win_path, newlen);
+	}
+    }
+#else
+    if (isutf8)
+	Perl_warn(aTHX_ "can't convert utf8 path");
+    win_path = (char *) safemalloc(len + 260 + 1001);
     if (absolute_flag)
-	err = cygwin_conv_to_full_win32_path(pathname, buf);
+	err = cygwin_conv_to_full_win32_path(src_path, win_path);
     else
-	err = cygwin_conv_to_win32_path(pathname, buf);
+	err = cygwin_conv_to_win32_path(src_path, win_path);
+#endif
     if (!err) {
-	ST(0) = sv_2mortal(newSVpv(buf, 0));
-	safefree(buf);
-       XSRETURN(1);
+	EXTEND(SP, 1);
+	ST(0) = sv_2mortal(newSVpv(win_path, 0));
+	if (isutf8) {
+	    SvUTF8_on(ST(0));
+	}
+	safefree(win_path);
+	XSRETURN(1);
     } else {
-	safefree(buf);
+	safefree(win_path);
 	XSRETURN_UNDEF;
     }
 }
@@ -293,24 +448,22 @@ XS(XS_Cygwin_mount_flags)
 {
     dXSARGS;
     char *pathname;
-    char flags[260];
+    char flags[PATH_MAX];
+    flags[0] = '\0';
 
     if (items != 1)
-        Perl_croak(aTHX_ "Usage: Cygwin::mount_flags(mnt_dir|'/cygwin')");
+        Perl_croak(aTHX_ "Usage: Cygwin::mount_flags( mnt_dir | '/cygdrive' )");
 
     pathname = SvPV_nolen(ST(0));
 
-    /* TODO: Check for cygdrive registry setting,
-     *       and then use CW_GET_CYGDRIVE_INFO
-     */
     if (!strcmp(pathname, "/cygdrive")) {
-	char user[260];
-	char system[260];
-	char user_flags[260];
-	char system_flags[260];
+	char user[PATH_MAX];
+	char system[PATH_MAX];
+	char user_flags[PATH_MAX];
+	char system_flags[PATH_MAX];
 
-	cygwin_internal (CW_GET_CYGDRIVE_INFO, user, system, user_flags,
-			 system_flags);
+	cygwin_internal (CW_GET_CYGDRIVE_INFO, user, system,
+			 user_flags, system_flags);
 
         if (strlen(user) > 0) {
             sprintf(flags, "%s,cygdrive,%s", user_flags, user);
@@ -323,6 +476,7 @@ XS(XS_Cygwin_mount_flags)
 
     } else {
 	struct mntent *mnt;
+	int found = 0;
 	setmntent (0, 0);
 	while ((mnt = getmntent (0))) {
 	    if (!strcmp(pathname, mnt->mnt_dir)) {
@@ -331,12 +485,42 @@ XS(XS_Cygwin_mount_flags)
 		    strcat(flags, ",");
 		    strcat(flags, mnt->mnt_opts);
 		}
+		found++;
 		break;
 	    }
 	}
 	endmntent (0);
-	ST(0) = sv_2mortal(newSVpv(flags, 0));
-	XSRETURN(1);
+
+	/* Check if arg is the current volume moint point if not default,
+	 * and then use CW_GET_CYGDRIVE_INFO also.
+	 */
+	if (!found) {
+	    char user[PATH_MAX];
+	    char system[PATH_MAX];
+	    char user_flags[PATH_MAX];
+	    char system_flags[PATH_MAX];
+
+	    cygwin_internal (CW_GET_CYGDRIVE_INFO, user, system,
+			     user_flags, system_flags);
+
+	    if (strlen(user) > 0) {
+		if (strcmp(user,pathname)) {
+		    sprintf(flags, "%s,cygdrive,%s", user_flags, user);
+		    found++;
+		}
+	    } else {
+		if (strcmp(user,pathname)) {
+		    sprintf(flags, "%s,cygdrive,%s", system_flags, system);
+		    found++;
+		}
+	    }
+	}
+	if (found) {
+	    ST(0) = sv_2mortal(newSVpv(flags, 0));
+	    XSRETURN(1);
+	} else {
+	    XSRETURN_UNDEF;
+	}
     }
 }