Simplify a string conversion WCS <==> MBS to stop using mbsnrtowcs and wcsnrtombs.

diff --git a/CMakeLists.txt b/CMakeLists.txt
index d16818ba9ed3897da46bccf856f052f6448fc26d..1714df0191417b7748404e47fa01f9be0699ec7a 100644 (file)
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -665,7 +665,6 @@ CHECK_FUNCTION_EXISTS_GLIBC(localtime_r HAVE_LOCALTIME_R)
 CHECK_FUNCTION_EXISTS_GLIBC(lstat HAVE_LSTAT)
 CHECK_FUNCTION_EXISTS_GLIBC(lutimes HAVE_LUTIMES)
 CHECK_FUNCTION_EXISTS_GLIBC(mbrtowc HAVE_MBRTOWC)
-CHECK_FUNCTION_EXISTS_GLIBC(mbsnrtowcs HAVE_MBSNRTOWCS)
 CHECK_FUNCTION_EXISTS_GLIBC(memmove HAVE_MEMMOVE)
 CHECK_FUNCTION_EXISTS_GLIBC(mkdir HAVE_MKDIR)
 CHECK_FUNCTION_EXISTS_GLIBC(mkfifo HAVE_MKFIFO)
@@ -699,7 +698,6 @@ CHECK_FUNCTION_EXISTS_GLIBC(wcrtomb HAVE_WCRTOMB)
 CHECK_FUNCTION_EXISTS_GLIBC(wcscmp HAVE_WCSCMP)
 CHECK_FUNCTION_EXISTS_GLIBC(wcscpy HAVE_WCSCPY)
 CHECK_FUNCTION_EXISTS_GLIBC(wcslen HAVE_WCSLEN)
-CHECK_FUNCTION_EXISTS_GLIBC(wcsnrtombs HAVE_WCSNRTOMBS)
 CHECK_FUNCTION_EXISTS_GLIBC(wctomb HAVE_WCTOMB)
 CHECK_FUNCTION_EXISTS_GLIBC(_ctime64_s HAVE__CTIME64_S)
 CHECK_FUNCTION_EXISTS_GLIBC(_fseeki64 HAVE__FSEEKI64)

diff --git a/build/cmake/config.h.in b/build/cmake/config.h.in
index 5ccf3681d310ccdc9ee7fe80ce0455de828328a7..4f1463c574805aa28103723dd9902c2d7af52a18 100644 (file)
--- a/build/cmake/config.h.in
+++ b/build/cmake/config.h.in
@@ -664,9 +664,6 @@ typedef uint64_t uintmax_t;
 /* Define to 1 if you have the `mbrtowc' function. */
 #cmakedefine HAVE_MBRTOWC 1
 
-/* Define to 1 if you have the `mbsnrtowcs' function. */
-#cmakedefine HAVE_MBSNRTOWCS 1
-
 /* Define to 1 if you have the `memmove' function. */
 #cmakedefine HAVE_MEMMOVE 1
 
@@ -949,9 +946,6 @@ typedef uint64_t uintmax_t;
 /* Define to 1 if you have the `wcslen' function. */
 #cmakedefine HAVE_WCSLEN 1
 
-/* Define to 1 if you have the `wcsnrtombs' function. */
-#cmakedefine HAVE_WCSNRTOMBS 1
-
 /* Define to 1 if you have the `wctomb' function. */
 #cmakedefine HAVE_WCTOMB 1
 

diff --git a/configure.ac b/configure.ac
index 4108cb54eca7d60fe0c51edddbd53c27536868d2..c15d432720eaf8060aa87f481b6b1b17f6822f96 100644 (file)
--- a/configure.ac
+++ b/configure.ac
@@ -437,13 +437,13 @@ AC_CHECK_FUNCS([futimens futimes futimesat])
 AC_CHECK_FUNCS([geteuid getpid getgrgid_r getgrnam_r])
 AC_CHECK_FUNCS([getpwnam_r getpwuid_r getvfsbyname gmtime_r])
 AC_CHECK_FUNCS([lchflags lchmod lchown link localtime_r lstat lutimes])
-AC_CHECK_FUNCS([mbrtowc mbsnrtowcs memmove memset])
+AC_CHECK_FUNCS([mbrtowc memmove memset])
 AC_CHECK_FUNCS([mkdir mkfifo mknod mkstemp])
 AC_CHECK_FUNCS([nl_langinfo openat pipe poll readlink readlinkat])
 AC_CHECK_FUNCS([select setenv setlocale sigaction statfs statvfs])
 AC_CHECK_FUNCS([strchr strdup strerror strncpy_s strrchr symlink timegm])
 AC_CHECK_FUNCS([tzset unsetenv utime utimensat utimes vfork])
-AC_CHECK_FUNCS([wcrtomb wcscmp wcscpy wcslen wcsnrtombs wctomb wmemcmp wmemcpy])
+AC_CHECK_FUNCS([wcrtomb wcscmp wcscpy wcslen wctomb wmemcmp wmemcpy])
 AC_CHECK_FUNCS([_ctime64_s _fseeki64])
 AC_CHECK_FUNCS([_get_timezone _localtime64_s _mkgmtime64])
 # detects cygwin-1.7, as opposed to older versions

diff --git a/libarchive/archive_string.c b/libarchive/archive_string.c
index b168e28da350df19967b4f58bbe0c2fa36ab93d6..37907433362ed9d1d90cd3f4daa6d221c18c5930 100644 (file)
--- a/libarchive/archive_string.c
+++ b/libarchive/archive_string.c
@@ -586,41 +586,6 @@ archive_wstring_append_from_mbs_in_codepage(struct archive_wstring *dest,
        return (ret);
 }
 
-#elif defined(HAVE_MBSNRTOWCS)
-
-/*
- * Convert MBS to WCS.
- * Note: returns -1 if conversion fails.
- */
-int
-archive_wstring_append_from_mbs(struct archive_wstring *dest,
-    const char *p, size_t len)
-{
-       size_t r;
-       /*
-        * No single byte will be more than one wide character,
-        * so this length estimate will always be big enough.
-        */
-       size_t wcs_length = len;
-       size_t mbs_length = len;
-       const char *mbs = p;
-       wchar_t *wcs;
-       mbstate_t shift_state;
-
-       memset(&shift_state, 0, sizeof(shift_state));
-       if (NULL == archive_wstring_ensure(dest, dest->length + wcs_length + 1))
-               return (-1);
-       wcs = dest->s + dest->length;
-       r = mbsnrtowcs(wcs, &mbs, mbs_length, wcs_length, &shift_state);
-       if (r != (size_t)-1) {
-               dest->length += r;
-               dest->s[dest->length] = L'\0';
-               return (0);
-       }
-       dest->s[dest->length] = L'\0';
-       return (-1);
-}
-
 #else
 
 /*
@@ -632,6 +597,7 @@ archive_wstring_append_from_mbs(struct archive_wstring *dest,
     const char *p, size_t len)
 {
        size_t r;
+       int ret_val = 0;
        /*
         * No single byte will be more than one wide character,
         * so this length estimate will always be big enough.
@@ -653,15 +619,29 @@ archive_wstring_append_from_mbs(struct archive_wstring *dest,
         * extra MBS when strlen(p) > len and one wide character consis of
         * multi bytes.
         */
-       while (wcs_length > 0 && *mbs && mbs_length > 0) {
+       while (*mbs && mbs_length > 0) {
+               if (wcs_length == 0) {
+                       dest->length = wcs - dest->s;
+                       dest->s[dest->length] = L'\0';
+                       wcs_length = mbs_length;
+                       if (NULL == archive_wstring_ensure(dest,
+                           dest->length + wcs_length + 1))
+                               return (-1);
+                       wcs = dest->s + dest->length;
+               }
 #if HAVE_MBRTOWC
                r = mbrtowc(wcs, mbs, wcs_length, &shift_state);
 #else
                r = mbtowc(wcs, mbs, wcs_length);
 #endif
                if (r == (size_t)-1 || r == (size_t)-2) {
-                       dest->s[dest->length] = L'\0';
-                       return (-1);
+                       ret_val = -1;
+                       if (errno == EILSEQ) {
+                               ++mbs;
+                               --mbs_length;
+                               continue;
+                       } else
+                               break;
                }
                if (r == 0 || r > mbs_length)
                        break;
@@ -672,7 +652,7 @@ archive_wstring_append_from_mbs(struct archive_wstring *dest,
        }
        dest->length = wcs - dest->s;
        dest->s[dest->length] = L'\0';
-       return (0);
+       return (ret_val);
 }
 
 #endif
@@ -784,73 +764,6 @@ archive_string_append_from_wcs_in_codepage(struct archive_string *as,
        return (defchar_used?-1:ret);
 }
 
-#elif defined(HAVE_WCSNRTOMBS)
-
-/*
- * Translates a wide character string into current locale character set
- * and appends to the archive_string.  Note: returns -1 if conversion
- * fails.
- */
-int
-archive_string_append_from_wcs(struct archive_string *as,
-    const wchar_t *w, size_t len)
-{
-       mbstate_t shift_state;
-       size_t r, ndest, nwc;
-       char *dest;
-       const wchar_t *wp, *wpp;
-       int ret_val = 0;
-
-       wp = w;
-       nwc = len;
-       ndest = len * 2;
-       /* Initialize the shift state. */
-       memset(&shift_state, 0, sizeof(shift_state));
-       while (nwc > 0) {
-               /* Allocate buffer for MBS. */
-               if (archive_string_ensure(as, as->length + ndest + 1) == NULL)
-                       return (-1);
-
-               dest = as->s + as->length;
-               wpp = wp;
-               r = wcsnrtombs(dest, &wp, nwc,
-                   as->buffer_length - as->length -1,
-                   &shift_state);
-               if (r == (size_t)-1) {
-                       if (errno == EILSEQ) {
-                               /* Retry conversion just for safe WCS. */
-                               size_t xwc = wp - wpp;
-                               wp = wpp;
-                               r = wcsnrtombs(dest, &wp, xwc,
-                                   as->buffer_length - as->length -1,
-                                   &shift_state);
-                               if (r == (size_t)-1)
-                                       /* This would not happen. */
-                                       return (-1);
-                               as->length += r;
-                               nwc -= wp - wpp;
-                               /* Skip an illegal wide char. */
-                               as->s[as->length++] = '?';
-                               wp++;
-                               nwc--;
-                               ret_val = -1;
-                               continue;
-                       } else {
-                               ret_val = -1;
-                               break;
-                       }
-               }
-               as->length += r;
-               if (wp == NULL || (wp - wpp) >= nwc)
-                       break;
-               /* Get a remaining WCS lenth. */
-               nwc -= wp - wpp;
-       }
-       /* All wide characters are translated to MBS. */
-       as->s[as->length] = '\0';
-       return (ret_val);
-}
-
 #elif defined(HAVE_WCTOMB) || defined(HAVE_WCRTOMB)
 
 /*
@@ -896,7 +809,7 @@ archive_string_append_from_wcs(struct archive_string *as,
                        /* Re-allocate buffer for MBS. */
                        if (archive_string_ensure(as,
                            as->length + len * 2 + 1) == NULL)
-                               __archive_errx(1, "Out of memory");
+                               return (-1);
                        p = as->s + as->length;
                        end = as->s + as->buffer_length - MB_CUR_MAX -1;
                }