Bug#5439 : mysql_server_init() crashes if ShiftJIS path is passed

(important for Adobe).

mf_pack.c, mf_dirname.c, charset.c, my_sys.h:

  - adding fs_character_set() function on Windows
  - ignoring fake slashes which are just multibyte
    parts in several functions in /mysys

Verified by Shu to work on WinXP and Win2k.
Test is not possible, or very hard to do.


include/my_sys.h:
  Bug#5439 : mysql_server_init() crashes if ShiftJIS path is passed
  (important for Adobe)
  - adding fs_character_set() function on Windows
  - ignoring fake slashes which are just multibyte
    parts in several functions in /mysys
mysys/charset.c:
  d
mysys/mf_dirname.c:
  d
mysys/mf_pack.c:
  d

4 files changed, 75 insertions, 1 deletions

diff --git a/include/my_sys.h b/include/my_sys.h
index e56f07a8140..8752aa30772 100644
--- a/include/my_sys.h
+++ b/include/my_sys.h
@@ -788,6 +788,11 @@ extern my_bool init_compiled_charsets(myf flags);
 extern void add_compiled_collation(CHARSET_INFO *cs);
 extern ulong escape_string_for_mysql(CHARSET_INFO *charset_info, char *to,
                                      const char *from, ulong length);
+#ifdef __WIN__
+#define BACKSLASH_MBTAIL
+/* File system character set */
+extern CHARSET_INFO *fs_character_set(void);
+#endif
 
 #ifdef __WIN__
 extern my_bool have_tcpip;		/* Is set if tcpip is used */
diff --git a/mysys/charset.c b/mysys/charset.c
index cabdbad3413..3a39fce9437 100644
--- a/mysys/charset.c
+++ b/mysys/charset.c
@@ -637,3 +637,29 @@ ulong escape_string_for_mysql(CHARSET_INFO *charset_info, char *to,
   *to= 0;
   return (ulong) (to - to_start);
 }
+
+
+#ifdef BACKSLASH_MBTAIL
+static CHARSET_INFO *fs_cset_cache= NULL;
+
+CHARSET_INFO *fs_character_set()
+{
+  if (!fs_cset_cache)
+  {
+    char buf[10]= "cp";
+    GetLocaleInfo(LOCALE_SYSTEM_DEFAULT, LOCALE_IDEFAULTANSICODEPAGE,
+                  buf+2, sizeof(buf)-3);
+    /*
+      We cannot call get_charset_by_name here
+      because fs_character_set() is executed before
+      LOCK_THD_charset mutex initialization, which
+      is used inside get_charset_by_name.
+      As we're now interested in cp932 only,
+      let's just detect it using strcmp().
+    */
+    fs_cset_cache= !strcmp(buf, "cp932") ?
+                   &my_charset_cp932_japanese_ci : &my_charset_bin;
+  }
+  return fs_cset_cache;
+}
+#endif
diff --git a/mysys/mf_dirname.c b/mysys/mf_dirname.c
index 3de82c05b87..45bf4d56c31 100644
--- a/mysys/mf_dirname.c
+++ b/mysys/mf_dirname.c
@@ -22,6 +22,9 @@
 uint dirname_length(const char *name)
 {
   register my_string pos,gpos;
+#ifdef BASKSLASH_MBTAIL
+  CHARSET_INFO *fs= fs_character_set();
+#endif
 #ifdef FN_DEVCHAR
   if ((pos=(char*)strrchr(name,FN_DEVCHAR)) == 0)
 #endif
@@ -29,12 +32,22 @@ uint dirname_length(const char *name)
 
   gpos= pos++;
   for ( ; *pos ; pos++)				/* Find last FN_LIBCHAR */
+  {
+#ifdef BASKSLASH_MBTAIL
+    uint l;
+    if (use_mb(fs) && (l= my_ismbchar(fs, pos, pos + 3)))
+    {
+      pos+= l - 1;
+      continue;
+    }
+#endif
     if (*pos == FN_LIBCHAR || *pos == '/'
 #ifdef FN_C_AFTER_DIR
 	|| *pos == FN_C_AFTER_DIR || *pos == FN_C_AFTER_DIR_2
 #endif
 	)
       gpos=pos;
+  }
   return ((uint) (uint) (gpos+1-(char*) name));
 }
 
@@ -85,6 +98,9 @@ uint dirname_part(my_string to, const char *name)
 char *convert_dirname(char *to, const char *from, const char *from_end)
 {
   char *to_org=to;
+#ifdef BACKSLASH_MBTAIL
+  CHARSET_INFO *fs= fs_character_set();
+#endif
 
   /* We use -2 here, becasue we need place for the last FN_LIBCHAR */
   if (!from_end || (from_end - from) > FN_REFLEN-2)
@@ -103,7 +119,22 @@ char *convert_dirname(char *to, const char *from, const char *from_end)
 	*to++= FN_C_AFTER_DIR;
 #endif
       else
-	*to++= *from;
+      {
+#ifdef BACKSLASH_MBTAIL
+        uint l;
+        if (use_mb(fs) && (l= my_ismbchar(fs, from, from + 3)))
+        {
+          memmove(to, from, l);
+          to+= l;
+          from+= l - 1;
+          to_org= to; /* Don't look inside mbchar */
+        }
+        else
+#endif
+        {
+          *to++= *from;
+        }
+      }
     }
     *to=0;
   }
diff --git a/mysys/mf_pack.c b/mysys/mf_pack.c
index 86172f648f4..049aa59a578 100644
--- a/mysys/mf_pack.c
+++ b/mysys/mf_pack.c
@@ -124,6 +124,9 @@ uint cleanup_dirname(register my_string to, const char *from)
   reg4 my_string start;
   char parent[5],				/* for "FN_PARENTDIR" */
        buff[FN_REFLEN+1],*end_parentdir;
+#ifdef BACKSLASH_MBTAIL
+  CHARSET_INFO *fs= fs_character_set();
+#endif
   DBUG_ENTER("cleanup_dirname");
   DBUG_PRINT("enter",("from: '%s'",from));
 
@@ -141,6 +144,15 @@ uint cleanup_dirname(register my_string to, const char *from)
   length=(uint) (strmov(parent+1,FN_PARENTDIR)-parent);
   for (pos=start ; (*pos= *from_ptr++) != 0 ; pos++)
   {
+#ifdef BACKSLASH_MBTAIL
+    uint l;
+    if (use_mb(fs) && (l= my_ismbchar(fs, from_ptr - 1, from_ptr + 2)))
+    {
+      for (l-- ; l ; *++pos= *from_ptr++, l--);
+      start= pos + 1; /* Don't look inside multi-byte char */
+      continue;
+    }
+#endif
     if (*pos == '/')
       *pos = FN_LIBCHAR;
     if (*pos == FN_LIBCHAR)