bug#7284: strnxfrm returns different results for equal strings

13 files changed, 117 insertions, 32 deletions

diff --git a/strings/ctype-big5.c b/strings/ctype-big5.c
index 8345c53202c..997b8ce93d6 100644
--- a/strings/ctype-big5.c
+++ b/strings/ctype-big5.c
@@ -298,6 +298,7 @@ static int my_strnxfrm_big5(CHARSET_INFO *cs __attribute__((unused)),
                      const uchar * src, uint srclen)
 {
   uint16 e;
+  uint dstlen= len;
 
   len = srclen;
   while (len--)
@@ -312,7 +313,9 @@ static int my_strnxfrm_big5(CHARSET_INFO *cs __attribute__((unused)),
     } else
       *dest++ = sort_order_big5[(uchar) *src++];
   }
-  return srclen;
+  if (dstlen > srclen)
+    bfill(dest, dstlen - srclen, ' ');
+  return dstlen;
 }
 
 #if 0
diff --git a/strings/ctype-bin.c b/strings/ctype-bin.c
index 7d17f62c8d0..95c52512243 100644
--- a/strings/ctype-bin.c
+++ b/strings/ctype-bin.c
@@ -341,13 +341,27 @@ static int my_wildcmp_bin(CHARSET_INFO *cs,
 
 
 static int my_strnxfrm_bin(CHARSET_INFO *cs __attribute__((unused)),
-			    uchar * dest, uint len,
-			    const uchar *src, 
-			    uint srclen __attribute__((unused)))
+                           uchar * dest, uint dstlen,
+                           const uchar *src, uint srclen)
 {
   if (dest != src)
-    memcpy(dest,src,len= min(len,srclen));
-  return len;
+    memcpy(dest, src, min(dstlen,srclen));
+  if (dstlen > srclen)
+    bfill(dest + srclen, dstlen - srclen, 0);
+  return dstlen;
+}
+
+
+static
+int my_strnxfrm_8bit_bin(CHARSET_INFO *cs __attribute__((unused)),
+                         uchar * dest, uint dstlen,
+                         const uchar *src, uint srclen)
+{
+  if (dest != src)
+    memcpy(dest, src, min(dstlen,srclen));
+  if (dstlen > srclen)
+    bfill(dest + srclen, dstlen - srclen, ' ');
+  return dstlen;
 }
 
 
@@ -417,7 +431,7 @@ MY_COLLATION_HANDLER my_collation_8bit_bin_handler =
     NULL,			/* init */
     my_strnncoll_8bit_bin,
     my_strnncollsp_8bit_bin,
-    my_strnxfrm_bin,
+    my_strnxfrm_8bit_bin,
     my_like_range_simple,
     my_wildcmp_bin,
     my_strcasecmp_bin,
diff --git a/strings/ctype-czech.c b/strings/ctype-czech.c
index 2177a18504e..5725e81b15e 100644
--- a/strings/ctype-czech.c
+++ b/strings/ctype-czech.c
@@ -296,16 +296,18 @@ static int my_strnxfrm_czech(CHARSET_INFO *cs __attribute__((unused)),
   int value;
   const uchar * p, * store;
   int pass = 0;
-  int totlen = 0;
+  uint totlen = 0;
   p = src;	store = src;
 
   do
   {
     NEXT_CMP_VALUE(src, p, store, pass, value, (int)srclen);
-    ADD_TO_RESULT(dest, (int)len, totlen, value);
+    ADD_TO_RESULT(dest, len, totlen, value);
   }
   while (value);
-  return totlen;
+  if (len > totlen)
+    bfill(dest + totlen, len - totlen, ' ');
+  return len;
 }
 
 #undef IS_END
diff --git a/strings/ctype-gbk.c b/strings/ctype-gbk.c
index 0be56e8d946..731ad58a2fb 100644
--- a/strings/ctype-gbk.c
+++ b/strings/ctype-gbk.c
@@ -2659,6 +2659,7 @@ static int my_strnxfrm_gbk(CHARSET_INFO *cs __attribute__((unused)),
                     const uchar * src, uint srclen)
 {
   uint16 e;
+  uint dstlen= len;
 
   len = srclen;
   while (len--)
@@ -2673,7 +2674,9 @@ static int my_strnxfrm_gbk(CHARSET_INFO *cs __attribute__((unused)),
     } else 
       *dest++ = sort_order_gbk[(uchar) *src++];
   }
-  return srclen;
+  if (dstlen > srclen)
+    bfill(dest, dstlen - srclen, ' ');
+  return dstlen;
 }
 
 
diff --git a/strings/ctype-latin1.c b/strings/ctype-latin1.c
index 5f1850b7772..32d9a227c2f 100644
--- a/strings/ctype-latin1.c
+++ b/strings/ctype-latin1.c
@@ -637,7 +637,6 @@ static int my_strnxfrm_latin1_de(CHARSET_INFO *cs __attribute__((unused)),
 				 uchar * dest, uint len,
 				 const uchar * src, uint srclen)
 {
-  const uchar *dest_orig = dest;
   const uchar *de = dest + len;
   const uchar *se = src + srclen;
   for ( ; src < se && dest < de ; src++)
@@ -647,7 +646,9 @@ static int my_strnxfrm_latin1_de(CHARSET_INFO *cs __attribute__((unused)),
     if ((chr=combo2map[*src]) && dest < de)
       *dest++=chr;
   }
-  return (int) (dest - dest_orig);
+  if (dest < de)
+    bfill(dest, de - dest, ' ');
+  return (int) len;
 }
 
 
diff --git a/strings/ctype-mb.c b/strings/ctype-mb.c
index 3cdf7f460cd..bf69fe683ae 100644
--- a/strings/ctype-mb.c
+++ b/strings/ctype-mb.c
@@ -412,13 +412,14 @@ static int my_strnncollsp_mb_bin(CHARSET_INFO * cs __attribute__((unused)),
 
 
 static int my_strnxfrm_mb_bin(CHARSET_INFO *cs __attribute__((unused)),
-			    uchar * dest, uint len,
-			    const uchar *src, 
-			    uint srclen __attribute__((unused)))
+                              uchar * dest, uint dstlen,
+                              const uchar *src, uint srclen)
 {
   if (dest != src)
-    memcpy(dest,src,len= min(len,srclen));
-  return len;
+    memcpy(dest, src, min(dstlen, srclen));
+  if (dstlen > srclen)
+    bfill(dest + srclen, dstlen - srclen, ' ');
+  return dstlen;
 }
 
 
diff --git a/strings/ctype-simple.c b/strings/ctype-simple.c
index a019665a235..5bfa9e52595 100644
--- a/strings/ctype-simple.c
+++ b/strings/ctype-simple.c
@@ -21,27 +21,68 @@
 
 #include "stdarg.h"
 
+/*
+  Converts a string into its sort key.
+  
+  SYNOPSIS
+     my_strnxfrm_xxx()
+     
+  IMPLEMENTATION
+     
+     The my_strxfrm_xxx() function transforms a string pointed to by
+     'src' with length 'srclen' according to the charset+collation 
+     pair 'cs' and copies the result key into 'dest'.
+     
+     Comparing two strings using memcmp() after my_strnxfrm_xxx()
+     is equal to comparing two original strings with my_strnncollsp_xxx().
+     
+     Not more than 'dstlen' bytes are written into 'dst'.
+     To garantee that the whole string is transformed, 'dstlen' must be
+     at least srclen*cs->strnxfrm_multiply bytes long. Otherwise,
+     consequent memcmp() may return a non-accurate result.
+     
+     If the source string is too short to fill whole 'dstlen' bytes,
+     then the 'dest' string is padded up to 'dstlen', ensuring that:
+     
+       "a"  == "a "
+       "a\0" < "a"
+       "a\0" < "a "
+     
+     my_strnxfrm_simple() is implemented for 8bit charsets and
+     simple collations with one-to-one string->key transformation.
+     
+     See also implementations for various charsets/collations in  
+     other ctype-xxx.c files.
+     
+  RETURN
+  
+    Target len 'dstlen'.
+  
+*/
+
 
 int my_strnxfrm_simple(CHARSET_INFO * cs, 
                        uchar *dest, uint len,
                        const uchar *src, uint srclen)
 {
   uchar *map= cs->sort_order;
+  uint dstlen= len;
   set_if_smaller(len, srclen);
   if (dest != src)
   {
     const uchar *end;
     for ( end=src+len; src < end ;  )
       *dest++= map[*src++];
-    return len;
   }
   else
   {
     const uchar *end;
     for ( end=dest+len; dest < end ; dest++)
       *dest= (char) map[(uchar) *dest];
-    return len;
   }
+  if (dstlen > len)
+    bfill(dest, dstlen - len, ' ');
+  return dstlen;
 }
 
 int my_strnncoll_simple(CHARSET_INFO * cs, const uchar *s, uint slen, 
diff --git a/strings/ctype-sjis.c b/strings/ctype-sjis.c
index a8b5394f8c5..c0b33a13cdd 100644
--- a/strings/ctype-sjis.c
+++ b/strings/ctype-sjis.c
@@ -291,7 +291,9 @@ static int my_strnxfrm_sjis(CHARSET_INFO *cs __attribute__((unused)),
     else
       *dest++ = sort_order_sjis[(uchar)*src++];
   }
-  return srclen;
+  if (len > srclen)
+    bfill(dest, len - srclen, ' ');
+  return len;
 }
 
 
diff --git a/strings/ctype-tis620.c b/strings/ctype-tis620.c
index 5d37aa965d9..3a43c556ac8 100644
--- a/strings/ctype-tis620.c
+++ b/strings/ctype-tis620.c
@@ -631,9 +631,13 @@ int my_strnxfrm_tis620(CHARSET_INFO *cs __attribute__((unused)),
                        uchar * dest, uint len,
                        const uchar * src, uint srclen)
 {
+  uint dstlen= len;
   len= (uint) (strmake((char*) dest, (char*) src, min(len, srclen)) -
 	       (char*) dest);
-  return (int) thai2sortable(dest, len);
+  len= thai2sortable(dest, len);
+  if (dstlen > len)
+    bfill(dest + len, dstlen - len, ' ');
+  return dstlen;
 }
 
 
diff --git a/strings/ctype-uca.c b/strings/ctype-uca.c
index 89c876ad10c..4992ef2dcd1 100644
--- a/strings/ctype-uca.c
+++ b/strings/ctype-uca.c
@@ -7214,8 +7214,7 @@ static int my_strnxfrm_uca(CHARSET_INFO *cs,
                            uchar *dst, uint dstlen,
                            const uchar *src, uint srclen)
 {
-  uchar *de = dst + dstlen;
-  const uchar *dst_orig = dst;
+  uchar *de = dst + (dstlen & (uint) ~1); // add even length for easier code
   int   s_res;
   my_uca_scanner scanner;
   scanner_handler->init(&scanner, cs, src, srclen);
@@ -7226,8 +7225,17 @@ static int my_strnxfrm_uca(CHARSET_INFO *cs,
     dst[1]= s_res & 0xFF;
     dst+= 2;
   }
-  for ( ; dst < de; *dst++='\0');
-  return dst - dst_orig;
+  s_res= cs->sort_order_big[0][0x20 * cs->sort_order[0]];
+  while (dst < de)
+  {
+    dst[0]= s_res >> 8;
+    dst[1]= s_res & 0xFF;
+    dst+= 2;
+  }
+  if (dstlen & 1) // if odd number then fill the last char
+    *dst= '\0';
+  
+  return dstlen;
 }
 
 
diff --git a/strings/ctype-ucs2.c b/strings/ctype-ucs2.c
index 403d31aa15b..936e2b6fdce 100644
--- a/strings/ctype-ucs2.c
+++ b/strings/ctype-ucs2.c
@@ -347,7 +347,6 @@ static int my_strnxfrm_ucs2(CHARSET_INFO *cs,
   int plane;
   uchar *de = dst + dstlen;
   const uchar *se = src + srclen;
-  const uchar *dst_orig = dst;
 
   while( src < se && dst < de )
   {
@@ -367,7 +366,9 @@ static int my_strnxfrm_ucs2(CHARSET_INFO *cs,
     }
     dst+=res;
   }
-  return dst - dst_orig;
+  if (dst < de)
+    cs->cset->fill(cs, dst, de - dst, ' ');
+  return dstlen;
 }
 
 
@@ -1377,7 +1378,9 @@ int my_strnxfrm_ucs2_bin(CHARSET_INFO *cs __attribute__((unused)),
 {
   if (dst != src)
     memcpy(dst,src,srclen= min(dstlen,srclen));
-  return srclen;
+  if (dstlen > srclen)
+    cs->cset->fill(cs, dst + srclen, dstlen - srclen, ' ');
+  return dstlen;
 }
 
 
diff --git a/strings/ctype-utf8.c b/strings/ctype-utf8.c
index ce9346eb475..502d0ec285e 100644
--- a/strings/ctype-utf8.c
+++ b/strings/ctype-utf8.c
@@ -2228,7 +2228,6 @@ static int my_strnxfrm_utf8(CHARSET_INFO *cs,
   int plane;
   uchar *de = dst + dstlen;
   const uchar *se = src + srclen;
-  const uchar *dst_orig = dst;
 
   while( src < se && dst < de )
   {
@@ -2248,7 +2247,9 @@ static int my_strnxfrm_utf8(CHARSET_INFO *cs,
     }
     dst+=res;
   }
-  return dst - dst_orig;
+  if (dst < de)
+    bfill(dst, de - dst, ' ');
+  return dstlen;
 }
 
 static int my_ismbchar_utf8(CHARSET_INFO *cs,const char *b, const char *e)
diff --git a/strings/ctype-win1250ch.c b/strings/ctype-win1250ch.c
index 4ada3d47bf5..98389a9a5a4 100644
--- a/strings/ctype-win1250ch.c
+++ b/strings/ctype-win1250ch.c
@@ -503,7 +503,9 @@ static int my_strnxfrm_win1250ch(CHARSET_INFO * cs  __attribute__((unused)),
       dest[totlen] = value;
     totlen++;
   } while (value) ;
-  return totlen;
+  if (len > totlen)
+    bfill(dest + totlen, len - totlen, ' ');
+  return len;
 }
 
 #undef IS_END