misc/libfreetype/src/base/ftcalc.c
changeset 9372 915436ff64ab
parent 9371 f3840de881bd
child 9373 b769a8e38cbd
equal deleted inserted replaced
9371:f3840de881bd 9372:915436ff64ab
     1 /***************************************************************************/
       
     2 /*                                                                         */
       
     3 /*  ftcalc.c                                                               */
       
     4 /*                                                                         */
       
     5 /*    Arithmetic computations (body).                                      */
       
     6 /*                                                                         */
       
     7 /*  Copyright 1996-2001, 2002, 2003, 2004, 2005, 2006, 2008 by             */
       
     8 /*  David Turner, Robert Wilhelm, and Werner Lemberg.                      */
       
     9 /*                                                                         */
       
    10 /*  This file is part of the FreeType project, and may only be used,       */
       
    11 /*  modified, and distributed under the terms of the FreeType project      */
       
    12 /*  license, LICENSE.TXT.  By continuing to use, modify, or distribute     */
       
    13 /*  this file you indicate that you have read the license and              */
       
    14 /*  understand and accept it fully.                                        */
       
    15 /*                                                                         */
       
    16 /***************************************************************************/
       
    17 
       
    18   /*************************************************************************/
       
    19   /*                                                                       */
       
    20   /* Support for 1-complement arithmetic has been totally dropped in this  */
       
    21   /* release.  You can still write your own code if you need it.           */
       
    22   /*                                                                       */
       
    23   /*************************************************************************/
       
    24 
       
    25   /*************************************************************************/
       
    26   /*                                                                       */
       
    27   /* Implementing basic computation routines.                              */
       
    28   /*                                                                       */
       
    29   /* FT_MulDiv(), FT_MulFix(), FT_DivFix(), FT_RoundFix(), FT_CeilFix(),   */
       
    30   /* and FT_FloorFix() are declared in freetype.h.                         */
       
    31   /*                                                                       */
       
    32   /*************************************************************************/
       
    33 
       
    34 
       
    35 #include <ft2build.h>
       
    36 #include FT_GLYPH_H
       
    37 #include FT_INTERNAL_CALC_H
       
    38 #include FT_INTERNAL_DEBUG_H
       
    39 #include FT_INTERNAL_OBJECTS_H
       
    40 
       
    41 #ifdef FT_MULFIX_INLINED
       
    42 #undef FT_MulFix
       
    43 #endif
       
    44 
       
    45 /* we need to define a 64-bits data type here */
       
    46 
       
    47 #ifdef FT_LONG64
       
    48 
       
    49   typedef FT_INT64  FT_Int64;
       
    50 
       
    51 #else
       
    52 
       
    53   typedef struct  FT_Int64_
       
    54   {
       
    55     FT_UInt32  lo;
       
    56     FT_UInt32  hi;
       
    57 
       
    58   } FT_Int64;
       
    59 
       
    60 #endif /* FT_LONG64 */
       
    61 
       
    62 
       
    63   /*************************************************************************/
       
    64   /*                                                                       */
       
    65   /* The macro FT_COMPONENT is used in trace mode.  It is an implicit      */
       
    66   /* parameter of the FT_TRACE() and FT_ERROR() macros, used to print/log  */
       
    67   /* messages during execution.                                            */
       
    68   /*                                                                       */
       
    69 #undef  FT_COMPONENT
       
    70 #define FT_COMPONENT  trace_calc
       
    71 
       
    72 
       
    73   /* The following three functions are available regardless of whether */
       
    74   /* FT_LONG64 is defined.                                             */
       
    75 
       
    76   /* documentation is in freetype.h */
       
    77 
       
    78   FT_EXPORT_DEF( FT_Fixed )
       
    79   FT_RoundFix( FT_Fixed  a )
       
    80   {
       
    81     return ( a >= 0 ) ?   ( a + 0x8000L ) & ~0xFFFFL
       
    82                       : -((-a + 0x8000L ) & ~0xFFFFL );
       
    83   }
       
    84 
       
    85 
       
    86   /* documentation is in freetype.h */
       
    87 
       
    88   FT_EXPORT_DEF( FT_Fixed )
       
    89   FT_CeilFix( FT_Fixed  a )
       
    90   {
       
    91     return ( a >= 0 ) ?   ( a + 0xFFFFL ) & ~0xFFFFL
       
    92                       : -((-a + 0xFFFFL ) & ~0xFFFFL );
       
    93   }
       
    94 
       
    95 
       
    96   /* documentation is in freetype.h */
       
    97 
       
    98   FT_EXPORT_DEF( FT_Fixed )
       
    99   FT_FloorFix( FT_Fixed  a )
       
   100   {
       
   101     return ( a >= 0 ) ?   a & ~0xFFFFL
       
   102                       : -((-a) & ~0xFFFFL );
       
   103   }
       
   104 
       
   105 
       
   106 #ifdef FT_CONFIG_OPTION_OLD_INTERNALS
       
   107 
       
   108   /* documentation is in ftcalc.h */
       
   109 
       
   110   FT_EXPORT_DEF( FT_Int32 )
       
   111   FT_Sqrt32( FT_Int32  x )
       
   112   {
       
   113     FT_UInt32  val, root, newroot, mask;
       
   114 
       
   115 
       
   116     root = 0;
       
   117     mask = (FT_UInt32)0x40000000UL;
       
   118     val  = (FT_UInt32)x;
       
   119 
       
   120     do
       
   121     {
       
   122       newroot = root + mask;
       
   123       if ( newroot <= val )
       
   124       {
       
   125         val -= newroot;
       
   126         root = newroot + mask;
       
   127       }
       
   128 
       
   129       root >>= 1;
       
   130       mask >>= 2;
       
   131 
       
   132     } while ( mask != 0 );
       
   133 
       
   134     return root;
       
   135   }
       
   136 
       
   137 #endif /* FT_CONFIG_OPTION_OLD_INTERNALS */
       
   138 
       
   139 
       
   140 #ifdef FT_LONG64
       
   141 
       
   142 
       
   143   /* documentation is in freetype.h */
       
   144 
       
   145   FT_EXPORT_DEF( FT_Long )
       
   146   FT_MulDiv( FT_Long  a,
       
   147              FT_Long  b,
       
   148              FT_Long  c )
       
   149   {
       
   150     FT_Int   s;
       
   151     FT_Long  d;
       
   152 
       
   153 
       
   154     s = 1;
       
   155     if ( a < 0 ) { a = -a; s = -1; }
       
   156     if ( b < 0 ) { b = -b; s = -s; }
       
   157     if ( c < 0 ) { c = -c; s = -s; }
       
   158 
       
   159     d = (FT_Long)( c > 0 ? ( (FT_Int64)a * b + ( c >> 1 ) ) / c
       
   160                          : 0x7FFFFFFFL );
       
   161 
       
   162     return ( s > 0 ) ? d : -d;
       
   163   }
       
   164 
       
   165 
       
   166 #ifdef TT_USE_BYTECODE_INTERPRETER
       
   167 
       
   168   /* documentation is in ftcalc.h */
       
   169 
       
   170   FT_BASE_DEF( FT_Long )
       
   171   FT_MulDiv_No_Round( FT_Long  a,
       
   172                       FT_Long  b,
       
   173                       FT_Long  c )
       
   174   {
       
   175     FT_Int   s;
       
   176     FT_Long  d;
       
   177 
       
   178 
       
   179     s = 1;
       
   180     if ( a < 0 ) { a = -a; s = -1; }
       
   181     if ( b < 0 ) { b = -b; s = -s; }
       
   182     if ( c < 0 ) { c = -c; s = -s; }
       
   183 
       
   184     d = (FT_Long)( c > 0 ? (FT_Int64)a * b / c
       
   185                          : 0x7FFFFFFFL );
       
   186 
       
   187     return ( s > 0 ) ? d : -d;
       
   188   }
       
   189 
       
   190 #endif /* TT_USE_BYTECODE_INTERPRETER */
       
   191 
       
   192 
       
   193   /* documentation is in freetype.h */
       
   194 
       
   195   FT_EXPORT_DEF( FT_Long )
       
   196   FT_MulFix( FT_Long  a,
       
   197              FT_Long  b )
       
   198   {
       
   199 #ifdef FT_MULFIX_ASSEMBLER
       
   200 
       
   201     return FT_MULFIX_ASSEMBLER( a, b );
       
   202 
       
   203 #else
       
   204 
       
   205     FT_Int   s = 1;
       
   206     FT_Long  c;
       
   207 
       
   208 
       
   209     if ( a < 0 )
       
   210     {
       
   211       a = -a;
       
   212       s = -1;
       
   213     }
       
   214 
       
   215     if ( b < 0 )
       
   216     {
       
   217       b = -b;
       
   218       s = -s;
       
   219     }
       
   220 
       
   221     c = (FT_Long)( ( (FT_Int64)a * b + 0x8000L ) >> 16 );
       
   222 
       
   223     return ( s > 0 ) ? c : -c;
       
   224 
       
   225 #endif /* FT_MULFIX_ASSEMBLER */
       
   226   }
       
   227 
       
   228 
       
   229   /* documentation is in freetype.h */
       
   230 
       
   231   FT_EXPORT_DEF( FT_Long )
       
   232   FT_DivFix( FT_Long  a,
       
   233              FT_Long  b )
       
   234   {
       
   235     FT_Int32   s;
       
   236     FT_UInt32  q;
       
   237 
       
   238     s = 1;
       
   239     if ( a < 0 ) { a = -a; s = -1; }
       
   240     if ( b < 0 ) { b = -b; s = -s; }
       
   241 
       
   242     if ( b == 0 )
       
   243       /* check for division by 0 */
       
   244       q = 0x7FFFFFFFL;
       
   245     else
       
   246       /* compute result directly */
       
   247       q = (FT_UInt32)( ( ( (FT_Int64)a << 16 ) + ( b >> 1 ) ) / b );
       
   248 
       
   249     return ( s < 0 ? -(FT_Long)q : (FT_Long)q );
       
   250   }
       
   251 
       
   252 
       
   253 #else /* !FT_LONG64 */
       
   254 
       
   255 
       
   256   static void
       
   257   ft_multo64( FT_UInt32  x,
       
   258               FT_UInt32  y,
       
   259               FT_Int64  *z )
       
   260   {
       
   261     FT_UInt32  lo1, hi1, lo2, hi2, lo, hi, i1, i2;
       
   262 
       
   263 
       
   264     lo1 = x & 0x0000FFFFU;  hi1 = x >> 16;
       
   265     lo2 = y & 0x0000FFFFU;  hi2 = y >> 16;
       
   266 
       
   267     lo = lo1 * lo2;
       
   268     i1 = lo1 * hi2;
       
   269     i2 = lo2 * hi1;
       
   270     hi = hi1 * hi2;
       
   271 
       
   272     /* Check carry overflow of i1 + i2 */
       
   273     i1 += i2;
       
   274     hi += (FT_UInt32)( i1 < i2 ) << 16;
       
   275 
       
   276     hi += i1 >> 16;
       
   277     i1  = i1 << 16;
       
   278 
       
   279     /* Check carry overflow of i1 + lo */
       
   280     lo += i1;
       
   281     hi += ( lo < i1 );
       
   282 
       
   283     z->lo = lo;
       
   284     z->hi = hi;
       
   285   }
       
   286 
       
   287 
       
   288   static FT_UInt32
       
   289   ft_div64by32( FT_UInt32  hi,
       
   290                 FT_UInt32  lo,
       
   291                 FT_UInt32  y )
       
   292   {
       
   293     FT_UInt32  r, q;
       
   294     FT_Int     i;
       
   295 
       
   296 
       
   297     q = 0;
       
   298     r = hi;
       
   299 
       
   300     if ( r >= y )
       
   301       return (FT_UInt32)0x7FFFFFFFL;
       
   302 
       
   303     i = 32;
       
   304     do
       
   305     {
       
   306       r <<= 1;
       
   307       q <<= 1;
       
   308       r  |= lo >> 31;
       
   309 
       
   310       if ( r >= (FT_UInt32)y )
       
   311       {
       
   312         r -= y;
       
   313         q |= 1;
       
   314       }
       
   315       lo <<= 1;
       
   316     } while ( --i );
       
   317 
       
   318     return q;
       
   319   }
       
   320 
       
   321 
       
   322   static void
       
   323   FT_Add64( FT_Int64*  x,
       
   324             FT_Int64*  y,
       
   325             FT_Int64  *z )
       
   326   {
       
   327     register FT_UInt32  lo, hi;
       
   328 
       
   329 
       
   330     lo = x->lo + y->lo;
       
   331     hi = x->hi + y->hi + ( lo < x->lo );
       
   332 
       
   333     z->lo = lo;
       
   334     z->hi = hi;
       
   335   }
       
   336 
       
   337 
       
   338   /* documentation is in freetype.h */
       
   339 
       
   340   /* The FT_MulDiv function has been optimized thanks to ideas from      */
       
   341   /* Graham Asher.  The trick is to optimize computation when everything */
       
   342   /* fits within 32-bits (a rather common case).                         */
       
   343   /*                                                                     */
       
   344   /*  we compute 'a*b+c/2', then divide it by 'c'. (positive values)     */
       
   345   /*                                                                     */
       
   346   /*  46340 is FLOOR(SQRT(2^31-1)).                                      */
       
   347   /*                                                                     */
       
   348   /*  if ( a <= 46340 && b <= 46340 ) then ( a*b <= 0x7FFEA810 )         */
       
   349   /*                                                                     */
       
   350   /*  0x7FFFFFFF - 0x7FFEA810 = 0x157F0                                  */
       
   351   /*                                                                     */
       
   352   /*  if ( c < 0x157F0*2 ) then ( a*b+c/2 <= 0x7FFFFFFF )                */
       
   353   /*                                                                     */
       
   354   /*  and 2*0x157F0 = 176096                                             */
       
   355   /*                                                                     */
       
   356 
       
   357   FT_EXPORT_DEF( FT_Long )
       
   358   FT_MulDiv( FT_Long  a,
       
   359              FT_Long  b,
       
   360              FT_Long  c )
       
   361   {
       
   362     long  s;
       
   363 
       
   364 
       
   365     /* XXX: this function does not allow 64-bit arguments */
       
   366     if ( a == 0 || b == c )
       
   367       return a;
       
   368 
       
   369     s  = a; a = FT_ABS( a );
       
   370     s ^= b; b = FT_ABS( b );
       
   371     s ^= c; c = FT_ABS( c );
       
   372 
       
   373     if ( a <= 46340L && b <= 46340L && c <= 176095L && c > 0 )
       
   374       a = ( a * b + ( c >> 1 ) ) / c;
       
   375 
       
   376     else if ( c > 0 )
       
   377     {
       
   378       FT_Int64  temp, temp2;
       
   379 
       
   380 
       
   381       ft_multo64( (FT_Int32)a, (FT_Int32)b, &temp );
       
   382 
       
   383       temp2.hi = 0;
       
   384       temp2.lo = (FT_UInt32)(c >> 1);
       
   385       FT_Add64( &temp, &temp2, &temp );
       
   386       a = ft_div64by32( temp.hi, temp.lo, (FT_Int32)c );
       
   387     }
       
   388     else
       
   389       a = 0x7FFFFFFFL;
       
   390 
       
   391     return ( s < 0 ? -a : a );
       
   392   }
       
   393 
       
   394 
       
   395 #ifdef TT_USE_BYTECODE_INTERPRETER
       
   396 
       
   397   FT_BASE_DEF( FT_Long )
       
   398   FT_MulDiv_No_Round( FT_Long  a,
       
   399                       FT_Long  b,
       
   400                       FT_Long  c )
       
   401   {
       
   402     long  s;
       
   403 
       
   404 
       
   405     if ( a == 0 || b == c )
       
   406       return a;
       
   407 
       
   408     s  = a; a = FT_ABS( a );
       
   409     s ^= b; b = FT_ABS( b );
       
   410     s ^= c; c = FT_ABS( c );
       
   411 
       
   412     if ( a <= 46340L && b <= 46340L && c > 0 )
       
   413       a = a * b / c;
       
   414 
       
   415     else if ( c > 0 )
       
   416     {
       
   417       FT_Int64  temp;
       
   418 
       
   419 
       
   420       ft_multo64( (FT_Int32)a, (FT_Int32)b, &temp );
       
   421       a = ft_div64by32( temp.hi, temp.lo, (FT_Int32)c );
       
   422     }
       
   423     else
       
   424       a = 0x7FFFFFFFL;
       
   425 
       
   426     return ( s < 0 ? -a : a );
       
   427   }
       
   428 
       
   429 #endif /* TT_USE_BYTECODE_INTERPRETER */
       
   430 
       
   431 
       
   432   /* documentation is in freetype.h */
       
   433 
       
   434   FT_EXPORT_DEF( FT_Long )
       
   435   FT_MulFix( FT_Long  a,
       
   436              FT_Long  b )
       
   437   {
       
   438 #ifdef FT_MULFIX_ASSEMBLER
       
   439 
       
   440     return FT_MULFIX_ASSEMBLER( a, b );
       
   441 
       
   442 #elif 0
       
   443 
       
   444     /*
       
   445      *  This code is nonportable.  See comment below.
       
   446      *
       
   447      *  However, on a platform where right-shift of a signed quantity fills
       
   448      *  the leftmost bits by copying the sign bit, it might be faster.
       
   449      */
       
   450 
       
   451     FT_Long   sa, sb;
       
   452     FT_ULong  ua, ub;
       
   453 
       
   454 
       
   455     if ( a == 0 || b == 0x10000L )
       
   456       return a;
       
   457 
       
   458     /*
       
   459      *  This is a clever way of converting a signed number `a' into its
       
   460      *  absolute value (stored back into `a') and its sign.  The sign is
       
   461      *  stored in `sa'; 0 means `a' was positive or zero, and -1 means `a'
       
   462      *  was negative.  (Similarly for `b' and `sb').
       
   463      *
       
   464      *  Unfortunately, it doesn't work (at least not portably).
       
   465      *
       
   466      *  It makes the assumption that right-shift on a negative signed value
       
   467      *  fills the leftmost bits by copying the sign bit.  This is wrong. 
       
   468      *  According to K&R 2nd ed, section `A7.8 Shift Operators' on page 206,
       
   469      *  the result of right-shift of a negative signed value is
       
   470      *  implementation-defined.  At least one implementation fills the
       
   471      *  leftmost bits with 0s (i.e., it is exactly the same as an unsigned
       
   472      *  right shift).  This means that when `a' is negative, `sa' ends up
       
   473      *  with the value 1 rather than -1.  After that, everything else goes
       
   474      *  wrong.
       
   475      */
       
   476     sa = ( a >> ( sizeof ( a ) * 8 - 1 ) );
       
   477     a  = ( a ^ sa ) - sa;
       
   478     sb = ( b >> ( sizeof ( b ) * 8 - 1 ) );
       
   479     b  = ( b ^ sb ) - sb;
       
   480 
       
   481     ua = (FT_ULong)a;
       
   482     ub = (FT_ULong)b;
       
   483 
       
   484     if ( ua <= 2048 && ub <= 1048576L )
       
   485       ua = ( ua * ub + 0x8000U ) >> 16;
       
   486     else
       
   487     {
       
   488       FT_ULong  al = ua & 0xFFFFU;
       
   489 
       
   490 
       
   491       ua = ( ua >> 16 ) * ub +  al * ( ub >> 16 ) +
       
   492            ( ( al * ( ub & 0xFFFFU ) + 0x8000U ) >> 16 );
       
   493     }
       
   494 
       
   495     sa ^= sb,
       
   496     ua  = (FT_ULong)(( ua ^ sa ) - sa);
       
   497 
       
   498     return (FT_Long)ua;
       
   499 
       
   500 #else /* 0 */
       
   501 
       
   502     FT_Long   s;
       
   503     FT_ULong  ua, ub;
       
   504 
       
   505 
       
   506     if ( a == 0 || b == 0x10000L )
       
   507       return a;
       
   508 
       
   509     s  = a; a = FT_ABS( a );
       
   510     s ^= b; b = FT_ABS( b );
       
   511 
       
   512     ua = (FT_ULong)a;
       
   513     ub = (FT_ULong)b;
       
   514 
       
   515     if ( ua <= 2048 && ub <= 1048576L )
       
   516       ua = ( ua * ub + 0x8000UL ) >> 16;
       
   517     else
       
   518     {
       
   519       FT_ULong  al = ua & 0xFFFFUL;
       
   520 
       
   521 
       
   522       ua = ( ua >> 16 ) * ub +  al * ( ub >> 16 ) +
       
   523            ( ( al * ( ub & 0xFFFFUL ) + 0x8000UL ) >> 16 );
       
   524     }
       
   525 
       
   526     return ( s < 0 ? -(FT_Long)ua : (FT_Long)ua );
       
   527 
       
   528 #endif /* 0 */
       
   529 
       
   530   }
       
   531 
       
   532 
       
   533   /* documentation is in freetype.h */
       
   534 
       
   535   FT_EXPORT_DEF( FT_Long )
       
   536   FT_DivFix( FT_Long  a,
       
   537              FT_Long  b )
       
   538   {
       
   539     FT_Int32   s;
       
   540     FT_UInt32  q;
       
   541 
       
   542 
       
   543     /* XXX: this function does not allow 64-bit arguments */
       
   544     s  = (FT_Int32)a; a = FT_ABS( a );
       
   545     s ^= (FT_Int32)b; b = FT_ABS( b );
       
   546 
       
   547     if ( b == 0 )
       
   548     {
       
   549       /* check for division by 0 */
       
   550       q = (FT_UInt32)0x7FFFFFFFL;
       
   551     }
       
   552     else if ( ( a >> 16 ) == 0 )
       
   553     {
       
   554       /* compute result directly */
       
   555       q = (FT_UInt32)( (a << 16) + (b >> 1) ) / (FT_UInt32)b;
       
   556     }
       
   557     else
       
   558     {
       
   559       /* we need more bits; we have to do it by hand */
       
   560       FT_Int64  temp, temp2;
       
   561 
       
   562       temp.hi  = (FT_Int32) (a >> 16);
       
   563       temp.lo  = (FT_UInt32)(a << 16);
       
   564       temp2.hi = 0;
       
   565       temp2.lo = (FT_UInt32)( b >> 1 );
       
   566       FT_Add64( &temp, &temp2, &temp );
       
   567       q = ft_div64by32( temp.hi, temp.lo, (FT_Int32)b );
       
   568     }
       
   569 
       
   570     return ( s < 0 ? -(FT_Int32)q : (FT_Int32)q );
       
   571   }
       
   572 
       
   573 
       
   574 #if 0
       
   575 
       
   576   /* documentation is in ftcalc.h */
       
   577 
       
   578   FT_EXPORT_DEF( void )
       
   579   FT_MulTo64( FT_Int32   x,
       
   580               FT_Int32   y,
       
   581               FT_Int64  *z )
       
   582   {
       
   583     FT_Int32  s;
       
   584 
       
   585 
       
   586     s  = x; x = FT_ABS( x );
       
   587     s ^= y; y = FT_ABS( y );
       
   588 
       
   589     ft_multo64( x, y, z );
       
   590 
       
   591     if ( s < 0 )
       
   592     {
       
   593       z->lo = (FT_UInt32)-(FT_Int32)z->lo;
       
   594       z->hi = ~z->hi + !( z->lo );
       
   595     }
       
   596   }
       
   597 
       
   598 
       
   599   /* apparently, the second version of this code is not compiled correctly */
       
   600   /* on Mac machines with the MPW C compiler..  tsk, tsk, tsk...           */
       
   601 
       
   602 #if 1
       
   603 
       
   604   FT_EXPORT_DEF( FT_Int32 )
       
   605   FT_Div64by32( FT_Int64*  x,
       
   606                 FT_Int32   y )
       
   607   {
       
   608     FT_Int32   s;
       
   609     FT_UInt32  q, r, i, lo;
       
   610 
       
   611 
       
   612     s  = x->hi;
       
   613     if ( s < 0 )
       
   614     {
       
   615       x->lo = (FT_UInt32)-(FT_Int32)x->lo;
       
   616       x->hi = ~x->hi + !x->lo;
       
   617     }
       
   618     s ^= y;  y = FT_ABS( y );
       
   619 
       
   620     /* Shortcut */
       
   621     if ( x->hi == 0 )
       
   622     {
       
   623       if ( y > 0 )
       
   624         q = x->lo / y;
       
   625       else
       
   626         q = 0x7FFFFFFFL;
       
   627 
       
   628       return ( s < 0 ? -(FT_Int32)q : (FT_Int32)q );
       
   629     }
       
   630 
       
   631     r  = x->hi;
       
   632     lo = x->lo;
       
   633 
       
   634     if ( r >= (FT_UInt32)y ) /* we know y is to be treated as unsigned here */
       
   635       return ( s < 0 ? 0x80000001UL : 0x7FFFFFFFUL );
       
   636                              /* Return Max/Min Int32 if division overflow. */
       
   637                              /* This includes division by zero!            */
       
   638     q = 0;
       
   639     for ( i = 0; i < 32; i++ )
       
   640     {
       
   641       r <<= 1;
       
   642       q <<= 1;
       
   643       r  |= lo >> 31;
       
   644 
       
   645       if ( r >= (FT_UInt32)y )
       
   646       {
       
   647         r -= y;
       
   648         q |= 1;
       
   649       }
       
   650       lo <<= 1;
       
   651     }
       
   652 
       
   653     return ( s < 0 ? -(FT_Int32)q : (FT_Int32)q );
       
   654   }
       
   655 
       
   656 #else /* 0 */
       
   657 
       
   658   FT_EXPORT_DEF( FT_Int32 )
       
   659   FT_Div64by32( FT_Int64*  x,
       
   660                 FT_Int32   y )
       
   661   {
       
   662     FT_Int32   s;
       
   663     FT_UInt32  q;
       
   664 
       
   665 
       
   666     s  = x->hi;
       
   667     if ( s < 0 )
       
   668     {
       
   669       x->lo = (FT_UInt32)-(FT_Int32)x->lo;
       
   670       x->hi = ~x->hi + !x->lo;
       
   671     }
       
   672     s ^= y;  y = FT_ABS( y );
       
   673 
       
   674     /* Shortcut */
       
   675     if ( x->hi == 0 )
       
   676     {
       
   677       if ( y > 0 )
       
   678         q = ( x->lo + ( y >> 1 ) ) / y;
       
   679       else
       
   680         q = 0x7FFFFFFFL;
       
   681 
       
   682       return ( s < 0 ? -(FT_Int32)q : (FT_Int32)q );
       
   683     }
       
   684 
       
   685     q = ft_div64by32( x->hi, x->lo, y );
       
   686 
       
   687     return ( s < 0 ? -(FT_Int32)q : (FT_Int32)q );
       
   688   }
       
   689 
       
   690 #endif /* 0 */
       
   691 
       
   692 #endif /* 0 */
       
   693 
       
   694 
       
   695 #endif /* FT_LONG64 */
       
   696 
       
   697 
       
   698   /* documentation is in ftglyph.h */
       
   699 
       
   700   FT_EXPORT_DEF( void )
       
   701   FT_Matrix_Multiply( const FT_Matrix*  a,
       
   702                       FT_Matrix        *b )
       
   703   {
       
   704     FT_Fixed  xx, xy, yx, yy;
       
   705 
       
   706 
       
   707     if ( !a || !b )
       
   708       return;
       
   709 
       
   710     xx = FT_MulFix( a->xx, b->xx ) + FT_MulFix( a->xy, b->yx );
       
   711     xy = FT_MulFix( a->xx, b->xy ) + FT_MulFix( a->xy, b->yy );
       
   712     yx = FT_MulFix( a->yx, b->xx ) + FT_MulFix( a->yy, b->yx );
       
   713     yy = FT_MulFix( a->yx, b->xy ) + FT_MulFix( a->yy, b->yy );
       
   714 
       
   715     b->xx = xx;  b->xy = xy;
       
   716     b->yx = yx;  b->yy = yy;
       
   717   }
       
   718 
       
   719 
       
   720   /* documentation is in ftglyph.h */
       
   721 
       
   722   FT_EXPORT_DEF( FT_Error )
       
   723   FT_Matrix_Invert( FT_Matrix*  matrix )
       
   724   {
       
   725     FT_Pos  delta, xx, yy;
       
   726 
       
   727 
       
   728     if ( !matrix )
       
   729       return FT_Err_Invalid_Argument;
       
   730 
       
   731     /* compute discriminant */
       
   732     delta = FT_MulFix( matrix->xx, matrix->yy ) -
       
   733             FT_MulFix( matrix->xy, matrix->yx );
       
   734 
       
   735     if ( !delta )
       
   736       return FT_Err_Invalid_Argument;  /* matrix can't be inverted */
       
   737 
       
   738     matrix->xy = - FT_DivFix( matrix->xy, delta );
       
   739     matrix->yx = - FT_DivFix( matrix->yx, delta );
       
   740 
       
   741     xx = matrix->xx;
       
   742     yy = matrix->yy;
       
   743 
       
   744     matrix->xx = FT_DivFix( yy, delta );
       
   745     matrix->yy = FT_DivFix( xx, delta );
       
   746 
       
   747     return FT_Err_Ok;
       
   748   }
       
   749 
       
   750 
       
   751   /* documentation is in ftcalc.h */
       
   752 
       
   753   FT_BASE_DEF( void )
       
   754   FT_Matrix_Multiply_Scaled( const FT_Matrix*  a,
       
   755                              FT_Matrix        *b,
       
   756                              FT_Long           scaling )
       
   757   {
       
   758     FT_Fixed  xx, xy, yx, yy;
       
   759 
       
   760     FT_Long   val = 0x10000L * scaling;
       
   761 
       
   762 
       
   763     if ( !a || !b )
       
   764       return;
       
   765 
       
   766     xx = FT_MulDiv( a->xx, b->xx, val ) + FT_MulDiv( a->xy, b->yx, val );
       
   767     xy = FT_MulDiv( a->xx, b->xy, val ) + FT_MulDiv( a->xy, b->yy, val );
       
   768     yx = FT_MulDiv( a->yx, b->xx, val ) + FT_MulDiv( a->yy, b->yx, val );
       
   769     yy = FT_MulDiv( a->yx, b->xy, val ) + FT_MulDiv( a->yy, b->yy, val );
       
   770 
       
   771     b->xx = xx;  b->xy = xy;
       
   772     b->yx = yx;  b->yy = yy;
       
   773   }
       
   774 
       
   775 
       
   776   /* documentation is in ftcalc.h */
       
   777 
       
   778   FT_BASE_DEF( void )
       
   779   FT_Vector_Transform_Scaled( FT_Vector*        vector,
       
   780                               const FT_Matrix*  matrix,
       
   781                               FT_Long           scaling )
       
   782   {
       
   783     FT_Pos   xz, yz;
       
   784 
       
   785     FT_Long  val = 0x10000L * scaling;
       
   786 
       
   787 
       
   788     if ( !vector || !matrix )
       
   789       return;
       
   790 
       
   791     xz = FT_MulDiv( vector->x, matrix->xx, val ) +
       
   792          FT_MulDiv( vector->y, matrix->xy, val );
       
   793 
       
   794     yz = FT_MulDiv( vector->x, matrix->yx, val ) +
       
   795          FT_MulDiv( vector->y, matrix->yy, val );
       
   796 
       
   797     vector->x = xz;
       
   798     vector->y = yz;
       
   799   }
       
   800 
       
   801 
       
   802   /* documentation is in ftcalc.h */
       
   803 
       
   804   FT_BASE_DEF( FT_Int32 )
       
   805   FT_SqrtFixed( FT_Int32  x )
       
   806   {
       
   807     FT_UInt32  root, rem_hi, rem_lo, test_div;
       
   808     FT_Int     count;
       
   809 
       
   810 
       
   811     root = 0;
       
   812 
       
   813     if ( x > 0 )
       
   814     {
       
   815       rem_hi = 0;
       
   816       rem_lo = x;
       
   817       count  = 24;
       
   818       do
       
   819       {
       
   820         rem_hi   = ( rem_hi << 2 ) | ( rem_lo >> 30 );
       
   821         rem_lo <<= 2;
       
   822         root   <<= 1;
       
   823         test_div = ( root << 1 ) + 1;
       
   824 
       
   825         if ( rem_hi >= test_div )
       
   826         {
       
   827           rem_hi -= test_div;
       
   828           root   += 1;
       
   829         }
       
   830       } while ( --count );
       
   831     }
       
   832 
       
   833     return (FT_Int32)root;
       
   834   }
       
   835 
       
   836 
       
   837   /* documentation is in ftcalc.h */
       
   838 
       
   839   FT_BASE_DEF( FT_Int )
       
   840   ft_corner_orientation( FT_Pos  in_x,
       
   841                          FT_Pos  in_y,
       
   842                          FT_Pos  out_x,
       
   843                          FT_Pos  out_y )
       
   844   {
       
   845     FT_Long  result; /* avoid overflow on 16-bit system */
       
   846 
       
   847 
       
   848     /* deal with the trivial cases quickly */
       
   849     if ( in_y == 0 )
       
   850     {
       
   851       if ( in_x >= 0 )
       
   852         result = out_y;
       
   853       else
       
   854         result = -out_y;
       
   855     }
       
   856     else if ( in_x == 0 )
       
   857     {
       
   858       if ( in_y >= 0 )
       
   859         result = -out_x;
       
   860       else
       
   861         result = out_x;
       
   862     }
       
   863     else if ( out_y == 0 )
       
   864     {
       
   865       if ( out_x >= 0 )
       
   866         result = in_y;
       
   867       else
       
   868         result = -in_y;
       
   869     }
       
   870     else if ( out_x == 0 )
       
   871     {
       
   872       if ( out_y >= 0 )
       
   873         result = -in_x;
       
   874       else
       
   875         result =  in_x;
       
   876     }
       
   877     else /* general case */
       
   878     {
       
   879 #ifdef FT_LONG64
       
   880 
       
   881       FT_Int64  delta = (FT_Int64)in_x * out_y - (FT_Int64)in_y * out_x;
       
   882 
       
   883 
       
   884       if ( delta == 0 )
       
   885         result = 0;
       
   886       else
       
   887         result = 1 - 2 * ( delta < 0 );
       
   888 
       
   889 #else
       
   890 
       
   891       FT_Int64  z1, z2;
       
   892 
       
   893 
       
   894       /* XXX: this function does not allow 64-bit arguments */
       
   895       ft_multo64( (FT_Int32)in_x, (FT_Int32)out_y, &z1 );
       
   896       ft_multo64( (FT_Int32)in_y, (FT_Int32)out_x, &z2 );
       
   897 
       
   898       if ( z1.hi > z2.hi )
       
   899         result = +1;
       
   900       else if ( z1.hi < z2.hi )
       
   901         result = -1;
       
   902       else if ( z1.lo > z2.lo )
       
   903         result = +1;
       
   904       else if ( z1.lo < z2.lo )
       
   905         result = -1;
       
   906       else
       
   907         result = 0;
       
   908 
       
   909 #endif
       
   910     }
       
   911 
       
   912     /* XXX: only the sign of return value, +1/0/-1 must be used */
       
   913     return (FT_Int)result;
       
   914   }
       
   915 
       
   916 
       
   917   /* documentation is in ftcalc.h */
       
   918 
       
   919   FT_BASE_DEF( FT_Int )
       
   920   ft_corner_is_flat( FT_Pos  in_x,
       
   921                      FT_Pos  in_y,
       
   922                      FT_Pos  out_x,
       
   923                      FT_Pos  out_y )
       
   924   {
       
   925     FT_Pos  ax = in_x;
       
   926     FT_Pos  ay = in_y;
       
   927 
       
   928     FT_Pos  d_in, d_out, d_corner;
       
   929 
       
   930 
       
   931     if ( ax < 0 )
       
   932       ax = -ax;
       
   933     if ( ay < 0 )
       
   934       ay = -ay;
       
   935     d_in = ax + ay;
       
   936 
       
   937     ax = out_x;
       
   938     if ( ax < 0 )
       
   939       ax = -ax;
       
   940     ay = out_y;
       
   941     if ( ay < 0 )
       
   942       ay = -ay;
       
   943     d_out = ax + ay;
       
   944 
       
   945     ax = out_x + in_x;
       
   946     if ( ax < 0 )
       
   947       ax = -ax;
       
   948     ay = out_y + in_y;
       
   949     if ( ay < 0 )
       
   950       ay = -ay;
       
   951     d_corner = ax + ay;
       
   952 
       
   953     return ( d_in + d_out - d_corner ) < ( d_corner >> 4 );
       
   954   }
       
   955 
       
   956 
       
   957 /* END */