unofficial mirror of guile-devel@gnu.org 
 help / color / mirror / Atom feed
* bit-extract using mpz
@ 2003-11-14 22:21 Kevin Ryde
  0 siblings, 0 replies; only message in thread
From: Kevin Ryde @ 2003-11-14 22:21 UTC (permalink / raw)


[-- Attachment #1: Type: text/plain, Size: 2761 bytes --]

        * numbers.c (scm_bit_extract): Use mpz functions, rearrange inum case
        to share some shifting.

The existing bit-operations.test does a good job on this, I couldn't
spot further cases that wanted exercising.

New code, for ease of contemplation:


#define MIN(x,y)  ((x) < (y) ? (x) : (y))

SCM_DEFINE (scm_bit_extract, "bit-extract", 3, 0, 0,
            (SCM n, SCM start, SCM end),
	    "Return the integer composed of the @var{start} (inclusive)\n"
	    "through @var{end} (exclusive) bits of @var{n}.  The\n"
	    "@var{start}th bit becomes the 0-th bit in the result.\n"
	    "\n"
	    "@lisp\n"
	    "(number->string (bit-extract #b1101101010 0 4) 2)\n"
	    "   @result{} \"1010\"\n"
	    "(number->string (bit-extract #b1101101010 4 9) 2)\n"
	    "   @result{} \"10110\"\n"
	    "@end lisp")
#define FUNC_NAME s_scm_bit_extract
{
  unsigned long int istart, iend, bits;
  SCM_VALIDATE_INUM_MIN_COPY (2, start,0, istart);
  SCM_VALIDATE_INUM_MIN_COPY (3, end, 0, iend);
  SCM_ASSERT_RANGE (3, end, (iend >= istart));

  /* how many bits to keep */
  bits = iend - istart;

  if (SCM_INUMP (n))
    {
      long int in = SCM_INUM (n);

      /* When istart>=SCM_I_FIXNUM_BIT we can just limit the shift to
         SCM_I_FIXNUM_BIT-1 to get either 0 or -1 per the sign of "in".
         FIXME: This shift relies on signed right shifts being arithmetic,
         which is not guaranteed by C99. */
      in >>= MIN (istart, SCM_I_FIXNUM_BIT-1);

      if (in < 0 && bits >= SCM_I_FIXNUM_BIT)
	{
	  /* Since we emulate two's complement encoded numbers, this
	   * special case requires us to produce a result that has
	   * more bits than can be stored in a fixnum.
	   */
          SCM result = scm_i_long2big (in);
          mpz_fdiv_r_2exp (SCM_I_BIG_MPZ (result), SCM_I_BIG_MPZ (result),
                           bits);
          return result;
	}

      /* mask down to requisite bits */
      bits = MIN (bits, SCM_I_FIXNUM_BIT);
      return SCM_MAKINUM (in & ((1L << bits) - 1));
    }
  else if (SCM_BIGP (n))
    {
      SCM result;
      if (bits == 1)
        {
          result = SCM_MAKINUM (mpz_tstbit (SCM_I_BIG_MPZ (n), istart));
        }
      else
        {
          /* ENHANCE-ME: It'd be nice not to allocate a new bignum when
             bits<SCM_I_FIXNUM_BIT.  Would want some help from GMP to get
             such bits into a ulong.  */
          result = scm_i_mkbig ();
          mpz_fdiv_q_2exp (SCM_I_BIG_MPZ(result), SCM_I_BIG_MPZ(n), istart);
          mpz_fdiv_r_2exp (SCM_I_BIG_MPZ(result), SCM_I_BIG_MPZ(result), bits);
          result = scm_i_normbig (result);
        }
      scm_remember_upto_here_1 (n);
      return result;
    }
  else
    SCM_WRONG_TYPE_ARG (SCM_ARG1, n);
}
#undef FUNC_NAME



[-- Attachment #2: numbers.c.bit-extract.diff --]
[-- Type: text/plain, Size: 3231 bytes --]

--- numbers.c.~1.206.~	1970-01-01 10:00:01.000000000 +1000
+++ numbers.c	2003-11-13 10:44:13.000000000 +1000
@@ -1475,6 +1475,8 @@
 #undef FUNC_NAME
 
 
+#define MIN(x,y)  ((x) < (y) ? (x) : (y))
+
 SCM_DEFINE (scm_bit_extract, "bit-extract", 3, 0, 0,
             (SCM n, SCM start, SCM end),
 	    "Return the integer composed of the @var{start} (inclusive)\n"
@@ -1489,50 +1491,59 @@
 	    "@end lisp")
 #define FUNC_NAME s_scm_bit_extract
 {
-  unsigned long int istart, iend;
+  unsigned long int istart, iend, bits;
   SCM_VALIDATE_INUM_MIN_COPY (2, start,0, istart);
   SCM_VALIDATE_INUM_MIN_COPY (3, end, 0, iend);
   SCM_ASSERT_RANGE (3, end, (iend >= istart));
 
+  /* how many bits to keep */
+  bits = iend - istart;
+
   if (SCM_INUMP (n))
     {
       long int in = SCM_INUM (n);
-      unsigned long int bits = iend - istart;
+
+      /* When istart>=SCM_I_FIXNUM_BIT we can just limit the shift to
+         SCM_I_FIXNUM_BIT-1 to get either 0 or -1 per the sign of "in".
+         FIXME: This shift relies on signed right shifts being arithmetic,
+         which is not guaranteed by C99. */
+      in >>= MIN (istart, SCM_I_FIXNUM_BIT-1);
 
       if (in < 0 && bits >= SCM_I_FIXNUM_BIT)
 	{
 	  /* Since we emulate two's complement encoded numbers, this
 	   * special case requires us to produce a result that has
-	   * more bits than can be stored in a fixnum.  Thus, we fall
-	   * back to the more general algorithm that is used for
-	   * bignums.
+	   * more bits than can be stored in a fixnum.
 	   */
-	  goto generalcase;
+          SCM result = scm_i_long2big (in);
+          mpz_fdiv_r_2exp (SCM_I_BIG_MPZ (result), SCM_I_BIG_MPZ (result),
+                           bits);
+          return result;
 	}
 
-      if (istart < SCM_I_FIXNUM_BIT)
-	{
-	  in = in >> istart;
-	  if (bits < SCM_I_FIXNUM_BIT)
-	    return SCM_MAKINUM (in & ((1L << bits) - 1));
-	  else /* we know: in >= 0 */
-	    return SCM_MAKINUM (in);
-	}
-      else if (in < 0)
-	return SCM_MAKINUM (-1L & ((1L << bits) - 1));
-      else
-	return SCM_MAKINUM (0);
+      /* mask down to requisite bits */
+      bits = MIN (bits, SCM_I_FIXNUM_BIT);
+      return SCM_MAKINUM (in & ((1L << bits) - 1));
     }
   else if (SCM_BIGP (n))
     {
-    generalcase:
-      {
-	SCM num1 = SCM_MAKINUM (1L);
-	SCM num2 = SCM_MAKINUM (2L);
-	SCM bits = SCM_MAKINUM (iend - istart);
-	SCM mask  = scm_difference (scm_integer_expt (num2, bits), num1);
-	return scm_logand (mask, scm_ash (n, SCM_MAKINUM (-istart)));
-      }
+      SCM result;
+      if (bits == 1)
+        {
+          result = SCM_MAKINUM (mpz_tstbit (SCM_I_BIG_MPZ (n), istart));
+        }
+      else
+        {
+          /* ENHANCE-ME: It'd be nice not to allocate a new bignum when
+             bits<SCM_I_FIXNUM_BIT.  Would want some help from GMP to get
+             such bits into a ulong.  */
+          result = scm_i_mkbig ();
+          mpz_fdiv_q_2exp (SCM_I_BIG_MPZ(result), SCM_I_BIG_MPZ(n), istart);
+          mpz_fdiv_r_2exp (SCM_I_BIG_MPZ(result), SCM_I_BIG_MPZ(result), bits);
+          result = scm_i_normbig (result);
+        }
+      scm_remember_upto_here_1 (n);
+      return result;
     }
   else
     SCM_WRONG_TYPE_ARG (SCM_ARG1, n);

[-- Attachment #3: Type: text/plain, Size: 142 bytes --]

_______________________________________________
Guile-devel mailing list
Guile-devel@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-devel

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2003-11-14 22:21 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-11-14 22:21 bit-extract using mpz Kevin Ryde

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).