Blame wctype/wchar-lookup.h

Packit 6c4009
/* Copyright (C) 2000-2018 Free Software Foundation, Inc.
Packit 6c4009
   This file is part of the GNU C Library.
Packit 6c4009
   Contributed by Bruno Haible <haible@clisp.cons.org>, 2000.
Packit 6c4009
Packit 6c4009
   The GNU C Library is free software; you can redistribute it and/or
Packit 6c4009
   modify it under the terms of the GNU Lesser General Public
Packit 6c4009
   License as published by the Free Software Foundation; either
Packit 6c4009
   version 2.1 of the License, or (at your option) any later version.
Packit 6c4009
Packit 6c4009
   The GNU C Library is distributed in the hope that it will be useful,
Packit 6c4009
   but WITHOUT ANY WARRANTY; without even the implied warranty of
Packit 6c4009
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
Packit 6c4009
   Lesser General Public License for more details.
Packit 6c4009
Packit 6c4009
   You should have received a copy of the GNU Lesser General Public
Packit 6c4009
   License along with the GNU C Library; if not, see
Packit 6c4009
   <http://www.gnu.org/licenses/>.  */
Packit 6c4009
Packit 6c4009
#include <stdint.h>
Packit 6c4009
Packit 6c4009
/* Tables indexed by a wide character are compressed through the use
Packit 6c4009
   of a multi-level lookup.  The compression effect comes from blocks
Packit 6c4009
   that don't need particular data and from blocks that can share their
Packit 6c4009
   data.  */
Packit 6c4009
Packit 6c4009
/* Bit tables are accessed by cutting wc in four blocks of bits:
Packit 6c4009
   - the high 32-q-p bits,
Packit 6c4009
   - the next q bits,
Packit 6c4009
   - the next p bits,
Packit 6c4009
   - the next 5 bits.
Packit 6c4009
Packit 6c4009
	    +------------------+-----+-----+-----+
Packit 6c4009
     wc  =  +     32-q-p-5     |  q  |  p  |  5  |
Packit 6c4009
	    +------------------+-----+-----+-----+
Packit 6c4009
Packit 6c4009
   p and q are variable.  For 16-bit Unicode it is sufficient to
Packit 6c4009
   choose p and q such that q+p+5 <= 16.
Packit 6c4009
Packit 6c4009
   The table contains the following uint32_t words:
Packit 6c4009
   - q+p+5,
Packit 6c4009
   - s = upper exclusive bound for wc >> (q+p+5),
Packit 6c4009
   - p+5,
Packit 6c4009
   - 2^q-1,
Packit 6c4009
   - 2^p-1,
Packit 6c4009
   - 1st-level table: s offsets, pointing into the 2nd-level table,
Packit 6c4009
   - 2nd-level table: k*2^q offsets, pointing into the 3rd-level table,
Packit 6c4009
   - 3rd-level table: j*2^p words, each containing 32 bits of data.
Packit 6c4009
*/
Packit 6c4009
Packit 6c4009
static __inline int
Packit 6c4009
__attribute ((always_inline))
Packit 6c4009
wctype_table_lookup (const char *table, uint32_t wc)
Packit 6c4009
{
Packit 6c4009
  uint32_t shift1 = ((const uint32_t *) table)[0];
Packit 6c4009
  uint32_t index1 = wc >> shift1;
Packit 6c4009
  uint32_t bound = ((const uint32_t *) table)[1];
Packit 6c4009
  if (index1 < bound)
Packit 6c4009
    {
Packit 6c4009
      uint32_t lookup1 = ((const uint32_t *) table)[5 + index1];
Packit 6c4009
      if (lookup1 != 0)
Packit 6c4009
	{
Packit 6c4009
	  uint32_t shift2 = ((const uint32_t *) table)[2];
Packit 6c4009
	  uint32_t mask2 = ((const uint32_t *) table)[3];
Packit 6c4009
	  uint32_t index2 = (wc >> shift2) & mask2;
Packit 6c4009
	  uint32_t lookup2 = ((const uint32_t *)(table + lookup1))[index2];
Packit 6c4009
	  if (lookup2 != 0)
Packit 6c4009
	    {
Packit 6c4009
	      uint32_t mask3 = ((const uint32_t *) table)[4];
Packit 6c4009
	      uint32_t index3 = (wc >> 5) & mask3;
Packit 6c4009
	      uint32_t lookup3 = ((const uint32_t *)(table + lookup2))[index3];
Packit 6c4009
Packit 6c4009
	      return (lookup3 >> (wc & 0x1f)) & 1;
Packit 6c4009
	    }
Packit 6c4009
	}
Packit 6c4009
    }
Packit 6c4009
  return 0;
Packit 6c4009
}
Packit 6c4009
Packit 6c4009
/* Byte tables are similar to bit tables, except that the addressing
Packit 6c4009
   unit is a single byte, and no 5 bits are used as a word index.  */
Packit 6c4009
Packit 6c4009
static __inline int
Packit 6c4009
__attribute ((always_inline))
Packit 6c4009
wcwidth_table_lookup (const char *table, uint32_t wc)
Packit 6c4009
{
Packit 6c4009
  uint32_t shift1 = ((const uint32_t *) table)[0];
Packit 6c4009
  uint32_t index1 = wc >> shift1;
Packit 6c4009
  uint32_t bound = ((const uint32_t *) table)[1];
Packit 6c4009
  if (index1 < bound)
Packit 6c4009
    {
Packit 6c4009
      uint32_t lookup1 = ((const uint32_t *) table)[5 + index1];
Packit 6c4009
      if (lookup1 != 0)
Packit 6c4009
	{
Packit 6c4009
	  uint32_t shift2 = ((const uint32_t *) table)[2];
Packit 6c4009
	  uint32_t mask2 = ((const uint32_t *) table)[3];
Packit 6c4009
	  uint32_t index2 = (wc >> shift2) & mask2;
Packit 6c4009
	  uint32_t lookup2 = ((const uint32_t *)(table + lookup1))[index2];
Packit 6c4009
	  if (lookup2 != 0)
Packit 6c4009
	    {
Packit 6c4009
	      uint32_t mask3 = ((const uint32_t *) table)[4];
Packit 6c4009
	      uint32_t index3 = wc & mask3;
Packit 6c4009
	      uint8_t lookup3 = ((const uint8_t *)(table + lookup2))[index3];
Packit 6c4009
Packit 6c4009
	      return lookup3;
Packit 6c4009
	    }
Packit 6c4009
	}
Packit 6c4009
    }
Packit 6c4009
  return 0xff;
Packit 6c4009
}
Packit 6c4009
Packit 6c4009
/* Mapping tables are similar to bit tables, except that the
Packit 6c4009
   addressing unit is a single signed 32-bit word, containing the
Packit 6c4009
   difference between the desired result and the argument, and no 5
Packit 6c4009
   bits are used as a word index.  */
Packit 6c4009
Packit 6c4009
static __inline uint32_t
Packit 6c4009
__attribute ((always_inline))
Packit 6c4009
wctrans_table_lookup (const char *table, uint32_t wc)
Packit 6c4009
{
Packit 6c4009
  uint32_t shift1 = ((const uint32_t *) table)[0];
Packit 6c4009
  uint32_t index1 = wc >> shift1;
Packit 6c4009
  uint32_t bound = ((const uint32_t *) table)[1];
Packit 6c4009
  if (index1 < bound)
Packit 6c4009
    {
Packit 6c4009
      uint32_t lookup1 = ((const uint32_t *) table)[5 + index1];
Packit 6c4009
      if (lookup1 != 0)
Packit 6c4009
	{
Packit 6c4009
	  uint32_t shift2 = ((const uint32_t *) table)[2];
Packit 6c4009
	  uint32_t mask2 = ((const uint32_t *) table)[3];
Packit 6c4009
	  uint32_t index2 = (wc >> shift2) & mask2;
Packit 6c4009
	  uint32_t lookup2 = ((const uint32_t *)(table + lookup1))[index2];
Packit 6c4009
	  if (lookup2 != 0)
Packit 6c4009
	    {
Packit 6c4009
	      uint32_t mask3 = ((const uint32_t *) table)[4];
Packit 6c4009
	      uint32_t index3 = wc & mask3;
Packit 6c4009
	      int32_t lookup3 = ((const int32_t *)(table + lookup2))[index3];
Packit 6c4009
Packit 6c4009
	      return wc + lookup3;
Packit 6c4009
	    }
Packit 6c4009
	}
Packit 6c4009
    }
Packit 6c4009
  return wc;
Packit 6c4009
}