src/hb-ot-shaper-indic-machine.rl

   1 /*
   2  * Copyright © 2011,2012  Google, Inc.
   3  *
   4  *  This is part of HarfBuzz, a text shaping library.
   5  *
   6  * Permission is hereby granted, without written agreement and without
   7  * license or royalty fees, to use, copy, modify, and distribute this
   8  * software and its documentation for any purpose, provided that the
   9  * above copyright notice and the following two paragraphs appear in
  10  * all copies of this software.
  11  *
  12  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
  13  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
  14  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
  15  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
  16  * DAMAGE.
  17  *
  18  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
  19  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  20  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
  21  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
  22  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
  23  *
  24  * Google Author(s): Behdad Esfahbod
  25  */
  26
  27 #ifndef HB_OT_SHAPER_INDIC_MACHINE_HH
  28 #define HB_OT_SHAPER_INDIC_MACHINE_HH
  29
  30 #include "hb.hh"
  31
  32 #include "hb-ot-layout.hh"
  33 #include "hb-ot-shaper-indic.hh"
  34
  35 /* buffer var allocations */
  36 #define indic_category() ot_shaper_var_u8_category() /* indic_category_t */
  37 #define indic_position() ot_shaper_var_u8_auxiliary() /* indic_position_t */
  38
  39 using indic_category_t = unsigned;
  40 using indic_position_t = ot_position_t;
  41
  42 #define I_Cat(Cat) indic_syllable_machine_ex_##Cat
  43
  44 enum indic_syllable_type_t {
  45   indic_consonant_syllable,
  46   indic_vowel_syllable,
  47   indic_standalone_cluster,
  48   indic_symbol_cluster,
  49   indic_broken_cluster,
  50   indic_non_indic_cluster,
  51 };
  52
  53 %%{
  54   machine indic_syllable_machine;
  55   alphtype unsigned char;
  56   write exports;
  57   write data;
  58 }%%
  59
  60 %%{
  61
  62
  63 export X    = 0;
  64 export C    = 1;
  65 export V    = 2;
  66 export N    = 3;
  67 export H    = 4;
  68 export ZWNJ = 5;
  69 export ZWJ  = 6;
  70 export M    = 7;
  71 export SM   = 8;
  72 export A    = 9;
  73 export VD   = 9;
  74 export PLACEHOLDER = 10;
  75 export DOTTEDCIRCLE = 11;
  76 export RS    = 12;
  77 export MPst  = 13;
  78 export Repha = 14;
  79 export Ra    = 15;
  80 export CM    = 16;
  81 export Symbol= 17;
  82 export CS    = 18;
  83
  84
  85 c = (C | Ra);                   # is_consonant
  86 n = ((ZWNJ?.RS)? (N.N?)?);      # is_consonant_modifier
  87 z = ZWJ|ZWNJ;                   # is_joiner
  88 reph = (Ra H | Repha);          # possible reph
  89
  90 cn = c.ZWJ?.n?;
  91 symbol = Symbol.N?;
  92 matra_group = z*.(M | SM? MPst).N?.H?;
  93 syllable_tail = (z?.SM.SM?.ZWNJ?)? (A | VD)*;
  94 halant_group = (z?.H.(ZWJ.N?)?);
  95 final_halant_group = halant_group | H.ZWNJ;
  96 medial_group = CM?;
  97 halant_or_matra_group = (final_halant_group | matra_group*);
  98
  99 complex_syllable_tail = (halant_group.cn)* medial_group halant_or_matra_group syllable_tail;
 100
 101 consonant_syllable =    (Repha|CS)? cn complex_syllable_tail;
 102 vowel_syllable =        reph? V.n? (ZWJ | complex_syllable_tail);
 103 standalone_cluster =    ((Repha|CS)? PLACEHOLDER | reph? DOTTEDCIRCLE).n? complex_syllable_tail;
 104 symbol_cluster =        symbol syllable_tail;
 105 broken_cluster =        reph? n? complex_syllable_tail;
 106 other =                 any;
 107
 108 main := |*
 109         consonant_syllable      => { found_syllable (indic_consonant_syllable); };
 110         vowel_syllable          => { found_syllable (indic_vowel_syllable); };
 111         standalone_cluster      => { found_syllable (indic_standalone_cluster); };
 112         symbol_cluster          => { found_syllable (indic_symbol_cluster); };
 113         broken_cluster          => { found_syllable (indic_broken_cluster); buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_BROKEN_SYLLABLE; };
 114         other                   => { found_syllable (indic_non_indic_cluster); };
 115 *|;
 116
 117
 118 }%%
 119
 120 #define found_syllable(syllable_type) \
 121   HB_STMT_START { \
 122     if (0) fprintf (stderr, "syllable %u..%u %s\n", ts, te, #syllable_type); \
 123     for (unsigned int i = ts; i < te; i++) \
 124       info[i].syllable() = (syllable_serial << 4) | syllable_type; \
 125     syllable_serial++; \
 126     if (syllable_serial == 16) syllable_serial = 1; \
 127   } HB_STMT_END
 128
 129 inline void
 130 find_syllables_indic (hb_buffer_t *buffer)
 131 {
 132   unsigned int p, pe, eof, ts, te, act;
 133   int cs;
 134   hb_glyph_info_t *info = buffer->info;
 135   %%{
 136     write init;
 137     getkey info[p].indic_category();
 138   }%%
 139
 140   p = 0;
 141   pe = eof = buffer->len;
 142
 143   unsigned int syllable_serial = 1;
 144   %%{
 145     write exec;
 146   }%%
 147 }
 148
 149 #undef found_syllable
 150
 151 #endif /* HB_OT_SHAPER_INDIC_MACHINE_HH */