src/hb-ot-shape-complex-use-machine.rl

   1 /*
   2  * Copyright © 2015  Mozilla Foundation.
   3  * Copyright © 2015  Google, Inc.
   4  *
   5  *  This is part of HarfBuzz, a text shaping library.
   6  *
   7  * Permission is hereby granted, without written agreement and without
   8  * license or royalty fees, to use, copy, modify, and distribute this
   9  * software and its documentation for any purpose, provided that the
  10  * above copyright notice and the following two paragraphs appear in
  11  * all copies of this software.
  12  *
  13  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
  14  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
  15  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
  16  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
  17  * DAMAGE.
  18  *
  19  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
  20  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
  21  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
  22  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
  23  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
  24  *
  25  * Mozilla Author(s): Jonathan Kew
  26  * Google Author(s): Behdad Esfahbod
  27  */
  28
  29 #ifndef HB_OT_SHAPE_COMPLEX_USE_MACHINE_HH
  30 #define HB_OT_SHAPE_COMPLEX_USE_MACHINE_HH
  31
  32 #include "hb-private.hh"
  33
  34 %%{
  35   machine use_syllable_machine;
  36   alphtype unsigned char;
  37   write data;
  38 }%%
  39
  40 %%{
  41
  42 # Same order as enum use_category_t.  Not sure how to avoid duplication.
  43
  44 O       = 0; # OTHER
  45
  46 B       = 1; # BASE
  47 IV      = 2; # BASE_VOWEL
  48 IND     = 3; # BASE_IND
  49 N       = 4; # BASE_NUM
  50 GB      = 5; # BASE_OTHER
  51 CGJ     = 6; # CGJ
  52 #F      = 7; # CONS_FINAL
  53 FM      = 8; # CONS_FINAL_MOD
  54 #M      = 9; # CONS_MED
  55 #CM     = 10; # CONS_MOD
  56 SUB     = 11; # CONS_SUB
  57 H       = 12; # HALANT
  58
  59 HN      = 13; # HALANT_NUM
  60 ZWNJ    = 14; # Zero width non-joiner
  61 ZWJ     = 15; # Zero width joiner
  62 WJ      = 16; # Word joiner
  63 Rsv     = 17; # Reserved characters
  64 R       = 18; # REPHA
  65 S       = 19; # SYM
  66 #SM     = 20; # SYM_MOD
  67 VS      = 21; # VARIATION_SELECTOR
  68 #V      = 36; # VOWEL
  69 #VM     = 40; # VOWEL_MOD
  70
  71 FAbv    = 24; # CONS_FINAL_ABOVE
  72 FBlw    = 25; # CONS_FINAL_BELOW
  73 FPst    = 26; # CONS_FINAL_POST
  74 MAbv    = 27; # CONS_MED_ABOVE
  75 MBlw    = 28; # CONS_MED_BELOW
  76 MPst    = 29; # CONS_MED_POST
  77 MPre    = 30; # CONS_MED_PRE
  78 CMAbv   = 31; # CONS_MOD_ABOVE
  79 CMBlw   = 32; # CONS_MOD_BELOW
  80 VAbv    = 33; # VOWEL_ABOVE / VOWEL_ABOVE_BELOW / VOWEL_ABOVE_BELOW_POST / VOWEL_ABOVE_POST
  81 VBlw    = 34; # VOWEL_BELOW / VOWEL_BELOW_POST
  82 VPst    = 35; # VOWEL_POST      UIPC = Right
  83 VPre    = 22; # VOWEL_PRE / VOWEL_PRE_ABOVE / VOWEL_PRE_ABOVE_POST / VOWEL_PRE_POST
  84 VMAbv   = 37; # VOWEL_MOD_ABOVE
  85 VMBlw   = 38; # VOWEL_MOD_BELOW
  86 VMPst   = 39; # VOWEL_MOD_POST
  87 VMPre   = 23; # VOWEL_MOD_PRE
  88 SMAbv   = 41; # SYM_MOD_ABOVE
  89 SMBlw   = 42; # SYM_MOD_BELOW
  90
  91
  92 consonant_modifiers = CMAbv* CMBlw* ((H B | SUB) VS? CMAbv? CMBlw*)*;
  93 medial_consonants = MPre? MAbv? MBlw? MPst?;
  94 dependent_vowels = VPre* VAbv* VBlw* VPst*;
  95 vowel_modifiers = VMPre* VMAbv* VMBlw* VMPst*;
  96 final_consonants = FAbv* FBlw* FPst* FM?;
  97
  98 virama_terminated_cluster =
  99         R? (B | GB | IV) VS?
 100         consonant_modifiers
 101         H
 102 ;
 103 consonant_cluster =
 104         R? (B | GB) VS?
 105         consonant_modifiers
 106         medial_consonants
 107         dependent_vowels
 108         vowel_modifiers
 109         final_consonants
 110 ;
 111 vowel_cluster =
 112         R? (IV) VS?
 113         consonant_modifiers
 114         medial_consonants
 115         vowel_modifiers
 116         final_consonants
 117 ;
 118
 119 broken_cluster =
 120         R?
 121         consonant_modifiers
 122         medial_consonants
 123         dependent_vowels
 124         vowel_modifiers
 125         final_consonants
 126 ;
 127
 128 number_joiner_terminated_cluster = N VS? (HN N VS?)* H;
 129 numeral_cluster = N VS? (HN N VS?)*;
 130 symbol_cluster = S VS? SMAbv* SMBlw*;
 131 independent_cluster = (IND | O | Rsv | WJ) VS?;
 132
 133 main := |*
 134         independent_cluster                     => { found_syllable (independent_cluster); };
 135         virama_terminated_cluster               => { found_syllable (virama_terminated_cluster); };
 136         consonant_cluster                       => { found_syllable (consonant_cluster); };
 137         vowel_cluster                           => { found_syllable (vowel_cluster); };
 138         number_joiner_terminated_cluster        => { found_syllable (number_joiner_terminated_cluster); };
 139         numeral_cluster                         => { found_syllable (numeral_cluster); };
 140         symbol_cluster                          => { found_syllable (symbol_cluster); };
 141         broken_cluster                          => { found_syllable (broken_cluster); };
 142 *|;
 143
 144
 145 }%%
 146
 147 #define found_syllable(syllable_type) \
 148   HB_STMT_START { \
 149     if (0) fprintf (stderr, "syllable %d..%d %s\n", last, p+1, #syllable_type); \
 150     for (unsigned int i = last; i < p+1; i++) \
 151       info[i].syllable() = (syllable_serial << 4) | syllable_type; \
 152     last = p+1; \
 153     syllable_serial++; \
 154     if (unlikely (syllable_serial == 16)) syllable_serial = 1; \
 155   } HB_STMT_END
 156
 157 static void
 158 find_syllables (hb_buffer_t *buffer)
 159 {
 160   unsigned int p, pe, eof, ts HB_UNUSED, te HB_UNUSED, act HB_UNUSED;
 161   int cs;
 162   hb_glyph_info_t *info = buffer->info;
 163   %%{
 164     write init;
 165     getkey info[p].use_category();
 166   }%%
 167
 168   p = 0;
 169   pe = eof = buffer->len;
 170
 171   unsigned int last = 0;
 172   unsigned int syllable_serial = 1;
 173   %%{
 174     write exec;
 175   }%%
 176 }
 177
 178 #undef found_syllable
 179
 180 #endif /* HB_OT_SHAPE_COMPLEX_USE_MACHINE_HH */