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 IND     = 3; # BASE_IND
  48 N       = 4; # BASE_NUM
  49 GB      = 5; # BASE_OTHER
  50 CGJ     = 6; # CGJ
  51 #F      = 7; # CONS_FINAL
  52 FM      = 8; # CONS_FINAL_MOD
  53 #M      = 9; # CONS_MED
  54 #CM     = 10; # CONS_MOD
  55 SUB     = 11; # CONS_SUB
  56 H       = 12; # HALANT
  57
  58 HN      = 13; # HALANT_NUM
  59 ZWNJ    = 14; # Zero width non-joiner
  60 ZWJ     = 15; # Zero width joiner
  61 WJ      = 16; # Word joiner
  62 Rsv     = 17; # Reserved characters
  63 R       = 18; # REPHA
  64 S       = 19; # SYM
  65 #SM     = 20; # SYM_MOD
  66 VS      = 21; # VARIATION_SELECTOR
  67 #V      = 36; # VOWEL
  68 #VM     = 40; # VOWEL_MOD
  69
  70 FAbv    = 24; # CONS_FINAL_ABOVE
  71 FBlw    = 25; # CONS_FINAL_BELOW
  72 FPst    = 26; # CONS_FINAL_POST
  73 MAbv    = 27; # CONS_MED_ABOVE
  74 MBlw    = 28; # CONS_MED_BELOW
  75 MPst    = 29; # CONS_MED_POST
  76 MPre    = 30; # CONS_MED_PRE
  77 CMAbv   = 31; # CONS_MOD_ABOVE
  78 CMBlw   = 32; # CONS_MOD_BELOW
  79 VAbv    = 33; # VOWEL_ABOVE / VOWEL_ABOVE_BELOW / VOWEL_ABOVE_BELOW_POST / VOWEL_ABOVE_POST
  80 VBlw    = 34; # VOWEL_BELOW / VOWEL_BELOW_POST
  81 VPst    = 35; # VOWEL_POST      UIPC = Right
  82 VPre    = 22; # VOWEL_PRE / VOWEL_PRE_ABOVE / VOWEL_PRE_ABOVE_POST / VOWEL_PRE_POST
  83 VMAbv   = 37; # VOWEL_MOD_ABOVE
  84 VMBlw   = 38; # VOWEL_MOD_BELOW
  85 VMPst   = 39; # VOWEL_MOD_POST
  86 VMPre   = 23; # VOWEL_MOD_PRE
  87 SMAbv   = 41; # SYM_MOD_ABOVE
  88 SMBlw   = 42; # SYM_MOD_BELOW
  89 CS      = 43; # CONS_WITH_STACKER
  90
  91
  92 # Override: Adhoc ZWJ placement. https://github.com/harfbuzz/harfbuzz/issues/542#issuecomment-353169729
  93 consonant_modifiers = CMAbv* CMBlw* ((ZWJ?.H.ZWJ? B | SUB) VS? CMAbv? CMBlw*)*;
  94 # Override: Allow two MBlw. https://github.com/harfbuzz/harfbuzz/issues/376
  95 medial_consonants = MPre? MAbv? MBlw?.MBlw? MPst?;
  96 dependent_vowels = VPre* VAbv* VBlw* VPst*;
  97 vowel_modifiers = VMPre* VMAbv* VMBlw* VMPst*;
  98 final_consonants = FAbv* FBlw* FPst* FM?;
  99
 100 virama_terminated_cluster =
 101         (R|CS)? (B | GB) VS?
 102         consonant_modifiers
 103         ZWJ?.H.ZWJ?
 104 ;
 105 standard_cluster =
 106         (R|CS)? (B | GB) VS?
 107         consonant_modifiers
 108         medial_consonants
 109         dependent_vowels
 110         vowel_modifiers
 111         final_consonants
 112 ;
 113
 114 broken_cluster =
 115         R?
 116         consonant_modifiers
 117         medial_consonants
 118         dependent_vowels
 119         vowel_modifiers
 120         final_consonants
 121 ;
 122
 123 number_joiner_terminated_cluster = N VS? (HN N VS?)* HN;
 124 numeral_cluster = N VS? (HN N VS?)*;
 125 symbol_cluster = S VS? SMAbv* SMBlw*;
 126 independent_cluster = (IND | O | Rsv | WJ) VS?;
 127 other = any;
 128
 129 main := |*
 130         independent_cluster                     => { found_syllable (independent_cluster); };
 131         virama_terminated_cluster               => { found_syllable (virama_terminated_cluster); };
 132         standard_cluster                        => { found_syllable (standard_cluster); };
 133         number_joiner_terminated_cluster        => { found_syllable (number_joiner_terminated_cluster); };
 134         numeral_cluster                         => { found_syllable (numeral_cluster); };
 135         symbol_cluster                          => { found_syllable (symbol_cluster); };
 136         broken_cluster                          => { found_syllable (broken_cluster); };
 137         other                                   => { found_syllable (non_cluster); };
 138 *|;
 139
 140
 141 }%%
 142
 143 #define found_syllable(syllable_type) \
 144   HB_STMT_START { \
 145     if (0) fprintf (stderr, "syllable %d..%d %s\n", last, p+1, #syllable_type); \
 146     for (unsigned int i = last; i < p+1; i++) \
 147       info[i].syllable() = (syllable_serial << 4) | syllable_type; \
 148     last = p+1; \
 149     syllable_serial++; \
 150     if (unlikely (syllable_serial == 16)) syllable_serial = 1; \
 151   } HB_STMT_END
 152
 153 static void
 154 find_syllables (hb_buffer_t *buffer)
 155 {
 156   unsigned int p, pe, eof, ts HB_UNUSED, te, act;
 157   int cs;
 158   hb_glyph_info_t *info = buffer->info;
 159   %%{
 160     write init;
 161     getkey info[p].use_category();
 162   }%%
 163
 164   p = 0;
 165   pe = eof = buffer->len;
 166
 167   unsigned int last = 0;
 168   unsigned int syllable_serial = 1;
 169   %%{
 170     write exec;
 171   }%%
 172 }
 173
 174 #undef found_syllable
 175
 176 #endif /* HB_OT_SHAPE_COMPLEX_USE_MACHINE_HH */