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.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 CS      = 43; # CONS_WITH_STACKER
  70 HVM     = 44; # HALANT_OR_VOWEL_MODIFIER
  71 Sk      = 48; # SAKOT
  72
  73 FAbv    = 24; # CONS_FINAL_ABOVE
  74 FBlw    = 25; # CONS_FINAL_BELOW
  75 FPst    = 26; # CONS_FINAL_POST
  76 MAbv    = 27; # CONS_MED_ABOVE
  77 MBlw    = 28; # CONS_MED_BELOW
  78 MPst    = 29; # CONS_MED_POST
  79 MPre    = 30; # CONS_MED_PRE
  80 CMAbv   = 31; # CONS_MOD_ABOVE
  81 CMBlw   = 32; # CONS_MOD_BELOW
  82 VAbv    = 33; # VOWEL_ABOVE / VOWEL_ABOVE_BELOW / VOWEL_ABOVE_BELOW_POST / VOWEL_ABOVE_POST
  83 VBlw    = 34; # VOWEL_BELOW / VOWEL_BELOW_POST
  84 VPst    = 35; # VOWEL_POST      UIPC = Right
  85 VPre    = 22; # VOWEL_PRE / VOWEL_PRE_ABOVE / VOWEL_PRE_ABOVE_POST / VOWEL_PRE_POST
  86 VMAbv   = 37; # VOWEL_MOD_ABOVE
  87 VMBlw   = 38; # VOWEL_MOD_BELOW
  88 VMPst   = 39; # VOWEL_MOD_POST
  89 VMPre   = 23; # VOWEL_MOD_PRE
  90 SMAbv   = 41; # SYM_MOD_ABOVE
  91 SMBlw   = 42; # SYM_MOD_BELOW
  92 FMAbv   = 45; # CONS_FINAL_MOD  UIPC = Top
  93 FMBlw   = 46; # CONS_FINAL_MOD  UIPC = Bottom
  94 FMPst   = 47; # CONS_FINAL_MOD  UIPC = Not_Applicable
  95
  96 h = H | HVM | Sk;
  97
  98 # Override: Adhoc ZWJ placement. https://github.com/harfbuzz/harfbuzz/issues/542#issuecomment-353169729
  99 consonant_modifiers = CMAbv* CMBlw* ((ZWJ?.h.ZWJ? B | SUB) VS? CMAbv? CMBlw*)*;
 100 # Override: Allow two MBlw. https://github.com/harfbuzz/harfbuzz/issues/376
 101 medial_consonants = MPre? MAbv? MBlw?.MBlw? MPst?;
 102 dependent_vowels = VPre* VAbv* VBlw* VPst*;
 103 vowel_modifiers = HVM? VMPre* VMAbv* VMBlw* VMPst*;
 104 final_consonants = FAbv* FBlw* FPst*;
 105 final_modifiers = FMAbv* FMBlw* | FMPst?;
 106
 107 complex_syllable_start = (R | CS)? (B | GB) VS?;
 108 complex_syllable_middle =
 109         consonant_modifiers
 110         medial_consonants
 111         dependent_vowels
 112         vowel_modifiers
 113         (Sk B)*
 114 ;
 115 complex_syllable_tail =
 116         complex_syllable_middle
 117         final_consonants
 118         final_modifiers
 119 ;
 120 number_joiner_terminated_cluster_tail = (HN N VS?)* HN;
 121 numeral_cluster_tail = (HN N VS?)+;
 122 symbol_cluster_tail = SMAbv+ SMBlw* | SMBlw+;
 123
 124 virama_terminated_cluster =
 125         complex_syllable_start
 126         consonant_modifiers
 127         ZWJ?.h.ZWJ?
 128 ;
 129 sakot_terminated_cluster =
 130         complex_syllable_start
 131         complex_syllable_middle
 132         Sk
 133 ;
 134 standard_cluster =
 135         complex_syllable_start
 136         complex_syllable_tail
 137 ;
 138 broken_cluster =
 139         R?
 140         (complex_syllable_tail | number_joiner_terminated_cluster_tail | numeral_cluster_tail | symbol_cluster_tail)
 141 ;
 142
 143 number_joiner_terminated_cluster = N VS? number_joiner_terminated_cluster_tail;
 144 numeral_cluster = N VS? numeral_cluster_tail?;
 145 symbol_cluster = (S | GB) VS? symbol_cluster_tail?;
 146 independent_cluster = (IND | O | Rsv | WJ) VS?;
 147 other = any;
 148
 149 main := |*
 150         independent_cluster                     => { found_syllable (independent_cluster); };
 151         virama_terminated_cluster               => { found_syllable (virama_terminated_cluster); };
 152         sakot_terminated_cluster                => { found_syllable (sakot_terminated_cluster); };
 153         standard_cluster                        => { found_syllable (standard_cluster); };
 154         number_joiner_terminated_cluster        => { found_syllable (number_joiner_terminated_cluster); };
 155         numeral_cluster                         => { found_syllable (numeral_cluster); };
 156         symbol_cluster                          => { found_syllable (symbol_cluster); };
 157         broken_cluster                          => { found_syllable (broken_cluster); };
 158         other                                   => { found_syllable (non_cluster); };
 159 *|;
 160
 161
 162 }%%
 163
 164 #define found_syllable(syllable_type) \
 165   HB_STMT_START { \
 166     if (0) fprintf (stderr, "syllable %d..%d %s\n", ts, te, #syllable_type); \
 167     for (unsigned int i = ts; i < te; i++) \
 168       info[i].syllable() = (syllable_serial << 4) | use_##syllable_type; \
 169     syllable_serial++; \
 170     if (unlikely (syllable_serial == 16)) syllable_serial = 1; \
 171   } HB_STMT_END
 172
 173 static void
 174 find_syllables_use (hb_buffer_t *buffer)
 175 {
 176   unsigned int p, pe, eof, ts, te, act;
 177   int cs;
 178   hb_glyph_info_t *info = buffer->info;
 179   %%{
 180     write init;
 181     getkey info[p].use_category();
 182   }%%
 183
 184   p = 0;
 185   pe = eof = buffer->len;
 186
 187   unsigned int syllable_serial = 1;
 188   %%{
 189     write exec;
 190   }%%
 191 }
 192
 193 #undef found_syllable
 194
 195 #endif /* HB_OT_SHAPE_COMPLEX_USE_MACHINE_HH */