third_party/harfbuzz/src/harfbuzz-hangul.c

   1 /*
   2  * Copyright (C) 2008 Nokia Corporation and/or its subsidiary(-ies)
   3  *
   4  * This is part of HarfBuzz, an OpenType Layout engine 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
  25 #include "harfbuzz-shaper.h"
  26 #include "harfbuzz-shaper-private.h"
  27
  28 #include <assert.h>
  29
  30 /*
  31 // Hangul is a syllable based script. Unicode reserves a large range
  32 // for precomposed hangul, where syllables are already precomposed to
  33 // their final glyph shape. In addition, a so called jamo range is
  34 // defined, that can be used to express old Hangul. Modern hangul
  35 // syllables can also be expressed as jamo, and should be composed
  36 // into syllables. The operation is rather simple and mathematical.
  37
  38 // Every hangul jamo is classified as being either a Leading consonant
  39 // (L), and intermediat Vowel (V) or a trailing consonant (T). Modern
  40 // hangul syllables (the ones in the precomposed area can be of type
  41 // LV or LVT.
  42 //
  43 // Syllable breaks do _not_ occur between:
  44 //
  45 // L              L, V or precomposed
  46 // V, LV          V, T
  47 // LVT, T         T
  48 //
  49 // A standard syllable is of the form L+V+T*. The above rules allow
  50 // nonstandard syllables L*V*T*. To transform them into standard
  51 // syllables fill characters L_f and V_f can be inserted.
  52 */
  53
  54 enum {
  55     Hangul_SBase = 0xac00,
  56     Hangul_LBase = 0x1100,
  57     Hangul_VBase = 0x1161,
  58     Hangul_TBase = 0x11a7,
  59     Hangul_SCount = 11172,
  60     Hangul_LCount = 19,
  61     Hangul_VCount = 21,
  62     Hangul_TCount = 28,
  63     Hangul_NCount = 21*28
  64 };
  65
  66 #define hangul_isPrecomposed(uc) \
  67     (uc >= Hangul_SBase && uc < Hangul_SBase + Hangul_SCount)
  68
  69 #define hangul_isLV(uc) \
  70     ((uc - Hangul_SBase) % Hangul_TCount == 0)
  71
  72 typedef enum {
  73     L,
  74     V,
  75     T,
  76     LV,
  77     LVT,
  78     X
  79 } HangulType;
  80
  81 static HangulType hangul_type(unsigned short uc) {
  82     if (uc > Hangul_SBase && uc < Hangul_SBase + Hangul_SCount)
  83         return hangul_isLV(uc) ? LV : LVT;
  84     if (uc < Hangul_LBase || uc > 0x11ff)
  85         return X;
  86     if (uc < Hangul_VBase)
  87         return L;
  88     if (uc < Hangul_TBase)
  89         return V;
  90     return T;
  91 }
  92
  93 static int hangul_nextSyllableBoundary(const HB_UChar16 *s, int start, int end)
  94 {
  95     const HB_UChar16 *uc = s + start;
  96
  97     HangulType state = hangul_type(*uc);
  98     int pos = 1;
  99
 100     while (pos < end - start) {
 101         HangulType newState = hangul_type(uc[pos]);
 102         switch(newState) {
 103         case X:
 104             goto finish;
 105         case L:
 106         case V:
 107         case T:
 108             if (state > newState)
 109                 goto finish;
 110             state = newState;
 111             break;
 112         case LV:
 113             if (state > L)
 114                 goto finish;
 115             state = V;
 116             break;
 117         case LVT:
 118             if (state > L)
 119                 goto finish;
 120             state = T;
 121         }
 122         ++pos;
 123     }
 124
 125  finish:
 126     return start+pos;
 127 }
 128
 129 #ifndef NO_OPENTYPE
 130 static const HB_OpenTypeFeature hangul_features [] = {
 131     { HB_MAKE_TAG('c', 'c', 'm', 'p'), CcmpProperty },
 132     { HB_MAKE_TAG('l', 'j', 'm', 'o'), CcmpProperty },
 133     { HB_MAKE_TAG('v', 'j', 'm', 'o'), CcmpProperty },
 134     { HB_MAKE_TAG('t', 'j', 'm', 'o'), CcmpProperty },
 135     { 0, 0 }
 136 };
 137 #endif
 138
 139 static HB_Bool hangul_shape_syllable(HB_ShaperItem *item, HB_Bool openType)
 140 {
 141     const HB_UChar16 *ch = item->string + item->item.pos;
 142     int len = item->item.length;
 143 #ifndef NO_OPENTYPE
 144     const int availableGlyphs = item->num_glyphs;
 145 #endif
 146
 147     int i;
 148     HB_UChar16 composed = 0;
 149     /* see if we can compose the syllable into a modern hangul */
 150     if (item->item.length == 2) {
 151         int LIndex = ch[0] - Hangul_LBase;
 152         int VIndex = ch[1] - Hangul_VBase;
 153         if (LIndex >= 0 && LIndex < Hangul_LCount &&
 154             VIndex >= 0 && VIndex < Hangul_VCount)
 155             composed = (LIndex * Hangul_VCount + VIndex) * Hangul_TCount + Hangul_SBase;
 156     } else if (item->item.length == 3) {
 157         int LIndex = ch[0] - Hangul_LBase;
 158         int VIndex = ch[1] - Hangul_VBase;
 159         int TIndex = ch[2] - Hangul_TBase;
 160         if (LIndex >= 0 && LIndex < Hangul_LCount &&
 161             VIndex >= 0 && VIndex < Hangul_VCount &&
 162             TIndex >= 0 && TIndex < Hangul_TCount)
 163             composed = (LIndex * Hangul_VCount + VIndex) * Hangul_TCount + TIndex + Hangul_SBase;
 164     }
 165
 166
 167
 168     /* if we have a modern hangul use the composed form */
 169     if (composed) {
 170         ch = &composed;
 171         len = 1;
 172     }
 173
 174     if (!item->font->klass->convertStringToGlyphIndices(item->font,
 175                                                         ch, len,
 176                                                         item->glyphs, &item->num_glyphs,
 177                                                         item->item.bidiLevel % 2))
 178         return FALSE;
 179     for (i = 0; i < len; i++) {
 180         item->attributes[i].mark = FALSE;
 181         item->attributes[i].clusterStart = FALSE;
 182         item->attributes[i].justification = 0;
 183         item->attributes[i].zeroWidth = FALSE;
 184         /*IDEBUG("    %d: %4x", i, ch[i].unicode()); */
 185     }
 186
 187 #ifndef NO_OPENTYPE
 188     if (!composed && openType) {
 189         HB_Bool positioned;
 190
 191         HB_STACKARRAY(unsigned short, logClusters, len);
 192         for (i = 0; i < len; ++i)
 193             logClusters[i] = i;
 194         item->log_clusters = logClusters;
 195
 196         HB_OpenTypeShape(item, /*properties*/0);
 197
 198         positioned = HB_OpenTypePosition(item, availableGlyphs, /*doLogClusters*/FALSE);
 199
 200         HB_FREE_STACKARRAY(logClusters);
 201
 202         if (!positioned)
 203             return FALSE;
 204     } else {
 205         HB_HeuristicPosition(item);
 206     }
 207 #endif
 208
 209     item->attributes[0].clusterStart = TRUE;
 210     return TRUE;
 211 }
 212
 213 HB_Bool HB_HangulShape(HB_ShaperItem *item)
 214 {
 215     const HB_UChar16 *uc = item->string + item->item.pos;
 216     HB_Bool allPrecomposed = TRUE;
 217     int i;
 218
 219     assert(item->item.script == HB_Script_Hangul);
 220
 221     for (i = 0; i < (int)item->item.length; ++i) {
 222         if (!hangul_isPrecomposed(uc[i])) {
 223             allPrecomposed = FALSE;
 224             break;
 225         }
 226     }
 227
 228     if (!allPrecomposed) {
 229         HB_Bool openType = FALSE;
 230         unsigned short *logClusters = item->log_clusters;
 231         HB_ShaperItem syllable;
 232         int first_glyph = 0;
 233         int sstart = item->item.pos;
 234         int end = sstart + item->item.length;
 235
 236 #ifndef NO_OPENTYPE
 237         openType = HB_SelectScript(item, hangul_features);
 238 #endif
 239         syllable = *item;
 240
 241         while (sstart < end) {
 242             int send = hangul_nextSyllableBoundary(item->string, sstart, end);
 243
 244             syllable.item.pos = sstart;
 245             syllable.item.length = send-sstart;
 246             syllable.glyphs = item->glyphs + first_glyph;
 247             syllable.attributes = item->attributes + first_glyph;
 248             syllable.offsets = item->offsets + first_glyph;
 249             syllable.advances = item->advances + first_glyph;
 250             syllable.num_glyphs = item->num_glyphs - first_glyph;
 251             if (!hangul_shape_syllable(&syllable, openType)) {
 252                 item->num_glyphs += syllable.num_glyphs;
 253                 return FALSE;
 254             }
 255             /* fix logcluster array */
 256             for (i = sstart; i < send; ++i)
 257                 logClusters[i-item->item.pos] = first_glyph;
 258             sstart = send;
 259             first_glyph += syllable.num_glyphs;
 260         }
 261         item->num_glyphs = first_glyph;
 262         return TRUE;
 263     }
 264
 265     return HB_BasicShape(item);
 266 }
 267
 268