2 * Copyright © 2010 Google, Inc.
4 * This is part of HarfBuzz, a text shaping library.
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.
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
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.
24 * Google Author(s): Behdad Esfahbod
27 #include "hb-ot-shape-complex-private.hh"
32 /* buffer var allocations */
33 #define arabic_shaping_action() var2.u32 /* arabic shaping action */
37 * Bits used in the joining tables
43 JOINING_TYPE_C = JOINING_TYPE_D,
44 JOINING_GROUP_ALAPH = 3,
45 JOINING_GROUP_DALATH_RISH = 4,
46 NUM_STATE_MACHINE_COLS = 5,
48 /* We deliberately don't have a JOINING_TYPE_L since that's unused in Unicode. */
51 JOINING_TYPE_X = 7 /* means: use general-category to choose between U or T. */
58 #include "hb-ot-shape-complex-arabic-table.hh"
60 static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_category_t gen_cat)
62 /* TODO Macroize the magic bit operations */
64 if (likely (JOINING_TABLE_FIRST <= u && u <= JOINING_TABLE_LAST)) {
65 unsigned int j_type = joining_table[u - JOINING_TABLE_FIRST];
66 if (likely (j_type != JOINING_TYPE_X))
70 /* Mongolian joining data is not in ArabicJoining.txt yet */
71 if (unlikely (0x1800 <= u && u <= 0x18AF))
73 /* All letters, SIBE SYLLABLE BOUNDARY MARKER, and NIRUGU are D */
74 if (gen_cat == HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER || u == 0x1807 || u == 0x180A)
75 return JOINING_TYPE_D;
78 if (unlikely ((u & ~(0x200C^0x200D)) == 0x200C)) {
79 return u == 0x200C ? JOINING_TYPE_U : JOINING_TYPE_C;
82 return ((1<<gen_cat) & ((1<<HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)|(1<<HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK)|(1<<HB_UNICODE_GENERAL_CATEGORY_FORMAT))) ?
83 JOINING_TYPE_T : JOINING_TYPE_U;
88 static const hb_tag_t arabic_syriac_features[] =
90 HB_TAG('i','n','i','t'),
91 HB_TAG('m','e','d','i'),
92 HB_TAG('f','i','n','a'),
93 HB_TAG('i','s','o','l'),
95 HB_TAG('m','e','d','2'),
96 HB_TAG('f','i','n','2'),
97 HB_TAG('f','i','n','3'),
102 /* Same order as the feature array */
116 COMMON_NUM_FEATURES = 4,
117 SYRIAC_NUM_FEATURES = 7,
118 TOTAL_NUM_FEATURES = NONE
121 static const struct arabic_state_table_entry {
126 } arabic_state_table[][NUM_STATE_MACHINE_COLS] =
128 /* jt_U, jt_R, jt_D, jg_ALAPH, jg_DALATH_RISH */
130 /* State 0: prev was U, not willing to join. */
131 { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,ISOL,1}, {NONE,ISOL,6}, },
133 /* State 1: prev was R or ISOL/ALAPH, not willing to join. */
134 { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN2,5}, {NONE,ISOL,6}, },
136 /* State 2: prev was D/ISOL, willing to join. */
137 { {NONE,NONE,0}, {INIT,FINA,1}, {INIT,FINA,3}, {INIT,FINA,4}, {INIT,FINA,6}, },
139 /* State 3: prev was D/FINA, willing to join. */
140 { {NONE,NONE,0}, {MEDI,FINA,1}, {MEDI,FINA,3}, {MEDI,FINA,4}, {MEDI,FINA,6}, },
142 /* State 4: prev was FINA ALAPH, not willing to join. */
143 { {NONE,NONE,0}, {MED2,ISOL,1}, {MED2,ISOL,2}, {MED2,FIN2,5}, {MED2,ISOL,6}, },
145 /* State 5: prev was FIN2/FIN3 ALAPH, not willing to join. */
146 { {NONE,NONE,0}, {ISOL,ISOL,1}, {ISOL,ISOL,2}, {ISOL,FIN2,5}, {ISOL,ISOL,6}, },
148 /* State 6: prev was DALATH/RISH, not willing to join. */
149 { {NONE,NONE,0}, {NONE,ISOL,1}, {NONE,ISOL,2}, {NONE,FIN3,5}, {NONE,ISOL,6}, }
155 _hb_ot_shape_complex_collect_features_arabic (hb_ot_shape_planner_t *planner, const hb_segment_properties_t *props)
157 /* ArabicOT spec enables 'cswh' for Arabic where as for basic shaper it's disabled by default. */
158 planner->map.add_bool_feature (HB_TAG('c','s','w','h'));
160 unsigned int num_features = props->script == HB_SCRIPT_SYRIAC ? SYRIAC_NUM_FEATURES : COMMON_NUM_FEATURES;
161 for (unsigned int i = 0; i < num_features; i++)
162 planner->map.add_bool_feature (arabic_syriac_features[i], false);
166 _hb_ot_shape_complex_setup_masks_arabic (hb_ot_shape_context_t *c)
168 unsigned int count = c->buffer->len;
169 unsigned int prev = 0, state = 0;
171 for (unsigned int i = 0; i < count; i++)
173 unsigned int this_type = get_joining_type (c->buffer->info[i].codepoint, (hb_unicode_general_category_t) c->buffer->info[i].general_category());
175 if (unlikely (this_type == JOINING_TYPE_T)) {
176 c->buffer->info[i].arabic_shaping_action() = NONE;
180 const arabic_state_table_entry *entry = &arabic_state_table[state][this_type];
182 if (entry->prev_action != NONE)
183 c->buffer->info[prev].arabic_shaping_action() = entry->prev_action;
185 c->buffer->info[i].arabic_shaping_action() = entry->curr_action;
188 state = entry->next_state;
191 hb_mask_t mask_array[TOTAL_NUM_FEATURES + 1] = {0};
192 unsigned int num_masks = c->buffer->props.script == HB_SCRIPT_SYRIAC ? SYRIAC_NUM_FEATURES : COMMON_NUM_FEATURES;
193 for (unsigned int i = 0; i < num_masks; i++)
194 mask_array[i] = c->plan->map.get_1_mask (arabic_syriac_features[i]);
196 for (unsigned int i = 0; i < count; i++)
197 c->buffer->info[i].mask |= mask_array[c->buffer->info[i].arabic_shaping_action()];