91
92 inline unsigned int decompose_compatibility (hb_codepoint_t u,
93 hb_codepoint_t *decomposed)
94 {
95 unsigned int ret = func.decompose_compatibility (this, u, decomposed, user_data.decompose_compatibility);
96 if (ret == 1 && u == decomposed[0]) {
97 decomposed[0] = 0;
98 return 0;
99 }
100 decomposed[ret] = 0;
101 return ret;
102 }
103
104
105 inline unsigned int
106 modified_combining_class (hb_codepoint_t unicode)
107 {
108 /* XXX This hack belongs to the Myanmar shaper. */
109 if (unlikely (unicode == 0x1037u)) unicode = 0x103Au;
110
111 /* XXX This hack belongs to the SEA shaper (for Tai Tham):
112 * Reorder SAKOT to ensure it comes after any tone marks. */
113 if (unlikely (unicode == 0x1A60u)) return 254;
114
115 /* XXX This hack belongs to the Tibetan shaper:
116 * Reorder PADMA to ensure it comes after any vowel marks. */
117 if (unlikely (unicode == 0x0FC6u)) return 254;
118 /* Reorder TSA -PHRU to reorder before U+0F74 */
119 if (unlikely (unicode == 0x0F39u)) return 127;
120
121 return _hb_modified_combining_class[combining_class (unicode)];
122 }
123
124 static inline hb_bool_t
125 is_variation_selector (hb_codepoint_t unicode)
126 {
127 /* U+180B..180D MONGOLIAN FREE VARIATION SELECTORs are handled in the
128 * Arabic shaper. No need to match them here. */
129 return unlikely (hb_in_ranges (unicode,
130 0xFE00u, 0xFE0Fu, /* VARIATION SELECTOR-1..16 */
131 0xE0100u, 0xE01EFu)); /* VARIATION SELECTOR-17..256 */
132 }
133
134 /* Default_Ignorable codepoints:
135 *
136 * Note: While U+115F, U+1160, U+3164 and U+FFA0 are Default_Ignorable,
137 * we do NOT want to hide them, as the way Uniscribe has implemented them
138 * is with regular spacing glyphs, and that's the way fonts are made to work.
139 * As such, we make exceptions for those four.
140 *
141 * Unicode 7.0:
142 * $ grep '; Default_Ignorable_Code_Point ' DerivedCoreProperties.txt | sed 's/;.*#/#/'
143 * 00AD # Cf SOFT HYPHEN
144 * 034F # Mn COMBINING GRAPHEME JOINER
145 * 061C # Cf ARABIC LETTER MARK
146 * 115F..1160 # Lo [2] HANGUL CHOSEONG FILLER..HANGUL JUNGSEONG FILLER
147 * 17B4..17B5 # Mn [2] KHMER VOWEL INHERENT AQ..KHMER VOWEL INHERENT AA
148 * 180B..180D # Mn [3] MONGOLIAN FREE VARIATION SELECTOR ONE..MONGOLIAN FREE VARIATION SELECTOR THREE
149 * 180E # Cf MONGOLIAN VOWEL SEPARATOR
150 * 200B..200F # Cf [5] ZERO WIDTH SPACE..RIGHT-TO-LEFT MARK
151 * 202A..202E # Cf [5] LEFT-TO-RIGHT EMBEDDING..RIGHT-TO-LEFT OVERRIDE
152 * 2060..2064 # Cf [5] WORD JOINER..INVISIBLE PLUS
153 * 2065 # Cn <reserved-2065>
154 * 2066..206F # Cf [10] LEFT-TO-RIGHT ISOLATE..NOMINAL DIGIT SHAPES
155 * 3164 # Lo HANGUL FILLER
156 * FE00..FE0F # Mn [16] VARIATION SELECTOR-1..VARIATION SELECTOR-16
157 * FEFF # Cf ZERO WIDTH NO-BREAK SPACE
158 * FFA0 # Lo HALFWIDTH HANGUL FILLER
159 * FFF0..FFF8 # Cn [9] <reserved-FFF0>..<reserved-FFF8>
162 * E0000 # Cn <reserved-E0000>
163 * E0001 # Cf LANGUAGE TAG
164 * E0002..E001F # Cn [30] <reserved-E0002>..<reserved-E001F>
165 * E0020..E007F # Cf [96] TAG SPACE..CANCEL TAG
166 * E0080..E00FF # Cn [128] <reserved-E0080>..<reserved-E00FF>
167 * E0100..E01EF # Mn [240] VARIATION SELECTOR-17..VARIATION SELECTOR-256
168 * E01F0..E0FFF # Cn [3600] <reserved-E01F0>..<reserved-E0FFF>
169 */
170 static inline hb_bool_t
171 is_default_ignorable (hb_codepoint_t ch)
172 {
173 hb_codepoint_t plane = ch >> 16;
174 if (likely (plane == 0))
175 {
176 /* BMP */
177 hb_codepoint_t page = ch >> 8;
178 switch (page) {
179 case 0x00: return unlikely (ch == 0x00ADu);
180 case 0x03: return unlikely (ch == 0x034Fu);
181 case 0x06: return unlikely (ch == 0x061Cu);
182 case 0x17: return hb_in_range (ch, 0x17B4u, 0x17B5u);
183 case 0x18: return hb_in_range (ch, 0x180Bu, 0x180Eu);
184 case 0x20: return hb_in_ranges (ch, 0x200Bu, 0x200Fu,
185 0x202Au, 0x202Eu,
186 0x2060u, 0x206Fu);
187 case 0xFE: return hb_in_range (ch, 0xFE00u, 0xFE0Fu) || ch == 0xFEFFu;
188 case 0xFF: return hb_in_range (ch, 0xFFF0u, 0xFFF8u);
189 default: return false;
190 }
191 }
192 else
193 {
194 /* Other planes */
195 switch (plane) {
196 case 0x01: return hb_in_ranges (ch, 0x1BCA0u, 0x1BCA3u,
197 0x1D173u, 0x1D17Au);
198 case 0x0E: return hb_in_range (ch, 0xE0000u, 0xE0FFFu);
199 default: return false;
200 }
201 }
202 }
203
204 /* Space estimates based on:
205 * http://www.unicode.org/charts/PDF/U2000.pdf
206 * https://www.microsoft.com/typography/developers/fdsspec/spaces.aspx
207 */
208 enum space_t {
209 NOT_SPACE = 0,
210 SPACE_EM = 1,
211 SPACE_EM_2 = 2,
212 SPACE_EM_3 = 3,
213 SPACE_EM_4 = 4,
214 SPACE_EM_5 = 5,
215 SPACE_EM_6 = 6,
216 SPACE_EM_16 = 16,
217 SPACE_4_EM_18, /* 4/18th of an EM! */
218 SPACE,
329 * a non-zero ccc. That makes them reorder with the Halant that is
330 * ccc=9. Just zero them, we don't need them in our Indic shaper.
331 */
332 #define HB_MODIFIED_COMBINING_CLASS_CCC84 0 /* length mark */
333 #define HB_MODIFIED_COMBINING_CLASS_CCC91 0 /* ai length mark */
334
335 /* Thai
336 *
337 * Modify U+0E38 and U+0E39 (ccc=103) to be reordered before U+0E3A (ccc=9).
338 * Assign 3, which is unassigned otherwise.
339 * Uniscribe does this reordering too.
340 */
341 #define HB_MODIFIED_COMBINING_CLASS_CCC103 3 /* sara u / sara uu */
342 #define HB_MODIFIED_COMBINING_CLASS_CCC107 107 /* mai * */
343
344 /* Lao */
345 #define HB_MODIFIED_COMBINING_CLASS_CCC118 118 /* sign u / sign uu */
346 #define HB_MODIFIED_COMBINING_CLASS_CCC122 122 /* mai * */
347
348 /* Tibetan
349 * Modify U+0F74 (ccc=132) to reorder before ccc=130 marks.
350 */
351 #define HB_MODIFIED_COMBINING_CLASS_CCC129 129 /* sign aa */
352 #define HB_MODIFIED_COMBINING_CLASS_CCC130 130 /* sign i */
353 #define HB_MODIFIED_COMBINING_CLASS_CCC132 128 /* sign u */
354
355
356 /* Misc */
357
358 #define HB_UNICODE_GENERAL_CATEGORY_IS_MARK(gen_cat) \
359 (FLAG_SAFE (gen_cat) & \
360 (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
361 FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \
362 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)))
363
364 #define HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL(gen_cat) \
365 (FLAG_SAFE (gen_cat) & \
366 (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
367 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | \
368 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL)))
369
370 #endif /* HB_UNICODE_PRIVATE_HH */
|
91
92 inline unsigned int decompose_compatibility (hb_codepoint_t u,
93 hb_codepoint_t *decomposed)
94 {
95 unsigned int ret = func.decompose_compatibility (this, u, decomposed, user_data.decompose_compatibility);
96 if (ret == 1 && u == decomposed[0]) {
97 decomposed[0] = 0;
98 return 0;
99 }
100 decomposed[ret] = 0;
101 return ret;
102 }
103
104
105 inline unsigned int
106 modified_combining_class (hb_codepoint_t unicode)
107 {
108 /* XXX This hack belongs to the Myanmar shaper. */
109 if (unlikely (unicode == 0x1037u)) unicode = 0x103Au;
110
111 /* XXX This hack belongs to the USE shaper (for Tai Tham):
112 * Reorder SAKOT to ensure it comes after any tone marks. */
113 if (unlikely (unicode == 0x1A60u)) return 254;
114
115 /* XXX This hack belongs to the Tibetan shaper:
116 * Reorder PADMA to ensure it comes after any vowel marks. */
117 if (unlikely (unicode == 0x0FC6u)) return 254;
118 /* Reorder TSA -PHRU to reorder before U+0F74 */
119 if (unlikely (unicode == 0x0F39u)) return 127;
120
121 return _hb_modified_combining_class[combining_class (unicode)];
122 }
123
124 static inline hb_bool_t
125 is_variation_selector (hb_codepoint_t unicode)
126 {
127 /* U+180B..180D MONGOLIAN FREE VARIATION SELECTORs are handled in the
128 * Arabic shaper. No need to match them here. */
129 return unlikely (hb_in_ranges<hb_codepoint_t> (unicode,
130 0xFE00u, 0xFE0Fu, /* VARIATION SELECTOR-1..16 */
131 0xE0100u, 0xE01EFu)); /* VARIATION SELECTOR-17..256 */
132 }
133
134 /* Default_Ignorable codepoints:
135 *
136 * Note: While U+115F, U+1160, U+3164 and U+FFA0 are Default_Ignorable,
137 * we do NOT want to hide them, as the way Uniscribe has implemented them
138 * is with regular spacing glyphs, and that's the way fonts are made to work.
139 * As such, we make exceptions for those four.
140 * Also ignoring U+1BCA0..1BCA3. https://github.com/behdad/harfbuzz/issues/503
141 *
142 * Unicode 7.0:
143 * $ grep '; Default_Ignorable_Code_Point ' DerivedCoreProperties.txt | sed 's/;.*#/#/'
144 * 00AD # Cf SOFT HYPHEN
145 * 034F # Mn COMBINING GRAPHEME JOINER
146 * 061C # Cf ARABIC LETTER MARK
147 * 115F..1160 # Lo [2] HANGUL CHOSEONG FILLER..HANGUL JUNGSEONG FILLER
148 * 17B4..17B5 # Mn [2] KHMER VOWEL INHERENT AQ..KHMER VOWEL INHERENT AA
149 * 180B..180D # Mn [3] MONGOLIAN FREE VARIATION SELECTOR ONE..MONGOLIAN FREE VARIATION SELECTOR THREE
150 * 180E # Cf MONGOLIAN VOWEL SEPARATOR
151 * 200B..200F # Cf [5] ZERO WIDTH SPACE..RIGHT-TO-LEFT MARK
152 * 202A..202E # Cf [5] LEFT-TO-RIGHT EMBEDDING..RIGHT-TO-LEFT OVERRIDE
153 * 2060..2064 # Cf [5] WORD JOINER..INVISIBLE PLUS
154 * 2065 # Cn <reserved-2065>
155 * 2066..206F # Cf [10] LEFT-TO-RIGHT ISOLATE..NOMINAL DIGIT SHAPES
156 * 3164 # Lo HANGUL FILLER
157 * FE00..FE0F # Mn [16] VARIATION SELECTOR-1..VARIATION SELECTOR-16
158 * FEFF # Cf ZERO WIDTH NO-BREAK SPACE
159 * FFA0 # Lo HALFWIDTH HANGUL FILLER
160 * FFF0..FFF8 # Cn [9] <reserved-FFF0>..<reserved-FFF8>
163 * E0000 # Cn <reserved-E0000>
164 * E0001 # Cf LANGUAGE TAG
165 * E0002..E001F # Cn [30] <reserved-E0002>..<reserved-E001F>
166 * E0020..E007F # Cf [96] TAG SPACE..CANCEL TAG
167 * E0080..E00FF # Cn [128] <reserved-E0080>..<reserved-E00FF>
168 * E0100..E01EF # Mn [240] VARIATION SELECTOR-17..VARIATION SELECTOR-256
169 * E01F0..E0FFF # Cn [3600] <reserved-E01F0>..<reserved-E0FFF>
170 */
171 static inline hb_bool_t
172 is_default_ignorable (hb_codepoint_t ch)
173 {
174 hb_codepoint_t plane = ch >> 16;
175 if (likely (plane == 0))
176 {
177 /* BMP */
178 hb_codepoint_t page = ch >> 8;
179 switch (page) {
180 case 0x00: return unlikely (ch == 0x00ADu);
181 case 0x03: return unlikely (ch == 0x034Fu);
182 case 0x06: return unlikely (ch == 0x061Cu);
183 case 0x17: return hb_in_range<hb_codepoint_t> (ch, 0x17B4u, 0x17B5u);
184 case 0x18: return hb_in_range<hb_codepoint_t> (ch, 0x180Bu, 0x180Eu);
185 case 0x20: return hb_in_ranges<hb_codepoint_t> (ch, 0x200Bu, 0x200Fu,
186 0x202Au, 0x202Eu,
187 0x2060u, 0x206Fu);
188 case 0xFE: return hb_in_range<hb_codepoint_t> (ch, 0xFE00u, 0xFE0Fu) || ch == 0xFEFFu;
189 case 0xFF: return hb_in_range<hb_codepoint_t> (ch, 0xFFF0u, 0xFFF8u);
190 default: return false;
191 }
192 }
193 else
194 {
195 /* Other planes */
196 switch (plane) {
197 case 0x01: return hb_in_range<hb_codepoint_t> (ch, 0x1D173u, 0x1D17Au);
198 case 0x0E: return hb_in_range<hb_codepoint_t> (ch, 0xE0000u, 0xE0FFFu);
199 default: return false;
200 }
201 }
202 }
203
204 /* Space estimates based on:
205 * http://www.unicode.org/charts/PDF/U2000.pdf
206 * https://www.microsoft.com/typography/developers/fdsspec/spaces.aspx
207 */
208 enum space_t {
209 NOT_SPACE = 0,
210 SPACE_EM = 1,
211 SPACE_EM_2 = 2,
212 SPACE_EM_3 = 3,
213 SPACE_EM_4 = 4,
214 SPACE_EM_5 = 5,
215 SPACE_EM_6 = 6,
216 SPACE_EM_16 = 16,
217 SPACE_4_EM_18, /* 4/18th of an EM! */
218 SPACE,
329 * a non-zero ccc. That makes them reorder with the Halant that is
330 * ccc=9. Just zero them, we don't need them in our Indic shaper.
331 */
332 #define HB_MODIFIED_COMBINING_CLASS_CCC84 0 /* length mark */
333 #define HB_MODIFIED_COMBINING_CLASS_CCC91 0 /* ai length mark */
334
335 /* Thai
336 *
337 * Modify U+0E38 and U+0E39 (ccc=103) to be reordered before U+0E3A (ccc=9).
338 * Assign 3, which is unassigned otherwise.
339 * Uniscribe does this reordering too.
340 */
341 #define HB_MODIFIED_COMBINING_CLASS_CCC103 3 /* sara u / sara uu */
342 #define HB_MODIFIED_COMBINING_CLASS_CCC107 107 /* mai * */
343
344 /* Lao */
345 #define HB_MODIFIED_COMBINING_CLASS_CCC118 118 /* sign u / sign uu */
346 #define HB_MODIFIED_COMBINING_CLASS_CCC122 122 /* mai * */
347
348 /* Tibetan
349 *
350 * In case of multiple vowel-signs, use u first (but after achung)
351 * this allows Dzongkha multi-vowel shortcuts to render correctly
352 */
353 #define HB_MODIFIED_COMBINING_CLASS_CCC129 129 /* sign aa */
354 #define HB_MODIFIED_COMBINING_CLASS_CCC130 132 /* sign i */
355 #define HB_MODIFIED_COMBINING_CLASS_CCC132 131 /* sign u */
356
357 /* Misc */
358
359 #define HB_UNICODE_GENERAL_CATEGORY_IS_MARK(gen_cat) \
360 (FLAG_UNSAFE (gen_cat) & \
361 (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
362 FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \
363 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)))
364
365 #define HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL(gen_cat) \
366 (FLAG_UNSAFE (gen_cat) & \
367 (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
368 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | \
369 FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL)))
370
371 #endif /* HB_UNICODE_PRIVATE_HH */
|