1 | /* |
2 | * Copyright © 2009 Red Hat, Inc. |
3 | * Copyright © 2011 Codethink Limited |
4 | * Copyright © 2010,2011,2012 Google, Inc. |
5 | * |
6 | * This is part of HarfBuzz, a text shaping library. |
7 | * |
8 | * Permission is hereby granted, without written agreement and without |
9 | * license or royalty fees, to use, copy, modify, and distribute this |
10 | * software and its documentation for any purpose, provided that the |
11 | * above copyright notice and the following two paragraphs appear in |
12 | * all copies of this software. |
13 | * |
14 | * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR |
15 | * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES |
16 | * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN |
17 | * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH |
18 | * DAMAGE. |
19 | * |
20 | * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, |
21 | * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND |
22 | * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS |
23 | * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO |
24 | * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS. |
25 | * |
26 | * Red Hat Author(s): Behdad Esfahbod |
27 | * Codethink Author(s): Ryan Lortie |
28 | * Google Author(s): Behdad Esfahbod |
29 | */ |
30 | |
31 | #ifndef HB_UNICODE_HH |
32 | #define HB_UNICODE_HH |
33 | |
34 | #include "hb.hh" |
35 | |
36 | |
37 | extern HB_INTERNAL const uint8_t _hb_modified_combining_class[256]; |
38 | |
39 | /* |
40 | * hb_unicode_funcs_t |
41 | */ |
42 | |
43 | #define HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS \ |
44 | HB_UNICODE_FUNC_IMPLEMENT (combining_class) \ |
45 | HB_UNICODE_FUNC_IMPLEMENT (eastasian_width) \ |
46 | HB_UNICODE_FUNC_IMPLEMENT (general_category) \ |
47 | HB_UNICODE_FUNC_IMPLEMENT (mirroring) \ |
48 | HB_UNICODE_FUNC_IMPLEMENT (script) \ |
49 | HB_UNICODE_FUNC_IMPLEMENT (compose) \ |
50 | HB_UNICODE_FUNC_IMPLEMENT (decompose) \ |
51 | HB_UNICODE_FUNC_IMPLEMENT (decompose_compatibility) \ |
52 | /* ^--- Add new callbacks here */ |
53 | |
54 | /* Simple callbacks are those taking a hb_codepoint_t and returning a hb_codepoint_t */ |
55 | #define HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS_SIMPLE \ |
56 | HB_UNICODE_FUNC_IMPLEMENT (hb_unicode_combining_class_t, combining_class) \ |
57 | HB_UNICODE_FUNC_IMPLEMENT (unsigned int, eastasian_width) \ |
58 | HB_UNICODE_FUNC_IMPLEMENT (hb_unicode_general_category_t, general_category) \ |
59 | HB_UNICODE_FUNC_IMPLEMENT (hb_codepoint_t, mirroring) \ |
60 | HB_UNICODE_FUNC_IMPLEMENT (hb_script_t, script) \ |
61 | /* ^--- Add new simple callbacks here */ |
62 | |
63 | struct hb_unicode_funcs_t |
64 | { |
65 | hb_object_header_t ; |
66 | ASSERT_POD (); |
67 | |
68 | hb_unicode_funcs_t *parent; |
69 | |
70 | bool immutable; |
71 | |
72 | #define HB_UNICODE_FUNC_IMPLEMENT(return_type, name) \ |
73 | inline return_type name (hb_codepoint_t unicode) { return func.name (this, unicode, user_data.name); } |
74 | HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS_SIMPLE |
75 | #undef HB_UNICODE_FUNC_IMPLEMENT |
76 | |
77 | inline hb_bool_t compose (hb_codepoint_t a, hb_codepoint_t b, |
78 | hb_codepoint_t *ab) |
79 | { |
80 | *ab = 0; |
81 | if (unlikely (!a || !b)) return false; |
82 | return func.compose (this, a, b, ab, user_data.compose); |
83 | } |
84 | |
85 | inline hb_bool_t decompose (hb_codepoint_t ab, |
86 | hb_codepoint_t *a, hb_codepoint_t *b) |
87 | { |
88 | *a = ab; *b = 0; |
89 | return func.decompose (this, ab, a, b, user_data.decompose); |
90 | } |
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/harfbuzz/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> |
161 | * 1BCA0..1BCA3 # Cf [4] SHORTHAND FORMAT LETTER OVERLAP..SHORTHAND FORMAT UP STEP |
162 | * 1D173..1D17A # Cf [8] MUSICAL SYMBOL BEGIN BEAM..MUSICAL SYMBOL END PHRASE |
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 | * https://unicode.org/charts/PDF/U2000.pdf |
206 | * https://docs.microsoft.com/en-us/typography/develop/character-design-standards/whitespace |
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, |
219 | SPACE_FIGURE, |
220 | SPACE_PUNCTUATION, |
221 | SPACE_NARROW, |
222 | }; |
223 | static inline space_t |
224 | space_fallback_type (hb_codepoint_t u) |
225 | { |
226 | switch (u) |
227 | { |
228 | /* All GC=Zs chars that can use a fallback. */ |
229 | default: return NOT_SPACE; /* U+1680 OGHAM SPACE MARK */ |
230 | case 0x0020u: return SPACE; /* U+0020 SPACE */ |
231 | case 0x00A0u: return SPACE; /* U+00A0 NO-BREAK SPACE */ |
232 | case 0x2000u: return SPACE_EM_2; /* U+2000 EN QUAD */ |
233 | case 0x2001u: return SPACE_EM; /* U+2001 EM QUAD */ |
234 | case 0x2002u: return SPACE_EM_2; /* U+2002 EN SPACE */ |
235 | case 0x2003u: return SPACE_EM; /* U+2003 EM SPACE */ |
236 | case 0x2004u: return SPACE_EM_3; /* U+2004 THREE-PER-EM SPACE */ |
237 | case 0x2005u: return SPACE_EM_4; /* U+2005 FOUR-PER-EM SPACE */ |
238 | case 0x2006u: return SPACE_EM_6; /* U+2006 SIX-PER-EM SPACE */ |
239 | case 0x2007u: return SPACE_FIGURE; /* U+2007 FIGURE SPACE */ |
240 | case 0x2008u: return SPACE_PUNCTUATION; /* U+2008 PUNCTUATION SPACE */ |
241 | case 0x2009u: return SPACE_EM_5; /* U+2009 THIN SPACE */ |
242 | case 0x200Au: return SPACE_EM_16; /* U+200A HAIR SPACE */ |
243 | case 0x202Fu: return SPACE_NARROW; /* U+202F NARROW NO-BREAK SPACE */ |
244 | case 0x205Fu: return SPACE_4_EM_18; /* U+205F MEDIUM MATHEMATICAL SPACE */ |
245 | case 0x3000u: return SPACE_EM; /* U+3000 IDEOGRAPHIC SPACE */ |
246 | } |
247 | } |
248 | |
249 | struct { |
250 | #define HB_UNICODE_FUNC_IMPLEMENT(name) hb_unicode_##name##_func_t name; |
251 | HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS |
252 | #undef HB_UNICODE_FUNC_IMPLEMENT |
253 | } func; |
254 | |
255 | struct { |
256 | #define HB_UNICODE_FUNC_IMPLEMENT(name) void *name; |
257 | HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS |
258 | #undef HB_UNICODE_FUNC_IMPLEMENT |
259 | } user_data; |
260 | |
261 | struct { |
262 | #define HB_UNICODE_FUNC_IMPLEMENT(name) hb_destroy_func_t name; |
263 | HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS |
264 | #undef HB_UNICODE_FUNC_IMPLEMENT |
265 | } destroy; |
266 | }; |
267 | DECLARE_NULL_INSTANCE (hb_unicode_funcs_t); |
268 | |
269 | |
270 | /* Modified combining marks */ |
271 | |
272 | /* Hebrew |
273 | * |
274 | * We permute the "fixed-position" classes 10-26 into the order |
275 | * described in the SBL Hebrew manual: |
276 | * |
277 | * https://www.sbl-site.org/Fonts/SBLHebrewUserManual1.5x.pdf |
278 | * |
279 | * (as recommended by: |
280 | * https://forum.fontlab.com/archive-old-microsoft-volt-group/vista-and-diacritic-ordering/msg22823/) |
281 | * |
282 | * More details here: |
283 | * https://bugzilla.mozilla.org/show_bug.cgi?id=662055 |
284 | */ |
285 | #define HB_MODIFIED_COMBINING_CLASS_CCC10 22 /* sheva */ |
286 | #define HB_MODIFIED_COMBINING_CLASS_CCC11 15 /* hataf segol */ |
287 | #define HB_MODIFIED_COMBINING_CLASS_CCC12 16 /* hataf patah */ |
288 | #define HB_MODIFIED_COMBINING_CLASS_CCC13 17 /* hataf qamats */ |
289 | #define HB_MODIFIED_COMBINING_CLASS_CCC14 23 /* hiriq */ |
290 | #define HB_MODIFIED_COMBINING_CLASS_CCC15 18 /* tsere */ |
291 | #define HB_MODIFIED_COMBINING_CLASS_CCC16 19 /* segol */ |
292 | #define HB_MODIFIED_COMBINING_CLASS_CCC17 20 /* patah */ |
293 | #define HB_MODIFIED_COMBINING_CLASS_CCC18 21 /* qamats */ |
294 | #define HB_MODIFIED_COMBINING_CLASS_CCC19 14 /* holam */ |
295 | #define HB_MODIFIED_COMBINING_CLASS_CCC20 24 /* qubuts */ |
296 | #define HB_MODIFIED_COMBINING_CLASS_CCC21 12 /* dagesh */ |
297 | #define HB_MODIFIED_COMBINING_CLASS_CCC22 25 /* meteg */ |
298 | #define HB_MODIFIED_COMBINING_CLASS_CCC23 13 /* rafe */ |
299 | #define HB_MODIFIED_COMBINING_CLASS_CCC24 10 /* shin dot */ |
300 | #define HB_MODIFIED_COMBINING_CLASS_CCC25 11 /* sin dot */ |
301 | #define HB_MODIFIED_COMBINING_CLASS_CCC26 26 /* point varika */ |
302 | |
303 | /* |
304 | * Arabic |
305 | * |
306 | * Modify to move Shadda (ccc=33) before other marks. See: |
307 | * https://unicode.org/faq/normalization.html#8 |
308 | * https://unicode.org/faq/normalization.html#9 |
309 | */ |
310 | #define HB_MODIFIED_COMBINING_CLASS_CCC27 28 /* fathatan */ |
311 | #define HB_MODIFIED_COMBINING_CLASS_CCC28 29 /* dammatan */ |
312 | #define HB_MODIFIED_COMBINING_CLASS_CCC29 30 /* kasratan */ |
313 | #define HB_MODIFIED_COMBINING_CLASS_CCC30 31 /* fatha */ |
314 | #define HB_MODIFIED_COMBINING_CLASS_CCC31 32 /* damma */ |
315 | #define HB_MODIFIED_COMBINING_CLASS_CCC32 33 /* kasra */ |
316 | #define HB_MODIFIED_COMBINING_CLASS_CCC33 27 /* shadda */ |
317 | #define HB_MODIFIED_COMBINING_CLASS_CCC34 34 /* sukun */ |
318 | #define HB_MODIFIED_COMBINING_CLASS_CCC35 35 /* superscript alef */ |
319 | |
320 | /* Syriac */ |
321 | #define HB_MODIFIED_COMBINING_CLASS_CCC36 36 /* superscript alaph */ |
322 | |
323 | /* Telugu |
324 | * |
325 | * Modify Telugu length marks (ccc=84, ccc=91). |
326 | * These are the only matras in the main Indic scripts range that have |
327 | * a non-zero ccc. That makes them reorder with the Halant that is |
328 | * ccc=9. Just zero them, we don't need them in our Indic shaper. |
329 | */ |
330 | #define HB_MODIFIED_COMBINING_CLASS_CCC84 0 /* length mark */ |
331 | #define HB_MODIFIED_COMBINING_CLASS_CCC91 0 /* ai length mark */ |
332 | |
333 | /* Thai |
334 | * |
335 | * Modify U+0E38 and U+0E39 (ccc=103) to be reordered before U+0E3A (ccc=9). |
336 | * Assign 3, which is unassigned otherwise. |
337 | * Uniscribe does this reordering too. |
338 | */ |
339 | #define HB_MODIFIED_COMBINING_CLASS_CCC103 3 /* sara u / sara uu */ |
340 | #define HB_MODIFIED_COMBINING_CLASS_CCC107 107 /* mai * */ |
341 | |
342 | /* Lao */ |
343 | #define HB_MODIFIED_COMBINING_CLASS_CCC118 118 /* sign u / sign uu */ |
344 | #define HB_MODIFIED_COMBINING_CLASS_CCC122 122 /* mai * */ |
345 | |
346 | /* Tibetan |
347 | * |
348 | * In case of multiple vowel-signs, use u first (but after achung) |
349 | * this allows Dzongkha multi-vowel shortcuts to render correctly |
350 | */ |
351 | #define HB_MODIFIED_COMBINING_CLASS_CCC129 129 /* sign aa */ |
352 | #define HB_MODIFIED_COMBINING_CLASS_CCC130 132 /* sign i */ |
353 | #define HB_MODIFIED_COMBINING_CLASS_CCC132 131 /* sign u */ |
354 | |
355 | /* Misc */ |
356 | |
357 | #define HB_UNICODE_GENERAL_CATEGORY_IS_MARK(gen_cat) \ |
358 | (FLAG_UNSAFE (gen_cat) & \ |
359 | (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \ |
360 | FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \ |
361 | FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK))) |
362 | |
363 | #define HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL(gen_cat) \ |
364 | (FLAG_UNSAFE (gen_cat) & \ |
365 | (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \ |
366 | FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) | \ |
367 | FLAG (HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL))) |
368 | |
369 | #endif /* HB_UNICODE_HH */ |
370 | |