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
37extern 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
63struct hb_unicode_funcs_t
64{
65 hb_object_header_t header;
66
67 hb_unicode_funcs_t *parent;
68
69#define HB_UNICODE_FUNC_IMPLEMENT(return_type, name) \
70 return_type name (hb_codepoint_t unicode) { return func.name (this, unicode, user_data.name); }
71HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS_SIMPLE
72#undef HB_UNICODE_FUNC_IMPLEMENT
73
74 hb_bool_t compose (hb_codepoint_t a, hb_codepoint_t b,
75 hb_codepoint_t *ab)
76 {
77 *ab = 0;
78 if (unlikely (!a || !b)) return false;
79 return func.compose (this, a, b, ab, user_data.compose);
80 }
81
82 hb_bool_t decompose (hb_codepoint_t ab,
83 hb_codepoint_t *a, hb_codepoint_t *b)
84 {
85 *a = ab; *b = 0;
86 return func.decompose (this, ab, a, b, user_data.decompose);
87 }
88
89 unsigned int decompose_compatibility (hb_codepoint_t u,
90 hb_codepoint_t *decomposed)
91 {
92 unsigned int ret = func.decompose_compatibility (this, u, decomposed, user_data.decompose_compatibility);
93 if (ret == 1 && u == decomposed[0]) {
94 decomposed[0] = 0;
95 return 0;
96 }
97 decomposed[ret] = 0;
98 return ret;
99 }
100
101 unsigned int
102 modified_combining_class (hb_codepoint_t u)
103 {
104 /* XXX This hack belongs to the Myanmar shaper. */
105 if (unlikely (u == 0x1037u)) u = 0x103Au;
106
107 /* XXX This hack belongs to the USE shaper (for Tai Tham):
108 * Reorder SAKOT to ensure it comes after any tone marks. */
109 if (unlikely (u == 0x1A60u)) return 254;
110
111 /* XXX This hack belongs to the Tibetan shaper:
112 * Reorder PADMA to ensure it comes after any vowel marks. */
113 if (unlikely (u == 0x0FC6u)) return 254;
114 /* Reorder TSA -PHRU to reorder before U+0F74 */
115 if (unlikely (u == 0x0F39u)) return 127;
116
117 return _hb_modified_combining_class[combining_class (u)];
118 }
119
120 static hb_bool_t
121 is_variation_selector (hb_codepoint_t unicode)
122 {
123 /* U+180B..180D MONGOLIAN FREE VARIATION SELECTORs are handled in the
124 * Arabic shaper. No need to match them here. */
125 return unlikely (hb_in_ranges<hb_codepoint_t> (unicode,
126 0xFE00u, 0xFE0Fu, /* VARIATION SELECTOR-1..16 */
127 0xE0100u, 0xE01EFu)); /* VARIATION SELECTOR-17..256 */
128 }
129
130 /* Default_Ignorable codepoints:
131 *
132 * Note: While U+115F, U+1160, U+3164 and U+FFA0 are Default_Ignorable,
133 * we do NOT want to hide them, as the way Uniscribe has implemented them
134 * is with regular spacing glyphs, and that's the way fonts are made to work.
135 * As such, we make exceptions for those four.
136 * Also ignoring U+1BCA0..1BCA3. https://github.com/harfbuzz/harfbuzz/issues/503
137 *
138 * Unicode 7.0:
139 * $ grep '; Default_Ignorable_Code_Point ' DerivedCoreProperties.txt | sed 's/;.*#/#/'
140 * 00AD # Cf SOFT HYPHEN
141 * 034F # Mn COMBINING GRAPHEME JOINER
142 * 061C # Cf ARABIC LETTER MARK
143 * 115F..1160 # Lo [2] HANGUL CHOSEONG FILLER..HANGUL JUNGSEONG FILLER
144 * 17B4..17B5 # Mn [2] KHMER VOWEL INHERENT AQ..KHMER VOWEL INHERENT AA
145 * 180B..180D # Mn [3] MONGOLIAN FREE VARIATION SELECTOR ONE..MONGOLIAN FREE VARIATION SELECTOR THREE
146 * 180E # Cf MONGOLIAN VOWEL SEPARATOR
147 * 200B..200F # Cf [5] ZERO WIDTH SPACE..RIGHT-TO-LEFT MARK
148 * 202A..202E # Cf [5] LEFT-TO-RIGHT EMBEDDING..RIGHT-TO-LEFT OVERRIDE
149 * 2060..2064 # Cf [5] WORD JOINER..INVISIBLE PLUS
150 * 2065 # Cn <reserved-2065>
151 * 2066..206F # Cf [10] LEFT-TO-RIGHT ISOLATE..NOMINAL DIGIT SHAPES
152 * 3164 # Lo HANGUL FILLER
153 * FE00..FE0F # Mn [16] VARIATION SELECTOR-1..VARIATION SELECTOR-16
154 * FEFF # Cf ZERO WIDTH NO-BREAK SPACE
155 * FFA0 # Lo HALFWIDTH HANGUL FILLER
156 * FFF0..FFF8 # Cn [9] <reserved-FFF0>..<reserved-FFF8>
157 * 1BCA0..1BCA3 # Cf [4] SHORTHAND FORMAT LETTER OVERLAP..SHORTHAND FORMAT UP STEP
158 * 1D173..1D17A # Cf [8] MUSICAL SYMBOL BEGIN BEAM..MUSICAL SYMBOL END PHRASE
159 * E0000 # Cn <reserved-E0000>
160 * E0001 # Cf LANGUAGE TAG
161 * E0002..E001F # Cn [30] <reserved-E0002>..<reserved-E001F>
162 * E0020..E007F # Cf [96] TAG SPACE..CANCEL TAG
163 * E0080..E00FF # Cn [128] <reserved-E0080>..<reserved-E00FF>
164 * E0100..E01EF # Mn [240] VARIATION SELECTOR-17..VARIATION SELECTOR-256
165 * E01F0..E0FFF # Cn [3600] <reserved-E01F0>..<reserved-E0FFF>
166 */
167 static hb_bool_t
168 is_default_ignorable (hb_codepoint_t ch)
169 {
170 hb_codepoint_t plane = ch >> 16;
171 if (likely (plane == 0))
172 {
173 /* BMP */
174 hb_codepoint_t page = ch >> 8;
175 switch (page) {
176 case 0x00: return unlikely (ch == 0x00ADu);
177 case 0x03: return unlikely (ch == 0x034Fu);
178 case 0x06: return unlikely (ch == 0x061Cu);
179 case 0x17: return hb_in_range<hb_codepoint_t> (ch, 0x17B4u, 0x17B5u);
180 case 0x18: return hb_in_range<hb_codepoint_t> (ch, 0x180Bu, 0x180Eu);
181 case 0x20: return hb_in_ranges<hb_codepoint_t> (ch, 0x200Bu, 0x200Fu,
182 0x202Au, 0x202Eu,
183 0x2060u, 0x206Fu);
184 case 0xFE: return hb_in_range<hb_codepoint_t> (ch, 0xFE00u, 0xFE0Fu) || ch == 0xFEFFu;
185 case 0xFF: return hb_in_range<hb_codepoint_t> (ch, 0xFFF0u, 0xFFF8u);
186 default: return false;
187 }
188 }
189 else
190 {
191 /* Other planes */
192 switch (plane) {
193 case 0x01: return hb_in_range<hb_codepoint_t> (ch, 0x1D173u, 0x1D17Au);
194 case 0x0E: return hb_in_range<hb_codepoint_t> (ch, 0xE0000u, 0xE0FFFu);
195 default: return false;
196 }
197 }
198 }
199
200 /* Space estimates based on:
201 * https://unicode.org/charts/PDF/U2000.pdf
202 * https://docs.microsoft.com/en-us/typography/develop/character-design-standards/whitespace
203 */
204 enum space_t {
205 NOT_SPACE = 0,
206 SPACE_EM = 1,
207 SPACE_EM_2 = 2,
208 SPACE_EM_3 = 3,
209 SPACE_EM_4 = 4,
210 SPACE_EM_5 = 5,
211 SPACE_EM_6 = 6,
212 SPACE_EM_16 = 16,
213 SPACE_4_EM_18, /* 4/18th of an EM! */
214 SPACE,
215 SPACE_FIGURE,
216 SPACE_PUNCTUATION,
217 SPACE_NARROW,
218 };
219 static space_t
220 space_fallback_type (hb_codepoint_t u)
221 {
222 switch (u)
223 {
224 /* All GC=Zs chars that can use a fallback. */
225 default: return NOT_SPACE; /* U+1680 OGHAM SPACE MARK */
226 case 0x0020u: return SPACE; /* U+0020 SPACE */
227 case 0x00A0u: return SPACE; /* U+00A0 NO-BREAK SPACE */
228 case 0x2000u: return SPACE_EM_2; /* U+2000 EN QUAD */
229 case 0x2001u: return SPACE_EM; /* U+2001 EM QUAD */
230 case 0x2002u: return SPACE_EM_2; /* U+2002 EN SPACE */
231 case 0x2003u: return SPACE_EM; /* U+2003 EM SPACE */
232 case 0x2004u: return SPACE_EM_3; /* U+2004 THREE-PER-EM SPACE */
233 case 0x2005u: return SPACE_EM_4; /* U+2005 FOUR-PER-EM SPACE */
234 case 0x2006u: return SPACE_EM_6; /* U+2006 SIX-PER-EM SPACE */
235 case 0x2007u: return SPACE_FIGURE; /* U+2007 FIGURE SPACE */
236 case 0x2008u: return SPACE_PUNCTUATION; /* U+2008 PUNCTUATION SPACE */
237 case 0x2009u: return SPACE_EM_5; /* U+2009 THIN SPACE */
238 case 0x200Au: return SPACE_EM_16; /* U+200A HAIR SPACE */
239 case 0x202Fu: return SPACE_NARROW; /* U+202F NARROW NO-BREAK SPACE */
240 case 0x205Fu: return SPACE_4_EM_18; /* U+205F MEDIUM MATHEMATICAL SPACE */
241 case 0x3000u: return SPACE_EM; /* U+3000 IDEOGRAPHIC SPACE */
242 }
243 }
244
245 struct {
246#define HB_UNICODE_FUNC_IMPLEMENT(name) hb_unicode_##name##_func_t name;
247 HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS
248#undef HB_UNICODE_FUNC_IMPLEMENT
249 } func;
250
251 struct {
252#define HB_UNICODE_FUNC_IMPLEMENT(name) void *name;
253 HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS
254#undef HB_UNICODE_FUNC_IMPLEMENT
255 } user_data;
256
257 struct {
258#define HB_UNICODE_FUNC_IMPLEMENT(name) hb_destroy_func_t name;
259 HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS
260#undef HB_UNICODE_FUNC_IMPLEMENT
261 } destroy;
262};
263DECLARE_NULL_INSTANCE (hb_unicode_funcs_t);
264
265
266/*
267 * Modified combining marks
268 */
269
270/* Hebrew
271 *
272 * We permute the "fixed-position" classes 10-26 into the order
273 * described in the SBL Hebrew manual:
274 *
275 * https://www.sbl-site.org/Fonts/SBLHebrewUserManual1.5x.pdf
276 *
277 * (as recommended by:
278 * https://forum.fontlab.com/archive-old-microsoft-volt-group/vista-and-diacritic-ordering/msg22823/)
279 *
280 * More details here:
281 * https://bugzilla.mozilla.org/show_bug.cgi?id=662055
282 */
283#define HB_MODIFIED_COMBINING_CLASS_CCC10 22 /* sheva */
284#define HB_MODIFIED_COMBINING_CLASS_CCC11 15 /* hataf segol */
285#define HB_MODIFIED_COMBINING_CLASS_CCC12 16 /* hataf patah */
286#define HB_MODIFIED_COMBINING_CLASS_CCC13 17 /* hataf qamats */
287#define HB_MODIFIED_COMBINING_CLASS_CCC14 23 /* hiriq */
288#define HB_MODIFIED_COMBINING_CLASS_CCC15 18 /* tsere */
289#define HB_MODIFIED_COMBINING_CLASS_CCC16 19 /* segol */
290#define HB_MODIFIED_COMBINING_CLASS_CCC17 20 /* patah */
291#define HB_MODIFIED_COMBINING_CLASS_CCC18 21 /* qamats */
292#define HB_MODIFIED_COMBINING_CLASS_CCC19 14 /* holam */
293#define HB_MODIFIED_COMBINING_CLASS_CCC20 24 /* qubuts */
294#define HB_MODIFIED_COMBINING_CLASS_CCC21 12 /* dagesh */
295#define HB_MODIFIED_COMBINING_CLASS_CCC22 25 /* meteg */
296#define HB_MODIFIED_COMBINING_CLASS_CCC23 13 /* rafe */
297#define HB_MODIFIED_COMBINING_CLASS_CCC24 10 /* shin dot */
298#define HB_MODIFIED_COMBINING_CLASS_CCC25 11 /* sin dot */
299#define HB_MODIFIED_COMBINING_CLASS_CCC26 26 /* point varika */
300
301/*
302 * Arabic
303 *
304 * Modify to move Shadda (ccc=33) before other marks. See:
305 * https://unicode.org/faq/normalization.html#8
306 * https://unicode.org/faq/normalization.html#9
307 */
308#define HB_MODIFIED_COMBINING_CLASS_CCC27 28 /* fathatan */
309#define HB_MODIFIED_COMBINING_CLASS_CCC28 29 /* dammatan */
310#define HB_MODIFIED_COMBINING_CLASS_CCC29 30 /* kasratan */
311#define HB_MODIFIED_COMBINING_CLASS_CCC30 31 /* fatha */
312#define HB_MODIFIED_COMBINING_CLASS_CCC31 32 /* damma */
313#define HB_MODIFIED_COMBINING_CLASS_CCC32 33 /* kasra */
314#define HB_MODIFIED_COMBINING_CLASS_CCC33 27 /* shadda */
315#define HB_MODIFIED_COMBINING_CLASS_CCC34 34 /* sukun */
316#define HB_MODIFIED_COMBINING_CLASS_CCC35 35 /* superscript alef */
317
318/* Syriac */
319#define HB_MODIFIED_COMBINING_CLASS_CCC36 36 /* superscript alaph */
320
321/* Telugu
322 *
323 * Modify Telugu length marks (ccc=84, ccc=91).
324 * These are the only matras in the main Indic scripts range that have
325 * a non-zero ccc. That makes them reorder with the Halant that is
326 * ccc=9. Just zero them, we don't need them in our Indic shaper.
327 */
328#define HB_MODIFIED_COMBINING_CLASS_CCC84 0 /* length mark */
329#define HB_MODIFIED_COMBINING_CLASS_CCC91 0 /* ai length mark */
330
331/* Thai
332 *
333 * Modify U+0E38 and U+0E39 (ccc=103) to be reordered before U+0E3A (ccc=9).
334 * Assign 3, which is unassigned otherwise.
335 * Uniscribe does this reordering too.
336 */
337#define HB_MODIFIED_COMBINING_CLASS_CCC103 3 /* sara u / sara uu */
338#define HB_MODIFIED_COMBINING_CLASS_CCC107 107 /* mai * */
339
340/* Lao */
341#define HB_MODIFIED_COMBINING_CLASS_CCC118 118 /* sign u / sign uu */
342#define HB_MODIFIED_COMBINING_CLASS_CCC122 122 /* mai * */
343
344/* Tibetan
345 *
346 * In case of multiple vowel-signs, use u first (but after achung)
347 * this allows Dzongkha multi-vowel shortcuts to render correctly
348 */
349#define HB_MODIFIED_COMBINING_CLASS_CCC129 129 /* sign aa */
350#define HB_MODIFIED_COMBINING_CLASS_CCC130 132 /* sign i */
351#define HB_MODIFIED_COMBINING_CLASS_CCC132 131 /* sign u */
352
353/* Misc */
354
355#define HB_UNICODE_GENERAL_CATEGORY_IS_MARK(gen_cat) \
356 (FLAG_UNSAFE (gen_cat) & \
357 (FLAG (HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK) | \
358 FLAG (HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK) | \
359 FLAG (HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)))
360
361
362/*
363 * Ranges, used for bsearch tables.
364 */
365
366struct hb_unicode_range_t
367{
368 static int
369 cmp (const void *_key, const void *_item)
370 {
371 hb_codepoint_t cp = *((hb_codepoint_t *) _key);
372 const hb_unicode_range_t *range = (hb_unicode_range_t *) _item;
373
374 if (cp < range->start)
375 return -1;
376 else if (cp <= range->end)
377 return 0;
378 else
379 return +1;
380 }
381
382 hb_codepoint_t start;
383 hb_codepoint_t end;
384};
385
386/*
387 * Emoji.
388 */
389
390HB_INTERNAL bool
391_hb_unicode_is_emoji_Extended_Pictographic (hb_codepoint_t cp);
392
393
394#endif /* HB_UNICODE_HH */
395