2 * Copyright © 2014 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 #ifndef HB_OT_CMAP_TABLE_HH
28 #define HB_OT_CMAP_TABLE_HH
30 #include "hb-open-type-private.hh"
37 * cmap -- Character To Glyph Index Mapping Table
40 #define HB_OT_TAG_cmap HB_TAG('c','m','a','p')
43 struct CmapSubtableFormat0
45 inline bool get_glyph (hb_codepoint_t codepoint
, hb_codepoint_t
*glyph
) const
47 hb_codepoint_t gid
= codepoint
< 256 ? glyphIdArray
[codepoint
] : 0;
54 inline bool sanitize (hb_sanitize_context_t
*c
) {
55 TRACE_SANITIZE (this);
56 return TRACE_RETURN (c
->check_struct (this));
60 USHORT format
; /* Format number is set to 0. */
61 USHORT lengthZ
; /* Byte length of this subtable. */
62 USHORT languageZ
; /* Ignore. */
63 BYTE glyphIdArray
[256];/* An array that maps character
64 * code to glyph index values. */
66 DEFINE_SIZE_STATIC (6 + 256);
69 struct CmapSubtableFormat4
71 inline bool get_glyph (hb_codepoint_t codepoint
, hb_codepoint_t
*glyph
) const
73 unsigned int segCount
;
74 const USHORT
*endCount
;
75 const USHORT
*startCount
;
76 const USHORT
*idDelta
;
77 const USHORT
*idRangeOffset
;
78 const USHORT
*glyphIdArray
;
79 unsigned int glyphIdArrayLength
;
81 segCount
= this->segCountX2
/ 2;
82 endCount
= this->values
;
83 startCount
= endCount
+ segCount
+ 1;
84 idDelta
= startCount
+ segCount
;
85 idRangeOffset
= idDelta
+ segCount
;
86 glyphIdArray
= idRangeOffset
+ segCount
;
87 glyphIdArrayLength
= (this->length
- 16 - 8 * segCount
) / 2;
89 /* Custom two-array bsearch. */
90 int min
= 0, max
= (int) segCount
- 1;
94 int mid
= (min
+ max
) / 2;
95 if (codepoint
< startCount
[mid
])
97 else if (codepoint
> endCount
[mid
])
109 unsigned int rangeOffset
= idRangeOffset
[i
];
110 if (rangeOffset
== 0)
111 gid
= codepoint
+ idDelta
[i
];
114 /* Somebody has been smoking... */
115 unsigned int index
= rangeOffset
/ 2 + (codepoint
- startCount
[i
]) + i
- segCount
;
116 if (unlikely (index
>= glyphIdArrayLength
))
118 gid
= glyphIdArray
[index
];
124 *glyph
= gid
& 0xFFFFu
;
128 inline bool sanitize (hb_sanitize_context_t
*c
)
130 TRACE_SANITIZE (this);
131 if (unlikely (!c
->check_struct (this)))
132 return TRACE_RETURN (false);
134 if (unlikely (!c
->check_range (this, length
)))
136 /* Some broken fonts have too long of a "length" value.
137 * If that is the case, just change the value to truncate
138 * the subtable at the end of the blob. */
139 uint16_t new_length
= (uint16_t) MIN ((uintptr_t) 65535,
140 (uintptr_t) (c
->end
-
142 if (!c
->try_set (&length
, new_length
))
143 return TRACE_RETURN (false);
146 return TRACE_RETURN (16 + 4 * (unsigned int) segCountX2
<= length
);
150 USHORT format
; /* Format number is set to 4. */
151 USHORT length
; /* This is the length in bytes of the
153 USHORT languageZ
; /* Ignore. */
154 USHORT segCountX2
; /* 2 x segCount. */
155 USHORT searchRangeZ
; /* 2 * (2**floor(log2(segCount))) */
156 USHORT entrySelectorZ
; /* log2(searchRange/2) */
157 USHORT rangeShiftZ
; /* 2 x segCount - searchRange */
161 USHORT endCount
[segCount
]; /* End characterCode for each segment,
163 USHORT reservedPad
; /* Set to 0. */
164 USHORT startCount
[segCount
]; /* Start character code for each segment. */
165 SHORT idDelta
[segCount
]; /* Delta for all character codes in segment. */
166 USHORT idRangeOffset
[segCount
];/* Offsets into glyphIdArray or 0 */
167 USHORT glyphIdArray
[VAR
]; /* Glyph index array (arbitrary length) */
171 DEFINE_SIZE_ARRAY (14, values
);
174 struct CmapSubtableLongGroup
176 friend struct CmapSubtableFormat12
;
177 friend struct CmapSubtableFormat13
;
179 int cmp (hb_codepoint_t codepoint
) const
181 if (codepoint
< startCharCode
) return -1;
182 if (codepoint
> endCharCode
) return +1;
186 inline bool sanitize (hb_sanitize_context_t
*c
) {
187 TRACE_SANITIZE (this);
188 return TRACE_RETURN (c
->check_struct (this));
192 ULONG startCharCode
; /* First character code in this group. */
193 ULONG endCharCode
; /* Last character code in this group. */
194 ULONG glyphID
; /* Glyph index; interpretation depends on
195 * subtable format. */
197 DEFINE_SIZE_STATIC (12);
200 template <typename UINT
>
201 struct CmapSubtableTrimmed
203 inline bool get_glyph (hb_codepoint_t codepoint
, hb_codepoint_t
*glyph
) const
205 /* Rely on our implicit array bound-checking. */
206 hb_codepoint_t gid
= glyphIdArray
[codepoint
- startCharCode
];
213 inline bool sanitize (hb_sanitize_context_t
*c
) {
214 TRACE_SANITIZE (this);
215 return TRACE_RETURN (c
->check_struct (this) && glyphIdArray
.sanitize (c
));
219 UINT formatReserved
; /* Subtable format and (maybe) padding. */
220 UINT lengthZ
; /* Byte length of this subtable. */
221 UINT languageZ
; /* Ignore. */
222 UINT startCharCode
; /* First character code covered. */
223 ArrayOf
<GlyphID
, UINT
>
224 glyphIdArray
; /* Array of glyph index values for character
225 * codes in the range. */
227 DEFINE_SIZE_ARRAY (5 * sizeof (UINT
), glyphIdArray
);
230 struct CmapSubtableFormat6
: CmapSubtableTrimmed
<USHORT
> {};
231 struct CmapSubtableFormat10
: CmapSubtableTrimmed
<ULONG
> {};
233 template <typename T
>
234 struct CmapSubtableLongSegmented
236 inline bool get_glyph (hb_codepoint_t codepoint
, hb_codepoint_t
*glyph
) const
238 int i
= groups
.bsearch (codepoint
);
241 *glyph
= T::group_get_glyph (groups
[i
], codepoint
);
245 inline bool sanitize (hb_sanitize_context_t
*c
) {
246 TRACE_SANITIZE (this);
247 return TRACE_RETURN (c
->check_struct (this) && groups
.sanitize (c
));
251 USHORT format
; /* Subtable format; set to 12. */
252 USHORT reservedZ
; /* Reserved; set to 0. */
253 ULONG lengthZ
; /* Byte length of this subtable. */
254 ULONG languageZ
; /* Ignore. */
255 SortedArrayOf
<CmapSubtableLongGroup
, ULONG
>
256 groups
; /* Groupings. */
258 DEFINE_SIZE_ARRAY (16, groups
);
261 struct CmapSubtableFormat12
: CmapSubtableLongSegmented
<CmapSubtableFormat12
>
263 static inline hb_codepoint_t
group_get_glyph (const CmapSubtableLongGroup
&group
,
265 { return group
.glyphID
+ (u
- group
.startCharCode
); }
268 struct CmapSubtableFormat13
: CmapSubtableLongSegmented
<CmapSubtableFormat13
>
270 static inline hb_codepoint_t
group_get_glyph (const CmapSubtableLongGroup
&group
,
271 hb_codepoint_t u HB_UNUSED
)
272 { return group
.glyphID
; }
277 GLYPH_VARIANT_NOT_FOUND
= 0,
278 GLYPH_VARIANT_FOUND
= 1,
279 GLYPH_VARIANT_USE_DEFAULT
= 2
282 struct UnicodeValueRange
284 inline int cmp (const hb_codepoint_t
&codepoint
) const
286 if (codepoint
< startUnicodeValue
) return -1;
287 if (codepoint
> startUnicodeValue
+ additionalCount
) return +1;
291 inline bool sanitize (hb_sanitize_context_t
*c
) {
292 TRACE_SANITIZE (this);
293 return TRACE_RETURN (c
->check_struct (this));
296 UINT24 startUnicodeValue
; /* First value in this range. */
297 BYTE additionalCount
; /* Number of additional values in this
300 DEFINE_SIZE_STATIC (4);
303 typedef SortedArrayOf
<UnicodeValueRange
, ULONG
> DefaultUVS
;
307 inline int cmp (const hb_codepoint_t
&codepoint
) const
309 return unicodeValue
.cmp (codepoint
);
312 inline bool sanitize (hb_sanitize_context_t
*c
) {
313 TRACE_SANITIZE (this);
314 return TRACE_RETURN (c
->check_struct (this));
317 UINT24 unicodeValue
; /* Base Unicode value of the UVS */
318 GlyphID glyphID
; /* Glyph ID of the UVS */
320 DEFINE_SIZE_STATIC (5);
323 typedef SortedArrayOf
<UVSMapping
, ULONG
> NonDefaultUVS
;
325 struct VariationSelectorRecord
327 inline glyph_variant_t
get_glyph (hb_codepoint_t codepoint
,
328 hb_codepoint_t
*glyph
,
329 const void *base
) const
332 const DefaultUVS
&defaults
= base
+defaultUVS
;
333 i
= defaults
.bsearch (codepoint
);
335 return GLYPH_VARIANT_USE_DEFAULT
;
336 const NonDefaultUVS
&nonDefaults
= base
+nonDefaultUVS
;
337 i
= nonDefaults
.bsearch (codepoint
);
340 *glyph
= nonDefaults
[i
].glyphID
;
341 return GLYPH_VARIANT_FOUND
;
343 return GLYPH_VARIANT_NOT_FOUND
;
346 inline int cmp (const hb_codepoint_t
&variation_selector
) const
348 return varSelector
.cmp (variation_selector
);
351 inline bool sanitize (hb_sanitize_context_t
*c
, void *base
) {
352 TRACE_SANITIZE (this);
353 return TRACE_RETURN (c
->check_struct (this) &&
354 defaultUVS
.sanitize (c
, base
) &&
355 nonDefaultUVS
.sanitize (c
, base
));
358 UINT24 varSelector
; /* Variation selector. */
359 OffsetTo
<DefaultUVS
, ULONG
>
360 defaultUVS
; /* Offset to Default UVS Table. May be 0. */
361 OffsetTo
<NonDefaultUVS
, ULONG
>
362 nonDefaultUVS
; /* Offset to Non-Default UVS Table. May be 0. */
364 DEFINE_SIZE_STATIC (11);
367 struct CmapSubtableFormat14
369 inline glyph_variant_t
get_glyph_variant (hb_codepoint_t codepoint
,
370 hb_codepoint_t variation_selector
,
371 hb_codepoint_t
*glyph
) const
373 return record
[record
.bsearch(variation_selector
)].get_glyph (codepoint
, glyph
, this);
376 inline bool sanitize (hb_sanitize_context_t
*c
) {
377 TRACE_SANITIZE (this);
378 return TRACE_RETURN (c
->check_struct (this) &&
379 record
.sanitize (c
, this));
383 USHORT format
; /* Format number is set to 0. */
384 ULONG lengthZ
; /* Byte length of this subtable. */
385 SortedArrayOf
<VariationSelectorRecord
, ULONG
>
386 record
; /* Variation selector records; sorted
387 * in increasing order of `varSelector'. */
389 DEFINE_SIZE_ARRAY (10, record
);
394 /* Note: We intentionally do NOT implement subtable formats 2 and 8. */
396 inline bool get_glyph (hb_codepoint_t codepoint
,
397 hb_codepoint_t
*glyph
) const
400 case 0: return u
.format0
.get_glyph(codepoint
, glyph
);
401 case 4: return u
.format4
.get_glyph(codepoint
, glyph
);
402 case 6: return u
.format6
.get_glyph(codepoint
, glyph
);
403 case 10: return u
.format10
.get_glyph(codepoint
, glyph
);
404 case 12: return u
.format12
.get_glyph(codepoint
, glyph
);
405 case 13: return u
.format13
.get_glyph(codepoint
, glyph
);
407 default: return false;
411 inline glyph_variant_t
get_glyph_variant (hb_codepoint_t codepoint
,
412 hb_codepoint_t variation_selector
,
413 hb_codepoint_t
*glyph
) const
416 case 14: return u
.format14
.get_glyph_variant(codepoint
, variation_selector
, glyph
);
417 default: return GLYPH_VARIANT_NOT_FOUND
;
421 inline bool sanitize (hb_sanitize_context_t
*c
) {
422 TRACE_SANITIZE (this);
423 if (!u
.format
.sanitize (c
)) return TRACE_RETURN (false);
425 case 0: return TRACE_RETURN (u
.format0
.sanitize (c
));
426 case 4: return TRACE_RETURN (u
.format4
.sanitize (c
));
427 case 6: return TRACE_RETURN (u
.format6
.sanitize (c
));
428 case 10: return TRACE_RETURN (u
.format10
.sanitize (c
));
429 case 12: return TRACE_RETURN (u
.format12
.sanitize (c
));
430 case 13: return TRACE_RETURN (u
.format13
.sanitize (c
));
431 case 14: return TRACE_RETURN (u
.format14
.sanitize (c
));
432 default:return TRACE_RETURN (true);
438 USHORT format
; /* Format identifier */
439 CmapSubtableFormat0 format0
;
440 CmapSubtableFormat4 format4
;
441 CmapSubtableFormat6 format6
;
442 CmapSubtableFormat10 format10
;
443 CmapSubtableFormat12 format12
;
444 CmapSubtableFormat13 format13
;
445 CmapSubtableFormat14 format14
;
448 DEFINE_SIZE_UNION (2, format
);
452 struct EncodingRecord
454 inline int cmp (const EncodingRecord
&other
) const
457 ret
= platformID
.cmp (other
.platformID
);
459 ret
= encodingID
.cmp (other
.encodingID
);
464 inline bool sanitize (hb_sanitize_context_t
*c
, void *base
) {
465 TRACE_SANITIZE (this);
466 return TRACE_RETURN (c
->check_struct (this) &&
467 subtable
.sanitize (c
, base
));
470 USHORT platformID
; /* Platform ID. */
471 USHORT encodingID
; /* Platform-specific encoding ID. */
472 OffsetTo
<CmapSubtable
, ULONG
>
473 subtable
; /* Byte offset from beginning of table to the subtable for this encoding. */
475 DEFINE_SIZE_STATIC (8);
480 static const hb_tag_t tableTag
= HB_OT_TAG_cmap
;
482 inline const CmapSubtable
*find_subtable (unsigned int platform_id
,
483 unsigned int encoding_id
) const
486 key
.platformID
.set (platform_id
);
487 key
.encodingID
.set (encoding_id
);
489 /* Note: We can use bsearch, but since it has no performance
490 * implications, we use lsearch and as such accept fonts with
491 * unsorted subtable list. */
492 int result
= encodingRecord
./*bsearch*/lsearch (key
);
493 if (result
== -1 || !encodingRecord
[result
].subtable
)
496 return &(this+encodingRecord
[result
].subtable
);
499 inline bool sanitize (hb_sanitize_context_t
*c
) {
500 TRACE_SANITIZE (this);
501 return TRACE_RETURN (c
->check_struct (this) &&
502 likely (version
== 0) &&
503 encodingRecord
.sanitize (c
, this));
506 USHORT version
; /* Table version number (0). */
507 SortedArrayOf
<EncodingRecord
>
508 encodingRecord
; /* Encoding tables. */
510 DEFINE_SIZE_ARRAY (4, encodingRecord
);
517 #endif /* HB_OT_CMAP_TABLE_HH */