Never skip over CGJ

We might want to tweak this some more.  For now, never skipping over
it is better behavior than always skipping.

Part of https://github.com/behdad/harfbuzz/issues/554
This commit is contained in:
Behdad Esfahbod 2017-10-15 10:48:19 +02:00
parent 25846cc39f
commit 81e2b9b8a9
3 changed files with 12 additions and 8 deletions

View File

@ -226,7 +226,9 @@ _next_syllable (hb_buffer_t *buffer, unsigned int start)
* - General_Category: 5 bits. * - General_Category: 5 bits.
* - A bit each for: * - A bit each for:
* * Is it Default_Ignorable(); we have a modified Default_Ignorable(). * * Is it Default_Ignorable(); we have a modified Default_Ignorable().
* * Whether it's one of the three Mongolian Free Variation Selectors. * * Whether it's one of the three Mongolian Free Variation Selectors,
* CGJ, or other characters that are hidden but should not be ignored
* like most other Default_Ignorable()s do during matching.
* * One free bit right now. * * One free bit right now.
* *
* The high-byte has different meanings, switched by the Gen-Cat: * The high-byte has different meanings, switched by the Gen-Cat:
@ -264,20 +266,21 @@ _hb_glyph_info_set_unicode_props (hb_glyph_info_t *info, hb_buffer_t *buffer)
buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES; buffer->scratch_flags |= HB_BUFFER_SCRATCH_FLAG_HAS_DEFAULT_IGNORABLES;
props |= UPROPS_MASK_IGNORABLE; props |= UPROPS_MASK_IGNORABLE;
if (u == 0x200Cu) props |= UPROPS_MASK_Cf_ZWNJ; if (u == 0x200Cu) props |= UPROPS_MASK_Cf_ZWNJ;
if (u == 0x200Du) props |= UPROPS_MASK_Cf_ZWJ; else if (u == 0x200Du) props |= UPROPS_MASK_Cf_ZWJ;
/* Mongolian Free Variation Selectors need to be remembered /* Mongolian Free Variation Selectors need to be remembered
* because although we need to hide them like default-ignorables, * because although we need to hide them like default-ignorables,
* they need to non-ignorable during shaping. This is similar to * they need to non-ignorable during shaping. This is similar to
* what we do for joiners in Indic-like shapers, but since the * what we do for joiners in Indic-like shapers, but since the
* FVSes are GC=Mn, we have use a separate bit to remember them. * FVSes are GC=Mn, we have use a separate bit to remember them.
* Fixes: * Fixes:
* https://github.com/behdad/harfbuzz/issues/234 * https://github.com/behdad/harfbuzz/issues/234 */
*/ else if (unlikely (hb_in_range (u, 0x180Bu, 0x180Du))) props |= UPROPS_MASK_HIDDEN;
if (unlikely (hb_in_range (u, 0x180Bu, 0x180Du))) props |= UPROPS_MASK_HIDDEN;
/* TAG characters need similar treatment. Fixes: /* TAG characters need similar treatment. Fixes:
* https://github.com/behdad/harfbuzz/issues/463 * https://github.com/behdad/harfbuzz/issues/463 */
*/ else if (unlikely (hb_in_range (u, 0xE0020u, 0xE007Fu))) props |= UPROPS_MASK_HIDDEN;
if (unlikely (hb_in_range (u, 0xE0020u, 0xE007Fu))) props |= UPROPS_MASK_HIDDEN; /* COMBINING GRAPHEME JOINER should not be skipped; at least some times.
* https://github.com/behdad/harfbuzz/issues/554 */
else if (unlikely (u == 0x034Fu)) props |= UPROPS_MASK_HIDDEN;
} }
else if (unlikely (HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL (gen_cat))) else if (unlikely (HB_UNICODE_GENERAL_CATEGORY_IS_NON_ENCLOSING_MARK_OR_MODIFIER_SYMBOL (gen_cat)))
{ {

View File

@ -1 +1,2 @@
fonts/sha1sum/051d92f8bc6ff724511b296c27623f824de256e9.ttf::U+0075,U+0361,U+034F,U+0301,U+0069:[gid2=0+1266|gid7=0@-617,442+0|gid5=0@-7,0+0|gid1=4+528] fonts/sha1sum/051d92f8bc6ff724511b296c27623f824de256e9.ttf::U+0075,U+0361,U+034F,U+0301,U+0069:[gid2=0+1266|gid7=0@-617,442+0|gid5=0@-7,0+0|gid1=4+528]
fonts/sha1sum/bf962d3202883a820aed019d9b5c1838c2ff69c6.ttf::U+0020,U+06CC,U+064E,U+034F,U+0651:[uni0651=1+0|space=1+0|uni064E=1@236,-432+0|uni06CC=1+1266|space=0+452]