2011-06-02 21:43:12 +00:00
|
|
|
|
/*
|
2012-05-24 19:00:01 +00:00
|
|
|
|
* Copyright © 2011,2012 Google, Inc.
|
2011-06-02 21:43:12 +00:00
|
|
|
|
*
|
|
|
|
|
* This is part of HarfBuzz, a text shaping library.
|
|
|
|
|
*
|
|
|
|
|
* Permission is hereby granted, without written agreement and without
|
|
|
|
|
* license or royalty fees, to use, copy, modify, and distribute this
|
|
|
|
|
* software and its documentation for any purpose, provided that the
|
|
|
|
|
* above copyright notice and the following two paragraphs appear in
|
|
|
|
|
* all copies of this software.
|
|
|
|
|
*
|
|
|
|
|
* IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
|
|
|
|
|
* DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
|
|
|
|
|
* ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
|
|
|
|
|
* IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
|
|
|
|
|
* DAMAGE.
|
|
|
|
|
*
|
|
|
|
|
* THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
|
|
|
|
|
* BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
|
|
|
|
|
* FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
|
|
|
|
|
* ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
|
|
|
|
|
* PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
|
|
|
|
|
*
|
|
|
|
|
* Google Author(s): Behdad Esfahbod
|
|
|
|
|
*/
|
|
|
|
|
|
2012-04-20 02:34:06 +00:00
|
|
|
|
#include "hb-ot-shape-complex-indic-private.hh"
|
2012-05-11 18:59:26 +00:00
|
|
|
|
#include "hb-ot-shape-private.hh"
|
2012-07-24 00:14:13 +00:00
|
|
|
|
#include "hb-ot-layout-private.hh"
|
2011-07-30 23:04:02 +00:00
|
|
|
|
|
2012-07-16 17:47:19 +00:00
|
|
|
|
#define OLD_INDIC_TAG(script) (((hb_tag_t) script) | 0x20000000)
|
|
|
|
|
#define IS_OLD_INDIC_TAG(tag) ( \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_BENGALI) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_DEVANAGARI) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_GUJARATI) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_GURMUKHI) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_KANNADA) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_MALAYALAM) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_ORIYA) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_TAMIL) || \
|
|
|
|
|
(tag) == OLD_INDIC_TAG (HB_SCRIPT_TELUGU) \
|
|
|
|
|
)
|
2012-06-05 19:17:44 +00:00
|
|
|
|
struct indic_options_t
|
2012-05-11 14:43:12 +00:00
|
|
|
|
{
|
2012-06-05 19:17:44 +00:00
|
|
|
|
int initialized : 1;
|
|
|
|
|
int uniscribe_bug_compatible : 1;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
union indic_options_union_t {
|
|
|
|
|
int i;
|
|
|
|
|
indic_options_t opts;
|
|
|
|
|
};
|
|
|
|
|
ASSERT_STATIC (sizeof (int) == sizeof (indic_options_union_t));
|
|
|
|
|
|
|
|
|
|
static indic_options_union_t
|
|
|
|
|
indic_options_init (void)
|
|
|
|
|
{
|
|
|
|
|
indic_options_union_t u;
|
|
|
|
|
u.i = 0;
|
|
|
|
|
u.opts.initialized = 1;
|
|
|
|
|
|
|
|
|
|
char *c = getenv ("HB_OT_INDIC_OPTIONS");
|
|
|
|
|
u.opts.uniscribe_bug_compatible = c && strstr (c, "uniscribe-bug-compatible");
|
|
|
|
|
|
|
|
|
|
return u;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
inline indic_options_t
|
|
|
|
|
indic_options (void)
|
|
|
|
|
{
|
|
|
|
|
static indic_options_union_t options;
|
|
|
|
|
|
|
|
|
|
if (unlikely (!options.i)) {
|
|
|
|
|
/* This is idempotent and threadsafe. */
|
|
|
|
|
options = indic_options_init ();
|
2012-05-11 14:43:12 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-06-05 19:17:44 +00:00
|
|
|
|
return options.opts;
|
|
|
|
|
}
|
|
|
|
|
|
2012-05-11 14:43:12 +00:00
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
static int
|
|
|
|
|
compare_codepoint (const void *pa, const void *pb)
|
|
|
|
|
{
|
|
|
|
|
hb_codepoint_t a = * (hb_codepoint_t *) pa;
|
|
|
|
|
hb_codepoint_t b = * (hb_codepoint_t *) pb;
|
|
|
|
|
|
|
|
|
|
return a < b ? -1 : a == b ? 0 : +1;
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-19 20:20:21 +00:00
|
|
|
|
static bool
|
|
|
|
|
would_substitute (hb_codepoint_t *glyphs, unsigned int glyphs_count,
|
|
|
|
|
hb_tag_t feature_tag, hb_ot_map_t *map, hb_face_t *face)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
2012-07-19 20:20:21 +00:00
|
|
|
|
unsigned int lookup_indices[32];
|
|
|
|
|
unsigned int offset, len;
|
|
|
|
|
|
|
|
|
|
offset = 0;
|
|
|
|
|
do {
|
|
|
|
|
len = ARRAY_LENGTH (lookup_indices);
|
|
|
|
|
hb_ot_layout_feature_get_lookup_indexes (face, HB_OT_TAG_GSUB,
|
|
|
|
|
map->get_feature_index (0/*GSUB*/, feature_tag),
|
|
|
|
|
offset,
|
|
|
|
|
&len,
|
|
|
|
|
lookup_indices);
|
|
|
|
|
|
|
|
|
|
for (unsigned int i = 0; i < len; i++)
|
|
|
|
|
if (hb_ot_layout_would_substitute_lookup (face, glyphs, glyphs_count, lookup_indices[i]))
|
|
|
|
|
return true;
|
|
|
|
|
|
|
|
|
|
offset += len;
|
|
|
|
|
} while (len == ARRAY_LENGTH (lookup_indices));
|
|
|
|
|
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
2012-07-19 20:20:21 +00:00
|
|
|
|
static indic_position_t
|
|
|
|
|
consonant_position (hb_codepoint_t u, hb_ot_map_t *map, hb_font_t *font)
|
|
|
|
|
{
|
2012-07-20 00:30:22 +00:00
|
|
|
|
if ((u & ~0x007F) == 0x1780)
|
|
|
|
|
return POS_BELOW_C; /* In Khmer coeng model, all are subjoining. */
|
|
|
|
|
|
2012-07-19 20:20:21 +00:00
|
|
|
|
hb_codepoint_t virama = (u & ~0x007F) | 0x004D;
|
|
|
|
|
if ((u & ~0x007F) == 0x0D80) virama = 0x0DCA; /* Sinahla */
|
|
|
|
|
hb_codepoint_t glyphs[2];
|
|
|
|
|
|
2012-07-20 14:30:24 +00:00
|
|
|
|
unsigned int virama_pos = IS_OLD_INDIC_TAG (map->get_chosen_script (0)) ? 1 : 0;
|
|
|
|
|
hb_font_get_glyph (font, virama, 0, &glyphs[virama_pos]);
|
|
|
|
|
hb_font_get_glyph (font, u, 0, &glyphs[1-virama_pos]);
|
2012-07-19 20:20:21 +00:00
|
|
|
|
|
|
|
|
|
hb_face_t *face = hb_font_get_face (font);
|
|
|
|
|
if (would_substitute (glyphs, ARRAY_LENGTH (glyphs), HB_TAG('p','r','e','f'), map, face)) return POS_BELOW_C;
|
|
|
|
|
if (would_substitute (glyphs, ARRAY_LENGTH (glyphs), HB_TAG('b','l','w','f'), map, face)) return POS_BELOW_C;
|
|
|
|
|
if (would_substitute (glyphs, ARRAY_LENGTH (glyphs), HB_TAG('p','s','t','f'), map, face)) return POS_POST_C;
|
|
|
|
|
return POS_BASE_C;
|
2011-07-29 20:37:02 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-07-19 16:59:06 +00:00
|
|
|
|
#define MATRA_POS_LEFT(u) POS_PRE_M
|
2012-07-19 17:25:08 +00:00
|
|
|
|
#define MATRA_POS_RIGHT(u) ( \
|
|
|
|
|
IS_DEVA(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_BENG(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_GURM(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_GUJA(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_ORYA(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_TAML(u) ? POS_AFTER_POST : \
|
2012-07-19 23:52:19 +00:00
|
|
|
|
IS_TELU(u) ? (u <= 0x0C42 ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
|
2012-07-20 01:17:48 +00:00
|
|
|
|
IS_KNDA(u) ? (u < 0x0CC3 || u > 0xCD6 ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
|
2012-07-19 17:25:08 +00:00
|
|
|
|
IS_MLYM(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_SINH(u) ? POS_AFTER_SUB : \
|
|
|
|
|
/*default*/ POS_AFTER_SUB \
|
|
|
|
|
)
|
|
|
|
|
#define MATRA_POS_TOP(u) ( /* BENG and MLYM don't have top matras. */ \
|
|
|
|
|
IS_DEVA(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_GURM(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_GUJA(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_ORYA(u) ? POS_AFTER_MAIN : \
|
|
|
|
|
IS_TAML(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_TELU(u) ? POS_BEFORE_SUB : \
|
|
|
|
|
IS_KNDA(u) ? POS_BEFORE_SUB : \
|
|
|
|
|
IS_SINH(u) ? POS_AFTER_SUB : \
|
|
|
|
|
/*default*/ POS_AFTER_SUB \
|
|
|
|
|
)
|
|
|
|
|
#define MATRA_POS_BOTTOM(u) ( \
|
|
|
|
|
IS_DEVA(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_BENG(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_GURM(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_GUJA(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_ORYA(u) ? POS_AFTER_SUB : \
|
|
|
|
|
IS_TAML(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_TELU(u) ? POS_BEFORE_SUB : \
|
|
|
|
|
IS_KNDA(u) ? POS_BEFORE_SUB : \
|
|
|
|
|
IS_MLYM(u) ? POS_AFTER_POST : \
|
|
|
|
|
IS_SINH(u) ? POS_AFTER_SUB : \
|
|
|
|
|
/*default*/ POS_AFTER_SUB \
|
|
|
|
|
)
|
2012-07-19 16:59:06 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static indic_position_t
|
|
|
|
|
matra_position (hb_codepoint_t u, indic_position_t side)
|
|
|
|
|
{
|
|
|
|
|
switch ((int) side)
|
|
|
|
|
{
|
|
|
|
|
case POS_PRE_C: return MATRA_POS_LEFT (u);
|
|
|
|
|
case POS_POST_C: return MATRA_POS_RIGHT (u);
|
|
|
|
|
case POS_ABOVE_C: return MATRA_POS_TOP (u);
|
|
|
|
|
case POS_BELOW_C: return MATRA_POS_BOTTOM (u);
|
|
|
|
|
};
|
|
|
|
|
abort ();
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-24 00:07:50 +00:00
|
|
|
|
static inline bool
|
2011-07-30 23:04:02 +00:00
|
|
|
|
is_ra (hb_codepoint_t u)
|
|
|
|
|
{
|
|
|
|
|
return !!bsearch (&u, ra_chars,
|
|
|
|
|
ARRAY_LENGTH (ra_chars),
|
|
|
|
|
sizeof (ra_chars[0]),
|
|
|
|
|
compare_codepoint);
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-24 00:07:50 +00:00
|
|
|
|
static inline bool
|
|
|
|
|
is_one_of (const hb_glyph_info_t &info, unsigned int flags)
|
|
|
|
|
{
|
|
|
|
|
/* If it ligated, all bets are off. */
|
2012-07-24 00:11:42 +00:00
|
|
|
|
if (is_a_ligature (info)) return false;
|
2012-07-24 00:07:50 +00:00
|
|
|
|
return !!(FLAG (info.indic_category()) & flags);
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-19 16:32:16 +00:00
|
|
|
|
#define JOINER_FLAGS (FLAG (OT_ZWJ) | FLAG (OT_ZWNJ))
|
2012-07-24 00:07:50 +00:00
|
|
|
|
static inline bool
|
2011-07-31 17:46:44 +00:00
|
|
|
|
is_joiner (const hb_glyph_info_t &info)
|
2011-07-31 15:10:14 +00:00
|
|
|
|
{
|
2012-07-24 00:07:50 +00:00
|
|
|
|
return is_one_of (info, JOINER_FLAGS);
|
2011-07-31 17:46:44 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-07-24 00:07:50 +00:00
|
|
|
|
/* Note:
|
|
|
|
|
*
|
|
|
|
|
* We treat Vowels and placeholders as if they were consonants. This is safe because Vowels
|
|
|
|
|
* cannot happen in a consonant syllable. The plus side however is, we can call the
|
|
|
|
|
* consonant syllable logic from the vowel syllable function and get it all right! */
|
2012-07-19 16:32:16 +00:00
|
|
|
|
#define CONSONANT_FLAGS (FLAG (OT_C) | FLAG (OT_Ra) | FLAG (OT_V) | FLAG (OT_NBSP) | FLAG (OT_DOTTEDCIRCLE))
|
2012-07-24 00:07:50 +00:00
|
|
|
|
static inline bool
|
2011-07-31 17:46:44 +00:00
|
|
|
|
is_consonant (const hb_glyph_info_t &info)
|
|
|
|
|
{
|
2012-07-24 00:07:50 +00:00
|
|
|
|
return is_one_of (info, CONSONANT_FLAGS);
|
2011-07-31 15:10:14 +00:00
|
|
|
|
}
|
2011-06-24 23:05:34 +00:00
|
|
|
|
|
2012-07-19 16:32:16 +00:00
|
|
|
|
#define HALANT_OR_COENG_FLAGS (FLAG (OT_H) | FLAG (OT_Coeng))
|
2012-07-24 00:07:50 +00:00
|
|
|
|
static inline bool
|
2012-07-17 15:37:32 +00:00
|
|
|
|
is_halant_or_coeng (const hb_glyph_info_t &info)
|
|
|
|
|
{
|
2012-07-24 00:07:50 +00:00
|
|
|
|
return is_one_of (info, HALANT_OR_COENG_FLAGS);
|
2012-07-23 23:51:48 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-07-19 16:32:16 +00:00
|
|
|
|
static inline void
|
2012-07-19 20:20:21 +00:00
|
|
|
|
set_indic_properties (hb_glyph_info_t &info, hb_ot_map_t *map, hb_font_t *font)
|
2012-07-19 16:32:16 +00:00
|
|
|
|
{
|
|
|
|
|
hb_codepoint_t u = info.codepoint;
|
|
|
|
|
unsigned int type = get_indic_categories (u);
|
2012-07-24 06:22:18 +00:00
|
|
|
|
indic_category_t cat = (indic_category_t) (type & 0x0F);
|
|
|
|
|
indic_position_t pos = (indic_position_t) (type >> 4);
|
2012-07-19 16:59:06 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
2012-07-24 06:22:18 +00:00
|
|
|
|
* Re-assign category
|
2012-07-19 16:59:06 +00:00
|
|
|
|
*/
|
|
|
|
|
|
2012-07-19 16:32:16 +00:00
|
|
|
|
|
|
|
|
|
/* The spec says U+0952 is OT_A. However, testing shows that Uniscribe
|
|
|
|
|
* treats U+0951..U+0952 all as OT_VD.
|
|
|
|
|
* TESTS:
|
|
|
|
|
* U+092E,U+0947,U+0952
|
|
|
|
|
* U+092E,U+0952,U+0947
|
|
|
|
|
* U+092E,U+0947,U+0951
|
|
|
|
|
* U+092E,U+0951,U+0947
|
|
|
|
|
* */
|
|
|
|
|
if (unlikely (hb_in_range<hb_codepoint_t> (u, 0x0951, 0x0954)))
|
|
|
|
|
cat = OT_VD;
|
|
|
|
|
|
|
|
|
|
if (cat == OT_X &&
|
2012-07-24 06:22:18 +00:00
|
|
|
|
unlikely (hb_in_range<hb_codepoint_t> (u, 0x17CB, 0x17D3))) /* Khmer Various signs */
|
|
|
|
|
{
|
|
|
|
|
/* These are like Top Matras. */
|
|
|
|
|
cat = OT_M;
|
|
|
|
|
pos = POS_ABOVE_C;
|
|
|
|
|
}
|
2012-07-24 05:04:18 +00:00
|
|
|
|
if (u == 0x17C6) /* Khmer Bindu doesn't like to be repositioned. */
|
|
|
|
|
cat = OT_N;
|
2012-07-19 16:32:16 +00:00
|
|
|
|
|
2012-07-19 16:59:06 +00:00
|
|
|
|
if (unlikely (u == 0x17D2)) cat = OT_Coeng; /* Khmer coeng */
|
|
|
|
|
else if (unlikely (u == 0x200C)) cat = OT_ZWNJ;
|
|
|
|
|
else if (unlikely (u == 0x200D)) cat = OT_ZWJ;
|
|
|
|
|
else if (unlikely (u == 0x25CC)) cat = OT_DOTTEDCIRCLE;
|
2012-07-20 01:13:04 +00:00
|
|
|
|
else if (unlikely (u == 0x0A71)) cat = OT_SM; /* GURMUKHI ADDAK. More like consonant medial. like 0A75. */
|
2012-07-19 16:32:16 +00:00
|
|
|
|
|
|
|
|
|
if (cat == OT_Repha) {
|
|
|
|
|
/* There are two kinds of characters marked as Repha:
|
|
|
|
|
* - The ones that are GenCat=Mn are already positioned visually, ie. after base. (eg. Khmer)
|
|
|
|
|
* - The ones that are GenCat=Lo is encoded logically, ie. beginning of syllable. (eg. Malayalam)
|
|
|
|
|
*
|
|
|
|
|
* We recategorize the first kind to look like a Nukta and attached to the base directly.
|
|
|
|
|
*/
|
|
|
|
|
if (_hb_glyph_info_get_general_category (&info) == HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)
|
|
|
|
|
cat = OT_N;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2012-07-19 16:59:06 +00:00
|
|
|
|
|
|
|
|
|
/*
|
2012-07-24 06:22:18 +00:00
|
|
|
|
* Re-assign position.
|
2012-07-19 16:59:06 +00:00
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
if ((FLAG (cat) & CONSONANT_FLAGS))
|
|
|
|
|
{
|
2012-07-19 20:20:21 +00:00
|
|
|
|
pos = consonant_position (u, map, font);
|
2012-07-19 16:32:16 +00:00
|
|
|
|
if (is_ra (u))
|
|
|
|
|
cat = OT_Ra;
|
2012-07-19 16:59:06 +00:00
|
|
|
|
}
|
|
|
|
|
else if (cat == OT_M)
|
|
|
|
|
{
|
|
|
|
|
pos = matra_position (u, pos);
|
|
|
|
|
}
|
|
|
|
|
else if (cat == OT_SM || cat == OT_VD)
|
|
|
|
|
{
|
2012-07-19 16:32:16 +00:00
|
|
|
|
pos = POS_SMVD;
|
2012-07-19 16:59:06 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-07-20 20:03:09 +00:00
|
|
|
|
if (unlikely (u == 0x0B01)) pos = POS_BEFORE_SUB; /* Oriya Bindu is BeforeSub in the spec. */
|
2012-07-19 16:59:06 +00:00
|
|
|
|
|
2012-07-24 05:32:07 +00:00
|
|
|
|
|
2012-07-19 16:32:16 +00:00
|
|
|
|
|
|
|
|
|
info.indic_category() = cat;
|
|
|
|
|
info.indic_position() = pos;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2012-05-11 18:50:53 +00:00
|
|
|
|
struct feature_list_t {
|
2011-06-24 23:05:34 +00:00
|
|
|
|
hb_tag_t tag;
|
|
|
|
|
hb_bool_t is_global;
|
2012-05-11 18:50:53 +00:00
|
|
|
|
};
|
|
|
|
|
|
2012-07-17 14:40:59 +00:00
|
|
|
|
/* These features are applied one at a time, given the order in this table. */
|
2012-05-11 18:50:53 +00:00
|
|
|
|
static const feature_list_t
|
|
|
|
|
indic_basic_features[] =
|
2011-06-02 21:43:12 +00:00
|
|
|
|
{
|
2011-06-24 23:05:34 +00:00
|
|
|
|
{HB_TAG('n','u','k','t'), true},
|
2012-07-20 00:24:14 +00:00
|
|
|
|
{HB_TAG('a','k','h','n'), true},
|
2011-06-24 23:05:34 +00:00
|
|
|
|
{HB_TAG('r','p','h','f'), false},
|
2012-05-09 09:06:47 +00:00
|
|
|
|
{HB_TAG('r','k','r','f'), true},
|
2011-06-24 23:05:34 +00:00
|
|
|
|
{HB_TAG('p','r','e','f'), false},
|
|
|
|
|
{HB_TAG('b','l','w','f'), false},
|
|
|
|
|
{HB_TAG('h','a','l','f'), false},
|
2012-07-16 16:05:35 +00:00
|
|
|
|
{HB_TAG('a','b','v','f'), false},
|
2011-06-24 23:05:34 +00:00
|
|
|
|
{HB_TAG('p','s','t','f'), false},
|
2012-07-17 17:55:10 +00:00
|
|
|
|
{HB_TAG('c','f','a','r'), false},
|
2012-07-20 14:47:46 +00:00
|
|
|
|
{HB_TAG('c','j','c','t'), true},
|
2012-05-13 16:09:29 +00:00
|
|
|
|
{HB_TAG('v','a','t','u'), true},
|
2011-06-24 23:05:34 +00:00
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* Same order as the indic_basic_features array */
|
|
|
|
|
enum {
|
|
|
|
|
_NUKT,
|
2012-07-20 00:24:14 +00:00
|
|
|
|
_AKHN,
|
2011-06-24 23:05:34 +00:00
|
|
|
|
RPHF,
|
2012-05-09 09:38:31 +00:00
|
|
|
|
_RKRF,
|
2011-06-24 23:05:34 +00:00
|
|
|
|
PREF,
|
|
|
|
|
BLWF,
|
|
|
|
|
HALF,
|
2012-07-16 16:05:35 +00:00
|
|
|
|
ABVF,
|
2011-06-24 23:05:34 +00:00
|
|
|
|
PSTF,
|
2012-07-17 17:55:10 +00:00
|
|
|
|
CFAR,
|
2012-07-20 14:47:46 +00:00
|
|
|
|
_CJCT,
|
2012-05-13 16:09:29 +00:00
|
|
|
|
VATU
|
2011-06-02 21:43:12 +00:00
|
|
|
|
};
|
|
|
|
|
|
2012-07-17 14:40:59 +00:00
|
|
|
|
/* These features are applied all at once. */
|
2012-05-11 18:50:53 +00:00
|
|
|
|
static const feature_list_t
|
|
|
|
|
indic_other_features[] =
|
2011-06-02 21:43:12 +00:00
|
|
|
|
{
|
2012-05-11 18:50:53 +00:00
|
|
|
|
{HB_TAG('i','n','i','t'), false},
|
|
|
|
|
{HB_TAG('p','r','e','s'), true},
|
|
|
|
|
{HB_TAG('a','b','v','s'), true},
|
|
|
|
|
{HB_TAG('b','l','w','s'), true},
|
|
|
|
|
{HB_TAG('p','s','t','s'), true},
|
|
|
|
|
{HB_TAG('h','a','l','n'), true},
|
|
|
|
|
|
|
|
|
|
{HB_TAG('d','i','s','t'), true},
|
|
|
|
|
{HB_TAG('a','b','v','m'), true},
|
|
|
|
|
{HB_TAG('b','l','w','m'), true},
|
|
|
|
|
};
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
initial_reordering (const hb_ot_map_t *map,
|
|
|
|
|
hb_face_t *face,
|
|
|
|
|
hb_buffer_t *buffer,
|
|
|
|
|
void *user_data HB_UNUSED);
|
2011-07-08 04:22:40 +00:00
|
|
|
|
static void
|
|
|
|
|
final_reordering (const hb_ot_map_t *map,
|
|
|
|
|
hb_face_t *face,
|
|
|
|
|
hb_buffer_t *buffer,
|
2011-07-29 20:37:02 +00:00
|
|
|
|
void *user_data HB_UNUSED);
|
2011-06-02 21:43:12 +00:00
|
|
|
|
|
|
|
|
|
void
|
2012-05-13 14:13:44 +00:00
|
|
|
|
_hb_ot_shape_complex_collect_features_indic (hb_ot_map_builder_t *map,
|
|
|
|
|
const hb_segment_properties_t *props HB_UNUSED)
|
2011-06-02 21:43:12 +00:00
|
|
|
|
{
|
2011-07-08 04:22:40 +00:00
|
|
|
|
map->add_bool_feature (HB_TAG('l','o','c','l'));
|
2011-07-20 20:42:10 +00:00
|
|
|
|
/* The Indic specs do not require ccmp, but we apply it here since if
|
|
|
|
|
* there is a use of it, it's typically at the beginning. */
|
2011-07-08 04:22:40 +00:00
|
|
|
|
map->add_bool_feature (HB_TAG('c','c','m','p'));
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
map->add_gsub_pause (initial_reordering, NULL);
|
2011-07-08 04:22:40 +00:00
|
|
|
|
|
2012-05-09 09:07:18 +00:00
|
|
|
|
for (unsigned int i = 0; i < ARRAY_LENGTH (indic_basic_features); i++) {
|
2011-07-08 02:25:25 +00:00
|
|
|
|
map->add_bool_feature (indic_basic_features[i].tag, indic_basic_features[i].is_global);
|
2012-05-09 09:07:18 +00:00
|
|
|
|
map->add_gsub_pause (NULL, NULL);
|
|
|
|
|
}
|
2011-06-02 21:43:12 +00:00
|
|
|
|
|
2011-07-08 04:22:40 +00:00
|
|
|
|
map->add_gsub_pause (final_reordering, NULL);
|
|
|
|
|
|
2012-07-17 14:40:59 +00:00
|
|
|
|
for (unsigned int i = 0; i < ARRAY_LENGTH (indic_other_features); i++)
|
2012-05-11 18:50:53 +00:00
|
|
|
|
map->add_bool_feature (indic_other_features[i].tag, indic_other_features[i].is_global);
|
2011-06-02 21:43:12 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-07-17 00:26:57 +00:00
|
|
|
|
void
|
|
|
|
|
_hb_ot_shape_complex_override_features_indic (hb_ot_map_builder_t *map,
|
|
|
|
|
const hb_segment_properties_t *props HB_UNUSED)
|
|
|
|
|
{
|
2012-07-17 00:31:24 +00:00
|
|
|
|
/* Uniscribe does not apply 'kern'. */
|
|
|
|
|
if (indic_options ().uniscribe_bug_compatible)
|
|
|
|
|
map->add_feature (HB_TAG('k','e','r','n'), 0, true);
|
2012-07-17 00:26:57 +00:00
|
|
|
|
}
|
|
|
|
|
|
2011-06-17 22:35:46 +00:00
|
|
|
|
|
2012-04-05 21:25:19 +00:00
|
|
|
|
hb_ot_shape_normalization_mode_t
|
|
|
|
|
_hb_ot_shape_complex_normalization_preference_indic (void)
|
2011-07-21 16:23:12 +00:00
|
|
|
|
{
|
|
|
|
|
/* We want split matras decomposed by the common shaping logic. */
|
2012-07-20 22:13:04 +00:00
|
|
|
|
/* XXX sort this out after adding per-shaper normalizers. */
|
|
|
|
|
return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS;
|
2011-07-21 16:23:12 +00:00
|
|
|
|
}
|
|
|
|
|
|
2011-06-17 22:35:46 +00:00
|
|
|
|
|
2011-06-02 21:43:12 +00:00
|
|
|
|
void
|
2012-07-19 20:20:21 +00:00
|
|
|
|
_hb_ot_shape_complex_setup_masks_indic (hb_ot_map_t *map,
|
2012-05-13 14:13:44 +00:00
|
|
|
|
hb_buffer_t *buffer,
|
2012-07-19 20:20:21 +00:00
|
|
|
|
hb_font_t *font)
|
2011-06-02 21:43:12 +00:00
|
|
|
|
{
|
2011-07-28 21:06:46 +00:00
|
|
|
|
HB_BUFFER_ALLOCATE_VAR (buffer, indic_category);
|
|
|
|
|
HB_BUFFER_ALLOCATE_VAR (buffer, indic_position);
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
/* We cannot setup masks here. We save information about characters
|
|
|
|
|
* and setup masks later on in a pause-callback. */
|
|
|
|
|
|
|
|
|
|
unsigned int count = buffer->len;
|
2011-06-02 21:43:12 +00:00
|
|
|
|
for (unsigned int i = 0; i < count; i++)
|
2012-07-19 20:20:21 +00:00
|
|
|
|
set_indic_properties (buffer->info[i], map, font);
|
2011-07-29 20:37:02 +00:00
|
|
|
|
}
|
2011-06-02 21:43:12 +00:00
|
|
|
|
|
2011-07-30 18:44:30 +00:00
|
|
|
|
static int
|
|
|
|
|
compare_indic_order (const hb_glyph_info_t *pa, const hb_glyph_info_t *pb)
|
|
|
|
|
{
|
|
|
|
|
int a = pa->indic_position();
|
|
|
|
|
int b = pb->indic_position();
|
|
|
|
|
|
|
|
|
|
return a < b ? -1 : a == b ? 0 : +1;
|
|
|
|
|
}
|
2011-06-17 22:35:46 +00:00
|
|
|
|
|
2012-05-11 16:58:57 +00:00
|
|
|
|
/* Rules from:
|
|
|
|
|
* https://www.microsoft.com/typography/otfntdev/devanot/shaping.aspx */
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
static void
|
2012-07-16 18:52:18 +00:00
|
|
|
|
initial_reordering_consonant_syllable (const hb_ot_map_t *map, hb_buffer_t *buffer, hb_mask_t *basic_mask_array,
|
2012-05-09 15:56:03 +00:00
|
|
|
|
unsigned int start, unsigned int end)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
2011-07-30 23:15:53 +00:00
|
|
|
|
hb_glyph_info_t *info = buffer->info;
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
2012-05-13 14:48:03 +00:00
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
/* 1. Find base consonant:
|
|
|
|
|
*
|
|
|
|
|
* The shaping engine finds the base consonant of the syllable, using the
|
|
|
|
|
* following algorithm: starting from the end of the syllable, move backwards
|
|
|
|
|
* until a consonant is found that does not have a below-base or post-base
|
|
|
|
|
* form (post-base forms have to follow below-base forms), or that is not a
|
|
|
|
|
* pre-base reordering Ra, or arrive at the first consonant. The consonant
|
|
|
|
|
* stopped at will be the base.
|
|
|
|
|
*
|
|
|
|
|
* o If the syllable starts with Ra + Halant (in a script that has Reph)
|
|
|
|
|
* and has more than one consonant, Ra is excluded from candidates for
|
|
|
|
|
* base consonants.
|
|
|
|
|
*/
|
|
|
|
|
|
2011-07-31 21:51:50 +00:00
|
|
|
|
unsigned int base = end;
|
2012-05-09 09:43:43 +00:00
|
|
|
|
bool has_reph = false;
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
2012-05-09 09:43:43 +00:00
|
|
|
|
{
|
2012-05-13 14:48:03 +00:00
|
|
|
|
/* -> If the syllable starts with Ra + Halant (in a script that has Reph)
|
|
|
|
|
* and has more than one consonant, Ra is excluded from candidates for
|
|
|
|
|
* base consonants. */
|
|
|
|
|
unsigned int limit = start;
|
2012-07-16 18:52:18 +00:00
|
|
|
|
if (basic_mask_array[RPHF] &&
|
2012-05-13 14:48:03 +00:00
|
|
|
|
start + 3 <= end &&
|
|
|
|
|
info[start].indic_category() == OT_Ra &&
|
|
|
|
|
info[start + 1].indic_category() == OT_H &&
|
2012-07-20 18:13:35 +00:00
|
|
|
|
(unlikely (buffer->props.script == HB_SCRIPT_SINHALA || buffer->props.script == HB_SCRIPT_TELUGU) ?
|
|
|
|
|
info[start + 2].indic_category() == OT_ZWJ /* In Sinhala & Telugu, form Reph only if ZWJ is present */:
|
2012-07-18 21:22:14 +00:00
|
|
|
|
!is_joiner (info[start + 2] /* In other scripts, any joiner blocks Reph formation */ )
|
|
|
|
|
))
|
2012-05-13 14:48:03 +00:00
|
|
|
|
{
|
|
|
|
|
limit += 2;
|
2012-07-18 21:22:14 +00:00
|
|
|
|
while (limit < end && is_joiner (info[limit]))
|
|
|
|
|
limit++;
|
2012-05-13 14:48:03 +00:00
|
|
|
|
base = start;
|
|
|
|
|
has_reph = true;
|
|
|
|
|
};
|
2012-05-09 09:43:43 +00:00
|
|
|
|
|
2012-07-18 17:13:03 +00:00
|
|
|
|
enum base_position_t {
|
|
|
|
|
BASE_FIRST,
|
|
|
|
|
BASE_LAST
|
|
|
|
|
} base_pos;
|
|
|
|
|
|
|
|
|
|
switch ((hb_tag_t) buffer->props.script)
|
|
|
|
|
{
|
2012-07-24 03:51:29 +00:00
|
|
|
|
case HB_SCRIPT_SINHALA:
|
2012-07-18 17:13:03 +00:00
|
|
|
|
case HB_SCRIPT_KHMER:
|
|
|
|
|
base_pos = BASE_FIRST;
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
default:
|
|
|
|
|
base_pos = BASE_LAST;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (base_pos == BASE_LAST)
|
2012-07-17 18:23:28 +00:00
|
|
|
|
{
|
|
|
|
|
/* -> starting from the end of the syllable, move backwards */
|
|
|
|
|
unsigned int i = end;
|
2012-07-20 22:38:27 +00:00
|
|
|
|
bool seen_below = false;
|
2012-07-17 18:23:28 +00:00
|
|
|
|
do {
|
|
|
|
|
i--;
|
|
|
|
|
/* -> until a consonant is found */
|
|
|
|
|
if (is_consonant (info[i]))
|
2012-05-11 14:23:38 +00:00
|
|
|
|
{
|
2012-07-17 18:23:28 +00:00
|
|
|
|
/* -> that does not have a below-base or post-base form
|
|
|
|
|
* (post-base forms have to follow below-base forms), */
|
|
|
|
|
if (info[i].indic_position() != POS_BELOW_C &&
|
2012-07-20 22:38:27 +00:00
|
|
|
|
(info[i].indic_position() != POS_POST_C || seen_below))
|
2012-07-17 18:23:28 +00:00
|
|
|
|
{
|
|
|
|
|
base = i;
|
|
|
|
|
break;
|
|
|
|
|
}
|
2012-07-20 22:38:27 +00:00
|
|
|
|
if (info[i].indic_position() == POS_BELOW_C)
|
|
|
|
|
seen_below = true;
|
2012-07-17 18:23:28 +00:00
|
|
|
|
|
|
|
|
|
/* -> or that is not a pre-base reordering Ra,
|
|
|
|
|
*
|
2012-07-17 18:25:14 +00:00
|
|
|
|
* IMPLEMENTATION NOTES:
|
|
|
|
|
*
|
|
|
|
|
* Our pre-base reordering Ra's are marked POS_BELOW, so will be skipped
|
|
|
|
|
* by the logic above already.
|
2012-07-17 18:23:28 +00:00
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* -> or arrive at the first consonant. The consonant stopped at will
|
|
|
|
|
* be the base. */
|
2012-05-11 14:23:38 +00:00
|
|
|
|
base = i;
|
|
|
|
|
}
|
2012-07-17 18:23:28 +00:00
|
|
|
|
else
|
2012-07-20 15:04:15 +00:00
|
|
|
|
{
|
2012-07-20 17:48:03 +00:00
|
|
|
|
/* A ZWJ stops the base search, and requests an explicit half form. */
|
|
|
|
|
if (info[i].indic_category() == OT_ZWJ)
|
2012-07-17 18:23:28 +00:00
|
|
|
|
break;
|
2012-07-20 15:04:15 +00:00
|
|
|
|
}
|
2012-07-17 18:23:28 +00:00
|
|
|
|
} while (i > limit);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
/* In scripts without half forms (eg. Khmer), the first consonant is always the base. */
|
|
|
|
|
|
|
|
|
|
if (!has_reph)
|
|
|
|
|
base = limit;
|
2012-07-24 03:51:29 +00:00
|
|
|
|
|
|
|
|
|
/* Find the last base consonant that is not blocked by ZWJ. If there is
|
[Indic] Further adjust base algorithm for Sinhala
Apparently if there is C,V,ZWJ,C, the first C will be base, but if
it's C,ZWJ,V,C, the second one will be.
Note that Uniscribe implements this differently, by breaking syllable in
the case of C,ZWJ,V,C and putting the first consonant in one syllable
and the rest in the next syllable.
Sinhala failures down from 208 to 158 (0.0581209%). No changes to
Khmer.
2012-07-24 04:21:16 +00:00
|
|
|
|
* a ZWJ right before a base consonant, that would request a subjoined form. */
|
2012-07-24 03:51:29 +00:00
|
|
|
|
for (unsigned int i = limit; i < end; i++)
|
|
|
|
|
if (is_consonant (info[i]) && info[i].indic_position() == POS_BASE_C)
|
[Indic] Further adjust base algorithm for Sinhala
Apparently if there is C,V,ZWJ,C, the first C will be base, but if
it's C,ZWJ,V,C, the second one will be.
Note that Uniscribe implements this differently, by breaking syllable in
the case of C,ZWJ,V,C and putting the first consonant in one syllable
and the rest in the next syllable.
Sinhala failures down from 208 to 158 (0.0581209%). No changes to
Khmer.
2012-07-24 04:21:16 +00:00
|
|
|
|
{
|
|
|
|
|
if (limit < i && info[i - 1].indic_category() == OT_ZWJ)
|
|
|
|
|
break;
|
|
|
|
|
else
|
|
|
|
|
base = i;
|
|
|
|
|
}
|
2012-07-24 03:51:29 +00:00
|
|
|
|
|
|
|
|
|
/* Mark all subsequent consonants as below. */
|
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
|
|
|
|
if (is_consonant (info[i]) && info[i].indic_position() == POS_BASE_C)
|
|
|
|
|
info[i].indic_position() = POS_BELOW_C;
|
2012-07-17 18:23:28 +00:00
|
|
|
|
}
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
2012-05-13 14:48:03 +00:00
|
|
|
|
/* -> If the syllable starts with Ra + Halant (in a script that has Reph)
|
|
|
|
|
* and has more than one consonant, Ra is excluded from candidates for
|
2012-07-24 04:26:43 +00:00
|
|
|
|
* base consonants.
|
|
|
|
|
*
|
|
|
|
|
* Only do this for unforced Reph. (ie. not for Ra,H,ZWJ. */
|
|
|
|
|
if (has_reph && base == start && start + 2 == limit) {
|
2012-05-13 14:48:03 +00:00
|
|
|
|
/* Have no other consonant, so Reph is not formed and Ra becomes base. */
|
|
|
|
|
has_reph = false;
|
|
|
|
|
}
|
2012-05-13 14:46:08 +00:00
|
|
|
|
}
|
2012-07-24 04:26:43 +00:00
|
|
|
|
|
2012-07-24 03:51:29 +00:00
|
|
|
|
if (base < end)
|
|
|
|
|
info[base].indic_position() = POS_BASE_C;
|
2012-05-10 09:37:42 +00:00
|
|
|
|
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
/* 2. Decompose and reorder Matras:
|
|
|
|
|
*
|
|
|
|
|
* Each matra and any syllable modifier sign in the cluster are moved to the
|
|
|
|
|
* appropriate position relative to the consonant(s) in the cluster. The
|
|
|
|
|
* shaping engine decomposes two- or three-part matras into their constituent
|
|
|
|
|
* parts before any repositioning. Matra characters are classified by which
|
|
|
|
|
* consonant in a conjunct they have affinity for and are reordered to the
|
|
|
|
|
* following positions:
|
|
|
|
|
*
|
|
|
|
|
* o Before first half form in the syllable
|
|
|
|
|
* o After subjoined consonants
|
|
|
|
|
* o After post-form consonant
|
|
|
|
|
* o After main consonant (for above marks)
|
|
|
|
|
*
|
|
|
|
|
* IMPLEMENTATION NOTES:
|
|
|
|
|
*
|
|
|
|
|
* The normalize() routine has already decomposed matras for us, so we don't
|
|
|
|
|
* need to worry about that.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* 3. Reorder marks to canonical order:
|
|
|
|
|
*
|
|
|
|
|
* Adjacent nukta and halant or nukta and vedic sign are always repositioned
|
|
|
|
|
* if necessary, so that the nukta is first.
|
|
|
|
|
*
|
|
|
|
|
* IMPLEMENTATION NOTES:
|
|
|
|
|
*
|
|
|
|
|
* We don't need to do this: the normalize() routine already did this for us.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
2011-07-30 18:44:30 +00:00
|
|
|
|
/* Reorder characters */
|
|
|
|
|
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = start; i < base; i++)
|
2012-07-20 14:18:23 +00:00
|
|
|
|
info[i].indic_position() = MIN (POS_PRE_C, (indic_position_t) info[i].indic_position());
|
2012-07-17 16:50:13 +00:00
|
|
|
|
|
2012-07-18 19:41:53 +00:00
|
|
|
|
if (base < end)
|
|
|
|
|
info[base].indic_position() = POS_BASE_C;
|
2011-07-30 18:44:30 +00:00
|
|
|
|
|
2012-07-17 16:50:13 +00:00
|
|
|
|
/* Mark final consonants. A final consonant is one appearing after a matra,
|
|
|
|
|
* like in Khmer. */
|
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
|
|
|
|
if (info[i].indic_category() == OT_M) {
|
|
|
|
|
for (unsigned int j = i + 1; j < end; j++)
|
|
|
|
|
if (is_consonant (info[j])) {
|
|
|
|
|
info[j].indic_position() = POS_FINAL_C;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
2011-07-31 00:14:44 +00:00
|
|
|
|
/* Handle beginning Ra */
|
2012-05-13 14:46:08 +00:00
|
|
|
|
if (has_reph)
|
2012-05-10 11:45:52 +00:00
|
|
|
|
info[start].indic_position() = POS_RA_TO_BECOME_REPH;
|
2011-07-31 00:14:44 +00:00
|
|
|
|
|
2011-07-31 01:08:10 +00:00
|
|
|
|
/* For old-style Indic script tags, move the first post-base Halant after
|
|
|
|
|
* last consonant. */
|
2012-07-16 17:47:19 +00:00
|
|
|
|
if (IS_OLD_INDIC_TAG (map->get_chosen_script (0))) {
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
2011-07-31 01:08:10 +00:00
|
|
|
|
if (info[i].indic_category() == OT_H) {
|
|
|
|
|
unsigned int j;
|
|
|
|
|
for (j = end - 1; j > i; j--)
|
2012-05-10 10:17:16 +00:00
|
|
|
|
if (is_consonant (info[j]))
|
2011-07-31 01:08:10 +00:00
|
|
|
|
break;
|
|
|
|
|
if (j > i) {
|
|
|
|
|
/* Move Halant to after last consonant. */
|
|
|
|
|
hb_glyph_info_t t = info[i];
|
|
|
|
|
memmove (&info[i], &info[i + 1], (j - i) * sizeof (info[0]));
|
|
|
|
|
info[j] = t;
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-20 19:10:02 +00:00
|
|
|
|
/* Attach misc marks to previous char to move with them. */
|
2012-05-11 14:43:12 +00:00
|
|
|
|
{
|
2012-07-20 19:10:02 +00:00
|
|
|
|
indic_position_t last_pos = POS_START;
|
|
|
|
|
for (unsigned int i = start; i < end; i++)
|
|
|
|
|
{
|
|
|
|
|
if ((FLAG (info[i].indic_category()) & (JOINER_FLAGS | FLAG (OT_N) | FLAG (OT_RS) | HALANT_OR_COENG_FLAGS)))
|
|
|
|
|
{
|
|
|
|
|
info[i].indic_position() = last_pos;
|
|
|
|
|
if (unlikely (indic_options ().uniscribe_bug_compatible &&
|
|
|
|
|
info[i].indic_category() == OT_H &&
|
|
|
|
|
info[i].indic_position() == POS_PRE_M))
|
|
|
|
|
{
|
|
|
|
|
/*
|
|
|
|
|
* Uniscribe doesn't move the Halant with Left Matra.
|
|
|
|
|
* TEST: U+092B,U+093F,U+094DE
|
|
|
|
|
*/
|
2012-05-11 14:43:12 +00:00
|
|
|
|
for (unsigned int j = i; j > start; j--)
|
2012-05-11 19:42:27 +00:00
|
|
|
|
if (info[j - 1].indic_position() != POS_PRE_M) {
|
2012-05-11 14:43:12 +00:00
|
|
|
|
info[i].indic_position() = info[j - 1].indic_position();
|
|
|
|
|
break;
|
|
|
|
|
}
|
2012-07-20 19:10:02 +00:00
|
|
|
|
}
|
|
|
|
|
} else if (info[i].indic_position() != POS_SMVD) {
|
|
|
|
|
last_pos = (indic_position_t) info[i].indic_position();
|
2012-05-11 14:43:12 +00:00
|
|
|
|
}
|
2012-07-20 19:10:02 +00:00
|
|
|
|
}
|
2012-05-11 14:43:12 +00:00
|
|
|
|
}
|
2012-07-17 15:16:19 +00:00
|
|
|
|
/* Re-attach ZWJ, ZWNJ, and halant to next char, for after-base consonants. */
|
|
|
|
|
{
|
|
|
|
|
unsigned int last_halant = end;
|
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
2012-07-17 15:37:32 +00:00
|
|
|
|
if (is_halant_or_coeng (info[i]))
|
2012-07-17 15:16:19 +00:00
|
|
|
|
last_halant = i;
|
|
|
|
|
else if (is_consonant (info[i])) {
|
|
|
|
|
for (unsigned int j = last_halant; j < i; j++)
|
2012-07-20 19:10:02 +00:00
|
|
|
|
if (info[j].indic_position() != POS_SMVD)
|
|
|
|
|
info[j].indic_position() = info[i].indic_position();
|
2012-07-17 15:16:19 +00:00
|
|
|
|
}
|
|
|
|
|
}
|
2011-07-30 18:44:30 +00:00
|
|
|
|
|
2012-05-10 09:31:20 +00:00
|
|
|
|
{
|
2012-07-23 03:58:55 +00:00
|
|
|
|
/* Things are out-of-control for post base positions, they may shuffle
|
|
|
|
|
* around like crazy, so merge clusters. For pre-base stuff, we handle
|
|
|
|
|
* cluster issues in final reordering. */
|
|
|
|
|
buffer->merge_clusters (base, end);
|
2012-05-10 09:31:20 +00:00
|
|
|
|
/* Sit tight, rock 'n roll! */
|
2012-05-10 08:51:38 +00:00
|
|
|
|
hb_bubble_sort (info + start, end - start, compare_indic_order);
|
2012-05-10 09:31:20 +00:00
|
|
|
|
/* Find base again */
|
|
|
|
|
base = end;
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = start; i < end; i++)
|
2012-05-10 09:31:20 +00:00
|
|
|
|
if (info[i].indic_position() == POS_BASE_C) {
|
|
|
|
|
base = i;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
2011-07-30 18:44:30 +00:00
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
/* Setup masks now */
|
|
|
|
|
|
2011-07-31 20:00:35 +00:00
|
|
|
|
{
|
|
|
|
|
hb_mask_t mask;
|
|
|
|
|
|
2012-05-10 11:45:52 +00:00
|
|
|
|
/* Reph */
|
2012-05-11 13:34:13 +00:00
|
|
|
|
for (unsigned int i = start; i < end && info[i].indic_position() == POS_RA_TO_BECOME_REPH; i++)
|
2012-07-16 18:52:18 +00:00
|
|
|
|
info[i].mask |= basic_mask_array[RPHF];
|
2012-05-10 11:45:52 +00:00
|
|
|
|
|
2011-07-31 20:00:35 +00:00
|
|
|
|
/* Pre-base */
|
2012-07-20 14:47:46 +00:00
|
|
|
|
mask = basic_mask_array[HALF];
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = start; i < base; i++)
|
2011-07-31 20:00:35 +00:00
|
|
|
|
info[i].mask |= mask;
|
|
|
|
|
/* Base */
|
2012-07-20 14:47:46 +00:00
|
|
|
|
mask = 0;
|
2012-07-18 19:41:53 +00:00
|
|
|
|
if (base < end)
|
|
|
|
|
info[base].mask |= mask;
|
2011-07-31 20:00:35 +00:00
|
|
|
|
/* Post-base */
|
2012-07-20 14:47:46 +00:00
|
|
|
|
mask = basic_mask_array[BLWF] | basic_mask_array[ABVF] | basic_mask_array[PSTF];
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
2011-07-31 20:00:35 +00:00
|
|
|
|
info[i].mask |= mask;
|
|
|
|
|
}
|
2011-07-31 17:46:44 +00:00
|
|
|
|
|
2012-07-16 19:20:15 +00:00
|
|
|
|
/* XXX This will not match for old-Indic spec since the Halant-Ra order is reversed already. */
|
2012-07-20 19:47:24 +00:00
|
|
|
|
if (basic_mask_array[PREF] && base + 2 < end)
|
2012-07-16 19:20:15 +00:00
|
|
|
|
{
|
2012-07-24 00:07:50 +00:00
|
|
|
|
/* Find a Halant,Ra sequence and mark it for pre-base reordering processing. */
|
2012-07-16 21:04:46 +00:00
|
|
|
|
for (unsigned int i = base + 1; i + 1 < end; i++)
|
2012-07-17 15:37:32 +00:00
|
|
|
|
if (is_halant_or_coeng (info[i]) &&
|
2012-07-16 21:04:46 +00:00
|
|
|
|
info[i + 1].indic_category() == OT_Ra)
|
|
|
|
|
{
|
2012-07-17 17:55:10 +00:00
|
|
|
|
info[i++].mask |= basic_mask_array[PREF];
|
|
|
|
|
info[i++].mask |= basic_mask_array[PREF];
|
|
|
|
|
|
|
|
|
|
/* Mark the subsequent stuff with 'cfar'. Used in Khmer.
|
|
|
|
|
* Read the feature spec.
|
|
|
|
|
* This allows distinguishing the following cases with MS Khmer fonts:
|
|
|
|
|
* U+1784,U+17D2,U+179A,U+17D2,U+1782
|
|
|
|
|
* U+1784,U+17D2,U+1782,U+17D2,U+179A
|
|
|
|
|
*/
|
|
|
|
|
for (; i < end; i++)
|
|
|
|
|
info[i].mask |= basic_mask_array[CFAR];
|
|
|
|
|
|
2012-07-16 21:04:46 +00:00
|
|
|
|
break;
|
|
|
|
|
}
|
2012-07-16 19:20:15 +00:00
|
|
|
|
}
|
|
|
|
|
|
2011-07-31 17:46:44 +00:00
|
|
|
|
/* Apply ZWJ/ZWNJ effects */
|
2012-05-11 14:23:38 +00:00
|
|
|
|
for (unsigned int i = start + 1; i < end; i++)
|
2011-07-31 17:46:44 +00:00
|
|
|
|
if (is_joiner (info[i])) {
|
|
|
|
|
bool non_joiner = info[i].indic_category() == OT_ZWNJ;
|
2011-07-31 19:57:00 +00:00
|
|
|
|
unsigned int j = i;
|
2011-07-31 17:46:44 +00:00
|
|
|
|
|
|
|
|
|
do {
|
|
|
|
|
j--;
|
2011-07-31 19:57:00 +00:00
|
|
|
|
|
2012-07-20 14:47:46 +00:00
|
|
|
|
/* A ZWJ disables CJCT, however, it's mere presence is enough
|
|
|
|
|
* to disable ligation. No explicit action needed. */
|
|
|
|
|
|
|
|
|
|
/* A ZWNJ disables HALF. */
|
2011-07-31 19:57:00 +00:00
|
|
|
|
if (non_joiner)
|
2012-07-16 18:52:18 +00:00
|
|
|
|
info[j].mask &= ~basic_mask_array[HALF];
|
2011-07-31 19:57:00 +00:00
|
|
|
|
|
2011-07-31 17:46:44 +00:00
|
|
|
|
} while (j > start && !is_consonant (info[j]));
|
|
|
|
|
}
|
2011-07-29 20:37:02 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
static void
|
2012-05-13 14:13:44 +00:00
|
|
|
|
initial_reordering_vowel_syllable (const hb_ot_map_t *map,
|
|
|
|
|
hb_buffer_t *buffer,
|
2012-07-16 18:52:18 +00:00
|
|
|
|
hb_mask_t *basic_mask_array,
|
2012-05-09 15:56:03 +00:00
|
|
|
|
unsigned int start, unsigned int end)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
2012-05-10 10:07:33 +00:00
|
|
|
|
/* We made the vowels look like consonants. So let's call the consonant logic! */
|
2012-07-16 18:52:18 +00:00
|
|
|
|
initial_reordering_consonant_syllable (map, buffer, basic_mask_array, start, end);
|
2011-07-29 20:37:02 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
2012-05-13 14:13:44 +00:00
|
|
|
|
initial_reordering_standalone_cluster (const hb_ot_map_t *map,
|
|
|
|
|
hb_buffer_t *buffer,
|
2012-07-16 18:52:18 +00:00
|
|
|
|
hb_mask_t *basic_mask_array,
|
2012-05-09 15:56:03 +00:00
|
|
|
|
unsigned int start, unsigned int end)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
2012-05-11 18:02:14 +00:00
|
|
|
|
/* We treat NBSP/dotted-circle as if they are consonants, so we should just chain.
|
|
|
|
|
* Only if not in compatibility mode that is... */
|
|
|
|
|
|
2012-06-05 19:17:44 +00:00
|
|
|
|
if (indic_options ().uniscribe_bug_compatible)
|
2012-05-11 18:02:14 +00:00
|
|
|
|
{
|
|
|
|
|
/* For dotted-circle, this is what Uniscribe does:
|
|
|
|
|
* If dotted-circle is the last glyph, it just does nothing.
|
|
|
|
|
* Ie. It doesn't form Reph. */
|
|
|
|
|
if (buffer->info[end - 1].indic_category() == OT_DOTTEDCIRCLE)
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-16 18:52:18 +00:00
|
|
|
|
initial_reordering_consonant_syllable (map, buffer, basic_mask_array, start, end);
|
2011-07-29 20:37:02 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
2012-05-13 14:13:44 +00:00
|
|
|
|
initial_reordering_non_indic (const hb_ot_map_t *map HB_UNUSED,
|
|
|
|
|
hb_buffer_t *buffer HB_UNUSED,
|
2012-07-16 18:52:18 +00:00
|
|
|
|
hb_mask_t *basic_mask_array HB_UNUSED,
|
2012-05-13 14:20:10 +00:00
|
|
|
|
unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
|
|
|
|
/* Nothing to do right now. If we ever switch to using the output
|
|
|
|
|
* buffer in the reordering process, we'd need to next_glyph() here. */
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#include "hb-ot-shape-complex-indic-machine.hh"
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
initial_reordering (const hb_ot_map_t *map,
|
2012-05-13 14:20:10 +00:00
|
|
|
|
hb_face_t *face HB_UNUSED,
|
2011-07-29 20:37:02 +00:00
|
|
|
|
hb_buffer_t *buffer,
|
|
|
|
|
void *user_data HB_UNUSED)
|
|
|
|
|
{
|
2012-07-16 18:52:18 +00:00
|
|
|
|
hb_mask_t basic_mask_array[ARRAY_LENGTH (indic_basic_features)] = {0};
|
2011-06-02 21:43:12 +00:00
|
|
|
|
unsigned int num_masks = ARRAY_LENGTH (indic_basic_features);
|
|
|
|
|
for (unsigned int i = 0; i < num_masks; i++)
|
2012-07-16 18:52:18 +00:00
|
|
|
|
basic_mask_array[i] = map->get_1_mask (indic_basic_features[i].tag);
|
2011-07-29 20:37:02 +00:00
|
|
|
|
|
2012-07-16 18:52:18 +00:00
|
|
|
|
find_syllables (map, buffer, basic_mask_array);
|
2011-06-02 21:43:12 +00:00
|
|
|
|
}
|
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
static void
|
2012-07-16 19:30:05 +00:00
|
|
|
|
final_reordering_syllable (hb_buffer_t *buffer,
|
|
|
|
|
hb_mask_t init_mask, hb_mask_t pref_mask,
|
2012-05-09 15:56:03 +00:00
|
|
|
|
unsigned int start, unsigned int end)
|
2011-07-29 20:37:02 +00:00
|
|
|
|
{
|
2012-05-10 10:53:53 +00:00
|
|
|
|
hb_glyph_info_t *info = buffer->info;
|
|
|
|
|
|
2011-07-31 19:18:57 +00:00
|
|
|
|
/* 4. Final reordering:
|
|
|
|
|
*
|
|
|
|
|
* After the localized forms and basic shaping forms GSUB features have been
|
|
|
|
|
* applied (see below), the shaping engine performs some final glyph
|
|
|
|
|
* reordering before applying all the remaining font features to the entire
|
|
|
|
|
* cluster.
|
2012-05-10 10:53:53 +00:00
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* Find base again */
|
2012-07-20 19:47:24 +00:00
|
|
|
|
unsigned int base;
|
|
|
|
|
for (base = start; base < end; base++)
|
|
|
|
|
if (info[base].indic_position() >= POS_BASE_C) {
|
|
|
|
|
if (start < base && info[base].indic_position() > POS_BASE_C)
|
|
|
|
|
base--;
|
|
|
|
|
break;
|
|
|
|
|
}
|
2012-05-10 10:53:53 +00:00
|
|
|
|
|
2012-05-10 11:09:08 +00:00
|
|
|
|
|
2012-05-10 10:53:53 +00:00
|
|
|
|
/* o Reorder matras:
|
2011-07-31 19:18:57 +00:00
|
|
|
|
*
|
|
|
|
|
* If a pre-base matra character had been reordered before applying basic
|
|
|
|
|
* features, the glyph can be moved closer to the main consonant based on
|
|
|
|
|
* whether half-forms had been formed. Actual position for the matra is
|
|
|
|
|
* defined as “after last standalone halant glyph, after initial matra
|
|
|
|
|
* position and before the main consonant”. If ZWJ or ZWNJ follow this
|
|
|
|
|
* halant, position is moved after it.
|
2012-05-10 10:53:53 +00:00
|
|
|
|
*/
|
|
|
|
|
|
2012-07-16 19:15:28 +00:00
|
|
|
|
if (start < base) /* Otherwise there can't be any pre-base matra characters. */
|
2012-05-11 19:36:32 +00:00
|
|
|
|
{
|
2012-07-16 19:26:56 +00:00
|
|
|
|
unsigned int new_pos = base - 1;
|
|
|
|
|
while (new_pos > start &&
|
2012-07-23 23:51:48 +00:00
|
|
|
|
!(is_one_of (info[new_pos], (FLAG (OT_M) | FLAG (OT_H) | FLAG (OT_Coeng)))))
|
2012-07-16 19:26:56 +00:00
|
|
|
|
new_pos--;
|
2012-07-23 03:55:19 +00:00
|
|
|
|
/* If we found no Halant we are done (just need to update clusters).
|
|
|
|
|
* Otherwise only proceed if the Halant does
|
2012-05-11 19:36:32 +00:00
|
|
|
|
* not belong to the Matra itself! */
|
2012-07-17 15:37:32 +00:00
|
|
|
|
if (is_halant_or_coeng (info[new_pos]) &&
|
2012-07-23 04:11:26 +00:00
|
|
|
|
info[new_pos].indic_position() != POS_PRE_M)
|
|
|
|
|
{
|
2012-05-11 19:36:32 +00:00
|
|
|
|
/* -> If ZWJ or ZWNJ follow this halant, position is moved after it. */
|
2012-07-16 19:26:56 +00:00
|
|
|
|
if (new_pos + 1 < end && is_joiner (info[new_pos + 1]))
|
|
|
|
|
new_pos++;
|
2012-05-11 19:36:32 +00:00
|
|
|
|
|
|
|
|
|
/* Now go see if there's actually any matras... */
|
2012-07-16 19:26:56 +00:00
|
|
|
|
for (unsigned int i = new_pos; i > start; i--)
|
2012-05-11 19:42:27 +00:00
|
|
|
|
if (info[i - 1].indic_position () == POS_PRE_M)
|
2012-05-11 19:36:32 +00:00
|
|
|
|
{
|
2012-07-16 19:40:33 +00:00
|
|
|
|
unsigned int old_pos = i - 1;
|
|
|
|
|
hb_glyph_info_t tmp = info[old_pos];
|
|
|
|
|
memmove (&info[old_pos], &info[old_pos + 1], (new_pos - old_pos) * sizeof (info[0]));
|
|
|
|
|
info[new_pos] = tmp;
|
2012-07-16 19:26:56 +00:00
|
|
|
|
new_pos--;
|
2012-05-11 19:36:32 +00:00
|
|
|
|
}
|
2012-07-23 12:22:55 +00:00
|
|
|
|
buffer->merge_clusters (new_pos, MIN (end, base + 1));
|
2012-07-23 03:55:19 +00:00
|
|
|
|
} else {
|
2012-07-23 04:11:26 +00:00
|
|
|
|
for (unsigned int i = start; i < base; i++)
|
2012-07-23 03:55:19 +00:00
|
|
|
|
if (info[i].indic_position () == POS_PRE_M) {
|
2012-07-23 12:22:55 +00:00
|
|
|
|
buffer->merge_clusters (i, MIN (end, base + 1));
|
2012-07-23 03:55:19 +00:00
|
|
|
|
break;
|
|
|
|
|
}
|
2012-05-11 19:36:32 +00:00
|
|
|
|
}
|
2012-05-10 10:53:53 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* o Reorder reph:
|
2011-07-31 19:18:57 +00:00
|
|
|
|
*
|
|
|
|
|
* Reph’s original position is always at the beginning of the syllable,
|
|
|
|
|
* (i.e. it is not reordered at the character reordering stage). However,
|
|
|
|
|
* it will be reordered according to the basic-forms shaping results.
|
|
|
|
|
* Possible positions for reph, depending on the script, are; after main,
|
|
|
|
|
* before post-base consonant forms, and after post-base consonant forms.
|
2012-05-10 11:45:52 +00:00
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* If there's anything after the Ra that has the REPH pos, it ought to be halant.
|
|
|
|
|
* Which means that the font has failed to ligate the Reph. In which case, we
|
|
|
|
|
* shouldn't move. */
|
|
|
|
|
if (start + 1 < end &&
|
|
|
|
|
info[start].indic_position() == POS_RA_TO_BECOME_REPH &&
|
|
|
|
|
info[start + 1].indic_position() != POS_RA_TO_BECOME_REPH)
|
|
|
|
|
{
|
2012-05-10 19:44:50 +00:00
|
|
|
|
unsigned int new_reph_pos;
|
|
|
|
|
|
|
|
|
|
enum reph_position_t {
|
2012-05-11 19:10:03 +00:00
|
|
|
|
REPH_AFTER_MAIN,
|
|
|
|
|
REPH_BEFORE_SUBSCRIPT,
|
|
|
|
|
REPH_AFTER_SUBSCRIPT,
|
|
|
|
|
REPH_BEFORE_POSTSCRIPT,
|
2012-06-04 12:28:19 +00:00
|
|
|
|
REPH_AFTER_POSTSCRIPT
|
2012-05-11 19:10:03 +00:00
|
|
|
|
} reph_pos;
|
|
|
|
|
|
|
|
|
|
/* XXX Figure out old behavior too */
|
2012-05-11 21:10:31 +00:00
|
|
|
|
switch ((hb_tag_t) buffer->props.script)
|
2012-05-11 19:10:03 +00:00
|
|
|
|
{
|
|
|
|
|
case HB_SCRIPT_MALAYALAM:
|
|
|
|
|
case HB_SCRIPT_ORIYA:
|
2012-07-24 03:51:29 +00:00
|
|
|
|
case HB_SCRIPT_SINHALA:
|
2012-05-11 19:10:03 +00:00
|
|
|
|
reph_pos = REPH_AFTER_MAIN;
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case HB_SCRIPT_GURMUKHI:
|
|
|
|
|
reph_pos = REPH_BEFORE_SUBSCRIPT;
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case HB_SCRIPT_BENGALI:
|
|
|
|
|
reph_pos = REPH_AFTER_SUBSCRIPT;
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
default:
|
|
|
|
|
case HB_SCRIPT_DEVANAGARI:
|
|
|
|
|
case HB_SCRIPT_GUJARATI:
|
|
|
|
|
reph_pos = REPH_BEFORE_POSTSCRIPT;
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case HB_SCRIPT_KANNADA:
|
|
|
|
|
case HB_SCRIPT_TAMIL:
|
|
|
|
|
case HB_SCRIPT_TELUGU:
|
|
|
|
|
reph_pos = REPH_AFTER_POSTSCRIPT;
|
|
|
|
|
break;
|
|
|
|
|
}
|
2012-05-10 19:44:50 +00:00
|
|
|
|
|
2012-05-10 11:45:52 +00:00
|
|
|
|
/* 1. If reph should be positioned after post-base consonant forms,
|
|
|
|
|
* proceed to step 5.
|
2012-05-10 19:44:50 +00:00
|
|
|
|
*/
|
2012-05-11 19:36:32 +00:00
|
|
|
|
if (reph_pos == REPH_AFTER_POSTSCRIPT)
|
2012-05-10 19:44:50 +00:00
|
|
|
|
{
|
2012-05-11 19:36:32 +00:00
|
|
|
|
goto reph_step_5;
|
2012-05-10 19:44:50 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* 2. If the reph repositioning class is not after post-base: target
|
2012-05-10 11:45:52 +00:00
|
|
|
|
* position is after the first explicit halant glyph between the
|
|
|
|
|
* first post-reph consonant and last main consonant. If ZWJ or ZWNJ
|
|
|
|
|
* are following this halant, position is moved after it. If such
|
|
|
|
|
* position is found, this is the target position. Otherwise,
|
|
|
|
|
* proceed to the next step.
|
|
|
|
|
*
|
|
|
|
|
* Note: in old-implementation fonts, where classifications were
|
|
|
|
|
* fixed in shaping engine, there was no case where reph position
|
|
|
|
|
* will be found on this step.
|
2012-05-10 19:44:50 +00:00
|
|
|
|
*/
|
|
|
|
|
{
|
|
|
|
|
new_reph_pos = start + 1;
|
2012-07-17 15:37:32 +00:00
|
|
|
|
while (new_reph_pos < base && !is_halant_or_coeng (info[new_reph_pos]))
|
2012-05-10 19:44:50 +00:00
|
|
|
|
new_reph_pos++;
|
|
|
|
|
|
2012-07-17 15:37:32 +00:00
|
|
|
|
if (new_reph_pos < base && is_halant_or_coeng (info[new_reph_pos])) {
|
2012-05-10 19:44:50 +00:00
|
|
|
|
/* ->If ZWJ or ZWNJ are following this halant, position is moved after it. */
|
|
|
|
|
if (new_reph_pos + 1 < base && is_joiner (info[new_reph_pos + 1]))
|
|
|
|
|
new_reph_pos++;
|
|
|
|
|
goto reph_move;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* 3. If reph should be repositioned after the main consonant: find the
|
2012-05-10 11:45:52 +00:00
|
|
|
|
* first consonant not ligated with main, or find the first
|
|
|
|
|
* consonant that is not a potential pre-base reordering Ra.
|
2012-05-10 19:44:50 +00:00
|
|
|
|
*/
|
2012-05-11 19:36:32 +00:00
|
|
|
|
if (reph_pos == REPH_AFTER_MAIN)
|
2012-05-10 19:44:50 +00:00
|
|
|
|
{
|
2012-07-16 19:21:12 +00:00
|
|
|
|
new_reph_pos = base;
|
|
|
|
|
/* XXX Skip potential pre-base reordering Ra. */
|
2012-07-20 20:17:28 +00:00
|
|
|
|
while (new_reph_pos + 1 < end && info[new_reph_pos + 1].indic_position() <= POS_AFTER_MAIN)
|
2012-07-16 19:21:12 +00:00
|
|
|
|
new_reph_pos++;
|
|
|
|
|
if (new_reph_pos < end)
|
|
|
|
|
goto reph_move;
|
2012-05-10 19:44:50 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* 4. If reph should be positioned before post-base consonant, find
|
2012-05-10 11:45:52 +00:00
|
|
|
|
* first post-base classified consonant not ligated with main. If no
|
|
|
|
|
* consonant is found, the target position should be before the
|
|
|
|
|
* first matra, syllable modifier sign or vedic sign.
|
2012-05-10 19:44:50 +00:00
|
|
|
|
*/
|
2012-05-11 19:36:32 +00:00
|
|
|
|
/* This is our take on what step 4 is trying to say (and failing, BADLY). */
|
|
|
|
|
if (reph_pos == REPH_AFTER_SUBSCRIPT)
|
2012-05-10 19:44:50 +00:00
|
|
|
|
{
|
2012-05-11 19:36:32 +00:00
|
|
|
|
new_reph_pos = base;
|
|
|
|
|
while (new_reph_pos < end &&
|
2012-07-19 16:11:12 +00:00
|
|
|
|
!( FLAG (info[new_reph_pos + 1].indic_position()) & (FLAG (POS_POST_C) | FLAG (POS_AFTER_POST) | FLAG (POS_SMVD))))
|
2012-05-11 19:36:32 +00:00
|
|
|
|
new_reph_pos++;
|
|
|
|
|
if (new_reph_pos < end)
|
|
|
|
|
goto reph_move;
|
2012-05-10 19:44:50 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* 5. If no consonant is found in steps 3 or 4, move reph to a position
|
2012-05-10 11:45:52 +00:00
|
|
|
|
* immediately before the first post-base matra, syllable modifier
|
|
|
|
|
* sign or vedic sign that has a reordering class after the intended
|
|
|
|
|
* reph position. For example, if the reordering position for reph
|
|
|
|
|
* is post-main, it will skip above-base matras that also have a
|
|
|
|
|
* post-main position.
|
|
|
|
|
*/
|
2012-05-10 19:44:50 +00:00
|
|
|
|
reph_step_5:
|
|
|
|
|
{
|
2012-07-20 15:25:41 +00:00
|
|
|
|
/* Copied from step 2. */
|
|
|
|
|
new_reph_pos = start + 1;
|
|
|
|
|
while (new_reph_pos < base && !is_halant_or_coeng (info[new_reph_pos]))
|
|
|
|
|
new_reph_pos++;
|
|
|
|
|
|
|
|
|
|
if (new_reph_pos < base && is_halant_or_coeng (info[new_reph_pos])) {
|
|
|
|
|
/* ->If ZWJ or ZWNJ are following this halant, position is moved after it. */
|
|
|
|
|
if (new_reph_pos + 1 < base && is_joiner (info[new_reph_pos + 1]))
|
|
|
|
|
new_reph_pos++;
|
|
|
|
|
goto reph_move;
|
|
|
|
|
}
|
2012-05-10 19:44:50 +00:00
|
|
|
|
}
|
2012-05-10 11:45:52 +00:00
|
|
|
|
|
2012-05-10 19:44:50 +00:00
|
|
|
|
/* 6. Otherwise, reorder reph to the end of the syllable.
|
|
|
|
|
*/
|
|
|
|
|
{
|
|
|
|
|
new_reph_pos = end - 1;
|
|
|
|
|
while (new_reph_pos > start && info[new_reph_pos].indic_position() == POS_SMVD)
|
|
|
|
|
new_reph_pos--;
|
|
|
|
|
|
2012-05-11 14:54:40 +00:00
|
|
|
|
/*
|
|
|
|
|
* If the Reph is to be ending up after a Matra,Halant sequence,
|
|
|
|
|
* position it before that Halant so it can interact with the Matra.
|
|
|
|
|
* However, if it's a plain Consonant,Halant we shouldn't do that.
|
|
|
|
|
* Uniscribe doesn't do this.
|
|
|
|
|
* TEST: U+0930,U+094D,U+0915,U+094B,U+094D
|
|
|
|
|
*/
|
2012-06-05 19:17:44 +00:00
|
|
|
|
if (!indic_options ().uniscribe_bug_compatible &&
|
2012-07-17 15:37:32 +00:00
|
|
|
|
unlikely (is_halant_or_coeng (info[new_reph_pos]))) {
|
2012-05-10 19:44:50 +00:00
|
|
|
|
for (unsigned int i = base + 1; i < new_reph_pos; i++)
|
|
|
|
|
if (info[i].indic_category() == OT_M) {
|
|
|
|
|
/* Ok, got it. */
|
|
|
|
|
new_reph_pos--;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
goto reph_move;
|
2012-05-10 13:41:04 +00:00
|
|
|
|
}
|
|
|
|
|
|
2012-05-10 19:44:50 +00:00
|
|
|
|
reph_move:
|
|
|
|
|
{
|
2012-07-23 04:11:26 +00:00
|
|
|
|
/* Yay, one big cluster! Merge before moving. */
|
|
|
|
|
buffer->merge_clusters (start, end);
|
|
|
|
|
|
2012-05-10 19:44:50 +00:00
|
|
|
|
/* Move */
|
|
|
|
|
hb_glyph_info_t reph = info[start];
|
|
|
|
|
memmove (&info[start], &info[start + 1], (new_reph_pos - start) * sizeof (info[0]));
|
|
|
|
|
info[new_reph_pos] = reph;
|
|
|
|
|
}
|
2012-05-10 11:45:52 +00:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/* o Reorder pre-base reordering consonants:
|
2011-07-31 19:18:57 +00:00
|
|
|
|
*
|
|
|
|
|
* If a pre-base reordering consonant is found, reorder it according to
|
|
|
|
|
* the following rules:
|
|
|
|
|
*/
|
|
|
|
|
|
2012-07-16 19:30:05 +00:00
|
|
|
|
if (pref_mask && base + 1 < end) /* Otherwise there can't be any pre-base reordering Ra. */
|
|
|
|
|
{
|
2012-07-16 21:04:46 +00:00
|
|
|
|
for (unsigned int i = base + 1; i < end; i++)
|
|
|
|
|
if ((info[i].mask & pref_mask) != 0)
|
2012-07-16 19:49:08 +00:00
|
|
|
|
{
|
2012-07-16 21:04:46 +00:00
|
|
|
|
/* 1. Only reorder a glyph produced by substitution during application
|
|
|
|
|
* of the <pref> feature. (Note that a font may shape a Ra consonant with
|
|
|
|
|
* the feature generally but block it in certain contexts.)
|
|
|
|
|
*/
|
|
|
|
|
if (i + 1 == end || (info[i + 1].mask & pref_mask) == 0)
|
|
|
|
|
{
|
|
|
|
|
/*
|
|
|
|
|
* 2. Try to find a target position the same way as for pre-base matra.
|
|
|
|
|
* If it is found, reorder pre-base consonant glyph.
|
|
|
|
|
*
|
|
|
|
|
* 3. If position is not found, reorder immediately before main
|
|
|
|
|
* consonant.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
unsigned int new_pos = base;
|
2012-07-24 05:44:47 +00:00
|
|
|
|
while (new_pos > start &&
|
|
|
|
|
!(is_one_of (info[new_pos - 1], FLAG(OT_M) | HALANT_OR_COENG_FLAGS)))
|
2012-07-16 21:04:46 +00:00
|
|
|
|
new_pos--;
|
|
|
|
|
|
2012-07-24 06:10:20 +00:00
|
|
|
|
/* In Khmer coeng model, a V,Ra can go *after* matras. If it goes after a
|
|
|
|
|
* split matra, it should be reordered to *before* the left part of such matra. */
|
|
|
|
|
if (new_pos > start && info[new_pos - 1].indic_category() == OT_M)
|
|
|
|
|
{
|
|
|
|
|
unsigned int old_pos = i;
|
|
|
|
|
for (unsigned int i = base + 1; i < old_pos; i++)
|
|
|
|
|
if (info[i].indic_category() == OT_M)
|
|
|
|
|
{
|
|
|
|
|
new_pos--;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2012-07-17 15:37:32 +00:00
|
|
|
|
if (new_pos > start && is_halant_or_coeng (info[new_pos - 1]))
|
2012-07-16 21:04:46 +00:00
|
|
|
|
/* -> If ZWJ or ZWNJ follow this halant, position is moved after it. */
|
|
|
|
|
if (new_pos < end && is_joiner (info[new_pos]))
|
|
|
|
|
new_pos++;
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
unsigned int old_pos = i;
|
2012-07-23 04:11:26 +00:00
|
|
|
|
buffer->merge_clusters (new_pos, old_pos + 1);
|
2012-07-16 21:04:46 +00:00
|
|
|
|
hb_glyph_info_t tmp = info[old_pos];
|
|
|
|
|
memmove (&info[new_pos + 1], &info[new_pos], (old_pos - new_pos) * sizeof (info[0]));
|
|
|
|
|
info[new_pos] = tmp;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
break;
|
2012-07-16 19:49:08 +00:00
|
|
|
|
}
|
2012-07-16 19:30:05 +00:00
|
|
|
|
}
|
2012-05-11 18:50:53 +00:00
|
|
|
|
|
|
|
|
|
|
2012-05-11 18:59:26 +00:00
|
|
|
|
/* Apply 'init' to the Left Matra if it's a word start. */
|
2012-05-11 19:42:27 +00:00
|
|
|
|
if (info[start].indic_position () == POS_PRE_M &&
|
2012-05-11 18:59:26 +00:00
|
|
|
|
(!start ||
|
2012-05-13 13:54:43 +00:00
|
|
|
|
!(FLAG (_hb_glyph_info_get_general_category (&info[start - 1])) &
|
2012-07-20 17:37:48 +00:00
|
|
|
|
FLAG_RANGE (HB_UNICODE_GENERAL_CATEGORY_FORMAT, HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK))))
|
2012-07-16 19:30:05 +00:00
|
|
|
|
info[start].mask |= init_mask;
|
2012-05-11 18:59:26 +00:00
|
|
|
|
|
2012-05-11 18:50:53 +00:00
|
|
|
|
|
2012-07-20 15:42:24 +00:00
|
|
|
|
/*
|
|
|
|
|
* Finish off the clusters and go home!
|
|
|
|
|
*/
|
2012-07-20 17:51:31 +00:00
|
|
|
|
if (indic_options ().uniscribe_bug_compatible)
|
2012-05-11 14:43:12 +00:00
|
|
|
|
{
|
2012-07-20 17:56:32 +00:00
|
|
|
|
/* Uniscribe merges the entire cluster.
|
2012-05-10 16:34:34 +00:00
|
|
|
|
* This means, half forms are submerged into the main consonants cluster.
|
|
|
|
|
* This is unnecessary, and makes cursor positioning harder, but that's what
|
|
|
|
|
* Uniscribe does. */
|
2012-07-23 04:11:26 +00:00
|
|
|
|
buffer->merge_clusters (start, end);
|
2012-05-10 16:34:34 +00:00
|
|
|
|
}
|
2012-05-09 15:56:03 +00:00
|
|
|
|
}
|
2011-07-31 19:18:57 +00:00
|
|
|
|
|
|
|
|
|
|
2012-05-09 15:56:03 +00:00
|
|
|
|
static void
|
|
|
|
|
final_reordering (const hb_ot_map_t *map,
|
2012-05-13 14:20:10 +00:00
|
|
|
|
hb_face_t *face HB_UNUSED,
|
2012-05-09 15:56:03 +00:00
|
|
|
|
hb_buffer_t *buffer,
|
|
|
|
|
void *user_data HB_UNUSED)
|
|
|
|
|
{
|
|
|
|
|
unsigned int count = buffer->len;
|
|
|
|
|
if (!count) return;
|
|
|
|
|
|
2012-07-16 19:30:05 +00:00
|
|
|
|
hb_mask_t init_mask = map->get_1_mask (HB_TAG('i','n','i','t'));
|
|
|
|
|
hb_mask_t pref_mask = map->get_1_mask (HB_TAG('p','r','e','f'));
|
2012-05-11 18:50:53 +00:00
|
|
|
|
|
2012-05-09 15:56:03 +00:00
|
|
|
|
hb_glyph_info_t *info = buffer->info;
|
|
|
|
|
unsigned int last = 0;
|
2012-05-11 09:41:39 +00:00
|
|
|
|
unsigned int last_syllable = info[0].syllable();
|
2012-05-09 15:56:03 +00:00
|
|
|
|
for (unsigned int i = 1; i < count; i++)
|
2012-05-11 09:41:39 +00:00
|
|
|
|
if (last_syllable != info[i].syllable()) {
|
2012-07-16 19:30:05 +00:00
|
|
|
|
final_reordering_syllable (buffer, init_mask, pref_mask, last, i);
|
2012-05-09 15:56:03 +00:00
|
|
|
|
last = i;
|
2012-05-11 09:41:39 +00:00
|
|
|
|
last_syllable = info[last].syllable();
|
2012-05-09 15:56:03 +00:00
|
|
|
|
}
|
2012-07-16 19:30:05 +00:00
|
|
|
|
final_reordering_syllable (buffer, init_mask, pref_mask, last, count);
|
2011-07-31 19:18:57 +00:00
|
|
|
|
|
2011-07-29 20:37:02 +00:00
|
|
|
|
HB_BUFFER_DEALLOCATE_VAR (buffer, indic_category);
|
|
|
|
|
HB_BUFFER_DEALLOCATE_VAR (buffer, indic_position);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2011-06-02 21:43:12 +00:00
|
|
|
|
|