*/
-#define IN_HALF_BLOCK(u, Base) (((u) & ~0x7F) == (Base))
+#define IN_HALF_BLOCK(u, Base) (((u) & ~0x7Fu) == (Base))
-#define IS_DEVA(u) (IN_HALF_BLOCK (u, 0x0900))
-#define IS_BENG(u) (IN_HALF_BLOCK (u, 0x0980))
-#define IS_GURU(u) (IN_HALF_BLOCK (u, 0x0A00))
-#define IS_GUJR(u) (IN_HALF_BLOCK (u, 0x0A80))
-#define IS_ORYA(u) (IN_HALF_BLOCK (u, 0x0B00))
-#define IS_TAML(u) (IN_HALF_BLOCK (u, 0x0B80))
-#define IS_TELU(u) (IN_HALF_BLOCK (u, 0x0C00))
-#define IS_KNDA(u) (IN_HALF_BLOCK (u, 0x0C80))
-#define IS_MLYM(u) (IN_HALF_BLOCK (u, 0x0D00))
-#define IS_SINH(u) (IN_HALF_BLOCK (u, 0x0D80))
-#define IS_KHMR(u) (IN_HALF_BLOCK (u, 0x1780))
+#define IS_DEVA(u) (IN_HALF_BLOCK (u, 0x0900u))
+#define IS_BENG(u) (IN_HALF_BLOCK (u, 0x0980u))
+#define IS_GURU(u) (IN_HALF_BLOCK (u, 0x0A00u))
+#define IS_GUJR(u) (IN_HALF_BLOCK (u, 0x0A80u))
+#define IS_ORYA(u) (IN_HALF_BLOCK (u, 0x0B00u))
+#define IS_TAML(u) (IN_HALF_BLOCK (u, 0x0B80u))
+#define IS_TELU(u) (IN_HALF_BLOCK (u, 0x0C00u))
+#define IS_KNDA(u) (IN_HALF_BLOCK (u, 0x0C80u))
+#define IS_MLYM(u) (IN_HALF_BLOCK (u, 0x0D00u))
+#define IS_SINH(u) (IN_HALF_BLOCK (u, 0x0D80u))
+#define IS_KHMR(u) (IN_HALF_BLOCK (u, 0x1780u))
#define MATRA_POS_LEFT(u) POS_PRE_M
IS_GUJR(u) ? POS_AFTER_POST : \
IS_ORYA(u) ? POS_AFTER_POST : \
IS_TAML(u) ? POS_AFTER_POST : \
- IS_TELU(u) ? (u <= 0x0C42 ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
- IS_KNDA(u) ? (u < 0x0CC3 || u > 0xCD6 ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
+ IS_TELU(u) ? (u <= 0x0C42u ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
+ IS_KNDA(u) ? (u < 0x0CC3u || u > 0xCD6u ? POS_BEFORE_SUB : POS_AFTER_SUB) : \
IS_MLYM(u) ? POS_AFTER_POST : \
IS_SINH(u) ? POS_AFTER_SUB : \
IS_KHMR(u) ? POS_AFTER_POST : \
* Or completely remove it and just check in the tables.
*/
static const hb_codepoint_t ra_chars[] = {
- 0x0930, /* Devanagari */
- 0x09B0, /* Bengali */
- 0x09F0, /* Bengali */
- 0x0A30, /* Gurmukhi */ /* No Reph */
- 0x0AB0, /* Gujarati */
- 0x0B30, /* Oriya */
- 0x0BB0, /* Tamil */ /* No Reph */
- 0x0C30, /* Telugu */ /* Reph formed only with ZWJ */
- 0x0CB0, /* Kannada */
- 0x0D30, /* Malayalam */ /* No Reph, Logical Repha */
-
- 0x0DBB, /* Sinhala */ /* Reph formed only with ZWJ */
-
- 0x179A, /* Khmer */ /* No Reph, Visual Repha */
+ 0x0930u, /* Devanagari */
+ 0x09B0u, /* Bengali */
+ 0x09F0u, /* Bengali */
+ 0x0A30u, /* Gurmukhi */ /* No Reph */
+ 0x0AB0u, /* Gujarati */
+ 0x0B30u, /* Oriya */
+ 0x0BB0u, /* Tamil */ /* No Reph */
+ 0x0C30u, /* Telugu */ /* Reph formed only with ZWJ */
+ 0x0CB0u, /* Kannada */
+ 0x0D30u, /* Malayalam */ /* No Reph, Logical Repha */
+
+ 0x0DBBu, /* Sinhala */ /* Reph formed only with ZWJ */
+
+ 0x179Au, /* Khmer */ /* No Reph, Visual Repha */
};
-static inline indic_position_t
-consonant_position (hb_codepoint_t u)
-{
- if ((u & ~0x007F) == 0x1780)
- return POS_BELOW_C; /* In Khmer coeng model, post and below forms should not be reordered. */
- return POS_BASE_C; /* Will recategorize later based on font lookups. */
-}
-
static inline bool
is_ra (hb_codepoint_t u)
{
is_one_of (const hb_glyph_info_t &info, unsigned int flags)
{
/* If it ligated, all bets are off. */
- if (is_a_ligature (info)) return false;
+ if (_hb_glyph_info_ligated (&info)) return false;
return !!(FLAG (info.indic_category()) & flags);
}
-#define JOINER_FLAGS (FLAG (OT_ZWJ) | FLAG (OT_ZWNJ))
static inline bool
is_joiner (const hb_glyph_info_t &info)
{
return is_one_of (info, JOINER_FLAGS);
}
-/* 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! */
-#define CONSONANT_FLAGS (FLAG (OT_C) | FLAG (OT_CM) | FLAG (OT_Ra) | FLAG (OT_V) | FLAG (OT_NBSP) | FLAG (OT_DOTTEDCIRCLE))
static inline bool
is_consonant (const hb_glyph_info_t &info)
{
return is_one_of (info, CONSONANT_FLAGS);
}
-#define HALANT_OR_COENG_FLAGS (FLAG (OT_H) | FLAG (OT_Coeng))
static inline bool
is_halant_or_coeng (const hb_glyph_info_t &info)
{
{
hb_codepoint_t u = info.codepoint;
unsigned int type = hb_indic_get_categories (u);
- indic_category_t cat = (indic_category_t) (type & 0x7F);
+ indic_category_t cat = (indic_category_t) (type & 0x7Fu);
indic_position_t pos = (indic_position_t) (type >> 8);
/* The spec says U+0952 is OT_A. However, testing shows that Uniscribe
- * treats U+0951..U+0952 all as OT_VD.
- * TESTS:
+ * treats a whole bunch of characters similarly.
+ * TESTS: For example, for U+0951:
* 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 (unlikely (u == 0x17D1))
- cat = OT_X;
- if (cat == OT_X &&
- unlikely (hb_in_range<hb_codepoint_t> (u, 0x17CB, 0x17D3))) /* Khmer Various signs */
+ * U+092E,U+0951,U+0952
+ * U+092E,U+0952,U+0951
+ */
+ if (unlikely (hb_in_ranges (u, 0x0951u, 0x0952u,
+ 0x1CD0u, 0x1CD2u,
+ 0x1CD4u, 0x1CE1u) ||
+ u == 0x1CF4u))
+ cat = OT_A;
+ /* The following act more like the Bindus. */
+ else if (unlikely (hb_in_range (u, 0x0953u, 0x0954u)))
+ cat = OT_SM;
+ /* The following act like consonants. */
+ else if (unlikely (hb_in_ranges (u, 0x0A72u, 0x0A73u,
+ 0x1CF5u, 0x1CF6u)))
+ cat = OT_C;
+ /* TODO: The following should only be allowed after a Visarga.
+ * For now, just treat them like regular tone marks. */
+ else if (unlikely (hb_in_range (u, 0x1CE2u, 0x1CE8u)))
+ cat = OT_A;
+ /* TODO: The following should only be allowed after some of
+ * the nasalization marks, maybe only for U+1CE9..U+1CF1.
+ * For now, just treat them like tone marks. */
+ else if (unlikely (u == 0x1CEDu))
+ cat = OT_A;
+ /* The following take marks in standalone clusters, similar to Avagraha. */
+ else if (unlikely (hb_in_ranges (u, 0xA8F2u, 0xA8F7u,
+ 0x1CE9u, 0x1CECu,
+ 0x1CEEu, 0x1CF1u)))
+ {
+ cat = OT_Symbol;
+ ASSERT_STATIC ((int) INDIC_SYLLABIC_CATEGORY_AVAGRAHA == OT_Symbol);
+ }
+ else if (unlikely (hb_in_range (u, 0x17CDu, 0x17D1u) ||
+ u == 0x17CBu || u == 0x17D3u || u == 0x17DDu)) /* Khmer Various signs */
{
/* These are like Top Matras. */
cat = OT_M;
pos = POS_ABOVE_C;
}
- if (u == 0x17C6) /* Khmer Bindu doesn't like to be repositioned. */
- cat = OT_N;
-
- 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;
- else if (unlikely (u == 0x0A71)) cat = OT_SM; /* GURMUKHI ADDAK. More like consonant medial. like 0A75. */
-
- 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;
- }
-
+ else if (unlikely (u == 0x17C6u)) cat = OT_N; /* Khmer Bindu doesn't like to be repositioned. */
+ else if (unlikely (u == 0x17D2u)) cat = OT_Coeng; /* Khmer coeng */
+ else if (unlikely (hb_in_range (u, 0x2010u, 0x2011u)))
+ cat = OT_PLACEHOLDER;
+ else if (unlikely (u == 0x25CCu)) cat = OT_DOTTEDCIRCLE;
+ else if (unlikely (u == 0xA982u)) cat = OT_SM; /* Javanese repha. */
+ else if (unlikely (u == 0xA9BEu)) cat = OT_CM2; /* Javanese medial ya. */
+ else if (unlikely (u == 0xA9BDu)) { cat = OT_M; pos = POS_POST_C; } /* Javanese vocalic r. */
/*
if ((FLAG (cat) & CONSONANT_FLAGS))
{
- pos = consonant_position (u);
+ pos = POS_BASE_C;
if (is_ra (u))
cat = OT_Ra;
}
{
pos = matra_position (u, pos);
}
- else if (cat == OT_SM || cat == OT_VD)
+ else if ((FLAG (cat) & (FLAG (OT_SM) | FLAG (OT_VD) | FLAG (OT_A) | FLAG (OT_Symbol))))
{
pos = POS_SMVD;
}
- if (unlikely (u == 0x0B01)) pos = POS_BEFORE_SUB; /* Oriya Bindu is BeforeSub in the spec. */
+ if (unlikely (u == 0x0B01u)) pos = POS_BEFORE_SUB; /* Oriya Bindu is BeforeSub in the spec. */
enum base_position_t {
BASE_POS_FIRST,
+ BASE_POS_LAST_SINHALA,
BASE_POS_LAST
};
enum reph_position_t {
- REPH_POS_DEFAULT = POS_BEFORE_POST,
-
REPH_POS_AFTER_MAIN = POS_AFTER_MAIN,
REPH_POS_BEFORE_SUB = POS_BEFORE_SUB,
REPH_POS_AFTER_SUB = POS_AFTER_SUB,
REPH_POS_BEFORE_POST = POS_BEFORE_POST,
- REPH_POS_AFTER_POST = POS_AFTER_POST
+ REPH_POS_AFTER_POST = POS_AFTER_POST,
+ REPH_POS_DONT_CARE = POS_RA_TO_BECOME_REPH
};
enum reph_mode_t {
REPH_MODE_IMPLICIT, /* Reph formed out of initial Ra,H sequence. */
REPH_MODE_VIS_REPHA, /* Encoded Repha character, no reordering needed. */
REPH_MODE_LOG_REPHA /* Encoded Repha character, needs reordering. */
};
+enum blwf_mode_t {
+ BLWF_MODE_PRE_AND_POST, /* Below-forms feature applied to pre-base and post-base. */
+ BLWF_MODE_POST_ONLY /* Below-forms feature applied to post-base only. */
+};
+enum pref_len_t {
+ PREF_LEN_1 = 1,
+ PREF_LEN_2 = 2,
+ PREF_LEN_DONT_CARE = PREF_LEN_2
+};
struct indic_config_t
{
hb_script_t script;
base_position_t base_pos;
reph_position_t reph_pos;
reph_mode_t reph_mode;
+ blwf_mode_t blwf_mode;
+ pref_len_t pref_len;
};
static const indic_config_t indic_configs[] =
{
/* Default. Should be first. */
- {HB_SCRIPT_INVALID, false, 0,BASE_POS_LAST, REPH_POS_DEFAULT, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_DEVANAGARI,true, 0x094D,BASE_POS_LAST, REPH_POS_BEFORE_POST,REPH_MODE_IMPLICIT},
- {HB_SCRIPT_BENGALI, true, 0x09CD,BASE_POS_LAST, REPH_POS_AFTER_SUB, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_GURMUKHI, true, 0x0A4D,BASE_POS_LAST, REPH_POS_BEFORE_SUB, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_GUJARATI, true, 0x0ACD,BASE_POS_LAST, REPH_POS_BEFORE_POST,REPH_MODE_IMPLICIT},
- {HB_SCRIPT_ORIYA, true, 0x0B4D,BASE_POS_LAST, REPH_POS_AFTER_MAIN, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_TAMIL, true, 0x0BCD,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_TELUGU, true, 0x0C4D,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_EXPLICIT},
- {HB_SCRIPT_KANNADA, true, 0x0CCD,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_IMPLICIT},
- {HB_SCRIPT_MALAYALAM, true, 0x0D4D,BASE_POS_LAST, REPH_POS_AFTER_MAIN, REPH_MODE_LOG_REPHA},
- {HB_SCRIPT_SINHALA, false,0x0DCA,BASE_POS_FIRST,REPH_POS_AFTER_MAIN, REPH_MODE_EXPLICIT},
- {HB_SCRIPT_KHMER, false,0x17D2,BASE_POS_FIRST,REPH_POS_DEFAULT, REPH_MODE_VIS_REPHA},
+ {HB_SCRIPT_INVALID, false, 0,BASE_POS_LAST, REPH_POS_BEFORE_POST,REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_1},
+ {HB_SCRIPT_DEVANAGARI,true, 0x094Du,BASE_POS_LAST, REPH_POS_BEFORE_POST,REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_BENGALI, true, 0x09CDu,BASE_POS_LAST, REPH_POS_AFTER_SUB, REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_GURMUKHI, true, 0x0A4Du,BASE_POS_LAST, REPH_POS_BEFORE_SUB, REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_GUJARATI, true, 0x0ACDu,BASE_POS_LAST, REPH_POS_BEFORE_POST,REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_ORIYA, true, 0x0B4Du,BASE_POS_LAST, REPH_POS_AFTER_MAIN, REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_TAMIL, true, 0x0BCDu,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_IMPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_2},
+ {HB_SCRIPT_TELUGU, true, 0x0C4Du,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_EXPLICIT, BLWF_MODE_POST_ONLY, PREF_LEN_2},
+ {HB_SCRIPT_KANNADA, true, 0x0CCDu,BASE_POS_LAST, REPH_POS_AFTER_POST, REPH_MODE_IMPLICIT, BLWF_MODE_POST_ONLY, PREF_LEN_2},
+ {HB_SCRIPT_MALAYALAM, true, 0x0D4Du,BASE_POS_LAST, REPH_POS_AFTER_MAIN, REPH_MODE_LOG_REPHA,BLWF_MODE_PRE_AND_POST, PREF_LEN_2},
+ {HB_SCRIPT_SINHALA, false,0x0DCAu,BASE_POS_LAST_SINHALA,
+ REPH_POS_AFTER_MAIN, REPH_MODE_EXPLICIT, BLWF_MODE_PRE_AND_POST, PREF_LEN_DONT_CARE},
+ {HB_SCRIPT_KHMER, false,0x17D2u,BASE_POS_FIRST,REPH_POS_DONT_CARE, REPH_MODE_VIS_REPHA,BLWF_MODE_PRE_AND_POST, PREF_LEN_2},
+ {HB_SCRIPT_JAVANESE, false,0xA9C0u,BASE_POS_FIRST,REPH_POS_DONT_CARE, REPH_MODE_VIS_REPHA,BLWF_MODE_PRE_AND_POST, PREF_LEN_1},
};
{HB_TAG('r','k','r','f'), F_GLOBAL},
{HB_TAG('p','r','e','f'), F_NONE},
{HB_TAG('b','l','w','f'), F_NONE},
- {HB_TAG('h','a','l','f'), F_NONE},
{HB_TAG('a','b','v','f'), F_NONE},
+ {HB_TAG('h','a','l','f'), F_NONE},
{HB_TAG('p','s','t','f'), F_NONE},
- {HB_TAG('c','f','a','r'), F_NONE},
{HB_TAG('v','a','t','u'), F_GLOBAL},
{HB_TAG('c','j','c','t'), F_GLOBAL},
+ {HB_TAG('c','f','a','r'), F_NONE},
/*
* Other features.
* These features are applied all at once, after final_reordering.
+ * Default Bengali font in Windows for example has intermixed
+ * lookups for init,pres,abvs,blws features.
*/
{HB_TAG('i','n','i','t'), F_NONE},
{HB_TAG('p','r','e','s'), F_GLOBAL},
_RKRF,
PREF,
BLWF,
- HALF,
ABVF,
+ HALF,
PSTF,
- CFAR,
_VATU,
_CJCT,
+ CFAR,
INIT,
_PRES,
final_reordering (const hb_ot_shape_plan_t *plan,
hb_font_t *font,
hb_buffer_t *buffer);
+static void
+clear_syllables (const hb_ot_shape_plan_t *plan,
+ hb_font_t *font,
+ hb_buffer_t *buffer);
static void
collect_features_indic (hb_ot_shape_planner_t *plan)
for (; i < INDIC_NUM_FEATURES; i++) {
map->add_feature (indic_features[i].tag, 1, indic_features[i].flags | F_MANUAL_ZWJ);
}
+
+ map->add_global_bool_feature (HB_TAG('c','a','l','t'));
+ map->add_global_bool_feature (HB_TAG('c','l','i','g'));
+
+ map->add_gsub_pause (clear_syllables);
}
static void
override_features_indic (hb_ot_shape_planner_t *plan)
{
- /* Uniscribe does not apply 'kern'. */
+ /* Uniscribe does not apply 'kern' in Khmer. */
if (hb_options ().uniscribe_bug_compatible)
- plan->map.add_feature (HB_TAG('k','e','r','n'), 0, F_GLOBAL);
+ {
+ switch ((hb_tag_t) plan->props.script)
+ {
+ case HB_SCRIPT_KHMER:
+ plan->map.add_feature (HB_TAG('k','e','r','n'), 0, F_GLOBAL);
+ break;
+ }
+ }
plan->map.add_feature (HB_TAG('l','i','g','a'), 0, F_GLOBAL);
}
struct would_substitute_feature_t
{
- inline void init (const hb_ot_map_t *map, hb_tag_t feature_tag)
+ inline void init (const hb_ot_map_t *map, hb_tag_t feature_tag, bool zero_context_)
{
+ zero_context = zero_context_;
map->get_stage_lookups (0/*GSUB*/,
map->get_feature_stage (0/*GSUB*/, feature_tag),
&lookups, &count);
inline bool would_substitute (const hb_codepoint_t *glyphs,
unsigned int glyphs_count,
- bool zero_context,
hb_face_t *face) const
{
for (unsigned int i = 0; i < count; i++)
private:
const hb_ot_map_t::lookup_map_t *lookups;
unsigned int count;
+ bool zero_context;
};
struct indic_shape_plan_t
break;
}
- indic_plan->is_old_spec = indic_plan->config->has_old_spec && ((plan->map.chosen_script[0] & 0x000000FF) != '2');
+ indic_plan->is_old_spec = indic_plan->config->has_old_spec && ((plan->map.chosen_script[0] & 0x000000FFu) != '2');
indic_plan->virama_glyph = (hb_codepoint_t) -1;
- indic_plan->rphf.init (&plan->map, HB_TAG('r','p','h','f'));
- indic_plan->pref.init (&plan->map, HB_TAG('p','r','e','f'));
- indic_plan->blwf.init (&plan->map, HB_TAG('b','l','w','f'));
- indic_plan->pstf.init (&plan->map, HB_TAG('p','s','t','f'));
+ /* Use zero-context would_substitute() matching for new-spec of the main
+ * Indic scripts, and scripts with one spec only, but not for old-specs. */
+ bool zero_context = !indic_plan->is_old_spec;
+ indic_plan->rphf.init (&plan->map, HB_TAG('r','p','h','f'), zero_context);
+ indic_plan->pref.init (&plan->map, HB_TAG('p','r','e','f'), zero_context);
+ indic_plan->blwf.init (&plan->map, HB_TAG('b','l','w','f'), zero_context);
+ indic_plan->pstf.init (&plan->map, HB_TAG('p','s','t','f'), zero_context);
for (unsigned int i = 0; i < ARRAY_LENGTH (indic_plan->mask_array); i++)
indic_plan->mask_array[i] = (indic_features[i].flags & F_GLOBAL) ?
static indic_position_t
consonant_position_from_face (const indic_shape_plan_t *indic_plan,
- const hb_codepoint_t glyphs[2],
+ const hb_codepoint_t consonant,
+ const hb_codepoint_t virama,
hb_face_t *face)
{
/* For old-spec, the order of glyphs is Consonant,Virama,
* 930,94D in 'blwf', not the expected 94D,930 (with new-spec
* table). As such, we simply match both sequences. Seems
* to work. */
- bool zero_context = indic_plan->is_old_spec ? false : true;
- hb_codepoint_t glyphs_r[2] = {glyphs[1], glyphs[0]};
- if (indic_plan->pref.would_substitute (glyphs , 2, zero_context, face) ||
- indic_plan->pref.would_substitute (glyphs_r, 2, zero_context, face))
- return POS_POST_C;
- if (indic_plan->blwf.would_substitute (glyphs , 2, zero_context, face) ||
- indic_plan->blwf.would_substitute (glyphs_r, 2, zero_context, face))
+ hb_codepoint_t glyphs[3] = {virama, consonant, virama};
+ if (indic_plan->blwf.would_substitute (glyphs , 2, face) ||
+ indic_plan->blwf.would_substitute (glyphs+1, 2, face))
return POS_BELOW_C;
- if (indic_plan->pstf.would_substitute (glyphs , 2, zero_context, face) ||
- indic_plan->pstf.would_substitute (glyphs_r, 2, zero_context, face))
+ if (indic_plan->pstf.would_substitute (glyphs , 2, face) ||
+ indic_plan->pstf.would_substitute (glyphs+1, 2, face))
+ return POS_POST_C;
+ unsigned int pref_len = indic_plan->config->pref_len;
+ if ((pref_len == PREF_LEN_2 &&
+ (indic_plan->pref.would_substitute (glyphs , 2, face) ||
+ indic_plan->pref.would_substitute (glyphs+1, 2, face)))
+ || (pref_len == PREF_LEN_1 &&
+ indic_plan->pref.would_substitute (glyphs+1, 1, face)))
return POS_POST_C;
return POS_BASE_C;
}
consonant_syllable,
vowel_syllable,
standalone_cluster,
+ symbol_cluster,
broken_cluster,
non_indic_cluster,
};
* and setup masks later on in a pause-callback. */
unsigned int count = buffer->len;
+ hb_glyph_info_t *info = buffer->info;
for (unsigned int i = 0; i < count; i++)
- set_indic_properties (buffer->info[i]);
+ set_indic_properties (info[i]);
}
static void
{
const indic_shape_plan_t *indic_plan = (const indic_shape_plan_t *) plan->data;
- hb_codepoint_t glyphs[2];
- if (indic_plan->get_virama_glyph (font, &glyphs[0]))
+ if (indic_plan->config->base_pos != BASE_POS_LAST)
+ return;
+
+ hb_codepoint_t virama;
+ if (indic_plan->get_virama_glyph (font, &virama))
{
hb_face_t *face = font->face;
unsigned int count = buffer->len;
+ hb_glyph_info_t *info = buffer->info;
for (unsigned int i = 0; i < count; i++)
- if (buffer->info[i].indic_position() == POS_BASE_C) {
- glyphs[1] = buffer->info[i].codepoint;
- buffer->info[i].indic_position() = consonant_position_from_face (indic_plan, glyphs, face);
+ if (info[i].indic_position() == POS_BASE_C)
+ {
+ hb_codepoint_t consonant = info[i].codepoint;
+ info[i].indic_position() = consonant_position_from_face (indic_plan, consonant, virama, face);
}
}
}
* and has more than one consonant, Ra is excluded from candidates for
* base consonants. */
unsigned int limit = start;
- if (indic_plan->mask_array[RPHF] &&
+ if (indic_plan->config->reph_pos != REPH_POS_DONT_CARE &&
+ indic_plan->mask_array[RPHF] &&
start + 3 <= end &&
(
(indic_plan->config->reph_mode == REPH_MODE_IMPLICIT && !is_joiner (info[start + 2])) ||
))
{
/* See if it matches the 'rphf' feature. */
- hb_codepoint_t glyphs[2] = {info[start].codepoint, info[start + 1].codepoint};
- if (indic_plan->rphf.would_substitute (glyphs, ARRAY_LENGTH (glyphs), true, face))
+ hb_codepoint_t glyphs[3] = {info[start].codepoint,
+ info[start + 1].codepoint,
+ indic_plan->config->reph_mode == REPH_MODE_EXPLICIT ?
+ info[start + 2].codepoint : 0};
+ if (indic_plan->rphf.would_substitute (glyphs, 2, face) ||
+ (indic_plan->config->reph_mode == REPH_MODE_EXPLICIT &&
+ indic_plan->rphf.would_substitute (glyphs, 3, face)))
{
limit += 2;
while (limit < end && is_joiner (info[limit]))
}
break;
- case BASE_POS_FIRST:
+ case BASE_POS_LAST_SINHALA:
{
- /* In scripts without half forms (eg. Khmer), the first consonant is always the base. */
+ /* Sinhala base positioning is slightly different from main Indic, in that:
+ * 1. Its ZWJ behavior is different,
+ * 2. We don't need to look into the font for consonant positions.
+ */
if (!has_reph)
base = limit;
/* Find the last base consonant that is not blocked by ZWJ. If there is
* a ZWJ right before a base consonant, that would request a subjoined form. */
for (unsigned int i = limit; i < end; i++)
- if (is_consonant (info[i]) && info[i].indic_position() == POS_BASE_C)
+ if (is_consonant (info[i]))
{
if (limit < i && info[i - 1].indic_category() == OT_ZWJ)
break;
/* 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)
+ if (is_consonant (info[i]))
+ info[i].indic_position() = POS_BELOW_C;
+ }
+ break;
+
+ case BASE_POS_FIRST:
+ {
+ /* The first consonant is always the base. */
+
+ assert (indic_plan->config->reph_mode == REPH_MODE_VIS_REPHA);
+ assert (!has_reph);
+
+ base = start;
+
+ /* 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_BELOW_C;
}
break;
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)))
+ if ((FLAG (info[i].indic_category()) & (JOINER_FLAGS | FLAG (OT_N) | FLAG (OT_RS) | MEDIAL_FLAGS | HALANT_OR_COENG_FLAGS)))
{
info[i].indic_position() = last_pos;
if (unlikely (info[i].indic_category() == OT_H &&
}
}
}
- /* Re-attach ZWJ, ZWNJ, and halant to next char, for after-base consonants. */
+ /* For post-base consonants let them own anything before them
+ * since the last consonant or matra. */
{
- unsigned int last_halant = end;
+ unsigned int last = base;
for (unsigned int i = base + 1; i < end; i++)
- if (is_halant_or_coeng (info[i]))
- last_halant = i;
- else if (is_consonant (info[i])) {
- for (unsigned int j = last_halant; j < i; j++)
- if (info[j].indic_position() != POS_SMVD)
+ if (is_consonant (info[i]))
+ {
+ for (unsigned int j = last + 1; j < i; j++)
+ if (info[j].indic_position() < POS_SMVD)
info[j].indic_position() = info[i].indic_position();
- }
+ last = i;
+ } else if (info[i].indic_category() == OT_M)
+ last = i;
}
+
{
- /* 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);
+ /* Use syllable() for sort accounting temporarily. */
+ unsigned int syllable = info[start].syllable();
+ for (unsigned int i = start; i < end; i++)
+ info[i].syllable() = i - start;
+
/* Sit tight, rock 'n roll! */
hb_bubble_sort (info + start, end - start, compare_indic_order);
/* Find base again */
base = end;
for (unsigned int i = start; i < end; i++)
- if (info[i].indic_position() == POS_BASE_C) {
- base = i;
+ if (info[i].indic_position() == POS_BASE_C)
+ {
+ base = i;
break;
}
+ /* Things are out-of-control for post base positions, they may shuffle
+ * around like crazy. In old-spec mode, we move halants around, so in
+ * that case merge all clusters after base. Otherwise, check the sort
+ * order and merge as needed.
+ * For pre-base stuff, we handle cluster issues in final reordering. */
+ if (indic_plan->is_old_spec || end - base > 127)
+ buffer->merge_clusters (base, end);
+ else
+ {
+ /* Note! syllable() is a one-byte field. */
+ for (unsigned int i = base; i < end; i++)
+ if (info[i].syllable() != 255)
+ {
+ unsigned int max = i;
+ unsigned int j = start + info[i].syllable();
+ while (j != i)
+ {
+ max = MAX (max, j);
+ unsigned int next = start + info[j].syllable();
+ info[j].syllable() = 255; /* So we don't process j later again. */
+ j = next;
+ }
+ if (i != max)
+ buffer->merge_clusters (i, max + 1);
+ }
+ }
+
+ /* Put syllable back in. */
+ for (unsigned int i = start; i < end; i++)
+ info[i].syllable() = syllable;
}
/* Setup masks now */
/* Pre-base */
mask = indic_plan->mask_array[HALF];
+ if (!indic_plan->is_old_spec &&
+ indic_plan->config->blwf_mode == BLWF_MODE_PRE_AND_POST)
+ mask |= indic_plan->mask_array[BLWF];
for (unsigned int i = start; i < base; i++)
info[i].mask |= mask;
/* Base */
}
}
- if (indic_plan->mask_array[PREF] && base + 2 < end)
+ unsigned int pref_len = indic_plan->config->pref_len;
+ if (indic_plan->mask_array[PREF] && base + pref_len < end)
{
+ assert (1 <= pref_len && pref_len <= 2);
/* Find a Halant,Ra sequence and mark it for pre-base reordering processing. */
- for (unsigned int i = base + 1; i + 1 < end; i++) {
- hb_codepoint_t glyphs[2] = {info[i].codepoint, info[i + 1].codepoint};
- if (indic_plan->pref.would_substitute (glyphs, ARRAY_LENGTH (glyphs), true, face))
+ for (unsigned int i = base + 1; i + pref_len - 1 < end; i++) {
+ hb_codepoint_t glyphs[2];
+ for (unsigned int j = 0; j < pref_len; j++)
+ glyphs[j] = info[i + j].codepoint;
+ if (indic_plan->pref.would_substitute (glyphs, pref_len, face))
{
- info[i++].mask |= indic_plan->mask_array[PREF];
- info[i++].mask |= indic_plan->mask_array[PREF];
+ for (unsigned int j = 0; j < pref_len; j++)
+ info[i++].mask |= indic_plan->mask_array[PREF];
/* Mark the subsequent stuff with 'cfar'. Used in Khmer.
* Read the feature spec.
* 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 |= indic_plan->mask_array[CFAR];
+ if (indic_plan->mask_array[CFAR])
+ for (; i < end; i++)
+ info[i].mask |= indic_plan->mask_array[CFAR];
break;
}
hb_buffer_t *buffer,
unsigned int start, unsigned int end)
{
- /* We treat NBSP/dotted-circle as if they are consonants, so we should just chain.
- * Only if not in compatibility mode that is... */
+ /* We treat placeholder/dotted-circle as if they are consonants, so we
+ * should just chain. Only if not in compatibility mode that is... */
if (hb_options ().uniscribe_bug_compatible)
{
}
static void
+initial_reordering_symbol_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
+ hb_face_t *face HB_UNUSED,
+ hb_buffer_t *buffer HB_UNUSED,
+ unsigned int start HB_UNUSED, unsigned int end HB_UNUSED)
+{
+ /* 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. */
+}
+
+static void
initial_reordering_non_indic_cluster (const hb_ot_shape_plan_t *plan HB_UNUSED,
hb_face_t *face HB_UNUSED,
hb_buffer_t *buffer HB_UNUSED,
case consonant_syllable: initial_reordering_consonant_syllable (plan, face, buffer, start, end); return;
case vowel_syllable: initial_reordering_vowel_syllable (plan, face, buffer, start, end); return;
case standalone_cluster: initial_reordering_standalone_cluster (plan, face, buffer, start, end); return;
+ case symbol_cluster: initial_reordering_symbol_cluster (plan, face, buffer, start, end); return;
case broken_cluster: initial_reordering_broken_cluster (plan, face, buffer, start, end); return;
case non_indic_cluster: initial_reordering_non_indic_cluster (plan, face, buffer, start, end); return;
}
/* Note: This loop is extra overhead, but should not be measurable. */
bool has_broken_syllables = false;
unsigned int count = buffer->len;
+ hb_glyph_info_t *info = buffer->info;
for (unsigned int i = 0; i < count; i++)
- if ((buffer->info[i].syllable() & 0x0F) == broken_cluster) {
+ if ((info[i].syllable() & 0x0F) == broken_cluster)
+ {
has_broken_syllables = true;
break;
}
hb_codepoint_t dottedcircle_glyph;
- if (!font->get_glyph (0x25CC, 0, &dottedcircle_glyph))
+ if (!font->get_glyph (0x25CCu, 0, &dottedcircle_glyph))
return;
hb_glyph_info_t dottedcircle = {0};
- dottedcircle.codepoint = 0x25CC;
+ dottedcircle.codepoint = 0x25CCu;
set_indic_properties (dottedcircle);
dottedcircle.codepoint = dottedcircle_glyph;
const indic_shape_plan_t *indic_plan = (const indic_shape_plan_t *) plan->data;
hb_glyph_info_t *info = buffer->info;
+
+ /* This function relies heavily on halant glyphs. Lots of ligation
+ * and possibly multiplication substitutions happened prior to this
+ * phase, and that might have messed up our properties. Recover
+ * from a particular case of that where we're fairly sure that a
+ * class of OT_H is desired but has been lost. */
+ if (indic_plan->virama_glyph)
+ {
+ unsigned int virama_glyph = indic_plan->virama_glyph;
+ for (unsigned int i = start; i < end; i++)
+ if (info[i].codepoint == virama_glyph &&
+ _hb_glyph_info_ligated (&info[i]) &&
+ _hb_glyph_info_multiplied (&info[i]))
+ {
+ /* This will make sure that this glyph passes is_halant_or_coeng() test. */
+ info[i].indic_category() = OT_H;
+ _hb_glyph_info_clear_ligated_and_multiplied (&info[i]);
+ }
+ }
+
+
/* 4. Final reordering:
*
* After the localized forms and basic shaping forms GSUB features have been
* cluster.
*/
+ bool try_pref = !!indic_plan->mask_array[PREF];
+
/* Find base again */
unsigned int base;
for (base = start; base < end; base++)
- if (info[base].indic_position() >= POS_BASE_C) {
+ if (info[base].indic_position() >= POS_BASE_C)
+ {
+ if (try_pref && base + 1 < end && indic_plan->config->pref_len == 2)
+ {
+ for (unsigned int i = base + 1; i < end; i++)
+ if ((info[i].mask & indic_plan->mask_array[PREF]) != 0)
+ {
+ if (!(_hb_glyph_info_substituted (&info[i]) &&
+ _hb_glyph_info_ligated_and_didnt_multiply (&info[i])))
+ {
+ /* Ok, this was a 'pref' candidate but didn't form any.
+ * Base is around here... */
+ base = i;
+ while (base < end && is_halant_or_coeng (info[base]))
+ base++;
+ info[base].indic_position() = POS_BASE_C;
+
+ try_pref = false;
+ }
+ break;
+ }
+ }
+
if (start < base && info[base].indic_position() > POS_BASE_C)
base--;
break;
}
if (base == end && start < base &&
- info[base - 1].indic_category() != OT_ZWJ)
- base--;
- while (start < base &&
- (info[base].indic_category() == OT_H ||
- info[base].indic_category() == OT_N))
+ is_one_of (info[base - 1], FLAG (OT_ZWJ)))
base--;
+ if (base < end)
+ while (start < base &&
+ is_one_of (info[base], (FLAG (OT_N) | HALANT_OR_COENG_FLAGS)))
+ base--;
/* o Reorder matras:
if (buffer->props.script != HB_SCRIPT_MALAYALAM && buffer->props.script != HB_SCRIPT_TAMIL)
{
while (new_pos > start &&
- !(is_one_of (info[new_pos], (FLAG (OT_M) | FLAG (OT_H) | FLAG (OT_Coeng)))))
+ !(is_one_of (info[new_pos], (FLAG (OT_M) | HALANT_OR_COENG_FLAGS))))
new_pos--;
/* If we found no Halant we are done.
info[new_pos] = tmp;
if (old_pos < base && base <= new_pos) /* Shouldn't actually happen. */
base--;
+ buffer->merge_clusters (new_pos, MIN (end, base + 1));
new_pos--;
}
- buffer->merge_clusters (new_pos, MIN (end, base + 1));
} else {
for (unsigned int i = start; i < base; i++)
if (info[i].indic_position () == POS_PRE_M) {
* before post-base consonant forms, and after post-base consonant forms.
*/
- /* 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. */
+ /* Two cases:
+ *
+ * - If repha is encoded as a sequence of characters (Ra,H or Ra,H,ZWJ), then
+ * we should only move it if the sequence ligated to the repha form.
+ *
+ * - If repha is encoded separately and in the logical position, we should only
+ * move it if it did NOT ligate. If it ligated, it's probably the font trying
+ * to make it work without the reordering.
+ */
if (start + 1 < end &&
info[start].indic_position() == POS_RA_TO_BECOME_REPH &&
- info[start + 1].indic_position() != POS_RA_TO_BECOME_REPH)
+ ((info[start].indic_category() == OT_Repha) ^
+ _hb_glyph_info_ligated_and_didnt_multiply (&info[start])))
{
unsigned int new_reph_pos;
reph_position_t reph_pos = indic_plan->config->reph_pos;
- /* XXX Figure out old behavior too */
+ assert (reph_pos != REPH_POS_DONT_CARE);
/* 1. If reph should be positioned after post-base consonant forms,
* proceed to step 5.
if (reph_pos == REPH_POS_AFTER_MAIN)
{
new_reph_pos = base;
- /* XXX Skip potential pre-base reordering Ra. */
while (new_reph_pos + 1 < end && info[new_reph_pos + 1].indic_position() <= POS_AFTER_MAIN)
new_reph_pos++;
if (new_reph_pos < end)
reph_move:
{
- /* Yay, one big cluster! Merge before moving. */
- buffer->merge_clusters (start, end);
+ buffer->merge_clusters (start, new_reph_pos + 1);
/* Move */
hb_glyph_info_t reph = info[start];
* the following rules:
*/
- if (indic_plan->mask_array[PREF] && base + 1 < end) /* Otherwise there can't be any pre-base reordering Ra. */
+ if (try_pref && base + 1 < end) /* Otherwise there can't be any pre-base reordering Ra. */
{
+ unsigned int pref_len = indic_plan->config->pref_len;
for (unsigned int i = base + 1; i < end; i++)
if ((info[i].mask & indic_plan->mask_array[PREF]) != 0)
{
* 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 & indic_plan->mask_array[PREF]) == 0)
+ /* Note: We just check that something got substituted. We don't check that
+ * the <pref> feature actually did it...
+ *
+ * If pref len is longer than one, then only reorder if it ligated. If
+ * pref len is one, only reorder if it didn't ligate with other things. */
+ if (_hb_glyph_info_substituted (&info[i]) &&
+ ((pref_len == 1) ^ _hb_glyph_info_ligated_and_didnt_multiply (&info[i])))
{
/*
* 2. Try to find a target position the same way as for pre-base matra.
!(is_one_of (info[new_pos - 1], FLAG(OT_M) | HALANT_OR_COENG_FLAGS)))
new_pos--;
- /* In Khmer coeng model, a V,Ra can go *after* matras. If it goes after a
+ /* In Khmer coeng model, a H,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)
{
*/
if (hb_options ().uniscribe_bug_compatible)
{
- /* Uniscribe merges the entire cluster.
- * 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. */
- buffer->merge_clusters (start, end);
+ switch ((hb_tag_t) plan->props.script)
+ {
+ case HB_SCRIPT_TAMIL:
+ case HB_SCRIPT_SINHALA:
+ break;
+
+ default:
+ /* Uniscribe merges the entire cluster... Except for Tamil & Sinhala.
+ * 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. */
+ buffer->merge_clusters (start, end);
+ break;
+ }
}
}
}
final_reordering_syllable (plan, buffer, last, count);
- /* Zero syllables now... */
- for (unsigned int i = 0; i < count; i++)
- info[i].syllable() = 0;
-
HB_BUFFER_DEALLOCATE_VAR (buffer, indic_category);
HB_BUFFER_DEALLOCATE_VAR (buffer, indic_position);
}
-static hb_ot_shape_normalization_mode_t
-normalization_preference_indic (const hb_segment_properties_t *props HB_UNUSED)
+static void
+clear_syllables (const hb_ot_shape_plan_t *plan HB_UNUSED,
+ hb_font_t *font HB_UNUSED,
+ hb_buffer_t *buffer)
{
- return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT;
+ hb_glyph_info_t *info = buffer->info;
+ unsigned int count = buffer->len;
+ for (unsigned int i = 0; i < count; i++)
+ info[i].syllable() = 0;
}
+
static bool
decompose_indic (const hb_ot_shape_normalize_context_t *c,
hb_codepoint_t ab,
switch (ab)
{
/* Don't decompose these. */
- case 0x0931 : return false;
- case 0x0B94 : return false;
+ case 0x0931u : return false;
+ case 0x0B94u : return false;
/*
* Decompose split matras that don't have Unicode decompositions.
*/
- case 0x0F77 : *a = 0x0FB2; *b= 0x0F81; return true;
- case 0x0F79 : *a = 0x0FB3; *b= 0x0F81; return true;
- case 0x17BE : *a = 0x17C1; *b= 0x17BE; return true;
- case 0x17BF : *a = 0x17C1; *b= 0x17BF; return true;
- case 0x17C0 : *a = 0x17C1; *b= 0x17C0; return true;
- case 0x17C4 : *a = 0x17C1; *b= 0x17C4; return true;
- case 0x17C5 : *a = 0x17C1; *b= 0x17C5; return true;
- case 0x1925 : *a = 0x1920; *b= 0x1923; return true;
- case 0x1926 : *a = 0x1920; *b= 0x1924; return true;
- case 0x1B3C : *a = 0x1B42; *b= 0x1B3C; return true;
- case 0x1112E : *a = 0x11127; *b= 0x11131; return true;
- case 0x1112F : *a = 0x11127; *b= 0x11132; return true;
+ case 0x0F77u : *a = 0x0FB2u; *b= 0x0F81u; return true;
+ case 0x0F79u : *a = 0x0FB3u; *b= 0x0F81u; return true;
+ case 0x17BEu : *a = 0x17C1u; *b= 0x17BEu; return true;
+ case 0x17BFu : *a = 0x17C1u; *b= 0x17BFu; return true;
+ case 0x17C0u : *a = 0x17C1u; *b= 0x17C0u; return true;
+ case 0x17C4u : *a = 0x17C1u; *b= 0x17C4u; return true;
+ case 0x17C5u : *a = 0x17C1u; *b= 0x17C5u; return true;
+ case 0x1925u : *a = 0x1920u; *b= 0x1923u; return true;
+ case 0x1926u : *a = 0x1920u; *b= 0x1924u; return true;
+ case 0x1B3Cu : *a = 0x1B42u; *b= 0x1B3Cu; return true;
+ case 0x1112Eu : *a = 0x11127u; *b= 0x11131u; return true;
+ case 0x1112Fu : *a = 0x11127u; *b= 0x11132u; return true;
#if 0
/* This one has no decomposition in Unicode, but needs no decomposition either. */
- /* case 0x0AC9 : return false; */
- case 0x0B57 : *a = no decomp, -> RIGHT; return true;
- case 0x1C29 : *a = no decomp, -> LEFT; return true;
- case 0xA9C0 : *a = no decomp, -> RIGHT; return true;
- case 0x111BF : *a = no decomp, -> ABOVE; return true;
+ /* case 0x0AC9u : return false; */
+ case 0x0B57u : *a = no decomp, -> RIGHT; return true;
+ case 0x1C29u : *a = no decomp, -> LEFT; return true;
+ case 0xA9C0u : *a = no decomp, -> RIGHT; return true;
+ case 0x111BuF : *a = no decomp, -> ABOVE; return true;
#endif
}
- if ((ab == 0x0DDA || hb_in_range<hb_codepoint_t> (ab, 0x0DDC, 0x0DDE)))
+ if ((ab == 0x0DDAu || hb_in_range (ab, 0x0DDCu, 0x0DDEu)))
{
/*
* Sinhala split matras... Let the fun begin.
if (hb_options ().uniscribe_bug_compatible ||
(c->font->get_glyph (ab, 0, &glyph) &&
- indic_plan->pstf.would_substitute (&glyph, 1, true, c->font->face)))
+ indic_plan->pstf.would_substitute (&glyph, 1, c->font->face)))
{
/* Ok, safe to use Uniscribe-style decomposition. */
- *a = 0x0DD9;
+ *a = 0x0DD9u;
*b = ab;
return true;
}
return false;
/* Composition-exclusion exceptions that we want to recompose. */
- if (a == 0x09AF && b == 0x09BC) { *ab = 0x09DF; return true; }
+ if (a == 0x09AFu && b == 0x09BCu) { *ab = 0x09DFu; return true; }
return c->unicode->compose (a, b, ab);
}
data_create_indic,
data_destroy_indic,
NULL, /* preprocess_text */
- normalization_preference_indic,
+ HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS_NO_SHORT_CIRCUIT,
decompose_indic,
compose_indic,
setup_masks_indic,