summaryrefslogtreecommitdiff
path: root/src/hb-ot-shape-complex-indic-private.hh
diff options
context:
space:
mode:
Diffstat (limited to 'src/hb-ot-shape-complex-indic-private.hh')
-rw-r--r--src/hb-ot-shape-complex-indic-private.hh314
1 files changed, 113 insertions, 201 deletions
diff --git a/src/hb-ot-shape-complex-indic-private.hh b/src/hb-ot-shape-complex-indic-private.hh
index 64af0da..d8dfc65 100644
--- a/src/hb-ot-shape-complex-indic-private.hh
+++ b/src/hb-ot-shape-complex-indic-private.hh
@@ -31,13 +31,10 @@
#include "hb-ot-shape-complex-private.hh"
+#include "hb-ot-shape-private.hh" /* XXX Remove */
-/* buffer var allocations */
-#define indic_category() complex_var_persistent_u8_0() /* indic_category_t */
-#define indic_position() complex_var_persistent_u8_1() /* indic_matra_category_t */
-
-#define INDIC_TABLE_ELEMENT_TYPE uint8_t
+#define INDIC_TABLE_ELEMENT_TYPE uint16_t
/* Cateories used in the OpenType spec:
* https://www.microsoft.com/typography/otfntdev/devanot/shaping.aspx
@@ -46,225 +43,140 @@
* Not sure how to avoid duplication. */
enum indic_category_t {
OT_X = 0,
- OT_C,
- OT_Ra, /* Not explicitly listed in the OT spec, but used in the grammar. */
- OT_V,
- OT_N,
- OT_H,
- OT_ZWNJ,
- OT_ZWJ,
- OT_M,
- OT_SM,
- OT_VD,
- OT_A,
- OT_NBSP,
- OT_DOTTEDCIRCLE /* Not in the spec, but special in Uniscribe. /Very very/ special! */
+ OT_C = 1,
+ OT_V = 2,
+ OT_N = 3,
+ OT_H = 4,
+ OT_ZWNJ = 5,
+ OT_ZWJ = 6,
+ OT_M = 7,
+ OT_SM = 8,
+ OT_VD = 9,
+ OT_A = 10,
+ OT_PLACEHOLDER = 11,
+ OT_DOTTEDCIRCLE = 12,
+ OT_RS = 13, /* Register Shifter, used in Khmer OT spec. */
+ OT_Coeng = 14, /* Khmer-style Virama. */
+ OT_Repha = 15, /* Atomically-encoded logical or visual repha. */
+ OT_Ra = 16,
+ OT_CM = 17, /* Consonant-Medial. */
+ OT_Symbol = 18, /* Avagraha, etc that take marks (SM,A,VD). */
+ OT_CM2 = 31 /* Consonant-Medial, second slot. */
};
+#define MEDIAL_FLAGS (FLAG (OT_CM) | FLAG (OT_CM2))
+
+/* 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_Ra) | MEDIAL_FLAGS | FLAG (OT_V) | FLAG (OT_PLACEHOLDER) | FLAG (OT_DOTTEDCIRCLE))
+#define JOINER_FLAGS (FLAG (OT_ZWJ) | FLAG (OT_ZWNJ))
+#define HALANT_OR_COENG_FLAGS (FLAG (OT_H) | FLAG (OT_Coeng))
+
+
/* Visual positions in a syllable from left to right. */
enum indic_position_t {
+ POS_START,
+
POS_RA_TO_BECOME_REPH,
POS_PRE_M,
POS_PRE_C,
+
POS_BASE_C,
+ POS_AFTER_MAIN,
+
POS_ABOVE_C,
+
+ POS_BEFORE_SUB,
POS_BELOW_C,
- POS_ABOVE_M,
- POS_BELOW_M,
+ POS_AFTER_SUB,
+
+ POS_BEFORE_POST,
POS_POST_C,
- POS_POST_M,
- POS_SMVD
+ POS_AFTER_POST,
+
+ POS_FINAL_C,
+ POS_SMVD,
+
+ POS_END
};
/* Categories used in IndicSyllabicCategory.txt from UCD. */
enum indic_syllabic_category_t {
- INDIC_SYLLABIC_CATEGORY_OTHER = OT_X,
-
- INDIC_SYLLABIC_CATEGORY_AVAGRAHA = OT_X,
- INDIC_SYLLABIC_CATEGORY_BINDU = OT_SM,
- INDIC_SYLLABIC_CATEGORY_CONSONANT = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_DEAD = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_FINAL = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_HEAD_LETTER = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_MEDIAL = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_PLACEHOLDER = OT_NBSP,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_SUBJOINED = OT_C,
- INDIC_SYLLABIC_CATEGORY_CONSONANT_REPHA = OT_C,
- INDIC_SYLLABIC_CATEGORY_MODIFYING_LETTER = OT_X,
- INDIC_SYLLABIC_CATEGORY_NUKTA = OT_N,
- INDIC_SYLLABIC_CATEGORY_REGISTER_SHIFTER = OT_X,
- INDIC_SYLLABIC_CATEGORY_TONE_LETTER = OT_X,
- INDIC_SYLLABIC_CATEGORY_TONE_MARK = OT_X,
- INDIC_SYLLABIC_CATEGORY_VIRAMA = OT_H,
- INDIC_SYLLABIC_CATEGORY_VISARGA = OT_SM,
- INDIC_SYLLABIC_CATEGORY_VOWEL = OT_V,
- INDIC_SYLLABIC_CATEGORY_VOWEL_DEPENDENT = OT_M,
- INDIC_SYLLABIC_CATEGORY_VOWEL_INDEPENDENT = OT_V
+ INDIC_SYLLABIC_CATEGORY_OTHER = OT_X,
+
+ INDIC_SYLLABIC_CATEGORY_AVAGRAHA = OT_Symbol,
+ INDIC_SYLLABIC_CATEGORY_BINDU = OT_SM,
+ INDIC_SYLLABIC_CATEGORY_BRAHMI_JOINING_NUMBER = OT_PLACEHOLDER, /* TODO */
+ INDIC_SYLLABIC_CATEGORY_CANTILLATION_MARK = OT_A,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT = OT_C,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_DEAD = OT_C,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_FINAL = OT_CM,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_HEAD_LETTER = OT_C,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_MEDIAL = OT_CM,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_PLACEHOLDER = OT_PLACEHOLDER,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_PRECEDING_REPHA = OT_Repha,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_SUBJOINED = OT_CM,
+ INDIC_SYLLABIC_CATEGORY_CONSONANT_SUCCEEDING_REPHA = OT_N,
+ INDIC_SYLLABIC_CATEGORY_GEMINATION_MARK = OT_SM,
+ INDIC_SYLLABIC_CATEGORY_INVISIBLE_STACKER = OT_H, /* TODO */
+ INDIC_SYLLABIC_CATEGORY_JOINER = OT_ZWJ,
+ INDIC_SYLLABIC_CATEGORY_MODIFYING_LETTER = OT_X,
+ INDIC_SYLLABIC_CATEGORY_NON_JOINER = OT_ZWNJ,
+ INDIC_SYLLABIC_CATEGORY_NUKTA = OT_N,
+ INDIC_SYLLABIC_CATEGORY_NUMBER = OT_PLACEHOLDER,
+ INDIC_SYLLABIC_CATEGORY_NUMBER_JOINER = OT_PLACEHOLDER, /* TODO */
+ INDIC_SYLLABIC_CATEGORY_PURE_KILLER = OT_H, /* TODO */
+ INDIC_SYLLABIC_CATEGORY_REGISTER_SHIFTER = OT_RS,
+ INDIC_SYLLABIC_CATEGORY_TONE_LETTER = OT_X,
+ INDIC_SYLLABIC_CATEGORY_TONE_MARK = OT_N,
+ INDIC_SYLLABIC_CATEGORY_VIRAMA = OT_H,
+ INDIC_SYLLABIC_CATEGORY_VISARGA = OT_SM,
+ INDIC_SYLLABIC_CATEGORY_VOWEL = OT_V,
+ INDIC_SYLLABIC_CATEGORY_VOWEL_DEPENDENT = OT_M,
+ INDIC_SYLLABIC_CATEGORY_VOWEL_INDEPENDENT = OT_V
};
/* Categories used in IndicSMatraCategory.txt from UCD */
enum indic_matra_category_t {
- INDIC_MATRA_CATEGORY_NOT_APPLICABLE = POS_BASE_C,
-
- INDIC_MATRA_CATEGORY_LEFT = POS_PRE_M,
- INDIC_MATRA_CATEGORY_TOP = POS_ABOVE_M,
- INDIC_MATRA_CATEGORY_BOTTOM = POS_BELOW_M,
- INDIC_MATRA_CATEGORY_RIGHT = POS_POST_M,
-
- /* We don't really care much about these since we decompose them
- * in the generic pre-shaping layer. They will only be used if
- * the font does not cover the decomposition. In which case, we
- * define these as aliases to the place we want the split-matra
- * glyph to show up. Quite arbitrary.
- *
- * TODO: There are some split matras without Unicode decompositions.
- * We have to figure out what to do with them.
- */
- INDIC_MATRA_CATEGORY_BOTTOM_AND_RIGHT = POS_POST_M,
- INDIC_MATRA_CATEGORY_LEFT_AND_RIGHT = POS_PRE_M,
- INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM = POS_BELOW_M,
- INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM_AND_RIGHT = POS_POST_M,
- INDIC_MATRA_CATEGORY_TOP_AND_LEFT = POS_PRE_M,
- INDIC_MATRA_CATEGORY_TOP_AND_LEFT_AND_RIGHT = POS_PRE_M,
- INDIC_MATRA_CATEGORY_TOP_AND_RIGHT = POS_POST_M,
-
- INDIC_MATRA_CATEGORY_INVISIBLE = INDIC_MATRA_CATEGORY_NOT_APPLICABLE,
- INDIC_MATRA_CATEGORY_OVERSTRUCK = INDIC_MATRA_CATEGORY_NOT_APPLICABLE,
- INDIC_MATRA_CATEGORY_VISUAL_ORDER_LEFT = INDIC_MATRA_CATEGORY_NOT_APPLICABLE
+ INDIC_MATRA_CATEGORY_NOT_APPLICABLE = POS_END,
+
+ INDIC_MATRA_CATEGORY_LEFT = POS_PRE_C,
+ INDIC_MATRA_CATEGORY_TOP = POS_ABOVE_C,
+ INDIC_MATRA_CATEGORY_BOTTOM = POS_BELOW_C,
+ INDIC_MATRA_CATEGORY_RIGHT = POS_POST_C,
+
+ /* These should resolve to the position of the last part of the split sequence. */
+ INDIC_MATRA_CATEGORY_BOTTOM_AND_RIGHT = INDIC_MATRA_CATEGORY_RIGHT,
+ INDIC_MATRA_CATEGORY_LEFT_AND_RIGHT = INDIC_MATRA_CATEGORY_RIGHT,
+ INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM = INDIC_MATRA_CATEGORY_BOTTOM,
+ INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM_AND_RIGHT = INDIC_MATRA_CATEGORY_RIGHT,
+ INDIC_MATRA_CATEGORY_TOP_AND_LEFT = INDIC_MATRA_CATEGORY_TOP,
+ INDIC_MATRA_CATEGORY_TOP_AND_LEFT_AND_RIGHT = INDIC_MATRA_CATEGORY_RIGHT,
+ INDIC_MATRA_CATEGORY_TOP_AND_RIGHT = INDIC_MATRA_CATEGORY_RIGHT,
+
+ INDIC_MATRA_CATEGORY_OVERSTRUCK = POS_AFTER_MAIN,
+ INDIC_MATRA_CATEGORY_VISUAL_ORDER_LEFT = POS_PRE_M
};
/* Note: We use ASSERT_STATIC_EXPR_ZERO() instead of ASSERT_STATIC_EXPR() and the comma operation
* because gcc fails to optimize the latter and fills the table in at runtime. */
#define INDIC_COMBINE_CATEGORIES(S,M) \
- (ASSERT_STATIC_EXPR_ZERO (M == INDIC_MATRA_CATEGORY_NOT_APPLICABLE || (S == INDIC_SYLLABIC_CATEGORY_VIRAMA || S == INDIC_SYLLABIC_CATEGORY_VOWEL_DEPENDENT)) + \
- ASSERT_STATIC_EXPR_ZERO (S < 16 && M < 16) + \
- ((M << 4) | S))
-
-
-#include "hb-ot-shape-complex-indic-table.hh"
-
-/* XXX
- * This is a hack for now. We should:
- * 1. Move this data into the main Indic table,
- * and/or
- * 2. Probe font lookups to determine consonant positions.
- */
-static const struct consonant_position_t {
- hb_codepoint_t u;
- indic_position_t position;
-} consonant_positions[] = {
- {0x0930, POS_BELOW_C},
- {0x09AC, POS_BELOW_C},
- {0x09AF, POS_POST_C},
- {0x09B0, POS_BELOW_C},
- {0x09F0, POS_BELOW_C},
- {0x0A2F, POS_POST_C},
- {0x0A30, POS_BELOW_C},
- {0x0A35, POS_BELOW_C},
- {0x0A39, POS_BELOW_C},
- {0x0AB0, POS_BELOW_C},
- {0x0B24, POS_BELOW_C},
- {0x0B28, POS_BELOW_C},
- {0x0B2C, POS_BELOW_C},
- {0x0B2D, POS_BELOW_C},
- {0x0B2E, POS_BELOW_C},
- {0x0B2F, POS_POST_C},
- {0x0B30, POS_BELOW_C},
- {0x0B32, POS_BELOW_C},
- {0x0B33, POS_BELOW_C},
- {0x0B5F, POS_POST_C},
- {0x0B71, POS_BELOW_C},
- {0x0C15, POS_BELOW_C},
- {0x0C16, POS_BELOW_C},
- {0x0C17, POS_BELOW_C},
- {0x0C18, POS_BELOW_C},
- {0x0C19, POS_BELOW_C},
- {0x0C1A, POS_BELOW_C},
- {0x0C1B, POS_BELOW_C},
- {0x0C1C, POS_BELOW_C},
- {0x0C1D, POS_BELOW_C},
- {0x0C1E, POS_BELOW_C},
- {0x0C1F, POS_BELOW_C},
- {0x0C20, POS_BELOW_C},
- {0x0C21, POS_BELOW_C},
- {0x0C22, POS_BELOW_C},
- {0x0C23, POS_BELOW_C},
- {0x0C24, POS_BELOW_C},
- {0x0C25, POS_BELOW_C},
- {0x0C26, POS_BELOW_C},
- {0x0C27, POS_BELOW_C},
- {0x0C28, POS_BELOW_C},
- {0x0C2A, POS_BELOW_C},
- {0x0C2B, POS_BELOW_C},
- {0x0C2C, POS_BELOW_C},
- {0x0C2D, POS_BELOW_C},
- {0x0C2E, POS_BELOW_C},
- {0x0C2F, POS_BELOW_C},
- {0x0C30, POS_BELOW_C},
- {0x0C32, POS_BELOW_C},
- {0x0C33, POS_BELOW_C},
- {0x0C35, POS_BELOW_C},
- {0x0C36, POS_BELOW_C},
- {0x0C37, POS_BELOW_C},
- {0x0C38, POS_BELOW_C},
- {0x0C39, POS_BELOW_C},
- {0x0C95, POS_BELOW_C},
- {0x0C96, POS_BELOW_C},
- {0x0C97, POS_BELOW_C},
- {0x0C98, POS_BELOW_C},
- {0x0C99, POS_BELOW_C},
- {0x0C9A, POS_BELOW_C},
- {0x0C9B, POS_BELOW_C},
- {0x0C9C, POS_BELOW_C},
- {0x0C9D, POS_BELOW_C},
- {0x0C9E, POS_BELOW_C},
- {0x0C9F, POS_BELOW_C},
- {0x0CA0, POS_BELOW_C},
- {0x0CA1, POS_BELOW_C},
- {0x0CA2, POS_BELOW_C},
- {0x0CA3, POS_BELOW_C},
- {0x0CA4, POS_BELOW_C},
- {0x0CA5, POS_BELOW_C},
- {0x0CA6, POS_BELOW_C},
- {0x0CA7, POS_BELOW_C},
- {0x0CA8, POS_BELOW_C},
- {0x0CAA, POS_BELOW_C},
- {0x0CAB, POS_BELOW_C},
- {0x0CAC, POS_BELOW_C},
- {0x0CAD, POS_BELOW_C},
- {0x0CAE, POS_BELOW_C},
- {0x0CAF, POS_BELOW_C},
- {0x0CB0, POS_BELOW_C},
- {0x0CB2, POS_BELOW_C},
- {0x0CB3, POS_BELOW_C},
- {0x0CB5, POS_BELOW_C},
- {0x0CB6, POS_BELOW_C},
- {0x0CB7, POS_BELOW_C},
- {0x0CB8, POS_BELOW_C},
- {0x0CB9, POS_BELOW_C},
- {0x0CDE, POS_BELOW_C},
- {0x0D2F, POS_POST_C},
- {0x0D30, POS_POST_C},
- {0x0D32, POS_BELOW_C},
- {0x0D35, POS_POST_C},
-};
-
-/* XXX
- * This is a hack for now. We should move this data into the main Indic table.
- * 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 */ /* No Reph */
- 0x0CB0, /* Kannada */
- 0x0D30, /* Malayalam */ /* No Reph */
-};
-
+ (ASSERT_STATIC_EXPR_ZERO (M == INDIC_MATRA_CATEGORY_NOT_APPLICABLE || \
+ ( \
+ S == INDIC_SYLLABIC_CATEGORY_CONSONANT_MEDIAL || \
+ S == INDIC_SYLLABIC_CATEGORY_GEMINATION_MARK || \
+ S == INDIC_SYLLABIC_CATEGORY_REGISTER_SHIFTER || \
+ S == INDIC_SYLLABIC_CATEGORY_CONSONANT_SUCCEEDING_REPHA || \
+ S == INDIC_SYLLABIC_CATEGORY_VIRAMA || \
+ S == INDIC_SYLLABIC_CATEGORY_VOWEL_DEPENDENT || \
+ false)) + \
+ ASSERT_STATIC_EXPR_ZERO (S < 255 && M < 255) + \
+ ((M << 8) | S))
+
+HB_INTERNAL INDIC_TABLE_ELEMENT_TYPE
+hb_indic_get_categories (hb_codepoint_t u);
#endif /* HB_OT_SHAPE_COMPLEX_INDIC_PRIVATE_HH */