Chiebot-Mirror
/
harfbuzz
mirror of https://github.com/harfbuzz/harfbuzz.git
8
0
Fork 0
Code Issues Projects Releases Wiki Activity

[indic-generator] Move Ra handling to the generator

Browse Source
pull/3648/head
Behdad Esfahbod 2 years ago
parent 419d2146c2
commit f0269e0f1b
3 changed files with 87 additions and 45 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 14
      src/gen-indic-table.py
  2. 88
      src/hb-ot-shaper-indic-table.cc
  3. 30
      src/hb-ot-shaper-indic.hh

14
src/gen-indic-table.py
Unescape View File

@ -120,6 +120,7 @@ category_map = {
'Vowel_Dependent' : 'M',
'Vowel_Independent' : 'V',
'Dotted_Circle' : 'DOTTEDCIRCLE', # Ours, not Unicode's
'Ra' : 'Ra', # Ours, not Unicode's
}
position_map = {
'Not_Applicable' : 'END',
@ -145,6 +146,18 @@ position_map = {
category_overrides = {
0x0930: 'Ra', # Devanagari
0x09B0: 'Ra', # Bengali
0x09F0: 'Ra', # Bengali
0x0A30: 'Ra', # Gurmukhi No Reph
0x0AB0: 'Ra', # Gujarati
0x0B30: 'Ra', # Oriya
0x0BB0: 'Ra', # Tamil No Reph
0x0C30: 'Ra', # Telugu Reph formed only with ZWJ
0x0CB0: 'Ra', # Kannada
0x0D30: 'Ra', # Malayalam No Reph, Logical Repha
0x0DBB: 'Ra', # Sinhala Reph formed only with ZWJ
# The following act more like the Bindus.
0x0953: 'SM',
0x0954: 'SM',
@ -265,6 +278,7 @@ print ()
# Shorten values
short = [{
"Repha": 'Rf',
"Coeng": 'Co',
"PLACEHOLDER": 'GB',
"DOTTEDCIRCLE": 'DC',

88
src/hb-ot-shaper-indic-table.cc
Unescape View File

@ -24,7 +24,7 @@
#pragma GCC diagnostic ignored "-Wunused-macros"
#define ISC_A OT_A /* 51 chars; A */
#define ISC_C OT_C /* 532 chars; C */
#define ISC_C OT_C /* 521 chars; C */
#define ISC_CM OT_CM /* 10 chars; CM */
#define ISC_CS OT_CS /* 2 chars; CS */
#define ISC_Co OT_Coeng /* 2 chars; Coeng */
@ -34,7 +34,8 @@
#define ISC_N OT_N /* 35 chars; N */
#define ISC_GB OT_PLACEHOLDER /* 168 chars; PLACEHOLDER */
#define ISC_RS OT_RS /* 2 chars; RS */
#define ISC_R OT_Repha /* 1 chars; Repha */
#define ISC_R OT_Ra /* 11 chars; Ra */
#define ISC_Rf OT_Repha /* 1 chars; Repha */
#define ISC_SM OT_SM /* 56 chars; SM */
#define ISC_S OT_Symbol /* 22 chars; Symbol */
#define ISC_V OT_V /* 190 chars; V */
@ -43,9 +44,9 @@
#define ISC_ZWNJ OT_ZWNJ /* 1 chars; ZWNJ */
#define IMC_T POS_ABOVE_C /* 125 chars; ABOVE_C */
#define IMC_A POS_AFTER_MAIN /* 44 chars; AFTER_MAIN */
#define IMC_A POS_AFTER_MAIN /* 55 chars; AFTER_MAIN */
#define IMC_B POS_BELOW_C /* 79 chars; BELOW_C */
#define IMC_X POS_END /* 904 chars; END */
#define IMC_X POS_END /* 893 chars; END */
#define IMC_R POS_POST_C /* 122 chars; POST_C */
#define IMC_L POS_PRE_C /* 21 chars; PRE_C */
@ -88,7 +89,13 @@ static const uint16_t indic_table[] = {
/* 0918 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0920 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0928 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0930 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0930 */ _(R,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Devanagari */
/* 0938 */ _(C,X), _(C,X), _(M,T), _(M,R), _(N,X), _(S,X), _(M,R), _(M,L),
/* 0940 */ _(M,R), _(M,B), _(M,B), _(M,B), _(M,B), _(M,T), _(M,T), _(M,T),
/* 0948 */ _(M,T), _(M,R), _(M,R), _(M,R), _(M,R), _(H,B), _(M,L), _(M,R),
@ -110,7 +117,13 @@ static const uint16_t indic_table[] = {
/* 0998 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 09A0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 09A8 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 09B0 */ _(C,X), _(X,X), _(C,X), _(X,X), _(X,X), _(X,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 09B0 */ _(R,X), _(X,X), _(C,X), _(X,X), _(X,X), _(X,X), _(C,X), _(C,X),
/* Bengali */
/* 09B8 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(S,X), _(M,R), _(M,L),
/* 09C0 */ _(M,R), _(M,B), _(M,B), _(M,B), _(M,B), _(X,X), _(X,X), _(M,L),
/* 09C8 */ _(M,L), _(X,X), _(X,X), _(M,R), _(M,R), _(H,B), _(C,X), _(X,X),
@ -118,10 +131,10 @@ static const uint16_t indic_table[] = {
/* 09D8 */ _(X,X), _(X,X), _(X,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X),
/* 09E0 */ _(V,X), _(V,X), _(M,B), _(M,B), _(X,X), _(X,X), _(GB,X), _(GB,X),
/* 09E8 */ _(GB,X), _(GB,X), _(GB,X), _(GB,X), _(GB,X), _(GB,X), _(GB,X), _(GB,X),
/* 09F0 */ _(C,X), _(C,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X),
/* Vedic Extensions */
/* 09F0 */ _(R,X), _(C,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X),
/* 09F8 */ _(X,X), _(X,X), _(X,X), _(X,X), _(GB,X), _(X,X), _(SM,T), _(X,X),
/* Gurmukhi */
@ -132,7 +145,13 @@ static const uint16_t indic_table[] = {
/* 0A18 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0A20 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0A28 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0A30 */ _(C,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(X,X),
/* Vedic Extensions */
/* 0A30 */ _(R,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(X,X),
/* Gurmukhi */
/* 0A38 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(X,X), _(M,R), _(M,L),
/* 0A40 */ _(M,R), _(M,B), _(M,B), _(X,X), _(X,X), _(X,X), _(X,X), _(M,T),
/* 0A48 */ _(M,T), _(X,X), _(X,X), _(M,T), _(M,T), _(H,B), _(X,X), _(X,X),
@ -157,7 +176,13 @@ static const uint16_t indic_table[] = {
/* 0A98 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0AA0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0AA8 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0AB0 */ _(C,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0AB0 */ _(R,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Gujarati */
/* 0AB8 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(S,X), _(M,R), _(M,L),
/* 0AC0 */ _(M,R), _(M,B), _(M,B), _(M,B), _(M,B), _(M,T), _(X,X), _(M,T),
/* 0AC8 */ _(M,T), _(M,R), _(X,X), _(M,R), _(M,R), _(H,B), _(X,X), _(X,X),
@ -176,7 +201,13 @@ static const uint16_t indic_table[] = {
/* 0B18 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0B20 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0B28 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0B30 */ _(C,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0B30 */ _(R,X), _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Oriya */
/* 0B38 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(S,X), _(M,R), _(M,T),
/* 0B40 */ _(M,R), _(M,B), _(M,B), _(M,B), _(M,B), _(X,X), _(X,X), _(M,L),
/* 0B48 */ _(M,T), _(X,X), _(X,X), _(M,R), _(M,R), _(H,B), _(X,X), _(X,X),
@ -201,7 +232,13 @@ static const uint16_t indic_table[] = {
/* 0B98 */ _(X,X), _(C,X), _(C,X), _(X,X), _(C,X), _(X,X), _(C,X), _(C,X),
/* 0BA0 */ _(X,X), _(X,X), _(X,X), _(C,X), _(C,X), _(X,X), _(X,X), _(X,X),
/* 0BA8 */ _(C,X), _(C,X), _(C,X), _(X,X), _(X,X), _(X,X), _(C,X), _(C,X),
/* 0BB0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0BB0 */ _(R,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Tamil */
/* 0BB8 */ _(C,X), _(C,X), _(X,X), _(X,X), _(X,X), _(X,X), _(M,R), _(M,R),
/* 0BC0 */ _(M,T), _(M,R), _(M,R), _(X,X), _(X,X), _(X,X), _(M,L), _(M,L),
/* 0BC8 */ _(M,L), _(X,X), _(M,R), _(M,R), _(M,R), _(H,T), _(X,X), _(X,X),
@ -220,7 +257,13 @@ static const uint16_t indic_table[] = {
/* 0C18 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0C20 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0C28 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0C30 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0C30 */ _(R,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Telugu */
/* 0C38 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(S,X), _(M,T), _(M,T),
/* 0C40 */ _(M,T), _(M,R), _(M,R), _(M,R), _(M,R), _(X,X), _(M,T), _(M,T),
/* 0C48 */ _(M,B), _(X,X), _(M,T), _(M,T), _(M,T), _(H,T), _(X,X), _(X,X),
@ -242,7 +285,13 @@ static const uint16_t indic_table[] = {
/* 0C98 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0CA0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0CA8 */ _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0CB0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0CB0 */ _(R,X), _(C,X), _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X),
/* Kannada */
/* 0CB8 */ _(C,X), _(C,X), _(X,X), _(X,X), _(N,X), _(S,X), _(M,R), _(M,T),
/* 0CC0 */ _(M,R), _(M,R), _(M,R), _(M,R), _(M,R), _(X,X), _(M,T), _(M,R),
/* 0CC8 */ _(M,R), _(X,X), _(M,R), _(M,R), _(M,T), _(H,T), _(X,X), _(X,X),
@ -261,10 +310,16 @@ static const uint16_t indic_table[] = {
/* 0D18 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0D20 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0D28 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0D30 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Vedic Extensions */
/* 0D30 */ _(R,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* Malayalam */
/* 0D38 */ _(C,X), _(C,X), _(C,X), _(M,T), _(M,T), _(S,X), _(M,R), _(M,R),
/* 0D40 */ _(M,R), _(M,R), _(M,R), _(M,B), _(M,B), _(X,X), _(M,L), _(M,L),
/* 0D48 */ _(M,L), _(X,X), _(M,R), _(M,R), _(M,R), _(H,T), _(R,X), _(X,X),
/* 0D48 */ _(M,L), _(X,X), _(M,R), _(M,R), _(M,R), _(H,T), _(Rf,X), _(X,X),
/* 0D50 */ _(X,X), _(X,X), _(X,X), _(X,X), _(C,X), _(C,X), _(C,X), _(M,R),
/* 0D58 */ _(X,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X), _(X,X), _(V,X),
/* 0D60 */ _(V,X), _(V,X), _(M,B), _(M,B), _(X,X), _(X,X), _(GB,X), _(GB,X),
@ -281,7 +336,7 @@ static const uint16_t indic_table[] = {
/* 0DA0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0DA8 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0DB0 */ _(C,X), _(C,X), _(X,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X),
/* 0DB8 */ _(C,X), _(C,X), _(C,X), _(C,X), _(X,X), _(C,X), _(X,X), _(X,X),
/* 0DB8 */ _(C,X), _(C,X), _(C,X), _(R,X), _(X,X), _(C,X), _(X,X), _(X,X),
/* 0DC0 */ _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(C,X), _(X,X),
/* 0DC8 */ _(X,X), _(X,X), _(H,T), _(X,X), _(X,X), _(X,X), _(X,X), _(M,R),
/* 0DD0 */ _(M,R), _(M,R), _(M,T), _(M,T), _(M,B), _(X,X), _(M,B), _(X,X),
@ -468,6 +523,7 @@ hb_indic_get_categories (hb_codepoint_t u)
#undef ISC_GB
#undef ISC_RS
#undef ISC_R
#undef ISC_Rf
#undef ISC_SM
#undef ISC_S
#undef ISC_V

30
src/hb-ot-shaper-indic.hh
Unescape View File

@ -209,31 +209,6 @@ matra_position_indic (hb_codepoint_t u, indic_position_t side)
return side;
}
/* 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[] = {
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 */
};
static inline bool
is_ra (hb_codepoint_t u)
{
return hb_array (ra_chars).lfind (u);
}
static inline void
set_indic_properties (hb_glyph_info_t &info)
{
@ -253,11 +228,8 @@ set_indic_properties (hb_glyph_info_t &info)
}
if ((FLAG_UNSAFE (cat) & CONSONANT_FLAGS))
{
pos = POS_BASE_C;
if (is_ra (u))
cat = OT_Ra;
}
else if (cat == OT_M)
{
pos = matra_position_indic (u, pos);

Write Preview
Loading…
Cancel
Save