blob: 150cbc187a68cfc3598fec2dbc5136fbe02c85e7 [file] [log] [blame]
/*
* Copyright (C) 2008 Nokia Corporation and/or its subsidiary(-ies)
*
* This is part of HarfBuzz, an OpenType Layout engine 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.
*/
#include "harfbuzz-shaper.h"
#include "harfbuzz-shaper-private.h"
#include <assert.h>
#include <stdio.h>
/*
// Vocabulary
// Base -> A consonant or an independent vowel in its full (not subscript) form. It is the
// center of the syllable, it can be surrounded by coeng (subscript) consonants, vowels,
// split vowels, signs... but there is only one base in a syllable, it has to be coded as
// the first character of the syllable.
// split vowel --> vowel that has two parts placed separately (e.g. Before and after the consonant).
// Khmer language has five of them. Khmer split vowels either have one part before the
// base and one after the base or they have a part before the base and a part above the base.
// The first part of all Khmer split vowels is the same character, identical to
// the glyph of Khmer dependent vowel SRA EI
// coeng --> modifier used in Khmer to construct coeng (subscript) consonants
// Differently than indian languages, the coeng modifies the consonant that follows it,
// not the one preceding it Each consonant has two forms, the base form and the subscript form
// the base form is the normal one (using the consonants code-point), the subscript form is
// displayed when the combination coeng + consonant is encountered.
// Consonant of type 1 -> A consonant which has subscript for that only occupies space under a base consonant
// Consonant of type 2.-> Its subscript form occupies space under and before the base (only one, RO)
// Consonant of Type 3 -> Its subscript form occupies space under and after the base (KHO, CHHO, THHO, BA, YO, SA)
// Consonant shifter -> Khmer has to series of consonants. The same dependent vowel has different sounds
// if it is attached to a consonant of the first series or a consonant of the second series
// Most consonants have an equivalent in the other series, but some of theme exist only in
// one series (for example SA). If we want to use the consonant SA with a vowel sound that
// can only be done with a vowel sound that corresponds to a vowel accompanying a consonant
// of the other series, then we need to use a consonant shifter: TRIISAP or MUSIKATOAN
// x17C9 y x17CA. TRIISAP changes a first series consonant to second series sound and
// MUSIKATOAN a second series consonant to have a first series vowel sound.
// Consonant shifter are both normally supercript marks, but, when they are followed by a
// superscript, they change shape and take the form of subscript dependent vowel SRA U.
// If they are in the same syllable as a coeng consonant, Unicode 3.0 says that they
// should be typed before the coeng. Unicode 4.0 breaks the standard and says that it should
// be placed after the coeng consonant.
// Dependent vowel -> In khmer dependent vowels can be placed above, below, before or after the base
// Each vowel has its own position. Only one vowel per syllable is allowed.
// Signs -> Khmer has above signs and post signs. Only one above sign and/or one post sign are
// Allowed in a syllable.
//
//
// order is important here! This order must be the same that is found in each horizontal
// line in the statetable for Khmer (see khmerStateTable) .
*/
enum KhmerCharClassValues {
CC_RESERVED = 0,
CC_CONSONANT = 1, /* Consonant of type 1 or independent vowel */
CC_CONSONANT2 = 2, /* Consonant of type 2 */
CC_CONSONANT3 = 3, /* Consonant of type 3 */
CC_ZERO_WIDTH_NJ_MARK = 4, /* Zero Width non joiner character (0x200C) */
CC_CONSONANT_SHIFTER = 5,
CC_ROBAT = 6, /* Khmer special diacritic accent -treated differently in state table */
CC_COENG = 7, /* Subscript consonant combining character */
CC_DEPENDENT_VOWEL = 8,
CC_SIGN_ABOVE = 9,
CC_SIGN_AFTER = 10,
CC_ZERO_WIDTH_J_MARK = 11, /* Zero width joiner character */
CC_COUNT = 12 /* This is the number of character classes */
};
enum KhmerCharClassFlags {
CF_CLASS_MASK = 0x0000FFFF,
CF_CONSONANT = 0x01000000, /* flag to speed up comparing */
CF_SPLIT_VOWEL = 0x02000000, /* flag for a split vowel -> the first part is added in front of the syllable */
CF_DOTTED_CIRCLE = 0x04000000, /* add a dotted circle if a character with this flag is the first in a syllable */
CF_COENG = 0x08000000, /* flag to speed up comparing */
CF_SHIFTER = 0x10000000, /* flag to speed up comparing */
CF_ABOVE_VOWEL = 0x20000000, /* flag to speed up comparing */
/* position flags */
CF_POS_BEFORE = 0x00080000,
CF_POS_BELOW = 0x00040000,
CF_POS_ABOVE = 0x00020000,
CF_POS_AFTER = 0x00010000,
CF_POS_MASK = 0x000f0000
};
/* Characters that get referred to by name */
enum KhmerChar {
C_SIGN_ZWNJ = 0x200C,
C_SIGN_ZWJ = 0x200D,
C_RO = 0x179A,
C_VOWEL_AA = 0x17B6,
C_SIGN_NIKAHIT = 0x17C6,
C_VOWEL_E = 0x17C1,
C_COENG = 0x17D2
};
/*
// simple classes, they are used in the statetable (in this file) to control the length of a syllable
// they are also used to know where a character should be placed (location in reference to the base character)
// and also to know if a character, when independently displayed, should be displayed with a dotted-circle to
// indicate error in syllable construction
*/
enum {
_xx = CC_RESERVED,
_sa = CC_SIGN_ABOVE | CF_DOTTED_CIRCLE | CF_POS_ABOVE,
_sp = CC_SIGN_AFTER | CF_DOTTED_CIRCLE| CF_POS_AFTER,
_c1 = CC_CONSONANT | CF_CONSONANT,
_c2 = CC_CONSONANT2 | CF_CONSONANT,
_c3 = CC_CONSONANT3 | CF_CONSONANT,
_rb = CC_ROBAT | CF_POS_ABOVE | CF_DOTTED_CIRCLE,
_cs = CC_CONSONANT_SHIFTER | CF_DOTTED_CIRCLE | CF_SHIFTER,
_dl = CC_DEPENDENT_VOWEL | CF_POS_BEFORE | CF_DOTTED_CIRCLE,
_db = CC_DEPENDENT_VOWEL | CF_POS_BELOW | CF_DOTTED_CIRCLE,
_da = CC_DEPENDENT_VOWEL | CF_POS_ABOVE | CF_DOTTED_CIRCLE | CF_ABOVE_VOWEL,
_dr = CC_DEPENDENT_VOWEL | CF_POS_AFTER | CF_DOTTED_CIRCLE,
_co = CC_COENG | CF_COENG | CF_DOTTED_CIRCLE,
/* split vowel */
_va = _da | CF_SPLIT_VOWEL,
_vr = _dr | CF_SPLIT_VOWEL
};
/*
// Character class: a character class value
// ORed with character class flags.
*/
typedef unsigned long KhmerCharClass;
/*
// Character class tables
// _xx character does not combine into syllable, such as numbers, puntuation marks, non-Khmer signs...
// _sa Sign placed above the base
// _sp Sign placed after the base
// _c1 Consonant of type 1 or independent vowel (independent vowels behave as type 1 consonants)
// _c2 Consonant of type 2 (only RO)
// _c3 Consonant of type 3
// _rb Khmer sign robat u17CC. combining mark for subscript consonants
// _cd Consonant-shifter
// _dl Dependent vowel placed before the base (left of the base)
// _db Dependent vowel placed below the base
// _da Dependent vowel placed above the base
// _dr Dependent vowel placed behind the base (right of the base)
// _co Khmer combining mark COENG u17D2, combines with the consonant or independent vowel following
// it to create a subscript consonant or independent vowel
// _va Khmer split vowel in which the first part is before the base and the second one above the base
// _vr Khmer split vowel in which the first part is before the base and the second one behind (right of) the base
*/
static const KhmerCharClass khmerCharClasses[] = {
_c1, _c1, _c1, _c3, _c1, _c1, _c1, _c1, _c3, _c1, _c1, _c1, _c1, _c3, _c1, _c1, /* 1780 - 178F */
_c1, _c1, _c1, _c1, _c3, _c1, _c1, _c1, _c1, _c3, _c2, _c1, _c1, _c1, _c3, _c3, /* 1790 - 179F */
_c1, _c3, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, _c1, /* 17A0 - 17AF */
_c1, _c1, _c1, _c1, _dr, _dr, _dr, _da, _da, _da, _da, _db, _db, _db, _va, _vr, /* 17B0 - 17BF */
_vr, _dl, _dl, _dl, _vr, _vr, _sa, _sp, _sp, _cs, _cs, _sa, _rb, _sa, _sa, _sa, /* 17C0 - 17CF */
_sa, _sa, _co, _sa, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _sa, _xx, _xx /* 17D0 - 17DF */
};
/* this enum must reflect the range of khmerCharClasses */
enum KhmerCharClassesRange {
KhmerFirstChar = 0x1780,
KhmerLastChar = 0x17df
};
/*
// Below we define how a character in the input string is either in the khmerCharClasses table
// (in which case we get its type back), a ZWJ or ZWNJ (two characters that may appear
// within the syllable, but are not in the table) we also get their type back, or an unknown object
// in which case we get _xx (CC_RESERVED) back
*/
static KhmerCharClass getKhmerCharClass(HB_UChar16 uc)
{
if (uc == C_SIGN_ZWJ) {
return CC_ZERO_WIDTH_J_MARK;
}
if (uc == C_SIGN_ZWNJ) {
return CC_ZERO_WIDTH_NJ_MARK;
}
if (uc < KhmerFirstChar || uc > KhmerLastChar) {
return CC_RESERVED;
}
return khmerCharClasses[uc - KhmerFirstChar];
}
/*
// The stateTable is used to calculate the end (the length) of a well
// formed Khmer Syllable.
//
// Each horizontal line is ordered exactly the same way as the values in KhmerClassTable
// CharClassValues. This coincidence of values allows the follow up of the table.
//
// Each line corresponds to a state, which does not necessarily need to be a type
// of component... for example, state 2 is a base, with is always a first character
// in the syllable, but the state could be produced a consonant of any type when
// it is the first character that is analysed (in ground state).
//
// Differentiating 3 types of consonants is necessary in order to
// forbid the use of certain combinations, such as having a second
// coeng after a coeng RO,
// The inexistent possibility of having a type 3 after another type 3 is permitted,
// eliminating it would very much complicate the table, and it does not create typing
// problems, as the case above.
//
// The table is quite complex, in order to limit the number of coeng consonants
// to 2 (by means of the table).
//
// There a peculiarity, as far as Unicode is concerned:
// - The consonant-shifter is considered in two possible different
// locations, the one considered in Unicode 3.0 and the one considered in
// Unicode 4.0. (there is a backwards compatibility problem in this standard).
//
//
// xx independent character, such as a number, punctuation sign or non-khmer char
//
// c1 Khmer consonant of type 1 or an independent vowel
// that is, a letter in which the subscript for is only under the
// base, not taking any space to the right or to the left
//
// c2 Khmer consonant of type 2, the coeng form takes space under
// and to the left of the base (only RO is of this type)
//
// c3 Khmer consonant of type 3. Its subscript form takes space under
// and to the right of the base.
//
// cs Khmer consonant shifter
//
// rb Khmer robat
//
// co coeng character (u17D2)
//
// dv dependent vowel (including split vowels, they are treated in the same way).
// even if dv is not defined above, the component that is really tested for is
// KhmerClassTable::CC_DEPENDENT_VOWEL, which is common to all dependent vowels
//
// zwj Zero Width joiner
//
// zwnj Zero width non joiner
//
// sa above sign
//
// sp post sign
//
// there are lines with equal content but for an easier understanding
// (and maybe change in the future) we did not join them
*/
static const signed char khmerStateTable[][CC_COUNT] =
{
/* xx c1 c2 c3 zwnj cs rb co dv sa sp zwj */
{ 1, 2, 2, 2, 1, 1, 1, 6, 1, 1, 1, 2}, /* 0 - ground state */
{-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, /* 1 - exit state (or sign to the right of the syllable) */
{-1, -1, -1, -1, 3, 4, 5, 6, 16, 17, 1, -1}, /* 2 - Base consonant */
{-1, -1, -1, -1, -1, 4, -1, -1, 16, -1, -1, -1}, /* 3 - First ZWNJ before a register shifter It can only be followed by a shifter or a vowel */
{-1, -1, -1, -1, 15, -1, -1, 6, 16, 17, 1, 14}, /* 4 - First register shifter */
{-1, -1, -1, -1, -1, -1, -1, -1, 20, -1, 1, -1}, /* 5 - Robat */
{-1, 7, 8, 9, -1, -1, -1, -1, -1, -1, -1, -1}, /* 6 - First Coeng */
{-1, -1, -1, -1, 12, 13, -1, 10, 16, 17, 1, 14}, /* 7 - First consonant of type 1 after coeng */
{-1, -1, -1, -1, 12, 13, -1, -1, 16, 17, 1, 14}, /* 8 - First consonant of type 2 after coeng */
{-1, -1, -1, -1, 12, 13, -1, 10, 16, 17, 1, 14}, /* 9 - First consonant or type 3 after ceong */
{-1, 11, 11, 11, -1, -1, -1, -1, -1, -1, -1, -1}, /* 10 - Second Coeng (no register shifter before) */
{-1, -1, -1, -1, 15, -1, -1, -1, 16, 17, 1, 14}, /* 11 - Second coeng consonant (or ind. vowel) no register shifter before */
{-1, -1, -1, -1, -1, 13, -1, -1, 16, -1, -1, -1}, /* 12 - Second ZWNJ before a register shifter */
{-1, -1, -1, -1, 15, -1, -1, -1, 16, 17, 1, 14}, /* 13 - Second register shifter */
{-1, -1, -1, -1, -1, -1, -1, -1, 16, -1, -1, -1}, /* 14 - ZWJ before vowel */
{-1, -1, -1, -1, -1, -1, -1, -1, 16, -1, -1, -1}, /* 15 - ZWNJ before vowel */
{-1, -1, -1, -1, -1, -1, -1, -1, -1, 17, 1, 18}, /* 16 - dependent vowel */
{-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 1, 18}, /* 17 - sign above */
{-1, -1, -1, -1, -1, -1, -1, 19, -1, -1, -1, -1}, /* 18 - ZWJ after vowel */
{-1, 1, -1, 1, -1, -1, -1, -1, -1, -1, -1, -1}, /* 19 - Third coeng */
{-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 1, -1}, /* 20 - dependent vowel after a Robat */
};
/* #define KHMER_DEBUG */
#ifdef KHMER_DEBUG
#define KHDEBUG qDebug
#else
#define KHDEBUG if(0) printf
#endif
/*
// Given an input string of characters and a location in which to start looking
// calculate, using the state table, which one is the last character of the syllable
// that starts in the starting position.
*/
static int khmer_nextSyllableBoundary(const HB_UChar16 *s, int start, int end, HB_Bool *invalid)
{
const HB_UChar16 *uc = s + start;
int state = 0;
int pos = start;
*invalid = FALSE;
while (pos < end) {
KhmerCharClass charClass = getKhmerCharClass(*uc);
if (pos == start) {
*invalid = (charClass > 0) && ! (charClass & CF_CONSONANT);
}
state = khmerStateTable[state][charClass & CF_CLASS_MASK];
KHDEBUG("state[%d]=%d class=%8lx (uc=%4x)", pos - start, state,
charClass, *uc );
if (state < 0) {
break;
}
++uc;
++pos;
}
return pos;
}
#ifndef NO_OPENTYPE
static const HB_OpenTypeFeature khmer_features[] = {
{ HB_MAKE_TAG( 'p', 'r', 'e', 'f' ), PreFormProperty },
{ HB_MAKE_TAG( 'b', 'l', 'w', 'f' ), BelowFormProperty },
{ HB_MAKE_TAG( 'a', 'b', 'v', 'f' ), AboveFormProperty },
{ HB_MAKE_TAG( 'p', 's', 't', 'f' ), PostFormProperty },
{ HB_MAKE_TAG( 'p', 'r', 'e', 's' ), PreSubstProperty },
{ HB_MAKE_TAG( 'b', 'l', 'w', 's' ), BelowSubstProperty },
{ HB_MAKE_TAG( 'a', 'b', 'v', 's' ), AboveSubstProperty },
{ HB_MAKE_TAG( 'p', 's', 't', 's' ), PostSubstProperty },
{ HB_MAKE_TAG( 'c', 'l', 'i', 'g' ), CligProperty },
{ 0, 0 }
};
#endif
static HB_Bool khmer_shape_syllable(HB_Bool openType, HB_ShaperItem *item)
{
/* KHDEBUG("syllable from %d len %d, str='%s'", item->from, item->length,
item->string->mid(item->from, item->length).toUtf8().data()); */
int len = 0;
int syllableEnd = item->item.pos + item->item.length;
unsigned short reordered[16];
unsigned char properties[16];
enum {
AboveForm = 0x01,
PreForm = 0x02,
PostForm = 0x04,
BelowForm = 0x08
};
#ifndef NO_OPENTYPE
const int availableGlyphs = item->num_glyphs;
#endif
int coengRo;
int i;
/* according to the specs this is the max length one can get
### the real value should be smaller */
assert(item->item.length < 13);
memset(properties, 0, 16*sizeof(unsigned char));
#ifdef KHMER_DEBUG
qDebug("original:");
for (int i = from; i < syllableEnd; i++) {
qDebug(" %d: %4x", i, string[i]);
}
#endif
/*
// write a pre vowel or the pre part of a split vowel first
// and look out for coeng + ro. RO is the only vowel of type 2, and
// therefore the only one that requires saving space before the base.
*/
coengRo = -1; /* There is no Coeng Ro, if found this value will change */
for (i = item->item.pos; i < syllableEnd; i += 1) {
KhmerCharClass charClass = getKhmerCharClass(item->string[i]);
/* if a split vowel, write the pre part. In Khmer the pre part
is the same for all split vowels, same glyph as pre vowel C_VOWEL_E */
if (charClass & CF_SPLIT_VOWEL) {
reordered[len] = C_VOWEL_E;
properties[len] = PreForm;
++len;
break; /* there can be only one vowel */
}
/* if a vowel with pos before write it out */
if (charClass & CF_POS_BEFORE) {
reordered[len] = item->string[i];
properties[len] = PreForm;
++len;
break; /* there can be only one vowel */
}
/* look for coeng + ro and remember position
works because coeng + ro is always in front of a vowel (if there is a vowel)
and because CC_CONSONANT2 is enough to identify it, as it is the only consonant
with this flag */
if ( (charClass & CF_COENG) && (i + 1 < syllableEnd) &&
( (getKhmerCharClass(item->string[i+1]) & CF_CLASS_MASK) == CC_CONSONANT2) ) {
coengRo = i;
}
}
/* write coeng + ro if found */
if (coengRo > -1) {
reordered[len] = C_COENG;
properties[len] = PreForm;
++len;
reordered[len] = C_RO;
properties[len] = PreForm;
++len;
}
/*
shall we add a dotted circle?
If in the position in which the base should be (first char in the string) there is
a character that has the Dotted circle flag (a character that cannot be a base)
then write a dotted circle */
if (getKhmerCharClass(item->string[item->item.pos]) & CF_DOTTED_CIRCLE) {
reordered[len] = C_DOTTED_CIRCLE;
++len;
}
/* copy what is left to the output, skipping before vowels and
coeng Ro if they are present */
for (i = item->item.pos; i < syllableEnd; i += 1) {
HB_UChar16 uc = item->string[i];
KhmerCharClass charClass = getKhmerCharClass(uc);
/* skip a before vowel, it was already processed */
if (charClass & CF_POS_BEFORE) {
continue;
}
/* skip coeng + ro, it was already processed */
if (i == coengRo) {
i += 1;
continue;
}
switch (charClass & CF_POS_MASK)
{
case CF_POS_ABOVE :
reordered[len] = uc;
properties[len] = AboveForm;
++len;
break;
case CF_POS_AFTER :
reordered[len] = uc;
properties[len] = PostForm;
++len;
break;
case CF_POS_BELOW :
reordered[len] = uc;
properties[len] = BelowForm;
++len;
break;
default:
/* assign the correct flags to a coeng consonant
Consonants of type 3 are taged as Post forms and those type 1 as below forms */
if ( (charClass & CF_COENG) && i + 1 < syllableEnd ) {
unsigned char property = (getKhmerCharClass(item->string[i+1]) & CF_CLASS_MASK) == CC_CONSONANT3 ?
PostForm : BelowForm;
reordered[len] = uc;
properties[len] = property;
++len;
i += 1;
reordered[len] = item->string[i];
properties[len] = property;
++len;
break;
}
/* if a shifter is followed by an above vowel change the shifter to below form,
an above vowel can have two possible positions i + 1 or i + 3
(position i+1 corresponds to unicode 3, position i+3 to Unicode 4)
and there is an extra rule for C_VOWEL_AA + C_SIGN_NIKAHIT also for two
different positions, right after the shifter or after a vowel (Unicode 4) */
if ( (charClass & CF_SHIFTER) && (i + 1 < syllableEnd) ) {
if (getKhmerCharClass(item->string[i+1]) & CF_ABOVE_VOWEL ) {
reordered[len] = uc;
properties[len] = BelowForm;
++len;
break;
}
if (i + 2 < syllableEnd &&
(item->string[i+1] == C_VOWEL_AA) &&
(item->string[i+2] == C_SIGN_NIKAHIT) )
{
reordered[len] = uc;
properties[len] = BelowForm;
++len;
break;
}
if (i + 3 < syllableEnd && (getKhmerCharClass(item->string[i+3]) & CF_ABOVE_VOWEL) ) {
reordered[len] = uc;
properties[len] = BelowForm;
++len;
break;
}
if (i + 4 < syllableEnd &&
(item->string[i+3] == C_VOWEL_AA) &&
(item->string[i+4] == C_SIGN_NIKAHIT) )
{
reordered[len] = uc;
properties[len] = BelowForm;
++len;
break;
}
}
/* default - any other characters */
reordered[len] = uc;
++len;
break;
} /* switch */
} /* for */
if (!item->font->klass->convertStringToGlyphIndices(item->font,
reordered, len,
item->glyphs, &item->num_glyphs,
item->item.bidiLevel % 2))
return FALSE;
KHDEBUG("after shaping: len=%d", len);
for (i = 0; i < len; i++) {
item->attributes[i].mark = FALSE;
item->attributes[i].clusterStart = FALSE;
item->attributes[i].justification = 0;
item->attributes[i].zeroWidth = FALSE;
KHDEBUG(" %d: %4x property=%x", i, reordered[i], properties[i]);
}
/* now we have the syllable in the right order, and can start running it through open type. */
#ifndef NO_OPENTYPE
if (openType) {
hb_uint32 where[16];
for (i = 0; i < len; ++i) {
where[i] = ~(PreSubstProperty
| BelowSubstProperty
| AboveSubstProperty
| PostSubstProperty
| CligProperty
| PositioningProperties);
if (properties[i] == PreForm)
where[i] &= ~PreFormProperty;
else if (properties[i] == BelowForm)
where[i] &= ~BelowFormProperty;
else if (properties[i] == AboveForm)
where[i] &= ~AboveFormProperty;
else if (properties[i] == PostForm)
where[i] &= ~PostFormProperty;
}
HB_OpenTypeShape(item, where);
if (!HB_OpenTypePosition(item, availableGlyphs, /*doLogClusters*/FALSE))
return FALSE;
} else
#endif
{
KHDEBUG("Not using openType");
HB_HeuristicPosition(item);
}
item->attributes[0].clusterStart = TRUE;
return TRUE;
}
HB_Bool HB_KhmerShape(HB_ShaperItem *item)
{
HB_Bool openType = FALSE;
unsigned short *logClusters = item->log_clusters;
int i;
HB_ShaperItem syllable = *item;
int first_glyph = 0;
int sstart = item->item.pos;
int end = sstart + item->item.length;
assert(item->item.script == HB_Script_Khmer);
#ifndef NO_OPENTYPE
openType = HB_SelectScript(item, khmer_features);
#endif
KHDEBUG("khmer_shape: from %d length %d", item->item.pos, item->item.length);
while (sstart < end) {
HB_Bool invalid;
int send = khmer_nextSyllableBoundary(item->string, sstart, end, &invalid);
KHDEBUG("syllable from %d, length %d, invalid=%s", sstart, send-sstart,
invalid ? "TRUE" : "FALSE");
syllable.item.pos = sstart;
syllable.item.length = send-sstart;
syllable.glyphs = item->glyphs + first_glyph;
syllable.attributes = item->attributes + first_glyph;
syllable.offsets = item->offsets + first_glyph;
syllable.advances = item->advances + first_glyph;
syllable.num_glyphs = item->num_glyphs - first_glyph;
if (!khmer_shape_syllable(openType, &syllable)) {
KHDEBUG("syllable shaping failed, syllable requests %d glyphs", syllable.num_glyphs);
item->num_glyphs += syllable.num_glyphs;
return FALSE;
}
/* fix logcluster array */
KHDEBUG("syllable:");
for (i = first_glyph; i < first_glyph + (int)syllable.num_glyphs; ++i)
KHDEBUG(" %d -> glyph %x", i, item->glyphs[i]);
KHDEBUG(" logclusters:");
for (i = sstart; i < send; ++i) {
KHDEBUG(" %d -> glyph %d", i, first_glyph);
logClusters[i-item->item.pos] = first_glyph;
}
sstart = send;
first_glyph += syllable.num_glyphs;
}
item->num_glyphs = first_glyph;
return TRUE;
}