harfbuzz/src/hb-ot-shape-complex-thai.cc

384 lines
12 KiB
C++

/*
* Copyright © 2010,2012 Google, Inc.
*
* 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
*/
#include "hb-ot-shape-complex.hh"
/* Thai / Lao shaper */
/* PUA shaping */
enum thai_consonant_type_t
{
NC,
AC,
RC,
DC,
NOT_CONSONANT,
NUM_CONSONANT_TYPES = NOT_CONSONANT
};
static thai_consonant_type_t
get_consonant_type (hb_codepoint_t u)
{
if (u == 0x0E1Bu || u == 0x0E1Du || u == 0x0E1Fu/* || u == 0x0E2Cu*/)
return AC;
if (u == 0x0E0Du || u == 0x0E10u)
return RC;
if (u == 0x0E0Eu || u == 0x0E0Fu)
return DC;
if (hb_in_range<hb_codepoint_t> (u, 0x0E01u, 0x0E2Eu))
return NC;
return NOT_CONSONANT;
}
enum thai_mark_type_t
{
AV,
BV,
T,
NOT_MARK,
NUM_MARK_TYPES = NOT_MARK
};
static thai_mark_type_t
get_mark_type (hb_codepoint_t u)
{
if (u == 0x0E31u || hb_in_range<hb_codepoint_t> (u, 0x0E34u, 0x0E37u) ||
u == 0x0E47u || hb_in_range<hb_codepoint_t> (u, 0x0E4Du, 0x0E4Eu))
return AV;
if (hb_in_range<hb_codepoint_t> (u, 0x0E38u, 0x0E3Au))
return BV;
if (hb_in_range<hb_codepoint_t> (u, 0x0E48u, 0x0E4Cu))
return T;
return NOT_MARK;
}
enum thai_action_t
{
NOP,
SD, /* Shift combining-mark down */
SL, /* Shift combining-mark left */
SDL, /* Shift combining-mark down-left */
RD /* Remove descender from base */
};
static hb_codepoint_t
thai_pua_shape (hb_codepoint_t u, thai_action_t action, hb_font_t *font)
{
struct thai_pua_mapping_t {
hb_codepoint_t u;
hb_codepoint_t win_pua;
hb_codepoint_t mac_pua;
} const *pua_mappings = nullptr;
static const thai_pua_mapping_t SD_mappings[] = {
{0x0E48u, 0xF70Au, 0xF88Bu}, /* MAI EK */
{0x0E49u, 0xF70Bu, 0xF88Eu}, /* MAI THO */
{0x0E4Au, 0xF70Cu, 0xF891u}, /* MAI TRI */
{0x0E4Bu, 0xF70Du, 0xF894u}, /* MAI CHATTAWA */
{0x0E4Cu, 0xF70Eu, 0xF897u}, /* THANTHAKHAT */
{0x0E38u, 0xF718u, 0xF89Bu}, /* SARA U */
{0x0E39u, 0xF719u, 0xF89Cu}, /* SARA UU */
{0x0E3Au, 0xF71Au, 0xF89Du}, /* PHINTHU */
{0x0000u, 0x0000u, 0x0000u}
};
static const thai_pua_mapping_t SDL_mappings[] = {
{0x0E48u, 0xF705u, 0xF88Cu}, /* MAI EK */
{0x0E49u, 0xF706u, 0xF88Fu}, /* MAI THO */
{0x0E4Au, 0xF707u, 0xF892u}, /* MAI TRI */
{0x0E4Bu, 0xF708u, 0xF895u}, /* MAI CHATTAWA */
{0x0E4Cu, 0xF709u, 0xF898u}, /* THANTHAKHAT */
{0x0000u, 0x0000u, 0x0000u}
};
static const thai_pua_mapping_t SL_mappings[] = {
{0x0E48u, 0xF713u, 0xF88Au}, /* MAI EK */
{0x0E49u, 0xF714u, 0xF88Du}, /* MAI THO */
{0x0E4Au, 0xF715u, 0xF890u}, /* MAI TRI */
{0x0E4Bu, 0xF716u, 0xF893u}, /* MAI CHATTAWA */
{0x0E4Cu, 0xF717u, 0xF896u}, /* THANTHAKHAT */
{0x0E31u, 0xF710u, 0xF884u}, /* MAI HAN-AKAT */
{0x0E34u, 0xF701u, 0xF885u}, /* SARA I */
{0x0E35u, 0xF702u, 0xF886u}, /* SARA II */
{0x0E36u, 0xF703u, 0xF887u}, /* SARA UE */
{0x0E37u, 0xF704u, 0xF888u}, /* SARA UEE */
{0x0E47u, 0xF712u, 0xF889u}, /* MAITAIKHU */
{0x0E4Du, 0xF711u, 0xF899u}, /* NIKHAHIT */
{0x0000u, 0x0000u, 0x0000u}
};
static const thai_pua_mapping_t RD_mappings[] = {
{0x0E0Du, 0xF70Fu, 0xF89Au}, /* YO YING */
{0x0E10u, 0xF700u, 0xF89Eu}, /* THO THAN */
{0x0000u, 0x0000u, 0x0000u}
};
switch (action) {
case NOP: return u;
case SD: pua_mappings = SD_mappings; break;
case SDL: pua_mappings = SDL_mappings; break;
case SL: pua_mappings = SL_mappings; break;
case RD: pua_mappings = RD_mappings; break;
}
for (; pua_mappings->u; pua_mappings++)
if (pua_mappings->u == u)
{
hb_codepoint_t glyph;
if (hb_font_get_glyph (font, pua_mappings->win_pua, 0, &glyph))
return pua_mappings->win_pua;
if (hb_font_get_glyph (font, pua_mappings->mac_pua, 0, &glyph))
return pua_mappings->mac_pua;
break;
}
return u;
}
static enum thai_above_state_t
{ /* Cluster above looks like: */
T0, /* ⣤ */
T1, /* ⣼ */
T2, /* ⣾ */
T3, /* ⣿ */
NUM_ABOVE_STATES
} thai_above_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] =
{
T0, /* NC */
T1, /* AC */
T0, /* RC */
T0, /* DC */
T3, /* NOT_CONSONANT */
};
static const struct thai_above_state_machine_edge_t {
thai_action_t action;
thai_above_state_t next_state;
} thai_above_state_machine[NUM_ABOVE_STATES][NUM_MARK_TYPES] =
{ /*AV*/ /*BV*/ /*T*/
/*T0*/ {{NOP,T3}, {NOP,T0}, {SD, T3}},
/*T1*/ {{SL, T2}, {NOP,T1}, {SDL,T2}},
/*T2*/ {{NOP,T3}, {NOP,T2}, {SL, T3}},
/*T3*/ {{NOP,T3}, {NOP,T3}, {NOP,T3}},
};
static enum thai_below_state_t
{
B0, /* No descender */
B1, /* Removable descender */
B2, /* Strict descender */
NUM_BELOW_STATES
} thai_below_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] =
{
B0, /* NC */
B0, /* AC */
B1, /* RC */
B2, /* DC */
B2, /* NOT_CONSONANT */
};
static const struct thai_below_state_machine_edge_t {
thai_action_t action;
thai_below_state_t next_state;
} thai_below_state_machine[NUM_BELOW_STATES][NUM_MARK_TYPES] =
{ /*AV*/ /*BV*/ /*T*/
/*B0*/ {{NOP,B0}, {NOP,B2}, {NOP, B0}},
/*B1*/ {{NOP,B1}, {RD, B2}, {NOP, B1}},
/*B2*/ {{NOP,B2}, {SD, B2}, {NOP, B2}},
};
static void
do_thai_pua_shaping (const hb_ot_shape_plan_t *plan HB_UNUSED,
hb_buffer_t *buffer,
hb_font_t *font)
{
thai_above_state_t above_state = thai_above_start_state[NOT_CONSONANT];
thai_below_state_t below_state = thai_below_start_state[NOT_CONSONANT];
unsigned int base = 0;
hb_glyph_info_t *info = buffer->info;
unsigned int count = buffer->len;
for (unsigned int i = 0; i < count; i++)
{
thai_mark_type_t mt = get_mark_type (info[i].codepoint);
if (mt == NOT_MARK) {
thai_consonant_type_t ct = get_consonant_type (info[i].codepoint);
above_state = thai_above_start_state[ct];
below_state = thai_below_start_state[ct];
base = i;
continue;
}
const thai_above_state_machine_edge_t &above_edge = thai_above_state_machine[above_state][mt];
const thai_below_state_machine_edge_t &below_edge = thai_below_state_machine[below_state][mt];
above_state = above_edge.next_state;
below_state = below_edge.next_state;
/* At least one of the above/below actions is NOP. */
thai_action_t action = above_edge.action != NOP ? above_edge.action : below_edge.action;
buffer->unsafe_to_break (base, i);
if (action == RD)
info[base].codepoint = thai_pua_shape (info[base].codepoint, action, font);
else
info[i].codepoint = thai_pua_shape (info[i].codepoint, action, font);
}
}
static void
preprocess_text_thai (const hb_ot_shape_plan_t *plan,
hb_buffer_t *buffer,
hb_font_t *font)
{
/* This function implements the shaping logic documented here:
*
* https://linux.thai.net/~thep/th-otf/shaping.html
*
* The first shaping rule listed there is needed even if the font has Thai
* OpenType tables. The rest do fallback positioning based on PUA codepoints.
* We implement that only if there exist no Thai GSUB in the font.
*/
/* The following is NOT specified in the MS OT Thai spec, however, it seems
* to be what Uniscribe and other engines implement. According to Eric Muller:
*
* When you have a SARA AM, decompose it in NIKHAHIT + SARA AA, *and* move the
* NIKHAHIT backwards over any tone mark (0E48-0E4B).
*
* <0E14, 0E4B, 0E33> -> <0E14, 0E4D, 0E4B, 0E32>
*
* This reordering is legit only when the NIKHAHIT comes from a SARA AM, not
* when it's there to start with. The string <0E14, 0E4B, 0E4D> is probably
* not what a user wanted, but the rendering is nevertheless nikhahit above
* chattawa.
*
* Same for Lao.
*
* Note:
*
* Uniscribe also does some below-marks reordering. Namely, it positions U+0E3A
* after U+0E38 and U+0E39. We do that by modifying the ccc for U+0E3A.
* See unicode->modified_combining_class (). Lao does NOT have a U+0E3A
* equivalent.
*/
/*
* Here are the characters of significance:
*
* Thai Lao
* SARA AM: U+0E33 U+0EB3
* SARA AA: U+0E32 U+0EB2
* Nikhahit: U+0E4D U+0ECD
*
* Testing shows that Uniscribe reorder the following marks:
* Thai: <0E31,0E34..0E37,0E47..0E4E>
* Lao: <0EB1,0EB4..0EB7,0EC7..0ECE>
*
* Note how the Lao versions are the same as Thai + 0x80.
*/
/* We only get one script at a time, so a script-agnostic implementation
* is adequate here. */
#define IS_SARA_AM(x) (((x) & ~0x0080u) == 0x0E33u)
#define NIKHAHIT_FROM_SARA_AM(x) ((x) - 0x0E33u + 0x0E4Du)
#define SARA_AA_FROM_SARA_AM(x) ((x) - 1)
#define IS_TONE_MARK(x) (hb_in_ranges<hb_codepoint_t> ((x) & ~0x0080u, 0x0E34u, 0x0E37u, 0x0E47u, 0x0E4Eu, 0x0E31u, 0x0E31u))
buffer->clear_output ();
unsigned int count = buffer->len;
for (buffer->idx = 0; buffer->idx < count && buffer->successful;)
{
hb_codepoint_t u = buffer->cur().codepoint;
if (likely (!IS_SARA_AM (u))) {
buffer->next_glyph ();
continue;
}
/* Is SARA AM. Decompose and reorder. */
hb_glyph_info_t &nikhahit = buffer->output_glyph (NIKHAHIT_FROM_SARA_AM (u));
_hb_glyph_info_set_continuation (&nikhahit);
buffer->replace_glyph (SARA_AA_FROM_SARA_AM (u));
if (unlikely (!buffer->successful))
return;
/* Make Nikhahit be recognized as a ccc=0 mark when zeroing widths. */
unsigned int end = buffer->out_len;
_hb_glyph_info_set_general_category (&buffer->out_info[end - 2], HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK);
/* Ok, let's see... */
unsigned int start = end - 2;
while (start > 0 && IS_TONE_MARK (buffer->out_info[start - 1].codepoint))
start--;
if (start + 2 < end)
{
/* Move Nikhahit (end-2) to the beginning */
buffer->merge_out_clusters (start, end);
hb_glyph_info_t t = buffer->out_info[end - 2];
memmove (buffer->out_info + start + 1,
buffer->out_info + start,
sizeof (buffer->out_info[0]) * (end - start - 2));
buffer->out_info[start] = t;
}
else
{
/* Since we decomposed, and NIKHAHIT is combining, merge clusters with the
* previous cluster. */
if (start && buffer->cluster_level == HB_BUFFER_CLUSTER_LEVEL_MONOTONE_GRAPHEMES)
buffer->merge_out_clusters (start - 1, end);
}
}
buffer->swap_buffers ();
/* If font has Thai GSUB, we are done. */
if (plan->props.script == HB_SCRIPT_THAI && !plan->map.found_script[0])
do_thai_pua_shaping (plan, buffer, font);
}
const hb_ot_complex_shaper_t _hb_ot_complex_shaper_thai =
{
nullptr, /* collect_features */
nullptr, /* override_features */
nullptr, /* data_create */
nullptr, /* data_destroy */
preprocess_text_thai,
nullptr, /* postprocess_glyphs */
HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
nullptr, /* decompose */
nullptr, /* compose */
nullptr, /* setup_masks */
HB_TAG_NONE, /* gpos_tag */
nullptr, /* reorder_marks */
HB_OT_SHAPE_ZERO_WIDTH_MARKS_BY_GDEF_LATE,
false,/* fallback_position */
};