More glyph grouping fixes

This commit is contained in:
Kovid Goyal
2021-03-25 14:06:16 +05:30
parent 44df11c443
commit d28caf15b2
2 changed files with 18 additions and 4 deletions

View File

@@ -10,6 +10,7 @@
#include "state.h" #include "state.h"
#include "emoji.h" #include "emoji.h"
#include "unicode-data.h" #include "unicode-data.h"
#include "charsets.h"
#define MISSING_GLYPH 4 #define MISSING_GLYPH 4
#define MAX_NUM_EXTRA_GLYPHS 8u #define MAX_NUM_EXTRA_GLYPHS 8u
@@ -961,10 +962,12 @@ shape_run(CPUCell *first_cpu_cell, GPUCell *first_gpu_cell, index_type num_cells
bool is_special = is_special_glyph(glyph_id, font, &G(current_cell_data)); bool is_special = is_special_glyph(glyph_id, font, &G(current_cell_data));
bool is_empty = is_special && is_empty_glyph(glyph_id, font); bool is_empty = is_special && is_empty_glyph(glyph_id, font);
uint32_t num_codepoints_used_by_glyph = 0; uint32_t num_codepoints_used_by_glyph = 0;
bool is_last_glyph = G(glyph_idx) == G(num_glyphs) - 1, is_first_glyph = G(glyph_idx) == 0; bool is_last_glyph = G(glyph_idx) == G(num_glyphs) - 1;
Group *current_group = G(groups) + G(group_idx); Group *current_group = G(groups) + G(group_idx);
if (is_last_glyph) { if (is_last_glyph) {
num_codepoints_used_by_glyph = UINT32_MAX; num_codepoints_used_by_glyph = UINT32_MAX;
next_cluster = 0;
} else { } else {
next_cluster = G(info)[G(glyph_idx) + 1].cluster; next_cluster = G(info)[G(glyph_idx) + 1].cluster;
// RTL languages like Arabic have decreasing cluster numbers // RTL languages like Arabic have decreasing cluster numbers
@@ -977,16 +980,25 @@ shape_run(CPUCell *first_cpu_cell, GPUCell *first_gpu_cell, index_type num_cells
else add_to_current_group = ligature_type == INFINITE_LIGATURE_MIDDLE || ligature_type == INFINITE_LIGATURE_END || is_empty; else add_to_current_group = ligature_type == INFINITE_LIGATURE_MIDDLE || ligature_type == INFINITE_LIGATURE_END || is_empty;
} else { } else {
if (is_special) { if (is_special) {
if (!is_first_glyph && !current_group->num_cells) add_to_current_group = true; if (!current_group->num_cells) add_to_current_group = true;
else if (font->spacer_strategy == SPACERS_BEFORE) add_to_current_group = G(prev_was_empty); else if (font->spacer_strategy == SPACERS_BEFORE) add_to_current_group = G(prev_was_empty);
else add_to_current_group = is_empty; else add_to_current_group = is_empty;
} else { } else {
add_to_current_group = !G(prev_was_special); add_to_current_group = !G(prev_was_special) || !current_group->num_cells;
} }
} }
static const bool debug_grouping = false;
if (debug_grouping) {
char ch[8] = {0};
encode_utf8(G(current_cell_data).current_codepoint, ch);
printf("\x1b[32m→ %s\x1b[m glyph_idx: %lu glyph_id: %u (%s) group_idx: %lu cluster: %u -> %u is_special: %d\n"
" num_codepoints_used_by_glyph: %u current_group: (%u cells, %u glyphs) add_to_current_group: %d\n",
ch, G(glyph_idx), glyph_id, glyph_name, G(group_idx), cluster, next_cluster, is_special,
num_codepoints_used_by_glyph, current_group->num_cells, current_group->num_glyphs, add_to_current_group);
}
if (current_group->num_glyphs >= MAX_GLYPHS_IN_GROUP || current_group->num_cells >= MAX_GLYPHS_IN_GROUP) add_to_current_group = false; if (current_group->num_glyphs >= MAX_GLYPHS_IN_GROUP || current_group->num_cells >= MAX_GLYPHS_IN_GROUP) add_to_current_group = false;
if (!add_to_current_group) { G(group_idx)++; current_group = G(groups) + G(group_idx); } if (!add_to_current_group) { current_group = G(groups) + ++G(group_idx); }
if (!current_group->num_glyphs++) { if (!current_group->num_glyphs++) {
if (ligature_type == INFINITE_LIGATURE_START || ligature_type == INFINITE_LIGATURE_MIDDLE) current_group->started_with_infinite_ligature = true; if (ligature_type == INFINITE_LIGATURE_START || ligature_type == INFINITE_LIGATURE_MIDDLE) current_group->started_with_infinite_ligature = true;
current_group->first_glyph_idx = G(glyph_idx); current_group->first_glyph_idx = G(glyph_idx);

View File

@@ -116,7 +116,9 @@ class Rendering(BaseTest):
self.ae(groups('|\U0001F601|\U0001F64f|\U0001F63a|'), [(1, 1), (2, 1), (1, 1), (2, 1), (1, 1), (2, 1), (1, 1)]) self.ae(groups('|\U0001F601|\U0001F64f|\U0001F63a|'), [(1, 1), (2, 1), (1, 1), (2, 1), (1, 1), (2, 1), (1, 1)])
self.ae(groups('He\u0347\u0305llo\u0337,', font='LiberationMono-Regular.ttf'), self.ae(groups('He\u0347\u0305llo\u0337,', font='LiberationMono-Regular.ttf'),
[(1, 1), (1, 3), (1, 1), (1, 1), (1, 2), (1, 1)]) [(1, 1), (1, 3), (1, 1), (1, 1), (1, 2), (1, 1)])
self.ae(groups('i\u0332\u0308', font='LiberationMono-Regular.ttf'), [(1, 2)]) self.ae(groups('i\u0332\u0308', font='LiberationMono-Regular.ttf'), [(1, 2)])
self.ae(groups('u\u0332 u\u0332\u0301', font='LiberationMono-Regular.ttf'), [(1, 2), (1, 1), (1, 2)])
def test_emoji_presentation(self): def test_emoji_presentation(self):
s = self.create_screen() s = self.create_screen()