[subst/SingleSubstFormat1] Rewrite nicer
This commit is contained in:
parent
cf123e6a0d
commit
bababe1072
|
@ -42,32 +42,7 @@ struct SingleSubstFormat1_3
|
||||||
hb_codepoint_t d = deltaGlyphID;
|
hb_codepoint_t d = deltaGlyphID;
|
||||||
hb_codepoint_t mask = get_mask ();
|
hb_codepoint_t mask = get_mask ();
|
||||||
|
|
||||||
if (1)
|
if ((this+coverage).get_population () < c->parent_active_glyphs ().get_population ())
|
||||||
{
|
|
||||||
/* Walk the coverage and set together, to bail out when the
|
|
||||||
* coverage iterator runs over the end of glyphset. Otherwise,
|
|
||||||
* the coverage may cover many glyphs and we spend a lot of
|
|
||||||
* time here. Other subtables don't have this problem because
|
|
||||||
* they zip coverage iterator with some array... */
|
|
||||||
auto c_iter = hb_iter (this+coverage);
|
|
||||||
auto s_iter = hb_iter (c->parent_active_glyphs ());
|
|
||||||
while (c_iter && s_iter)
|
|
||||||
{
|
|
||||||
hb_codepoint_t cv = *c_iter;
|
|
||||||
hb_codepoint_t sv = *s_iter;
|
|
||||||
if (cv == sv)
|
|
||||||
{
|
|
||||||
c->output->add ((cv + d) & mask);
|
|
||||||
c_iter++;
|
|
||||||
s_iter++;
|
|
||||||
}
|
|
||||||
else if (cv < sv)
|
|
||||||
c_iter++;
|
|
||||||
else
|
|
||||||
s_iter++;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
else
|
|
||||||
{
|
{
|
||||||
+ hb_iter (this+coverage)
|
+ hb_iter (this+coverage)
|
||||||
| hb_filter (c->parent_active_glyphs ())
|
| hb_filter (c->parent_active_glyphs ())
|
||||||
|
@ -75,6 +50,14 @@ struct SingleSubstFormat1_3
|
||||||
| hb_sink (c->output)
|
| hb_sink (c->output)
|
||||||
;
|
;
|
||||||
}
|
}
|
||||||
|
else
|
||||||
|
{
|
||||||
|
+ hb_iter (c->parent_active_glyphs ())
|
||||||
|
| hb_filter (this+coverage)
|
||||||
|
| hb_map ([d, mask] (hb_codepoint_t g) { return (g + d) & mask; })
|
||||||
|
| hb_sink (c->output)
|
||||||
|
;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
void closure_lookups (hb_closure_lookups_context_t *c) const {}
|
void closure_lookups (hb_closure_lookups_context_t *c) const {}
|
||||||
|
@ -136,37 +119,18 @@ struct SingleSubstFormat1_3
|
||||||
hb_codepoint_t mask = get_mask ();
|
hb_codepoint_t mask = get_mask ();
|
||||||
|
|
||||||
hb_sorted_vector_t<hb_codepoint_t> intersection;
|
hb_sorted_vector_t<hb_codepoint_t> intersection;
|
||||||
if (1)
|
if ((this+coverage).get_population () < glyphset.get_population ())
|
||||||
{
|
|
||||||
/* Walk the coverage and set together, to bail out when the
|
|
||||||
* coverage iterator runs over the end of glyphset. Otherwise,
|
|
||||||
* the coverage may cover many glyphs and we spend a lot of
|
|
||||||
* time here. Other subtables don't have this problem because
|
|
||||||
* they zip coverage iterator with some array... */
|
|
||||||
auto c_iter = hb_iter (this+coverage);
|
|
||||||
auto s_iter = hb_iter (glyphset);
|
|
||||||
while (c_iter && s_iter)
|
|
||||||
{
|
|
||||||
hb_codepoint_t cv = *c_iter;
|
|
||||||
hb_codepoint_t sv = *s_iter;
|
|
||||||
if (cv == sv)
|
|
||||||
{
|
|
||||||
intersection.push (cv);
|
|
||||||
c_iter++;
|
|
||||||
s_iter++;
|
|
||||||
}
|
|
||||||
else if (cv < sv)
|
|
||||||
c_iter++;
|
|
||||||
else
|
|
||||||
s_iter++;
|
|
||||||
}
|
|
||||||
}
|
|
||||||
else
|
|
||||||
{
|
{
|
||||||
+ hb_iter (this+coverage)
|
+ hb_iter (this+coverage)
|
||||||
| hb_filter (glyphset)
|
| hb_filter (glyphset)
|
||||||
| hb_sink (intersection);
|
| hb_sink (intersection);
|
||||||
}
|
}
|
||||||
|
else
|
||||||
|
{
|
||||||
|
+ hb_iter (glyphset)
|
||||||
|
| hb_filter (this+coverage)
|
||||||
|
| hb_sink (intersection);
|
||||||
|
}
|
||||||
|
|
||||||
auto it =
|
auto it =
|
||||||
+ hb_iter (intersection)
|
+ hb_iter (intersection)
|
||||||
|
|
Loading…
Reference in New Issue