In cluster-level=0, group ZWJ/ZWNJ with previous cluster

This better emulates Unicode grapheme clusters.

Note that Uniscribe does NOT do this, but should be harmless with most clients,
and improve fallback with clients that use HarfBuzz cluster as unit of fallback.

Fixes https://github.com/behdad/harfbuzz/issues/217

4 files changed, 10 insertions, 3 deletions

diff --git a/src/hb-ot-layout-private.hh b/src/hb-ot-layout-private.hh
index 094db957..b5c670f6 100644
--- a/src/hb-ot-layout-private.hh
+++ b/src/hb-ot-layout-private.hh
@@ -365,6 +365,12 @@ _hb_glyph_info_is_zwj (const hb_glyph_info_t *info)
   return !!(info->unicode_props() & UPROPS_MASK_ZWJ);
 }
 
+static inline hb_bool_t
+_hb_glyph_info_is_joiner (const hb_glyph_info_t *info)
+{
+  return !!(info->unicode_props() & (UPROPS_MASK_ZWNJ | UPROPS_MASK_ZWJ));
+}
+
 static inline void
 _hb_glyph_info_flip_joiners (hb_glyph_info_t *info)
 {
diff --git a/src/hb-ot-shape.cc b/src/hb-ot-shape.cc
index 015707da..42b921c6 100644
--- a/src/hb-ot-shape.cc
+++ b/src/hb-ot-shape.cc
@@ -273,7 +273,8 @@ hb_form_clusters (hb_buffer_t *buffer)
   hb_glyph_info_t *info = buffer->info;
   for (unsigned int i = 1; i < count; i++)
   {
-    if (likely (!HB_UNICODE_GENERAL_CATEGORY_IS_MARK (_hb_glyph_info_get_general_category (&info[i]))))
+    if (likely (!HB_UNICODE_GENERAL_CATEGORY_IS_MARK (_hb_glyph_info_get_general_category (&info[i])) &&
+		!_hb_glyph_info_is_joiner (&info[i])))
     {
       buffer->merge_clusters (base, i);
       base = i;
diff --git a/test/shaping/tests/indic-joiner-candrabindu.tests b/test/shaping/tests/indic-joiner-candrabindu.tests
index 2ef3304b..80ad8ce2 100644
--- a/test/shaping/tests/indic-joiner-candrabindu.tests
+++ b/test/shaping/tests/indic-joiner-candrabindu.tests
@@ -1,2 +1,2 @@
 fonts/sha1sum/5028afb650b1bb718ed2131e872fbcce57828fff.ttf::U+0B13,U+200D,U+0B01:[omorya=0+1450]
-fonts/sha1sum/5028afb650b1bb718ed2131e872fbcce57828fff.ttf::U+0B13,U+200C,U+0B01:[oorya=0+1309|space=1+0|candrabinduorya=1+0]
+fonts/sha1sum/5028afb650b1bb718ed2131e872fbcce57828fff.ttf::U+0B13,U+200C,U+0B01:[oorya=0+1309|space=0+0|candrabinduorya=0+0]
diff --git a/test/shaping/tests/mongolian-variation-selector.tests b/test/shaping/tests/mongolian-variation-selector.tests
index 29d11f4c..c24c9ba9 100644
--- a/test/shaping/tests/mongolian-variation-selector.tests
+++ b/test/shaping/tests/mongolian-variation-selector.tests
@@ -1,3 +1,3 @@
 fonts/sha1sum/37033cc5cf37bb223d7355153016b6ccece93b28.ttf::U+1826,U+180B,U+1826:[uni1826.E85E_ue.init1=0+599|uni1826.E856_ue.fina=2+750]
 fonts/sha1sum/ef86fe710cfea877bbe0dbb6946a1f88d0661031.ttf::U+1820,U+180B:[uni1820.E821_a.isol1=0+1199]
-fonts/sha1sum/bb29ce50df2bdba2d10726427c6b7609bf460e04.ttf::U+183A,U+1823,U+182E,U+182B,U+1822,U+1826,U+180B,U+1832,U+180B,U+1827,U+1837,U+0020,U+182D,U+182D,U+180B,U+0020,U+182D,U+180C,U+0020,U+182D,U+180D,U+200D,U+0020,U+182D,U+200D,U+182D,U+180B,U+200D,U+0020,U+182D,U+180C,U+200D,U+0020,U+182D,U+180D,U+200D,U+0020,U+200D,U+182D,U+200D,U+200D,U+182D,U+180B,U+200D,U+0020,U+200D,U+182D,U+180C,U+200D,U+0020,U+200D,U+182D,U+180D,U+200D,U+0020,U+200D,U+182D,U+200D,U+182D,U+180B,U+0020,U+200D,U+182D,U+180C,U+0020,U+1820,U+200C,U+182D,U+1820,U+1837,U+0020,U+1830,U+1824,U+1837,U+200D,U+200D,U+182D,U+1820,U+200D,U+0020,U+200D,U+182D,U+1824,U+182F,U+1822,U+0020,U+182A,U+1820,U+1822,U+182D,U+180E,U+1820,U+202F,U+1836,U+1822,U+1828:[uni183A1823.E971_ko.init=0+950|uni182E.E904_m.medi=2+400|uni182B1822.E8A6_pi.medi=3+1150|uni1826.E854_ue.medi1=5+1100|uni1832.E916_t.medi1=7+1000|uni1827.E85C_ee.medi=9+750|uni1837.E931_r.fina=10+750|space=11+500|uni182D.E8E2_g.init=12+1000|uni182D.E8E8_g.fina1=13+1250|space=15+500|uni182D.EA1B_g.isol2=16+1000|space=18+500|uni182D.EA1E_g.init3=19+650|space=21+0|space=22+500|uni182D.E8E2_g.init=23+1000|space=24+0|uni182D.E8E5_g.medi1=25+800|space=27+0|space=28+500|uni182D.EA1D_g.init2=29+950|space=31+0|space=32+500|uni182D.EA1E_g.init3=33+650|space=35+0|space=36+500|space=37+0|uni182D.E8E4_g.medi=38+800|space=39+0|space=40+0|uni182D.E8E5_g.medi1=41+800|space=43+0|space=44+500|space=45+0|uni182D.E8E6_g.medi2=46+650|space=48+0|space=49+500|space=50+0|uni182D.E8E6_g.medi2=51+650|space=53+0|space=54+500|space=55+0|uni182D.E8E4_g.medi=56+800|space=57+0|uni182D.E8E8_g.fina1=58+1250|space=60+500|space=61+0|uni182D.E8E9_g.fina2=62+1050|space=64+500|uni1820.E820_a.isol=65+1550|space=66+0|uni182D.E8E2_g.init=67+1000|uni1820.E823_a.medi=68+400|uni1837.E931_r.fina=69+750|space=70+500|uni1830.E90B_s.init=71+850|uni1824.E844_u.medi=72+600|uni1837.E930_r.medi=73+600|space=74+0|space=75+0|uni182D.E8E5_g.medi1=76+800|uni1820.E823_a.medi=77+400|space=78+0|space=79+500|space=80+0|uni182D.E8E5_g.medi1=81+800|uni1824.E844_u.medi=82+600|uni182F.E908_l.medi=83+400|uni1822.E837_i.fina=84+600|space=85+500|uni182A1820.E875_ba.init=86+1000|uni1822.E836_i.medi2=88+1000|uni182D.E8E8_g.fina1=89+1250|space=90+0|uni1820.E827_a.fina2=91+600|uni202F.nobreak=92+500|uni1836.E92B_y.init1=93+500|uni1822.E834_i.medi=94+500|uni1828.E866_n.fina=95+850]
+fonts/sha1sum/bb29ce50df2bdba2d10726427c6b7609bf460e04.ttf::U+183A,U+1823,U+182E,U+182B,U+1822,U+1826,U+180B,U+1832,U+180B,U+1827,U+1837,U+0020,U+182D,U+182D,U+180B,U+0020,U+182D,U+180C,U+0020,U+182D,U+180D,U+200D,U+0020,U+182D,U+200D,U+182D,U+180B,U+200D,U+0020,U+182D,U+180C,U+200D,U+0020,U+182D,U+180D,U+200D,U+0020,U+200D,U+182D,U+200D,U+200D,U+182D,U+180B,U+200D,U+0020,U+200D,U+182D,U+180C,U+200D,U+0020,U+200D,U+182D,U+180D,U+200D,U+0020,U+200D,U+182D,U+200D,U+182D,U+180B,U+0020,U+200D,U+182D,U+180C,U+0020,U+1820,U+200C,U+182D,U+1820,U+1837,U+0020,U+1830,U+1824,U+1837,U+200D,U+200D,U+182D,U+1820,U+200D,U+0020,U+200D,U+182D,U+1824,U+182F,U+1822,U+0020,U+182A,U+1820,U+1822,U+182D,U+180E,U+1820,U+202F,U+1836,U+1822,U+1828:[uni183A1823.E971_ko.init=0+950|uni182E.E904_m.medi=2+400|uni182B1822.E8A6_pi.medi=3+1150|uni1826.E854_ue.medi1=5+1100|uni1832.E916_t.medi1=7+1000|uni1827.E85C_ee.medi=9+750|uni1837.E931_r.fina=10+750|space=11+500|uni182D.E8E2_g.init=12+1000|uni182D.E8E8_g.fina1=13+1250|space=15+500|uni182D.EA1B_g.isol2=16+1000|space=18+500|uni182D.EA1E_g.init3=19+650|space=19+0|space=22+500|uni182D.E8E2_g.init=23+1000|space=23+0|uni182D.E8E5_g.medi1=25+800|space=25+0|space=28+500|uni182D.EA1D_g.init2=29+950|space=29+0|space=32+500|uni182D.EA1E_g.init3=33+650|space=33+0|space=36+500|space=36+0|uni182D.E8E4_g.medi=38+800|space=38+0|space=38+0|uni182D.E8E5_g.medi1=41+800|space=41+0|space=44+500|space=44+0|uni182D.E8E6_g.medi2=46+650|space=46+0|space=49+500|space=49+0|uni182D.E8E6_g.medi2=51+650|space=51+0|space=54+500|space=54+0|uni182D.E8E4_g.medi=56+800|space=56+0|uni182D.E8E8_g.fina1=58+1250|space=60+500|space=60+0|uni182D.E8E9_g.fina2=62+1050|space=64+500|uni1820.E820_a.isol=65+1550|space=65+0|uni182D.E8E2_g.init=67+1000|uni1820.E823_a.medi=68+400|uni1837.E931_r.fina=69+750|space=70+500|uni1830.E90B_s.init=71+850|uni1824.E844_u.medi=72+600|uni1837.E930_r.medi=73+600|space=73+0|space=73+0|uni182D.E8E5_g.medi1=76+800|uni1820.E823_a.medi=77+400|space=77+0|space=79+500|space=79+0|uni182D.E8E5_g.medi1=81+800|uni1824.E844_u.medi=82+600|uni182F.E908_l.medi=83+400|uni1822.E837_i.fina=84+600|space=85+500|uni182A1820.E875_ba.init=86+1000|uni1822.E836_i.medi2=88+1000|uni182D.E8E8_g.fina1=89+1250|space=90+0|uni1820.E827_a.fina2=91+600|uni202F.nobreak=92+500|uni1836.E92B_y.init1=93+500|uni1822.E834_i.medi=94+500|uni1828.E866_n.fina=95+850]