Merge pull request #98 from LaurenzV/4.1

Sync with 4.1
harfbuzz · Feb 16, 2024 · 26baa9a · 26baa9a
2 parents 26fec9c + 550139c
commit 26baa9a
Show file tree

Hide file tree

Showing 12 changed files with 375 additions and 256 deletions.
diff --git a/scripts/gen-universal-table.py b/scripts/gen-universal-table.py
@@ -242,15 +242,7 @@ def is_CONS_WITH_STACKER(U, UISC, UDI, UGC, AJT):
 
 
 def is_HALANT(U, UISC, UDI, UGC, AJT):
-    return (UISC in [Virama, Invisible_Stacker]
-            and not is_HALANT_OR_VOWEL_MODIFIER(U, UISC, UDI, UGC, AJT)
-            and not is_SAKOT(U, UISC, UDI, UGC, AJT))
-
-
-def is_HALANT_OR_VOWEL_MODIFIER(U, UISC, UDI, UGC, AJT):
-    # Split off of HALANT
-    # https://github.com/harfbuzz/harfbuzz/issues/1379
-    return U == 0x1134D
+    return UISC == Virama
 
 
 def is_HALANT_NUM(U, UISC, UDI, UGC, AJT):
@@ -273,6 +265,13 @@ def is_HIEROGLYPH_SEGMENT_END(U, UISC, UDI, UGC, AJT):
     return UISC == Hieroglyph_Segment_End
 
 
+def is_INVISIBLE_STACKER(U, UISC, UDI, UGC, AJT):
+    # Split off of HALANT
+    return (UISC == Invisible_Stacker
+            and not is_SAKOT(U, UISC, UDI, UGC, AJT)
+            )
+
+
 def is_ZWNJ(U, UISC, UDI, UGC, AJT):
     return UISC == Non_Joiner
 
@@ -329,8 +328,8 @@ def is_Word_Joiner(U, UISC, UDI, UGC, AJT):
     'SUB': is_CONS_SUB,
     'CS': is_CONS_WITH_STACKER,
     'H': is_HALANT,
-    'HVM': is_HALANT_OR_VOWEL_MODIFIER,
     'HN': is_HALANT_NUM,
+    'IS': is_INVISIBLE_STACKER,
     'G': is_HIEROGLYPH,
     'J': is_HIEROGLYPH_JOINER,
     'SB': is_HIEROGLYPH_SEGMENT_BEGIN,
@@ -378,7 +377,7 @@ def is_Word_Joiner(U, UISC, UDI, UGC, AJT):
         'BLW': [Bottom],
     },
     'H': None,
-    'HVM': None,
+    'IS': None,
     'B': None,
     'FM': {
         'ABV': [Top],

diff --git a/src/buffer.rs b/src/buffer.rs
@@ -5,6 +5,7 @@ use std::cmp::min;
 use ttf_parser::GlyphId;
 
 use crate::buffer::glyph_flag::{UNSAFE_TO_BREAK, UNSAFE_TO_CONCAT};
+use crate::face::GlyphExtents;
 use crate::unicode::{CharExt, GeneralCategory, GeneralCategoryExt, Space};
 use crate::{script, Direction, Face, Language, Mask, Script};
 
@@ -1630,7 +1631,13 @@ impl Buffer {
 
     #[inline]
     pub fn allocate_lig_id(&mut self) -> u8 {
-        self.next_serial() & 0x07
+        let mut lig_id = self.next_serial() & 0x07;
+
+        if lig_id == 0 {
+            lig_id = self.allocate_lig_id();
+        }
+
+        lig_id
     }
 }
 
@@ -2012,7 +2019,8 @@ impl GlyphBuffer {
             }
 
             if flags.contains(SerializeFlags::GLYPH_EXTENTS) {
-                let extents = face.glyph_extents(info.as_glyph()).unwrap_or_default();
+                let mut extents = GlyphExtents::default();
+                face.glyph_extents(info.as_glyph(), &mut extents);
                 write!(
                     &mut s,
                     "<{},{},{},{}>",

diff --git a/src/complex/indic.rs b/src/complex/indic.rs
@@ -671,14 +671,13 @@ fn collect_features(planner: &mut ShapePlanner) {
     }
 
     planner.ot_map.add_gsub_pause(Some(final_reordering));
+    planner
+        .ot_map
+        .add_gsub_pause(Some(crate::ot::clear_syllables));
 
     for feature in INDIC_FEATURES.iter().skip(10) {
         planner.ot_map.add_feature(feature.0, feature.1, 1);
     }
-
-    planner
-        .ot_map
-        .add_gsub_pause(Some(crate::ot::clear_syllables));
 }
 
 fn override_features(planner: &mut ShapePlanner) {

diff --git a/src/complex/universal.rs b/src/complex/universal.rs
@@ -62,7 +62,7 @@ pub mod category {
     pub const CS: u8 = 43; // CONS_WITH_STACKER
 
     // https://github.com/harfbuzz/harfbuzz/issues/1102
-    pub const HVM: u8 = 44; // HALANT_OR_VOWEL_MODIFIER
+    pub const IS: u8 = 44; // HALANT_OR_VOWEL_MODIFIER
 
     pub const SK: u8 = 48; // SAKOT
 
@@ -141,7 +141,7 @@ impl GlyphInfo {
     }
 
     fn is_halant_use(&self) -> bool {
-        matches!(self.use_category(), category::H | category::HVM) && !self.is_ligated()
+        matches!(self.use_category(), category::H | category::IS) && !self.is_ligated()
     }
 }
 

diff --git a/src/complex/universal_machine.rl b/src/complex/universal_machine.rl
@@ -44,7 +44,7 @@ ZWNJ	= 14; # Zero width non-joiner
 WJ = 16; # Word joiner
 R	= 18; # REPHA
 CS	= 43; # CONS_WITH_STACKER
-HVM	= 44; # HALANT_OR_VOWEL_MODIFIER
+IS	= 44; # HALANT_OR_VOWEL_MODIFIER
 Sk	= 48; # SAKOT
 G	= 49; # HIEROGLYPH
 J	= 50; # HIEROGLYPH_JOINER
@@ -74,12 +74,12 @@ FMAbv	= 45; # CONS_FINAL_MOD	UIPC = Top
 FMBlw	= 46; # CONS_FINAL_MOD	UIPC = Bottom
 FMPst	= 47; # CONS_FINAL_MOD	UIPC = Not_Applicable
 
-h = H | HVM | Sk;
+h = H | IS | Sk;
 
 consonant_modifiers = CMAbv* CMBlw* ((h B | SUB) CMAbv? CMBlw*)*;
 medial_consonants = MPre? MAbv? MBlw? MPst?;
-dependent_vowels = VPre* VAbv* VBlw* VPst*;
-vowel_modifiers = HVM? VMPre* VMAbv* VMBlw* VMPst*;
+dependent_vowels = VPre* VAbv* VBlw* VPst* | H;
+vowel_modifiers = VMPre* VMAbv* VMBlw* VMPst*;
 final_consonants = FAbv* FBlw* FPst*;
 final_modifiers = FMAbv* FMBlw* | FMPst?;
 
@@ -102,7 +102,7 @@ symbol_cluster_tail = SMAbv+ SMBlw* | SMBlw+;
 
 virama_terminated_cluster_tail =
 	consonant_modifiers
-	h
+	IS
 ;
 virama_terminated_cluster =
 	complex_syllable_start
@@ -168,6 +168,7 @@ pub fn find_syllables(buffer: &mut Buffer) {
     let mut p = p0;
     let mut ts = p0;
     let mut te = p0;
+    let mut act = p0;
     let pe = p.end();
     let eof = p.end();
     let mut syllable_serial = 1u8;