WIP: improve colr emoji detection and testing

fontique/src/collection/mod.rs

@@ -318,22 +318,19 @@ impl Inner {
     /// Returns the family object for the given family identifier.
     pub fn family(&mut self, id: FamilyId) -> Option<FamilyInfo> {
         self.sync_shared();
+
         if let Some(family) = self.data.families.get(&id) {
-            family.as_ref().cloned()
-        } else {
-            #[cfg(feature = "system")]
-            if let Some(system) = &self.system {
-                let family = system.fonts.lock().unwrap().family(id);
-                self.data.families.insert(id, family.clone());
-                family
-            } else {
-                None
-            }
-            #[cfg(not(feature = "system"))]
-            {
-                None
-            }
+            return family.as_ref().cloned();
         }
+
+        #[cfg(feature = "system")]
+        if let Some(system) = &self.system {
+            let family = system.fonts.lock().unwrap().family(id);
+            self.data.families.insert(id, family.clone());
+            return family;
+        }
+
+        None
     }
 
     /// Returns the family object for the given name.

parley/src/analysis/cluster.rs

@@ -4,19 +4,19 @@
 use alloc::vec::Vec;
 use icu_normalizer::properties::Decomposed;
 
-use crate::analysis::AnalysisDataSources;
+use crate::{analysis::AnalysisDataSources, emoji::EmojiPresentationStyle};
 
 /// The maximum number of characters in a single cluster.
 const MAX_CLUSTER_SIZE: usize = 32;
 
 #[derive(Debug, Default)]
 pub(crate) struct CharCluster {
     pub chars: Vec<Char>,
-    pub is_emoji: bool,
     pub map_len: u8,
     pub start: u32,
     pub end: u32,
     pub force_normalize: bool,
+    pub emoji_presentation_style: EmojiPresentationStyle,
     comp: Form,
     decomp: Form,
     form: FormKind,
@@ -52,6 +52,8 @@ pub(crate) struct Char {
     /// Indexes into the list of styles for the containing text run, to find the style applicable
     /// to this character.
     pub style_index: u16,
+    /// Whether the emoji presentation selector
+    pub is_emoji_presentation_selector: bool,
 }
 
 pub(crate) type GlyphId = u16;
@@ -93,7 +95,6 @@ pub(crate) enum Status {
 impl CharCluster {
     pub(crate) fn clear(&mut self) {
         self.chars.clear();
-        self.is_emoji = false;
         self.map_len = 0;
         self.start = 0;
         self.end = 0;
@@ -102,6 +103,7 @@ impl CharCluster {
         self.decomp.clear();
         self.form = FormKind::Original;
         self.best_ratio = 0.;
+        self.emoji_presentation_style = EmojiPresentationStyle::Default;
     }
 
     #[inline(always)]
@@ -351,17 +353,23 @@ impl<'a> Mapper<'a> {
         }
         let mut mapped = 0;
         for (c, g) in self.chars.iter().zip(glyphs.iter_mut()) {
-            if !c.contributes_to_shaping {
-                *g = f(c.ch);
-                if self.map_len == 1 {
-                    mapped += 1;
-                }
-            } else {
-                let gid = f(c.ch);
-                *g = gid;
-                if gid != 0 {
+            *g = f(c.ch);
+
+            // If the color emoji has a presentation style, ignore the variation selector.
+            if c.is_emoji_presentation_selector {
+                mapped += 1;
+                continue;
+            }
+
+            if c.contributes_to_shaping {
+                if *g != 0 {
                     mapped += 1;
                 }
+                continue;
+            }
+
+            if self.map_len == 1 {
+                mapped += 1;
             }
         }
         let ratio = mapped as f32 / self.map_len as f32;

parley/src/analysis/mod.rs

@@ -13,9 +13,13 @@ use icu_normalizer::properties::{
     CanonicalComposition, CanonicalCompositionBorrowed, CanonicalDecomposition,
     CanonicalDecompositionBorrowed,
 };
-use icu_properties::props::{BidiMirroringGlyph, GeneralCategory, GraphemeClusterBreak, Script};
+use icu_properties::props::{
+    BidiMirroringGlyph, EmojiModifier, EmojiModifierBase, EmojiPresentation, GeneralCategory,
+    GraphemeClusterBreak, Script,
+};
 use icu_properties::{
-    CodePointMapData, CodePointMapDataBorrowed, PropertyNamesShort, PropertyNamesShortBorrowed,
+    CodePointMapData, CodePointMapDataBorrowed, CodePointSetData, CodePointSetDataBorrowed,
+    PropertyNamesShort, PropertyNamesShortBorrowed,
 };
 use icu_segmenter::options::{LineBreakOptions, LineBreakWordOption, WordBreakInvariantOptions};
 use icu_segmenter::{
@@ -92,6 +96,21 @@ impl AnalysisDataSources {
     fn brackets(&self) -> CodePointMapDataBorrowed<'_, BidiMirroringGlyph> {
         const { CodePointMapData::new() }
     }
+
+    #[inline(always)]
+    pub(crate) fn emoji_modifier(&self) -> CodePointSetDataBorrowed<'_> {
+        const { CodePointSetData::new::<EmojiModifier>() }
+    }
+
+    #[inline(always)]
+    pub(crate) fn emoji_modifier_base(&self) -> CodePointSetDataBorrowed<'_> {
+        const { CodePointSetData::new::<EmojiModifierBase>() }
+    }
+
+    #[inline(always)]
+    pub(crate) fn emoji_presentation(&self) -> CodePointSetDataBorrowed<'_> {
+        const { CodePointSetData::new::<EmojiPresentation>() }
+    }
 }
 
 #[derive(Copy, Clone, Debug, PartialEq, Eq)]