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

@@ -22,7 +22,7 @@ use icu_segmenter::{
     GraphemeClusterSegmenter, GraphemeClusterSegmenterBorrowed, LineSegmenter,
     LineSegmenterBorrowed, WordSegmenter, WordSegmenterBorrowed,
 };
-use parley_data::Properties;
+use parley_data::{Properties, emoji::EmojiProperties};
 
 pub(crate) struct AnalysisDataSources;
 
@@ -92,6 +92,11 @@ impl AnalysisDataSources {
     fn brackets(&self) -> CodePointMapDataBorrowed<'_, BidiMirroringGlyph> {
         const { CodePointMapData::new() }
     }
+
+    #[inline(always)]
+    pub(crate) const fn emoji_properties(&self, c: char) -> EmojiProperties {
+        EmojiProperties::get(c)
+    }
 }
 
 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
@@ -106,6 +111,8 @@ pub(crate) struct CharInfo {
     pub bidi_class: icu_properties::props::BidiClass,
     /// Whether or not the character is a bracket, plus mirror data if so.
     pub bracket: BidiMirroringGlyph,
+    /// The emoji properties of this character.
+    pub emoji_properties: EmojiProperties,
 
     flags: u8,
 }
@@ -139,6 +146,7 @@ impl CharInfo {
         grapheme_cluster_break: GraphemeClusterBreak,
         bidi_class: icu_properties::props::BidiClass,
         bracket: BidiMirroringGlyph,
+        emoji_properties: EmojiProperties,
         is_variation_selector: bool,
         is_region_indicator: bool,
         is_control: bool,
@@ -152,6 +160,7 @@ impl CharInfo {
             grapheme_cluster_break,
             bidi_class,
             bracket,
+            emoji_properties,
             flags: (is_variation_selector as u8) << Self::VARIATION_SELECTOR_SHIFT
                 | (is_region_indicator as u8) << Self::REGION_INDICATOR_SHIFT
                 | (is_control as u8) << Self::CONTROL_SHIFT
@@ -429,6 +438,7 @@ pub(crate) fn analyze_text<B: Brush>(lcx: &mut LayoutContext<B>, mut text: &str)
     });
 
     let properties = |c| lcx.analysis_data_sources.properties(c);
+    let emoji_properties = |c| lcx.analysis_data_sources.emoji_properties(c);
 
     let mut needs_bidi_resolution = false;
 
@@ -448,6 +458,11 @@ pub(crate) fn analyze_text<B: Brush>(lcx: &mut LayoutContext<B>, mut text: &str)
             let is_variation_selector = properties.is_variation_selector();
             let is_region_indicator = properties.is_region_indicator();
             let next_mandatory_linebreak = properties.is_mandatory_linebreak();
+            let emoji_properties = if is_emoji_or_pictograph {
+                emoji_properties(ch)
+            } else {
+                EmojiProperties::ZERO
+            };
 
             let boundary = if is_mandatory_linebreak {
                 Boundary::Mandatory
@@ -479,6 +494,7 @@ pub(crate) fn analyze_text<B: Brush>(lcx: &mut LayoutContext<B>, mut text: &str)
                     grapheme_cluster_break,
                     bidi_class,
                     bracket,
+                    emoji_properties,
                     is_variation_selector,
                     is_region_indicator,
                     general_category == GeneralCategory::Control,