address review: share the first inserted dictionary on a concurrent load

crates/engine/src/spell.rs

@@ -171,7 +171,9 @@ impl Dictionary {
                 ["COMPOUNDWORDMAX", n, ..] => d.compound_max = n.parse().unwrap_or(0),
                 ["REP", n] => {
                     for _ in 0..n.parse::<usize>().unwrap_or(0) {
-                        let r: Vec<&str> = lines.next().unwrap_or("").split_whitespace().collect();
+                        // The count is the file's claim: stop when the records run out.
+                        let Some(line) = lines.next() else { break };
+                        let r: Vec<&str> = line.split_whitespace().collect();
                         if let [_, a, b, ..] = r.as_slice() {
                             d.rep.push((a.replace('_', " "), b.replace('_', " ")));
                         }
@@ -182,7 +184,8 @@ impl Dictionary {
                     let Some(flag) = one_flag(Some(flag), mode) else { continue };
                     let cross = cross.eq_ignore_ascii_case("y");
                     for _ in 0..n.parse::<usize>().unwrap_or(0) {
-                        let r: Vec<&str> = lines.next().unwrap_or("").split_whitespace().collect();
+                        let Some(line) = lines.next() else { break };
+                        let r: Vec<&str> = line.split_whitespace().collect();
                         let [_, _, strip, add, rest @ ..] = r.as_slice() else { continue };
                         let (add, cont) = match add.split_once('/') {
                             Some((a, c)) => (a, parse_flags(c, mode)),
@@ -576,6 +579,15 @@ fn cache() -> &'static Cache {
     CACHE.get_or_init(Default::default)
 }
 
+/// Language strings remembered by [`dictionary`]; a string past it is resolved again on each call.
+const MAX_CACHED_LANGS: usize = 256;
+
+/// Parsed dictionaries by their affix file, so aliases of one language share it.
+fn by_files() -> &'static Mutex<HashMap<PathBuf, Arc<Dictionary>>> {
+    static M: OnceLock<Mutex<HashMap<PathBuf, Arc<Dictionary>>>> = OnceLock::new();
+    M.get_or_init(Default::default)
+}
+
 /// The dictionary for `lang` (a BCP-47 tag such as `en-US`), loaded once per process.
 pub fn dictionary(lang: &str) -> Option<Arc<Dictionary>> {
     let cache = cache();
@@ -586,10 +598,23 @@ pub fn dictionary(lang: &str) -> Option<Arc<Dictionary>> {
         return d.clone();
     }
     let loaded = find_in(&search_dirs(), lang).and_then(|(a, d)| {
-        let (a, d) = (std::fs::read(a).ok()?, std::fs::read(d).ok()?);
-        Some(Arc::new(Dictionary::parse(&a, &d)))
+        // Tags that resolve to the same files share one parsed dictionary.
+        if let Ok(m) = by_files().lock()
+            && let Some(dict) = m.get(&a)
+        {
+            return Some(dict.clone());
+        }
+        let (bytes_a, bytes_d) = (std::fs::read(&a).ok()?, std::fs::read(d).ok()?);
+        let dict = Arc::new(Dictionary::parse(&bytes_a, &bytes_d));
+        // A concurrent loader of the same files may have inserted first: share its copy.
+        if let Ok(mut m) = by_files().lock() {
+            return Some(m.entry(a).or_insert(dict).clone());
+        }
+        Some(dict)
     });
-    if let Ok(mut c) = cache.lock() {
+    if let Ok(mut c) = cache.lock()
+        && c.len() < MAX_CACHED_LANGS
+    {
         c.insert(key, loaded.clone());
     }
     loaded
@@ -880,6 +905,14 @@ SFX D Y 1\nSFX D 0 ed/S .\n";
         assert!(d.suggest("workingslides", 5).contains(&"working slides".to_string()));
     }
 
+    /// An affix header can claim billions of records; parsing stops where the file does.
+    #[test]
+    fn a_header_count_past_the_records_ends_with_them() {
+        let aff = b"SFX S Y 4000000000\nSFX S 0 s .\nREP 4000000000\nREP a b\n";
+        let d = Dictionary::parse(aff, b"1\ncat/S\n".as_slice());
+        assert!(d.check("cats"));
+    }
+
     /// #1175: compound splitting is quadratic in the word's length, so a word longer than
     /// `MAX_CHECK_CHARS` is not checked; shorter ones still are.
     #[test]

crates/pptx/src/custgeom.rs

@@ -15,10 +15,6 @@ use crate::xml::{A, El, W};
 
 const DEG: f64 = 60_000.0;
 
-/// Distinct path coordinate spaces whose guides are evaluated for one custom geometry; paths in
-/// further spaces are left out.
-const MAX_SPACES: usize = 16;
-
 /// Guide variables for one shape.
 pub struct Guides {
     vars: HashMap<String, f64>,
@@ -155,26 +151,14 @@ fn pt(g: &Guides, e: &El) -> (f64, f64) {
 pub fn read_cust_geom(cg: &El, w_emu: f64, h_emu: f64) -> Vec<CustomPath> {
     let mut out = vec![];
     let Some(pl) = cg.child("pathLst") else { return out };
-    let mut spaces: HashMap<(u64, u64), Guides> = HashMap::new();
-    // The cap on cached spaces saves guide evaluation: a geometry without guides has none to save.
-    let guided = ["avLst", "gdLst"].iter().any(|n| cg.child(n).is_some_and(|l| l.elements().next().is_some()));
     for p in pl.children_named("path").take(1000) {
         let pw = p.f64("w").unwrap_or(0.0);
         let ph = p.f64("h").unwrap_or(0.0);
         // Guides are evaluated in the path's own space when it has one.
         let (gw, gh) = if pw > 0.0 || ph > 0.0 { (pw, ph) } else { (w_emu, h_emu) };
-        // The guide lists are evaluated once per coordinate space, not once per path.
-        let key = (gw.to_bits(), gh.to_bits());
-        if !spaces.contains_key(&key) {
-            if guided && spaces.len() >= MAX_SPACES {
-                continue;
-            }
-            let mut g = Guides::new(gw, gh);
-            g.eval_list(cg.child("avLst"));
-            g.eval_list(cg.child("gdLst"));
-            spaces.insert(key, g);
-        }
-        let Some(g) = spaces.get(&key) else { continue };
+        let mut g = Guides::new(gw, gh);
+        g.eval_list(cg.child("avLst"));
+        g.eval_list(cg.child("gdLst"));
         let mut d = String::new();
         let mut cur = (0.0, 0.0);
         let mut start = (0.0, 0.0);
@@ -182,7 +166,7 @@ pub fn read_cust_geom(cg: &El, w_emu: f64, h_emu: f64) -> Vec<CustomPath> {
             match c.local() {
                 "moveTo" => {
                     if let Some(q) = c.child("pt") {
-                        cur = pt(g, q);
+                        cur = pt(&g, q);
                         start = cur;
                         d.push_str("M ");
                         num(&mut d, cur.0);
@@ -193,7 +177,7 @@ pub fn read_cust_geom(cg: &El, w_emu: f64, h_emu: f64) -> Vec<CustomPath> {
                 }
                 "lnTo" => {
                     if let Some(q) = c.child("pt") {
-                        cur = pt(g, q);
+                        cur = pt(&g, q);
                         d.push_str("L ");
                         num(&mut d, cur.0);
                         d.push(' ');
@@ -202,7 +186,7 @@ pub fn read_cust_geom(cg: &El, w_emu: f64, h_emu: f64) -> Vec<CustomPath> {
                     }
                 }
                 "cubicBezTo" => {
-                    let pts: Vec<(f64, f64)> = c.children_named("pt").map(|q| pt(g, q)).collect();
+                    let pts: Vec<(f64, f64)> = c.children_named("pt").map(|q| pt(&g, q)).collect();
                     if let [a, b, e] = pts.as_slice() {
                         d.push_str("C ");
                         for (x, y) in [a, b, e] {
@@ -215,7 +199,7 @@ pub fn read_cust_geom(cg: &El, w_emu: f64, h_emu: f64) -> Vec<CustomPath> {
                     }
                 }
                 "quadBezTo" => {
-                    let pts: Vec<(f64, f64)> = c.children_named("pt").map(|q| pt(g, q)).collect();
+                    let pts: Vec<(f64, f64)> = c.children_named("pt").map(|q| pt(&g, q)).collect();
                     if let [a, e] = pts.as_slice() {
                         d.push_str("Q ");
                         for (x, y) in [a, e] {
@@ -486,29 +470,6 @@ mod tests {
         assert_eq!(paths[1].d, "M 1 1 Q 2 2 3 1");
     }
 
-    #[test]
-    fn paths_share_guides_and_distinct_spaces_are_capped() {
-        let path = |w: usize| format!(r#"<a:path w="{w}" h="10"><a:moveTo><a:pt x="mid" y="1"/></a:moveTo></a:path>"#);
-        let paths: String = (1..=40).map(|i| path(10 * i)).collect();
-        let x = format!(
-            r#"<a:custGeom xmlns:a="x"><a:gdLst><a:gd name="mid" fmla="*/ w 1 2"/></a:gdLst><a:pathLst>{paths}{}</a:pathLst></a:custGeom>"#,
-            path(10)
-        );
-        let d = parse(x.as_bytes()).unwrap();
-        let out = read_cust_geom(&d.root, 1.0, 1.0);
-        assert_eq!(out.len(), MAX_SPACES + 1, "16 spaces plus the repeat of the first");
-        assert!(out[0].d.starts_with("M 5 1"), "{}", out[0].d);
-        assert!(out[MAX_SPACES].d.starts_with("M 5 1"), "{}", out[MAX_SPACES].d);
-    }
-
-    #[test]
-    fn guide_free_geometries_are_not_capped_by_space() {
-        let paths: String = (1..=40).map(|i| format!(r#"<a:path w="{}" h="10"><a:moveTo><a:pt x="1" y="1"/></a:moveTo></a:path>"#, 10 * i)).collect();
-        let x = format!(r#"<a:custGeom xmlns:a="x"><a:avLst/><a:gdLst/><a:pathLst>{paths}</a:pathLst></a:custGeom>"#);
-        let d = parse(x.as_bytes()).unwrap();
-        assert_eq!(read_cust_geom(&d.root, 1.0, 1.0).len(), 40);
-    }
-
     #[test]
     fn write_then_read_path() {
         let p = CustomPath { w: 100.0, h: 50.0, d: "M 0 0 L 100 0 C 100 10 90 50 50 50 Q 0 50 0 0 Z".into(), fill: FillMode::Norm, stroke: true };