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/read.rs

@@ -38,8 +38,6 @@ pub struct Imp<'a> {
     pub vml_previews: HashMap<String, HashMap<String, String>>,
     /// Aggregate byte count of retained chart sidecars across the presentation.
     pub kept_sidecar_bytes: usize,
-    /// Data slots of the charts decoded so far (see `chart::MAX_PRESENTATION_CHART_POINTS`).
-    pub chart_points: usize,
     /// Cached kept sidecars by normalized chart part name.
     pub kept_charts: HashMap<String, deckcraft_model::chart::ChartKept>,
     /// Add-in tags parts, parsed once each, by normalized part name.
@@ -444,7 +442,6 @@ pub fn import_theme(bytes: &[u8]) -> Result<(deckcraft_model::theme::Theme, Vec<
         table_style_ids: Vec::new(),
         vml_previews: HashMap::new(),
         kept_sidecar_bytes: 0,
-        chart_points: 0,
         kept_charts: HashMap::new(),
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
@@ -486,7 +483,6 @@ pub fn import_checked(bytes: &[u8]) -> Result<(Presentation, Vec<String>), PptxE
         table_style_ids: Vec::new(),
         vml_previews: HashMap::new(),
         kept_sidecar_bytes: 0,
-        chart_points: 0,
         kept_charts: HashMap::new(),
         tags_by_part: HashMap::new(),
         tag_bytes: 0,

crates/pptx/src/read/chart.rs

@@ -39,24 +39,6 @@ pub(crate) fn rich_text(t: &El) -> Option<String> {
 /// Most category/value/size slots one chart may decode across all its series and caches, so
 /// sparse `ptCount`/`idx` declarations can't expand far beyond the XML they came from.
 pub(crate) const MAX_CHART_POINTS: usize = 1_000_000;
-/// Data slots (values and categories) the charts of one presentation may hold in all; a chart past
-/// it is read as a placeholder frame. Each frame that references a chart decodes its own copy.
-pub(crate) const MAX_PRESENTATION_CHART_POINTS: usize = 4_000_000;
-
-/// Slots a decoded chart holds, counting the chart itself as at least one.
-pub(crate) fn point_count(c: &Chart) -> usize {
-    let series = |s: &deckcraft_model::chart::Series| {
-        let bars = s.error_bars.as_ref().map_or(0, |e| e.plus.len().saturating_add(e.minus.len()));
-        s.values.len().saturating_add(s.x_values.len()).saturating_add(s.sizes.len()).saturating_add(bars)
-    };
-    let levels: usize = c.category_levels.iter().map(Vec::len).sum();
-    1usize.saturating_add(c.categories.len()).saturating_add(levels).saturating_add(c.series.iter().map(series).fold(0, usize::saturating_add))
-}
-
-/// Bytes of the strings a kept relationship owns.
-fn rel_bytes(r: &deckcraft_model::chart::ChartRel) -> usize {
-    r.id.len() + r.rel_type.len() + r.target.len() + r.part.len()
-}
 
 /// Most slots one cache may hold.
 const MAX_CACHE_POINTS: usize = 100_000;
@@ -484,13 +466,6 @@ pub(crate) fn capture_kept(imp: &mut super::Imp, chart_part: &str) -> deckcraft_
 
     let key = crate::opc::norm(chart_part);
     if let Some(cached) = imp.kept_charts.get(&key) {
-        // Every frame that shows this chart gets its own copy, so each copy is charged like the first.
-        let bytes: usize = cached.parts.iter().map(|p| p.bytes.len()).sum::<usize>() + cached.rels.iter().map(rel_bytes).sum::<usize>();
-        if imp.kept_sidecar_bytes.saturating_add(bytes) > MAX_KEPT_TOTAL_BYTES {
-            log::warn!("pptx: chart sidecar retained byte budget exceeded for {chart_part}");
-            return ChartKept::default();
-        }
-        imp.kept_sidecar_bytes += bytes;
         return cached.clone();
     }
 
@@ -510,13 +485,6 @@ pub(crate) fn capture_kept(imp: &mut super::Imp, chart_part: &str) -> deckcraft_
             kept.rels.push(ChartRel { id: r.id, rel_type: r.ty, target: r.target_raw, part: r.target, external: false });
         }
     }
-    // The first capture's relationship strings are retained too (the parts were charged as copied).
-    let rel_total: usize = kept.rels.iter().map(rel_bytes).sum();
-    if imp.kept_sidecar_bytes.saturating_add(rel_total) > MAX_KEPT_TOTAL_BYTES {
-        log::warn!("pptx: chart sidecar retained byte budget exceeded for {chart_part}");
-        return ChartKept::default();
-    }
-    imp.kept_sidecar_bytes += rel_total;
     imp.kept_charts.insert(key, kept.clone());
     kept
 }
@@ -622,16 +590,6 @@ fn text_size(e: Option<&El>) -> Option<f64> {
 mod tests {
     use super::*;
 
-    #[test]
-    fn point_count_charges_every_retained_vector() {
-        let mut c = Chart { categories: vec!["a".into()], category_levels: vec![vec!["x".into(); 2], vec!["y".into(); 3]], ..Default::default() };
-        let mut s =
-            deckcraft_model::chart::Series { values: vec![Some(1.0); 4], x_values: vec![None; 5], sizes: vec![None; 6], ..Default::default() };
-        s.error_bars = Some(deckcraft_model::chart::ErrorBars { plus: vec![None; 7], minus: vec![None; 8], ..Default::default() });
-        c.series = vec![s];
-        assert_eq!(point_count(&c), 1 + 1 + 5 + 4 + 5 + 6 + 15);
-    }
-
     #[test]
     fn sparse_caches_share_one_chart_wide_budget() {
         // Every series declares 100000 points but writes one: without a chart-wide budget each

crates/pptx/src/read/shapes.rs

@@ -428,14 +428,10 @@ fn frame(imp: &mut Imp, part: &Part, ctx: &mut IdCtx, e: &El, depth: usize) -> S
             return s;
         }
         if let Some(c) = gd.child("chart")
-            && imp.chart_points < super::chart::MAX_PRESENTATION_CHART_POINTS
             && let Some(rel) = part.rels.get(c.attr("r:id").unwrap_or(""))
             && let Some(bytes) = imp.pkg.xml_bytes(&rel.target)
             && let Some(mut chart) = if uri.ends_with("/chartex") { super::read_chartex_xml(bytes) } else { super::read_chart_xml(bytes) }
-            // A chart that would cross the presentation limit is read as the placeholder frame instead.
-            && imp.chart_points.saturating_add(super::chart::point_count(&chart)) <= super::chart::MAX_PRESENTATION_CHART_POINTS
         {
-            imp.chart_points = imp.chart_points.saturating_add(super::chart::point_count(&chart));
             chart.kept = super::chart::capture_kept(imp, &rel.target);
             s.kind = ShapeKind::Chart(Box::new(chart));
             return s;