engine: index font names and embedded faces instead of rescanning

crates/engine/src/fonts.rs

@@ -1,6 +1,8 @@
 //! Fonts a presentation uses that this machine doesn't have, so opening it can say so (its text
 //! draws in a stand-in; see `FontDb::document_face`).
 
+use std::collections::HashSet;
+
 use deckcraft_fonts::FontDb;
 use deckcraft_model::resolve::{self, Ctx};
 use deckcraft_model::text::{Bullet, TextBody};
@@ -12,8 +14,13 @@ use deckcraft_model::{Presentation, Shape, ShapeKind};
 /// font to load and are left for the round trip.
 pub fn install_embedded_fonts(p: &Presentation) {
     let db = FontDb::global();
+    // Faces can name one media item many times; each is copied and enumerated once.
+    let mut seen = HashSet::new();
     for font in &p.embedded_fonts {
         for id in font.faces().into_iter().flatten() {
+            if !seen.insert(id) {
+                continue;
+            }
             let Some(item) = p.media(id) else { continue };
             let Some(sfnt) = deckcraft_fonts::embedded_sfnt(&item.data) else { continue };
             db.add_font(sfnt.to_vec());
@@ -27,12 +34,8 @@ pub fn install_embedded_fonts(p: &Presentation) {
 /// and can't be installed, so they aren't reported.
 pub fn missing_fonts(p: &Presentation) -> Vec<String> {
     let db = FontDb::global();
-    let mut out: Vec<String> = Vec::new();
-    for family in used_fonts(p) {
-        if !family.starts_with('.') && !db.has_family(&family) && !out.iter().any(|f| f.eq_ignore_ascii_case(&family)) {
-            out.push(family);
-        }
-    }
+    // used_fonts is already free of case-insensitive duplicates.
+    let mut out: Vec<String> = used_fonts(p).into_iter().filter(|f| !f.starts_with('.') && !db.has_family(f)).collect();
     out.sort_by_key(|f| f.to_lowercase());
     out
 }
@@ -40,9 +43,10 @@ pub fn missing_fonts(p: &Presentation) -> Vec<String> {
 /// The families the slides' text runs and bullets are drawn in.
 pub fn used_fonts(p: &Presentation) -> Vec<String> {
     let mut out: Vec<String> = Vec::new();
+    let mut seen: HashSet<String> = HashSet::new();
     let mut add = |f: String| {
         let f = f.trim().to_string();
-        if !f.is_empty() && !out.iter().any(|o| o.eq_ignore_ascii_case(&f)) {
+        if !f.is_empty() && seen.insert(f.to_ascii_lowercase()) {
             out.push(f);
         }
     };

crates/engine/src/tests.rs

@@ -6458,3 +6458,24 @@ fn master_body_placeholder_format_keeps_per_level_formatting() {
     assert_eq!((level(&s, 0).run.size, level(&s, 1).run.size), (Some(28.0), Some(24.0)));
     assert_eq!(shown(&s), vec![Some(28.0), Some(24.0)]);
 }
+
+#[test]
+fn many_distinct_fonts_are_listed_once_each_without_a_quadratic_scan() {
+    // 30,000 runs in 15,000 families, each written twice in different case.
+    let mut p = deckcraft_model::Presentation::default();
+    let slide = std::sync::Arc::make_mut(&mut p.slides[0]);
+    let mut body = deckcraft_model::text::TextBody::from_text("x");
+    let run = body.paragraphs[0].runs[0].clone();
+    body.paragraphs[0].runs = (0..30_000)
+        .map(|i| {
+            let mut r = run.clone();
+            let name = format!("Qqq Font {}", i / 2);
+            r.props.font = Some(if i % 2 == 0 { name } else { name.to_uppercase() });
+            r
+        })
+        .collect();
+    slide.shapes[0].text = Some(body);
+    let used = crate::fonts::used_fonts(&p);
+    assert_eq!(used.iter().filter(|f| f.starts_with("Qqq Font ")).count(), 15_000);
+    assert_eq!(crate::fonts::missing_fonts(&p).iter().filter(|f| f.to_ascii_lowercase().starts_with("qqq font ")).count(), 15_000);
+}

crates/pptx/src/read.rs

@@ -100,7 +100,8 @@ impl Part {
     }
     /// Like [`Part::keep`], also declaring prefixes that only an enclosing wrapper binds.
     pub fn keep_inner(&self, el: &El) -> String {
-        keep_with(el, &[&self.ns, &self.inner_ns])
+        let ns: Vec<(String, String)> = self.ns.iter().chain(&self.inner_ns).cloned().collect();
+        keep_xml(el, &ns)
     }
 }
 
@@ -135,58 +136,36 @@ fn inner_decls(root: &El, ns: &[(String, String)]) -> Vec<(String, String)> {
     out
 }
 
-/// Longest namespace name copied onto a kept fragment or a resolved `mc` branch. Real ones are
-/// URIs of well under 100 bytes; a longer one is not copied, so one declaration cannot be
-/// multiplied by every fragment that uses its prefix.
-const MAX_NS_VALUE: usize = 2048;
-
-/// Prefixes `el` and its descendants use in element and attribute names, and those listed in
-/// `mc:Ignorable` (which must stay declared for the markup-compatibility rules to hold).
-fn used_prefixes(el: &El) -> HashSet<&str> {
-    fn rec<'a>(e: &'a El, used: &mut HashSet<&'a str>, depth: usize) {
+/// Serialize `el`, declaring every prefix it uses that the source root declared.
+pub fn keep_xml(el: &El, ns: &[(String, String)]) -> String {
+    let mut used = HashSet::new();
+    fn rec(e: &El, used: &mut HashSet<String>, depth: usize) {
         if depth > crate::xml::MAX_DEPTH {
             return;
         }
         if let Some((p, _)) = e.name.split_once(':') {
-            used.insert(p);
+            used.insert(p.to_string());
         }
-        for (k, v) in &e.attrs {
-            if let Some((p, local)) = k.split_once(':')
+        for (k, _) in &e.attrs {
+            if let Some((p, _)) = k.split_once(':')
                 && p != "xmlns"
                 && p != "xml"
             {
-                used.insert(p);
-                if local == "Ignorable" {
-                    used.extend(v.split_whitespace());
-                }
+                used.insert(p.to_string());
             }
         }
         for c in e.elements() {
             rec(c, used, depth + 1);
         }
     }
-    let mut used = HashSet::new();
     rec(el, &mut used, 0);
-    used
-}
-
-/// Serialize `el`, declaring every prefix it uses that the source root declared.
-pub fn keep_xml(el: &El, ns: &[(String, String)]) -> String {
-    keep_with(el, &[ns])
-}
-
-fn keep_with(el: &El, decls: &[&[(String, String)]]) -> String {
-    let used = used_prefixes(el);
-    let mut have: HashSet<&str> = el.attrs.iter().map(|(a, _)| a.as_str()).collect();
-    let mut extra = Vec::new();
-    for (k, v) in decls.iter().flat_map(|d| d.iter()) {
+    let mut el = el.clone();
+    for (k, v) in ns {
         let p = k.strip_prefix("xmlns:").unwrap_or("");
-        if !p.is_empty() && used.contains(p) && v.len() <= MAX_NS_VALUE && have.insert(k) {
-            extra.push((k.clone(), v.clone()));
+        if !p.is_empty() && used.contains(p) && !el.attrs.iter().any(|(a, _)| a == k) {
+            el.attrs.push((k.clone(), v.clone()));
         }
     }
-    let mut el = el.clone();
-    el.attrs.extend(extra);
     el.to_xml()
 }
 
@@ -336,12 +315,9 @@ fn kept_parts(pkg: &Package, pres_path: &str) -> Vec<deckcraft_model::KeptPart>
             let rp = opc::rels_path(&name);
             if let Some(rels_xml) = text(&rp) {
                 for sub in pkg.rels(&name).items.iter().filter(|s| !s.external && keepable(&s.target)) {
-                    // Checked before the part is read: repeated relationships to one large part
-                    // must not copy it again each time.
-                    if out.len() >= 400 || out.iter().any(|k| k.name.eq_ignore_ascii_case(&sub.target)) {
-                        continue;
-                    }
-                    if let Some(x) = text(&sub.target) {
+                    if let Some(x) = text(&sub.target)
+                        && !out.iter().any(|k| k.name.eq_ignore_ascii_case(&sub.target))
+                    {
                         out.push(KeptPart {
                             content_type: pkg.content_type(&sub.target).unwrap_or_default(),
                             name: sub.target.clone(),
@@ -389,19 +365,13 @@ fn resolve_mc_rels(el: &mut El, depth: usize, rels: Option<&Rels>) {
                     if let Some(b) = branch {
                         for n in &b.children {
                             if let Node::El(e) = n {
-                                // Carry the namespace declarations from the branch that its content
-                                // uses (all of them per child made children x declarations copies).
-                                let used = used_prefixes(e);
-                                let mut have: HashSet<&str> = e.attrs.iter().map(|(a, _)| a.as_str()).collect();
-                                let mut extra = Vec::new();
+                                let mut e = e.clone();
+                                // Carry namespace declarations from the branch onto its content.
                                 for (k, v) in b.attrs.iter().chain(ac.attrs.iter()) {
-                                    let needed = k == "xmlns" || k.strip_prefix("xmlns:").is_some_and(|p| used.contains(p));
-                                    if needed && v.len() <= MAX_NS_VALUE && have.insert(k) {
-                                        extra.push((k.clone(), v.clone()));
+                                    if k.starts_with("xmlns") && !e.attrs.iter().any(|(a, _)| a == k) {
+                                        e.attrs.push((k.clone(), v.clone()));
                                     }
                                 }
-                                let mut e = e.clone();
-                                e.attrs.extend(extra);
                                 out.push(Node::El(e));
                             }
                         }
@@ -1444,36 +1414,3 @@ fn header_footer(p: &Presentation) -> HeaderFooter {
     }
     hf
 }
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn mc_branches_carry_only_the_declarations_their_content_uses() {
-        // 2,000 declarations on the branch and 2,000 children: each child gets the one it uses
-        // (and the default), not all 2,000.
-        let decls: String = (0..2000).map(|i| format!(r#" xmlns:n{i}="urn:n{i}""#)).collect();
-        let kids: String = (0..2000).map(|i| format!("<n{i}:x/>")).collect();
-        let x = format!(
-            r#"<r xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"><mc:AlternateContent><mc:Fallback xmlns="urn:d"{decls}>{kids}</mc:Fallback></mc:AlternateContent></r>"#
-        );
-        let mut d = crate::xml::parse(x.as_bytes()).unwrap();
-        resolve_mc(&mut d.root, 0);
-        let kids: Vec<&El> = d.root.elements().collect();
-        assert_eq!(kids.len(), 2000);
-        assert_eq!(kids[7].attrs, vec![("xmlns".to_string(), "urn:d".to_string()), ("xmlns:n7".to_string(), "urn:n7".to_string())]);
-    }
-
-    #[test]
-    fn kept_fragments_skip_oversized_namespace_names() {
-        let big = "u".repeat(MAX_NS_VALUE + 1);
-        let ns = vec![("xmlns:a".to_string(), "urn:a".to_string()), ("xmlns:b".to_string(), big)];
-        let d = crate::xml::parse(br#"<a:x xmlns:a="urn:a" xmlns:b="urn:b"><b:y/></a:x>"#).unwrap();
-        let mut el = d.root.clone();
-        el.attrs.clear();
-        let out = keep_xml(&el, &ns);
-        assert!(out.contains(r#"xmlns:a="urn:a""#), "{out}");
-        assert!(!out.contains("uuuu"), "the oversized declaration is not copied");
-    }
-}