pptx: copy only used, bounded namespace declarations onto imported fragments

crates/engine/src/compare.rs

@@ -121,10 +121,9 @@ pub fn diff(cur: &Presentation, rev: &Presentation, by_position: bool) -> Vec<Ch
     let cur_matched: HashSet<usize> = pairs.iter().map(|p| p.0).collect();
     let rev_matched: HashSet<usize> = pairs.iter().map(|p| p.1).collect();
     let mut out = vec![];
-    let mut media = Media::new(cur, rev);
     for (i, j) in &pairs {
         let (Some(a), Some(b)) = (cur.slides.get(*i), rev.slides.get(*j)) else { continue };
-        let details = slide_details(cur, a, rev, b, &mut media);
+        let details = slide_details(cur, a, rev, b);
         if !details.is_empty() {
             out.push(Change { kind: ChangeKind::Changed, slide: *i, number: i.saturating_add(1), only_in: None, revised: Some(*j), details });
         }
@@ -141,18 +140,10 @@ pub fn diff(cur: &Presentation, rev: &Presentation, by_position: bool) -> Vec<Ch
             });
         }
     }
-    // Each revised slide's open match (the first pair naming it).
-    let mut rev_to_cur: HashMap<usize, usize> = HashMap::new();
-    for p in &pairs {
-        rev_to_cur.entry(p.1).or_insert(p.0);
-    }
-    // After the open slide that matches the nearest matched revised slide before it.
-    let mut at = 0;
     for (j, s) in rev.slides.iter().enumerate() {
-        if let Some(i) = rev_to_cur.get(&j) {
-            at = i + 1;
-        }
         if !rev_matched.contains(&j) {
+            // After the open slide that matches the revised slide before it.
+            let at = (0..j).rev().find_map(|k| pairs.iter().find(|p| p.1 == k).map(|p| p.0 + 1)).unwrap_or(0);
             out.push(Change {
                 kind: ChangeKind::Inserted,
                 slide: at,
@@ -183,34 +174,8 @@ fn shape_label(s: &Shape) -> String {
     if s.name.trim().is_empty() { format!("Shape {}", s.id.0) } else { s.name.clone() }
 }
 
-/// Both decks' media bytes by id, and which picture pairs were already compared: a deck can
-/// show one picture many times, and its bytes are compared once per pair.
-struct Media<'a> {
-    cur: HashMap<MediaId, &'a [u8]>,
-    rev: HashMap<MediaId, &'a [u8]>,
-    same: HashMap<(MediaId, MediaId), bool>,
-}
-
-impl<'a> Media<'a> {
-    fn new(cur: &'a Presentation, rev: &'a Presentation) -> Self {
-        // The first item with an id wins, as in `Presentation::media`.
-        let index = |p: &'a Presentation| {
-            let mut m = HashMap::new();
-            for x in &p.media {
-                m.entry(x.id).or_insert(x.data.as_slice());
-            }
-            m
-        };
-        Media { cur: index(cur), rev: index(rev), same: HashMap::new() }
-    }
-    /// Whether the pictures (by media id, `None` for a shape that is not a picture) differ.
-    fn differ(&mut self, a: Option<MediaId>, b: Option<MediaId>) -> bool {
-        let (ba, bb) = (a.and_then(|x| self.cur.get(&x)), b.and_then(|y| self.rev.get(&y)));
-        match (a, b) {
-            (Some(x), Some(y)) if ba.is_some() && bb.is_some() => !*self.same.entry((x, y)).or_insert_with(|| ba == bb),
-            _ => ba.is_some() || bb.is_some(),
-        }
-    }
+fn media_bytes(p: &Presentation, id: MediaId) -> Option<&[u8]> {
+    p.media(id).map(|m| m.data.as_slice())
 }
 
 /// A shape without its ids (shape and media), for comparing across decks.
@@ -239,7 +204,7 @@ fn turn(ctx: Option<&Ctx>, s: &Shape) -> (f64, bool, bool) {
     drawn_box(ctx, s).map_or((0.0, false, false), |x| (x.rot, x.flip_h, x.flip_v))
 }
 
-fn shape_details(ca: Option<&Ctx>, a: &Shape, cb: Option<&Ctx>, b: &Shape, media: &mut Media, out: &mut Vec<String>) {
+fn shape_details(cp: &Presentation, ca: Option<&Ctx>, a: &Shape, rp: &Presentation, cb: Option<&Ctx>, b: &Shape, out: &mut Vec<String>) {
     let label = shape_label(b);
     let text = |s: &Shape| s.text.as_ref().map(|t| t.text()).unwrap_or_default();
     let mut what = vec![];
@@ -269,11 +234,11 @@ fn shape_details(ca: Option<&Ctx>, a: &Shape, cb: Option<&Ctx>, b: &Shape, media
     if a.fill != b.fill || a.line != b.line || a.style != b.style || a.effects != b.effects {
         what.push("formatting");
     }
-    let pic = |s: &Shape| match &s.kind {
-        ShapeKind::Picture { fill } => Some(fill.media),
+    let pic = |p: &Presentation, s: &Shape| match &s.kind {
+        ShapeKind::Picture { fill } => media_bytes(p, fill.media).map(|b| b.to_vec()),
         _ => None,
     };
-    if media.differ(pic(a), pic(b)) {
+    if pic(cp, a) != pic(rp, b) {
         what.push("picture");
     }
     // Anything the checks above do not cover (alt text, hidden, a link…) is reported too, also
@@ -300,32 +265,21 @@ fn shape_details(ca: Option<&Ctx>, a: &Shape, cb: Option<&Ctx>, b: &Shape, media
     }
 }
 
-fn slide_details(cp: &Presentation, a: &Slide, rp: &Presentation, b: &Slide, media: &mut Media) -> Vec<String> {
+fn slide_details(cp: &Presentation, a: &Slide, rp: &Presentation, b: &Slide) -> Vec<String> {
     let mut out = vec![];
     let mut used: HashSet<usize> = HashSet::new();
     let (ca, cb) = (Ctx::for_slide(cp, a), Ctx::for_slide(rp, b));
-    // The revised shapes by id (first wins) and by name (in order, with a cursor past used ones).
-    let mut by_id: HashMap<ShapeId, usize> = HashMap::new();
-    let mut by_name: HashMap<&str, (Vec<usize>, usize)> = HashMap::new();
-    for (k, x) in b.shapes.iter().enumerate() {
-        by_id.entry(x.id).or_insert(k);
-        if !x.name.is_empty() {
-            by_name.entry(x.name.as_str()).or_default().0.push(k);
-        }
-    }
     for sa in &a.shapes {
-        let k = by_id.get(&sa.id).copied().or_else(|| {
-            let (ks, next) = by_name.get_mut(sa.name.as_str())?;
-            while ks.get(*next).is_some_and(|k| used.contains(k)) {
-                *next += 1;
-            }
-            ks.get(*next).copied()
-        });
+        let k = b
+            .shapes
+            .iter()
+            .position(|x| x.id == sa.id)
+            .or_else(|| b.shapes.iter().enumerate().position(|(k, x)| !used.contains(&k) && x.name == sa.name && !x.name.is_empty()));
         match k {
             Some(k) if !used.contains(&k) => {
                 used.insert(k);
                 if let Some(sb) = b.shapes.get(k) {
-                    shape_details(ca.as_ref(), sa, cb.as_ref(), sb, media, &mut out);
+                    shape_details(cp, ca.as_ref(), sa, rp, cb.as_ref(), sb, &mut out);
                 }
             }
             _ => out.push(format!("{} deleted (only in the deck)", shape_label(sa))),
@@ -380,34 +334,19 @@ pub fn adopt(doc: &mut Presentation, rev: &Presentation, slide: &Slide, replacin
     let mut sl = slide.clone();
     // Media.
     let mut media_map: HashMap<MediaId, MediaId> = HashMap::new();
-    // Both decks' media by id (the first item with an id wins, as in `Presentation::media`),
-    // looked up once per reference instead of searched.
-    let mut rev_at: HashMap<MediaId, usize> = HashMap::new();
-    for (i, m) in rev.media.iter().enumerate() {
-        rev_at.entry(m.id).or_insert(i);
-    }
-    let mut doc_at: HashMap<MediaId, usize> = HashMap::new();
-    for (i, m) in doc.media.iter().enumerate() {
-        doc_at.entry(m.id).or_insert(i);
-    }
     for id in crate::media_refs::slide_media(&sl) {
-        let Some(m) = rev_at.get(&id).and_then(|&i| rev.media.get(i)) else { continue };
-        match doc_at.get(&id).and_then(|&i| doc.media.get(i)) {
+        let Some(m) = rev.media(id) else { continue };
+        match doc.media(id) {
             Some(x) if x.data == m.data => {}
             None => {
                 // Keeps the revision's id, so later ids must start past it.
-                doc_at.insert(id, doc.media.len());
                 doc.media.push(m.clone());
                 doc.next_id = doc.next_id.max(id.0);
             }
             Some(_) => {
-                let before = doc.media.len();
                 let new = doc
                     .add_media(&m.name, &m.content_type, m.data.as_ref().clone())
                     .ok_or_else(|| crate::cmd::bad("review.accept", crate::cmd::NO_ID_SPACE))?;
-                if doc.media.len() > before {
-                    doc_at.entry(new).or_insert(before);
-                }
                 media_map.insert(id, new);
             }
         }
@@ -580,39 +519,6 @@ mod tests {
         assert_eq!(s.doc().unwrap().doc.slides.len(), n, "comparing alone changes nothing");
     }
 
-    #[test]
-    fn many_inserted_slides_and_same_named_pictures_compare_by_index() {
-        use std::sync::Arc;
-
-        use deckcraft_model::style::PictureFill;
-        use deckcraft_model::{Shape, ShapeId, ShapeKind, SlideId, defaults};
-
-        // One slide of 3,000 pictures of one image, all named "Box"; the revision has the same
-        // pictures under other ids (matched by name, in order) and 3,000 slides inserted after it.
-        let mut cur = defaults::new_presentation(None);
-        let img = cur.add_media("a.png", "image/png", vec![7; 4096]).unwrap();
-        let mut rev = cur.clone();
-        let pic = |id: u32| Shape {
-            id: ShapeId(id),
-            name: "Box".into(),
-            xfrm: Some(deckcraft_model::Xfrm::new(10.0, 10.0, 50.0, 50.0)),
-            kind: ShapeKind::Picture { fill: PictureFill { media: img, ..Default::default() } },
-            ..Default::default()
-        };
-        Arc::make_mut(&mut cur.slides[0]).shapes = (0..3000).map(|k| pic(100_000 + k)).collect();
-        Arc::make_mut(&mut rev.slides[0]).shapes = (0..3000).map(|k| pic(200_000 + k)).collect();
-        let mut base = (*rev.slides[0]).clone();
-        base.shapes.clear();
-        for k in 0..3000 {
-            let mut s = base.clone();
-            s.id = SlideId(300_000 + k);
-            rev.slides.push(Arc::new(s));
-        }
-        let changes = super::diff(&cur, &rev, false);
-        assert_eq!(changes.len(), 3000, "only the inserted slides differ");
-        assert!(changes.iter().all(|c| c.kind == super::ChangeKind::Inserted && c.slide == 1));
-    }
-
     /// A deck with one slide: a table whose cell has a picture fill, a picture background and a
     /// transition sound. `tag` starts the bytes of each.
     fn media_deck(tag: &str) -> deckcraft_model::Presentation {

crates/pptx/src/read.rs

@@ -100,8 +100,7 @@ impl Part {
     }
     /// Like [`Part::keep`], also declaring prefixes that only an enclosing wrapper binds.
     pub fn keep_inner(&self, el: &El) -> String {
-        let ns: Vec<(String, String)> = self.ns.iter().chain(&self.inner_ns).cloned().collect();
-        keep_xml(el, &ns)
+        keep_with(el, &[&self.ns, &self.inner_ns])
     }
 }
 
@@ -136,36 +135,58 @@ fn inner_decls(root: &El, ns: &[(String, String)]) -> Vec<(String, String)> {
     out
 }
 
-/// 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) {
+/// 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) {
         if depth > crate::xml::MAX_DEPTH {
             return;
         }
         if let Some((p, _)) = e.name.split_once(':') {
-            used.insert(p.to_string());
+            used.insert(p);
         }
-        for (k, _) in &e.attrs {
-            if let Some((p, _)) = k.split_once(':')
+        for (k, v) in &e.attrs {
+            if let Some((p, local)) = k.split_once(':')
                 && p != "xmlns"
                 && p != "xml"
             {
-                used.insert(p.to_string());
+                used.insert(p);
+                if local == "Ignorable" {
+                    used.extend(v.split_whitespace());
+                }
             }
         }
         for c in e.elements() {
             rec(c, used, depth + 1);
         }
     }
+    let mut used = HashSet::new();
     rec(el, &mut used, 0);
-    let mut el = el.clone();
-    for (k, v) in ns {
+    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 p = k.strip_prefix("xmlns:").unwrap_or("");
-        if !p.is_empty() && used.contains(p) && !el.attrs.iter().any(|(a, _)| a == k) {
-            el.attrs.push((k.clone(), v.clone()));
+        if !p.is_empty() && used.contains(p) && v.len() <= MAX_NS_VALUE && have.insert(k) {
+            extra.push((k.clone(), v.clone()));
         }
     }
+    let mut el = el.clone();
+    el.attrs.extend(extra);
     el.to_xml()
 }
 
@@ -315,9 +336,12 @@ 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)) {
-                    if let Some(x) = text(&sub.target)
-                        && !out.iter().any(|k| k.name.eq_ignore_ascii_case(&sub.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) {
                         out.push(KeptPart {
                             content_type: pkg.content_type(&sub.target).unwrap_or_default(),
                             name: sub.target.clone(),
@@ -365,13 +389,19 @@ 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 {
-                                let mut e = e.clone();
-                                // Carry namespace declarations from the branch onto its content.
+                                // 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();
                                 for (k, v) in b.attrs.iter().chain(ac.attrs.iter()) {
-                                    if k.starts_with("xmlns") && !e.attrs.iter().any(|(a, _)| a == k) {
-                                        e.attrs.push((k.clone(), v.clone()));
+                                    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()));
                                     }
                                 }
+                                let mut e = e.clone();
+                                e.attrs.extend(extra);
                                 out.push(Node::El(e));
                             }
                         }
@@ -1414,3 +1444,36 @@ 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");
+    }
+}