address review: keep declarations for prefixes used in QName attribute values

crates/engine/src/file_access.rs

@@ -85,20 +85,14 @@ impl AutomationRoots {
 
     /// May the file or folder at `path` be read (or written)? `Err` says why not.
     pub fn check(&self, path: &str, access: Access) -> Result<(), String> {
-        self.check_path(Path::new(path), access)
-    }
-
-    /// [`Self::check`] for a path that may not be UTF-8: the checked path is the one opened.
-    pub fn check_path(&self, path: &Path, access: Access) -> Result<(), String> {
-        let path_text = path.display();
         let roots = match access {
             Access::Read => &self.read,
             Access::Write => &self.write,
         };
         if roots.is_empty() {
             return match access {
-                Access::Read => Err(format!("automation filesystem access is not granted: read authority is absent: {path_text}")),
-                Access::Write => Err(format!("automation filesystem access is not granted: write authority is absent: {path_text}")),
+                Access::Read => Err(format!("automation filesystem access is not granted: read authority is absent: {path}")),
+                Access::Write => Err(format!("automation filesystem access is not granted: write authority is absent: {path}")),
             };
         }
         let resolved = resolve(path)?;
@@ -107,8 +101,8 @@ impl AutomationRoots {
         }
         let roots = roots.iter().map(|root| root.shown.display().to_string()).collect::<Vec<_>>().join(", ");
         match access {
-            Access::Read => Err(format!("automation path rejected: outside the read roots {roots}: {path_text}")),
-            Access::Write => Err(format!("automation path rejected: outside the write roots {roots}: {path_text}")),
+            Access::Read => Err(format!("automation path rejected: outside the read roots {roots}: {path}")),
+            Access::Write => Err(format!("automation path rejected: outside the write roots {roots}: {path}")),
         }
     }
 }
@@ -197,23 +191,17 @@ pub fn check_write(path: &str) -> Result<(), String> {
     current().map_or(Ok(()), |r| r.check(path, Access::Write))
 }
 
-/// May `path` be written here? Always, without roots in force. For a path that may not be UTF-8.
-pub fn check_write_path(path: &Path) -> Result<(), String> {
-    current().map_or(Ok(()), |r| r.check_path(path, Access::Write))
-}
-
 /// `path` made absolute with its links followed; for a path that doesn't exist (yet), its nearest
 /// existing folder's links followed and the rest appended.
-fn resolve(path: &Path) -> Result<PathBuf, String> {
-    if path.as_os_str().is_empty() {
+fn resolve(path: &str) -> Result<PathBuf, String> {
+    if path.is_empty() {
         return Err("automation path rejected: path is empty".into());
     }
-    let shown = path.display();
     // Checked as given: making `…\NUL` absolute gives the device path `\\.\NUL` on Windows.
-    if cfg!(windows) && path.components().any(|c| matches!(c, Component::Normal(n) if is_device_name(&n.to_string_lossy()))) {
-        return Err(format!("automation path rejected: reserved device names are not allowed: {shown}"));
+    if cfg!(windows) && Path::new(path).components().any(|c| matches!(c, Component::Normal(n) if is_device_name(&n.to_string_lossy()))) {
+        return Err(format!("automation path rejected: reserved device names are not allowed: {path}"));
     }
-    let absolute = std::path::absolute(path).map_err(|e| format!("automation path rejected: cannot resolve {shown}: {e}"))?;
+    let absolute = std::path::absolute(path).map_err(|e| format!("automation path rejected: cannot resolve {path}: {e}"))?;
     // The names below the nearest existing folder, last first.
     let mut missing: Vec<OsString> = vec![];
     let mut at = absolute.as_path();
@@ -224,11 +212,11 @@ fn resolve(path: &Path) -> Result<PathBuf, String> {
                 return Ok(found);
             }
             Err(e) => match std::fs::symlink_metadata(at) {
-                Ok(m) if m.file_type().is_symlink() => return Err(format!("automation path rejected: a link that leads nowhere: {shown}")),
+                Ok(m) if m.file_type().is_symlink() => return Err(format!("automation path rejected: a link that leads nowhere: {path}")),
                 // There, yet its real path can't be had (no permission): nothing to compare.
-                Ok(_) => return Err(format!("automation path rejected: cannot resolve {shown}: {e}")),
+                Ok(_) => return Err(format!("automation path rejected: cannot resolve {path}: {e}")),
                 Err(e) if e.kind() == std::io::ErrorKind::NotFound => {}
-                Err(e) => return Err(format!("automation path rejected: cannot resolve {shown}: {e}")),
+                Err(e) => return Err(format!("automation path rejected: cannot resolve {path}: {e}")),
             },
         }
         let parent = at.parent();
@@ -238,12 +226,12 @@ fn resolve(path: &Path) -> Result<PathBuf, String> {
                 at = parent;
             }
             (Some(Component::ParentDir), _) => {
-                return Err(format!("automation path rejected: parent traversal below a folder that doesn't exist: {shown}"));
+                return Err(format!("automation path rejected: parent traversal below a folder that doesn't exist: {path}"));
             }
             // Down to the root (or drive) and it doesn't exist either.
             _ => {
                 let e = std::io::Error::from(std::io::ErrorKind::NotFound);
-                return Err(format!("automation path rejected: cannot resolve {shown}: {e}"));
+                return Err(format!("automation path rejected: cannot resolve {path}: {e}"));
             }
         }
     }

crates/engine/src/spell.rs

@@ -668,8 +668,6 @@ pub fn add_user_word(word: &str) -> std::io::Result<()> {
         return Ok(());
     }
     let path = user_dictionary_path().ok_or_else(|| std::io::Error::new(std::io::ErrorKind::NotFound, "no folder for the user dictionary"))?;
-    // A confined automation client may write only inside its write roots, this file included.
-    crate::file_access::check_write_path(&path).map_err(|e| std::io::Error::new(std::io::ErrorKind::PermissionDenied, e))?;
     if let Some(d) = path.parent() {
         std::fs::create_dir_all(d)?;
     }
@@ -955,20 +953,6 @@ SFX D Y 1\nSFX D 0 ed/S .\n";
         assert_eq!(w, vec!["check", "done"]);
     }
 
-    #[test]
-    fn a_confined_client_cannot_write_the_user_dictionary_outside_its_roots() {
-        let root = std::env::temp_dir().join(format!("deckcraft-dict-root-{}", std::process::id()));
-        std::fs::create_dir_all(&root).unwrap();
-        let roots = crate::file_access::AutomationRoots::new(std::slice::from_ref(&root), std::slice::from_ref(&root)).unwrap();
-        // The user dictionary is a file of its own outside `root`, not whatever HOME points at.
-        let outside = std::env::temp_dir().join(format!("deckcraft-dict-outside-{}", std::process::id()));
-        set_user_dictionary_path(outside.join("user.dic"));
-        let r = crate::file_access::confine(roots.as_ref(), || add_user_word("zqxconfinedword"));
-        assert_eq!(r.map_err(|e| e.kind()), Err(std::io::ErrorKind::PermissionDenied));
-        assert!(!outside.exists(), "the refused write created nothing");
-        let _ = std::fs::remove_dir_all(&root);
-    }
-
     #[test]
     fn long_and_numeric_flags_and_latin1() {
         let d = Dictionary::parse(b"FLAG long\nSFX Aa Y 1\nSFX Aa 0 s .\n", b"1\ncat/Aa\n");

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 +135,70 @@ 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` or `mc:MustUnderstand` (which must stay declared for the markup-compatibility rules to hold),
+/// and those starting a QName-like attribute value.
+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" || local == "MustUnderstand" {
+                    used.extend(v.split_whitespace());
+                }
+            }
+            // QName-valued attributes (`xsi:type="q:Item"`) need their prefix declared too.
+            if k != "xmlns" && !k.starts_with("xmlns:") {
+                for t in v.split_whitespace() {
+                    if let Some((p, _)) = t.split_once(':')
+                        && !p.is_empty()
+                        && p.bytes().all(|b| b.is_ascii_alphanumeric() || matches!(b, b'_' | b'-' | b'.'))
+                    {
+                        used.insert(p);
+                    }
+                }
             }
         }
         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 +348,12 @@ 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(),
@@ -363,15 +399,39 @@ fn resolve_mc_rels(el: &mut El, depth: usize, rels: Option<&Rels>) {
                     });
                     let branch = choice.or_else(|| ac.child("Fallback"));
                     if let Some(b) = branch {
+                        // The branch's namespace declarations (the first usable one of each name,
+                        // in order), indexed by prefix once: each child then looks up the prefixes
+                        // it uses instead of scanning every declaration.
+                        let mut decls: Vec<(&String, &String)> = Vec::new();
+                        // An oversized declaration still shadows an outer one of its name: its
+                        // content is left unbound rather than bound to the outer namespace.
+                        let mut by_prefix: HashMap<&str, Option<usize>> = HashMap::new();
+                        for (k, v) in b.attrs.iter().chain(ac.attrs.iter()) {
+                            let prefix = if k == "xmlns" { Some("") } else { k.strip_prefix("xmlns:") };
+                            if let Some(p) = prefix
+                                && !by_prefix.contains_key(p)
+                            {
+                                by_prefix.insert(p, (v.len() <= MAX_NS_VALUE).then_some(decls.len()));
+                                if v.len() <= MAX_NS_VALUE {
+                                    decls.push((k, v));
+                                }
+                            }
+                        }
                         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 have: HashSet<&str> = e.attrs.iter().map(|(a, _)| a.as_str()).collect();
+                                let mut wanted: Vec<usize> = std::iter::once("")
+                                    .chain(used.iter().copied())
+                                    .filter_map(|p| by_prefix.get(p).copied().flatten())
+                                    .filter(|&i| !have.contains(decls[i].0.as_str()))
+                                    .collect();
+                                wanted.sort_unstable();
+                                wanted.dedup();
                                 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()) {
-                                    if k.starts_with("xmlns") && !e.attrs.iter().any(|(a, _)| a == k) {
-                                        e.attrs.push((k.clone(), v.clone()));
-                                    }
-                                }
+                                e.attrs.extend(wanted.into_iter().map(|i| (decls[i].0.clone(), decls[i].1.clone())));
                                 out.push(Node::El(e));
                             }
                         }
@@ -1414,3 +1474,72 @@ 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())]);
+    }
+
+    /// The branch's own declaration of a prefix wins over the `AlternateContent`'s; an oversized
+    /// one still shadows it (its content stays unbound); a child's own declaration is left alone.
+    #[test]
+    fn mc_branch_declarations_keep_their_precedence() {
+        let long = "u".repeat(MAX_NS_VALUE + 1);
+        let x = format!(
+            r#"<r xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"><mc:AlternateContent xmlns:p="urn:ac" xmlns:q="urn:q-short"><mc:Fallback xmlns:p="urn:b" xmlns:q="{long}"><p:x/><q:y/><p:z xmlns:p="urn:own"/></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();
+        let attrs = |i: usize| kids[i].attrs.iter().map(|(k, v)| format!("{k}={v}")).collect::<Vec<_>>();
+        assert_eq!(attrs(0), ["xmlns:p=urn:b"]);
+        assert!(attrs(1).is_empty(), "{:?}", attrs(1));
+        assert_eq!(attrs(2), ["xmlns:p=urn:own"]);
+    }
+
+    #[test]
+    fn mc_branches_keep_prefixes_named_by_must_understand() {
+        let x = r#"<r xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"><mc:AlternateContent><mc:Fallback xmlns:q="urn:q"><a mc:MustUnderstand="q"/></mc:Fallback></mc:AlternateContent></r>"#;
+        let mut d = crate::xml::parse(x.as_bytes()).unwrap();
+        resolve_mc(&mut d.root, 0);
+        let kid = d.root.elements().next().unwrap();
+        assert!(kid.attrs.iter().any(|(k, v)| k == "xmlns:q" && v == "urn:q"), "{:?}", kid.attrs);
+    }
+
+    #[test]
+    fn mc_branches_keep_prefixes_used_in_qname_values() {
+        let x = r#"<r xmlns:mc="http://schemas.openxmlformats.org/markup-compatibility/2006"><mc:AlternateContent><mc:Fallback xmlns:q="urn:q" xmlns:z="urn:z"><n xmlns:xsi="urn:xsi" xsi:type="q:Item"/></mc:Fallback></mc:AlternateContent></r>"#;
+        let mut d = crate::xml::parse(x.as_bytes()).unwrap();
+        resolve_mc(&mut d.root, 0);
+        let kid = d.root.elements().next().unwrap();
+        assert!(kid.attrs.iter().any(|(k, v)| k == "xmlns:q" && v == "urn:q"), "{:?}", kid.attrs);
+        assert!(!kid.attrs.iter().any(|(k, _)| k == "xmlns:z"), "unused declarations stay out: {:?}", kid.attrs);
+    }
+
+    #[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");
+    }
+}