pptx: take nested math text once and check an object's part for relationships once

crates/engine/src/cmd.rs

@@ -433,17 +433,6 @@ pub(crate) fn shape_of(s: &Session, id: ShapeId) -> Result<deckcraft_model::Shap
 pub fn xfrm_of(doc: &deckcraft_model::Presentation, sel: &crate::Selection, shape: &deckcraft_model::Shape) -> Xfrm {
     let mut x = shape.xfrm.unwrap_or_else(|| ctx_xfrm(doc, sel, shape));
     if let deckcraft_model::ShapeKind::Table(t) = &shape.kind {
-        // A loaded table may not be normalized yet: measure the bounded grid an edit will
-        // leave, not every cell the file stored.
-        let normalized;
-        let t = if t.is_normalized() {
-            t
-        } else {
-            let mut n = t.clone();
-            n.normalize();
-            normalized = n;
-            &normalized
-        };
         let drawn: f64 = table_drawn_heights(doc, sel, t).iter().sum();
         if drawn.is_finite() && drawn > x.h {
             x.h = drawn;

crates/engine/src/cmd/table.rs

@@ -285,8 +285,6 @@ fn with_table(s: &mut Session, p: &Value, cmd: &str, f: impl Fn(&mut Table) -> R
     };
     let r = edit_shapes(s, &json!({"id": id}), cmd, |sh| match &mut sh.kind {
         ShapeKind::Table(t) => {
-            // Edits walk the grid: bound it first (a loaded table need not be normalized yet).
-            t.normalize();
             f(t)?;
             t.normalize();
             let (w, h) = (t.width(), t.height());

crates/model/src/table.rs

@@ -77,12 +77,6 @@ impl Default for Table {
     }
 }
 
-/// Cells a table may hold after [`Table::normalize`] (PowerPoint's own limit is 75 x 75).
-pub const MAX_CELLS: usize = 100_000;
-/// Cell visits [`Table::merge`] may spend growing its rectangle over existing merges: a few
-/// passes over the largest grid, where a real table needs one or two.
-const MAX_MERGE_WORK: usize = 16 * MAX_CELLS;
-
 impl Table {
     pub fn new(rows: usize, cols: usize, width: f64, row_height: f64) -> Self {
         let rows = rows.clamp(1, 1000);
@@ -170,12 +164,8 @@ impl Table {
         true
     }
     /// The merged areas: every uncovered cell spanning more than one cell, clipped to the grid.
-    /// Merges of a well-formed table do not overlap, so together they cover at most the grid; a
-    /// file's overlapping spans past that are dropped, so rebuilding them cannot revisit cells
-    /// without bound.
     fn merges(&self) -> Vec<Merge> {
         let (nr, nc) = (self.rows.len(), self.cols.len());
-        let mut area = nr.saturating_mul(nc);
         let mut out = vec![];
         for (r, row) in self.rows.iter().enumerate() {
             for (c, cell) in row.cells.iter().enumerate().take(nc) {
@@ -184,10 +174,7 @@ impl Table {
                 }
                 let rs = (cell.row_span.max(1) as usize).min(nr - r);
                 let gs = (cell.grid_span.max(1) as usize).min(nc - c);
-                if (rs > 1 || gs > 1)
-                    && let Some(left) = area.checked_sub(rs.saturating_mul(gs))
-                {
-                    area = left;
+                if rs > 1 || gs > 1 {
                     out.push(Merge { r, c, rs, gs });
                 }
             }
@@ -240,17 +227,9 @@ impl Table {
         if r1 >= self.rows.len() || c1 >= self.cols.len() || (r0 == r1 && c0 == c1) {
             return false;
         }
-        // Grow the rectangle until no existing merge sticks out of it. Each pass reads the whole
-        // grid and may grow it by one row, so the passes are charged against
-        // [`MAX_MERGE_WORK`]: past it the merge is refused, before anything changed.
+        // Grow the rectangle until no existing merge sticks out of it.
         let (mut r0, mut c0, mut r1, mut c1) = (r0, c0, r1, c1);
-        let pass = self.rows.len().saturating_mul(self.cols.len()).max(1);
-        let mut work = 0usize;
         loop {
-            work = work.saturating_add(pass);
-            if work > MAX_MERGE_WORK {
-                return false;
-            }
             let before = (r0, c0, r1, c1);
             for (r, row) in self.rows.iter().enumerate() {
                 for (c, cell) in row.cells.iter().enumerate().take(self.cols.len()) {
@@ -304,31 +283,21 @@ impl Table {
             return false;
         }
         // Spans come from the file and may reach past the table; only walk the cells that exist.
-        let end = r.saturating_add(rs).min(self.rows.len());
-        let gs = gs.min(self.cols.len().saturating_sub(c));
-        for row in self.rows.get_mut(r..end).unwrap_or_default() {
-            for x in row.cells.iter_mut().skip(c).take(gs) {
-                x.h_merge = false;
-                x.v_merge = false;
-                x.grid_span = 1;
-                x.row_span = 1;
+        for rr in r..r.saturating_add(rs).min(self.rows.len()) {
+            for cc in c..c.saturating_add(gs).min(self.cols.len()) {
+                if let Some(x) = self.cell_mut(rr, cc) {
+                    x.h_merge = false;
+                    x.v_merge = false;
+                    x.grid_span = 1;
+                    x.row_span = 1;
+                }
             }
         }
         true
     }
-    /// Whether the grid is as [`Table::normalize`] leaves it: within [`MAX_CELLS`], every row
-    /// exactly `cols.len()` cells wide.
-    pub fn is_normalized(&self) -> bool {
-        let n = self.cols.len();
-        n <= MAX_CELLS && self.rows.len() <= (MAX_CELLS / n.max(1)).max(1) && self.rows.iter().all(|r| r.cells.len() == n)
-    }
-    /// Make every row have exactly `cols.len()` cells (repair after loading). A grid over
-    /// [`MAX_CELLS`] loses its last rows (and columns) first, so a file naming many rows and many
-    /// columns but few cells cannot make this allocate their product.
+    /// Make every row have exactly `cols.len()` cells (repair after loading).
     pub fn normalize(&mut self) {
-        self.cols.truncate(MAX_CELLS);
         let n = self.cols.len();
-        self.rows.truncate((MAX_CELLS / n.max(1)).max(1));
         for r in &mut self.rows {
             r.cells.resize_with(n, || Cell::new(""));
             for c in &mut r.cells {
@@ -544,61 +513,6 @@ mod tests {
         }
     }
 
-    /// Each cell spans two rows, uncovered: merging the last two rows would grow the rectangle a
-    /// row per pass over the whole grid, 10,000 passes. The merge is refused at its work bound
-    /// instead, leaving the table as it was; a small merge still works.
-    #[test]
-    fn merge_expansion_stops_at_its_work_bound() {
-        let mut t = Table::new(10_000, 10, 400.0, 30.0);
-        for row in &mut t.rows {
-            for c in &mut row.cells {
-                c.row_span = 2;
-            }
-        }
-        let before = t.clone();
-        assert!(!t.merge(9_998, 0, 9_999, 9));
-        assert_eq!(t, before);
-        let mut small = Table::new(4, 4, 400.0, 30.0);
-        assert!(small.merge(0, 0, 1, 1));
-        assert!(small.is_normalized() && !Table { cols: vec![1.0; 3], rows: vec![Row::default(); 2], ..Default::default() }.is_normalized());
-    }
-
-    #[test]
-    fn normalize_caps_a_sparse_grid_at_max_cells() {
-        // 1,000 columns and 1,000 empty rows would be a million cells.
-        let mut t = Table { cols: vec![10.0; 1000], rows: vec![Row::default(); 1000], ..Default::default() };
-        t.normalize();
-        assert_eq!(t.cols.len(), 1000);
-        assert_eq!(t.rows.len(), MAX_CELLS / 1000);
-        assert!(t.rows.iter().all(|r| r.cells.len() == 1000));
-        // Columns alone past the cap keep one row of MAX_CELLS cells.
-        let mut t = Table { cols: vec![1.0; MAX_CELLS + 5], rows: vec![Row::default(); 3], ..Default::default() };
-        t.normalize();
-        assert_eq!((t.cols.len(), t.rows.len()), (MAX_CELLS, 1));
-        // A normal table is untouched.
-        let mut t = Table::new(4, 4, 400.0, 30.0);
-        t.normalize();
-        assert_eq!((t.n_rows(), t.n_cols()), (4, 4));
-    }
-
-    #[test]
-    fn overlapping_file_spans_rebuild_in_bounded_work() {
-        // Every cell an uncovered anchor spanning to the far corner: their areas add up to about
-        // n^4 / 4, but only what fits the grid is kept, so an edit's rebuild stays near n^2.
-        let n = 200;
-        let mut t = Table::new(n, n, 400.0, 30.0);
-        for row in &mut t.rows {
-            for c in &mut row.cells {
-                c.grid_span = u32::MAX;
-                c.row_span = u32::MAX;
-            }
-        }
-        assert_eq!(t.merges().len(), 1, "the first anchor already covers the whole grid");
-        t.insert_row(n);
-        assert_eq!(t.n_rows(), n + 1);
-        assert!(t.split(0, 0));
-    }
-
     fn merged() -> Table {
         // 4x4 with a 2x2 merge at (1,1) holding "M".
         let mut t = Table::new(4, 4, 400.0, 30.0);

crates/pptx/src/read.rs

@@ -46,6 +46,9 @@ pub struct Imp<'a> {
     pub tag_bytes: usize,
     /// The tag budget ran out and the warning was given.
     pub tags_capped: bool,
+    /// Whether a part has relationships of its own, by normalized part name: looked up once
+    /// however many embedded objects point at the part.
+    pub has_rels: HashMap<String, bool>,
 }
 
 impl Imp<'_> {
@@ -446,6 +449,7 @@ pub fn import_theme(bytes: &[u8]) -> Result<(deckcraft_model::theme::Theme, Vec<
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
+        has_rels: HashMap::new(),
     };
     let part = Part::new(&pkg, &theme_path, &doc);
     let theme = theme::read_theme(&mut imp, &part, &doc.root);
@@ -487,6 +491,7 @@ pub fn import_checked(bytes: &[u8]) -> Result<(Presentation, Vec<String>), PptxE
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
+        has_rels: HashMap::new(),
     };
     // Slide ids first so hyperlinks anywhere can point at slides.
     let slide_ids = register_slides(&mut imp, &docs, &slide_entries, &mut lost_slides);

crates/pptx/src/read/dml.rs

@@ -627,12 +627,22 @@ fn push_text_runs(para: &mut Paragraph, props: RunProps, text: &str) {
 /// Appends an Office Math (a14:m) run: its plain text and its OMML.
 fn push_math_run(para: &mut Paragraph, part: &Part, e: &El) {
     if para.runs.len() < MAX_PARAGRAPH_RUNS {
-        let mut text = String::new();
-        let mut ts = vec![];
-        e.find_all("t", &mut ts);
-        for t in ts {
-            text.push_str(&t.text());
+        // The text of the outermost `t` elements: a `t` nested in another is already part of
+        // the outer one's text, and taking it again would copy it once per nesting level.
+        fn outer_t(e: &El, depth: usize, text: &mut String) {
+            if depth > crate::xml::MAX_DEPTH {
+                return;
+            }
+            for c in e.elements() {
+                if c.is("t") {
+                    text.push_str(&c.text());
+                } else {
+                    outer_t(c, depth + 1, text);
+                }
+            }
         }
+        let mut text = String::new();
+        outer_t(e, 0, &mut text);
         let omml = e.elements().map(|c| part.keep_inner(c)).collect::<String>();
         para.runs.push(Run { text, props: RunProps::default(), kind: RunKind::Math { omml } });
     }
@@ -677,6 +687,17 @@ mod tests {
     use super::*;
     use crate::xml::parse;
 
+    #[test]
+    fn nested_math_text_is_taken_once() {
+        // 200 `m:t` nested in each other around one word: the word, not 200 copies of it.
+        let x = format!(r#"<a14:m xmlns:a14="a" xmlns:m="m"><m:oMath>{}word{}</m:oMath></a14:m>"#, "<m:t>".repeat(200), "</m:t>".repeat(200));
+        let d = parse(x.as_bytes()).unwrap();
+        let part = Part { name: "ppt/slides/slide1.xml".into(), rels: Default::default(), ns: vec![], inner_ns: vec![] };
+        let mut para = Paragraph::default();
+        push_math_run(&mut para, &part, &d.root);
+        assert_eq!(para.runs[0].text, "word");
+    }
+
     #[test]
     fn colors_and_transforms() {
         let d = parse(

crates/pptx/src/read/shapes.rs

@@ -475,7 +475,8 @@ fn frame(imp: &mut Imp, part: &Part, ctx: &mut IdCtx, e: &El, depth: usize) -> S
         .into_iter()
         .map(|id| {
             let rel = part.rels.get(&id).filter(|r| !r.external)?;
-            if !imp.pkg.rels(&rel.target).items.is_empty() {
+            let pkg = imp.pkg;
+            if *imp.has_rels.entry(crate::opc::norm(&rel.target)).or_insert_with(|| !pkg.rels(&rel.target).items.is_empty()) {
                 return None;
             }
             let (kind, target) = (rel.ty.clone(), rel.target.clone());