Analysis ToolPak: stop building a report once it outgrows the output limit

Report builders filled their whole output map before write_report
checked its size, so Descriptive Statistics grouped by rows over a
tall, nearly empty column allocated about 16 entries for each of up
to a million variables before refusing the report.

Grid now tracks its extent as values are stored and keeps nothing
once the report's rows x columns pass MAX_OUTPUT_CELLS; write_report
refuses such a report. extent() no longer rescans every entry.

crates/chart/src/resolve.rs

@@ -11,10 +11,6 @@ use crate::{ChartData, Rgba, SeriesData, TrendlineData};
 pub(crate) const MAX_POINTS: usize = 100_000;
 /// Cap on series per chart.
 pub(crate) const MAX_SERIES: usize = 255;
-/// Cap on the points of all series of one chart together (values, X values and bubble sizes):
-/// [`MAX_SERIES`] × [`MAX_POINTS`] would let a few short whole-column references retain and
-/// draw tens of millions of points. Series past it get only what is left, or no points.
-pub(crate) const MAX_CHART_POINTS: usize = 1_000_000;
 /// Cap on trendlines drawn per series.
 pub(crate) const MAX_TRENDLINES: usize = 16;
 
@@ -81,15 +77,10 @@ fn union_parts(f: &str) -> Option<Vec<&str>> {
 /// Evaluates a series formula to a flat list of values (row-major), capped. A union of
 /// references (`(A1:A3,A5)`, as PivotCharts use) is read part by part.
 fn eval_list(wb: &Workbook, sheet: usize, formula: &str) -> Vec<Value> {
-    eval_list_in(wb, sheet, formula, 0, MAX_POINTS)
+    eval_list_in(wb, sheet, formula, 0)
 }
 
-/// [`eval_list`] keeping at most `cap` values (and never more than [`MAX_POINTS`]).
-fn eval_list_capped(wb: &Workbook, sheet: usize, formula: &str, cap: usize) -> Vec<Value> {
-    eval_list_in(wb, sheet, formula, 0, cap.min(MAX_POINTS))
-}
-
-fn eval_list_in(wb: &Workbook, sheet: usize, formula: &str, depth: u32, cap: usize) -> Vec<Value> {
+fn eval_list_in(wb: &Workbook, sheet: usize, formula: &str, depth: u32) -> Vec<Value> {
     let f = formula.trim();
     if f.is_empty() {
         return vec![];
@@ -99,19 +90,16 @@ fn eval_list_in(wb: &Workbook, sheet: usize, formula: &str, depth: u32, cap: usi
     {
         let mut out = Vec::new();
         for p in parts {
-            if out.len() >= cap {
+            if out.len() >= MAX_POINTS {
                 break;
             }
-            let rest = cap - out.len();
-            out.extend(eval_list_in(wb, sheet, p, depth + 1, rest).into_iter().take(rest));
+            let rest = MAX_POINTS - out.len();
+            out.extend(eval_list_in(wb, sheet, p, depth + 1).into_iter().take(rest));
         }
         return out;
     }
-    if cap == 0 {
-        return vec![];
-    }
     match evaluate(wb, sheet, CellRef::default(), f) {
-        Value::Array(a) => a.iter().take(cap).cloned().collect(),
+        Value::Array(a) => a.iter().take(MAX_POINTS).cloned().collect(),
         v => vec![v],
     }
 }
@@ -123,11 +111,8 @@ fn to_num(v: &Value) -> Option<f64> {
     }
 }
 
-/// A series formula's numbers, at most `cap` of them; `cap` is reduced by what was taken.
-fn nums(wb: &Workbook, sheet: usize, formula: &str, cap: &mut usize) -> Vec<Option<f64>> {
-    let out: Vec<Option<f64>> = eval_list_capped(wb, sheet, formula, *cap).iter().map(to_num).collect();
-    *cap -= out.len().min(*cap);
-    out
+fn nums(wb: &Workbook, sheet: usize, formula: &str) -> Vec<Option<f64>> {
+    eval_list(wb, sheet, formula).iter().map(to_num).collect()
 }
 
 /// Number format code of the first cell of a reference ("General" when not a reference).
@@ -243,21 +228,15 @@ pub fn resolve(wb: &Workbook, sheet: usize, chart: &Chart) -> ChartData {
     let mut series = Vec::new();
     let mut categories: Option<Vec<String>> = None;
     let mut category_levels = Vec::new();
-    // One point budget for the whole chart, not one per series.
-    let mut budget = MAX_CHART_POINTS;
     for (i, s) in chart.series.iter().take(MAX_SERIES).enumerate() {
-        let values = nums(wb, sheet, &s.values, &mut budget);
+        let values = nums(wb, sheet, &s.values);
         let name = s.name.as_deref().map(|n| series_name(wb, sheet, n)).filter(|n| !n.is_empty()).unwrap_or_else(|| format!("Series{}", i + 1));
         let color =
             s.color.as_ref().and_then(|c| c.resolve(&wb.theme)).map(|[r, g, b]| [r, g, b, 0xFF]).unwrap_or_else(|| series_color(&wb.theme, i));
         let kind = effective_kind(chart.kind, s.kind, i);
         let x = match (&s.categories, numeric_x || matches!(kind, ChartKind::Scatter | ChartKind::ScatterLines | ChartKind::Bubble)) {
             // X values with text in them (labels) are ignored: the points are numbered 1..n.
-            (Some(f), true) => {
-                let xs = eval_list_capped(wb, sheet, f, budget);
-                budget -= xs.len().min(budget);
-                Some(xs).filter(|xs| !xs.iter().any(Value::is_text)).map(|xs| xs.iter().map(to_num).collect())
-            }
+            (Some(f), true) => Some(eval_list(wb, sheet, f)).filter(|xs| !xs.iter().any(Value::is_text)).map(|xs| xs.iter().map(to_num).collect()),
             _ => None,
         };
         if categories.is_none()
@@ -269,7 +248,7 @@ pub fn resolve(wb: &Workbook, sheet: usize, chart: &Chart) -> ChartData {
                 category_levels = outer;
             }
         }
-        let sizes = s.bubble_sizes.as_deref().map(|f| nums(wb, sheet, f, &mut budget));
+        let sizes = s.bubble_sizes.as_deref().map(|f| nums(wb, sheet, f));
         let trendlines = s
             .trendlines
             .iter()

crates/chart/src/tests.rs

@@ -1258,21 +1258,3 @@ fn box_whisker_draws_each_series_quartile_method() {
     // The boxes span 3.5 and 2.5 data units on the same value axis.
     assert!((heights[0] / heights[1] - 1.4).abs() < 0.001);
 }
-
-#[test]
-fn series_share_one_point_budget() {
-    // Many series over the same long column each get a share of one chart-wide budget, not
-    // `MAX_POINTS` apiece.
-    let mut wb = Workbook::new();
-    let s = wb.sheet_mut(0).unwrap();
-    s.set_value(CellRef::new(0, 0), Value::Number(1.0));
-    s.set_value(CellRef::new(59_999, 0), Value::Number(2.0));
-    let mut c = chart(ChartKind::ColumnClustered);
-    c.series = (0..40).map(|_| Series { values: "Sheet1!$A$1:$A$60000".into(), ..Default::default() }).collect();
-    let d = resolve(&wb, 0, &c);
-    assert_eq!(d.series.len(), 40);
-    assert_eq!(d.series[0].values.len(), 60_000);
-    let total: usize = d.series.iter().map(|s| s.values.len()).sum();
-    assert_eq!(total, crate::resolve::MAX_CHART_POINTS);
-    assert!(d.series.last().is_some_and(|s| s.values.is_empty()));
-}

crates/engine/src/chart_svg.rs

@@ -10,10 +10,6 @@ use crate::io::esc;
 /// Widest or tallest chart written; larger anchors are clamped.
 const MAX_SIDE: f32 = 16_384.0;
 
-/// Most bytes of SVG written for the charts of one web page together. A chart past what is
-/// left is left out of the page rather than grown without bound.
-pub(crate) const MAX_PAGE_SVG_BYTES: usize = 64 * 1024 * 1024;
-
 struct HelveticaMeasure;
 
 impl gridcraft_chart::Measure for HelveticaMeasure {
@@ -22,34 +18,24 @@ impl gridcraft_chart::Measure for HelveticaMeasure {
     }
 }
 
-/// `chart` of sheet `sheet` as an `<svg>` element placed by `style`, or `None` for a degenerate
-/// size or an element longer than `budget` bytes.
-pub(crate) fn chart_svg(wb: &Workbook, sheet: usize, chart: &Chart, style: &str, budget: usize) -> Option<String> {
+/// `chart` of sheet `sheet` as an `<svg>` element placed by `style`, or `None` for a degenerate size.
+pub(crate) fn chart_svg(wb: &Workbook, sheet: usize, chart: &Chart, style: &str) -> Option<String> {
     let (w, h) = (chart.anchor.width.min(MAX_SIDE), chart.anchor.height.min(MAX_SIDE));
     if !(w >= 1.0 && h >= 1.0) {
         return None;
     }
     let data = gridcraft_chart::resolve(wb, sheet, chart);
     let prims = gridcraft_chart::render(chart, &data, w, h, &HelveticaMeasure);
-    prims_svg_within(&prims, w, h, style, budget)
+    Some(prims_svg(&prims, w, h, style))
 }
 
-#[cfg(test)]
 fn prims_svg(prims: &[Prim], w: f32, h: f32, style: &str) -> String {
-    prims_svg_within(prims, w, h, style, usize::MAX).unwrap_or_default()
-}
-
-/// The `<svg>` element, or `None` as soon as it grows past `budget` bytes.
-fn prims_svg_within(prims: &[Prim], w: f32, h: f32, style: &str, budget: usize) -> Option<String> {
     let mut out = String::new();
     let _ = write!(
         out,
         "<svg xmlns=\"http://www.w3.org/2000/svg\" width=\"{w}\" height=\"{h}\" viewBox=\"0 0 {w} {h}\" font-family=\"Helvetica,Arial,sans-serif\" style=\"{style}\">"
     );
     for p in prims {
-        if out.len() > budget {
-            return None;
-        }
         match p {
             Prim::Rect { x, y, w, h, fill, stroke, radius } => {
                 let _ = write!(out, "<rect x=\"{}\" y=\"{}\" width=\"{}\" height=\"{}\"", num(*x), num(*y), num(*w), num(*h));
@@ -106,7 +92,7 @@ fn prims_svg_within(prims: &[Prim], w: f32, h: f32, style: &str, budget: usize)
         }
     }
     out.push_str("</svg>");
-    (out.len() <= budget).then_some(out)
+    out
 }
 
 /// Fill and stroke attributes of an open shape element; the caller closes it.
@@ -163,17 +149,6 @@ fn num(v: f32) -> String {
 mod tests {
     use super::*;
 
-    #[test]
-    fn an_svg_past_its_byte_budget_is_left_out() {
-        let rect = Prim::Rect { x: 1.0, y: 2.0, w: 3.0, h: 4.0, fill: Some([255, 0, 0, 255]), stroke: None, radius: 0.0 };
-        let prims = vec![rect; 10_000];
-        let full = prims_svg(&prims, 100.0, 50.0, "");
-        assert!(full.len() > 100_000);
-        assert_eq!(prims_svg_within(&prims, 100.0, 50.0, "", full.len()), Some(full.clone()));
-        assert_eq!(prims_svg_within(&prims, 100.0, 50.0, "", full.len() - 1), None);
-        assert_eq!(prims_svg_within(&prims, 100.0, 50.0, "", 1_000), None);
-    }
-
     #[test]
     fn every_primitive_becomes_an_svg_element() {
         let red = [255, 0, 0, 255];

crates/engine/src/cmd/analysis.rs

@@ -272,11 +272,25 @@ struct Grid {
     cells: BTreeMap<(u32, u32), Value>,
     /// Cells shown as percentages.
     pct: Vec<(u32, u32)>,
+    /// Rows and columns spanned by `cells`.
+    rows: u32,
+    cols: u32,
+    /// A cell was dropped because the report outgrew [`MAX_OUTPUT_CELLS`]: the report is refused.
+    too_large: bool,
 }
 
 impl Grid {
+    /// Stores a value. Past [`MAX_OUTPUT_CELLS`] (the most a report writes) nothing more is kept,
+    /// so a builder that would produce millions of entries stops allocating them, and
+    /// [`write_report`] refuses the report.
     fn put(&mut self, r: u32, c: u32, v: Value) {
         if !v.is_empty() {
+            let (rows, cols) = (self.rows.max(r.saturating_add(1)), self.cols.max(c.saturating_add(1)));
+            if self.too_large || u64::from(rows) * u64::from(cols) > MAX_OUTPUT_CELLS {
+                self.too_large = true;
+                return;
+            }
+            (self.rows, self.cols) = (rows, cols);
             self.cells.insert((r, c), v);
         }
     }
@@ -296,9 +310,7 @@ impl Grid {
     }
     /// (rows, cols) spanned.
     fn extent(&self) -> (u32, u32) {
-        let rows = self.cells.keys().map(|(r, _)| r + 1).max().unwrap_or(0);
-        let cols = self.cells.keys().map(|(_, c)| c + 1).max().unwrap_or(0);
-        (rows, cols)
+        (self.rows, self.cols)
     }
 }
 
@@ -1749,7 +1761,7 @@ fn sampling(s: &Session, p: &Json, seed: Option<u64>) -> Result<Grid> {
 fn write_report(s: &mut Session, p: &Json, tool: &str, g: Grid) -> Result<Json> {
     let (rows, cols) = g.extent();
     let (rows, cols) = (rows.max(1), cols.max(1));
-    if u64::from(rows) * u64::from(cols) > MAX_OUTPUT_CELLS {
+    if g.too_large || u64::from(rows) * u64::from(cols) > MAX_OUTPUT_CELLS {
         return Err(bad(CMD, "the report is too large"));
     }
     let d = s.doc()?;
@@ -1914,6 +1926,28 @@ mod tests {
         x
     }
 
+    #[test]
+    fn a_report_stops_growing_past_the_output_limit() {
+        // Entries past the limit are not kept, and the report is refused.
+        let mut g = super::Grid::default();
+        g.num(1999, 1999, 1.0);
+        assert!(!g.too_large);
+        g.num(0, 2000, 1.0);
+        assert!(g.too_large);
+        assert_eq!(g.cells.len(), 1);
+        assert_eq!(g.extent(), (2000, 2000));
+        // Row-grouped statistics of a tall, almost empty column: one variable per row would make
+        // millions of report entries; the report is refused instead of built.
+        let mut s = s();
+        s.execute("range.setValues", json!({"range": "A1", "values": [[1]]})).unwrap();
+        s.execute("range.setValues", json!({"range": "A1000000", "values": [[2]]})).unwrap();
+        let err = s
+            .execute("data.analysisToolPak", json!({"tool": "descriptive", "range": "A1:A1000000", "groupedBy": "rows", "output": "C1"}))
+            .expect_err("report too large");
+        assert!(err.to_string().contains("too large"), "{err}");
+        assert_eq!(v(&s, "C1"), Value::Empty);
+    }
+
     #[test]
     fn forecast_standard_errors_match_the_full_window_rescan() {
         // #1043: the running sum must give what re-adding every window gave, gaps included.

crates/engine/src/io.rs

@@ -736,10 +736,8 @@ pub fn to_html(wb: &Workbook, sheet: usize) -> Result<String> {
                 place(&im.anchor)
             );
         }
-        let mut svg_budget = crate::chart_svg::MAX_PAGE_SVG_BYTES;
         for ch in &sh.charts {
-            if let Some(svg) = crate::chart_svg::chart_svg(wb, sheet, ch, &place(&ch.anchor), svg_budget) {
-                svg_budget -= svg.len();
+            if let Some(svg) = crate::chart_svg::chart_svg(wb, sheet, ch, &place(&ch.anchor)) {
                 out.push_str(&svg);
             }
         }