mirror of
https://github.com/storytold/gridcraft.git
synced 2026-11-07 02:49:43 -06:00
Compare commits
1
Commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
fc1bc1f474 |
+11
-37
@@ -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).
|
||||
@@ -243,26 +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);
|
||||
// Only X values that are kept are charged to the budget.
|
||||
if xs.iter().any(Value::is_text) {
|
||||
None
|
||||
} else {
|
||||
budget -= xs.len().min(budget);
|
||||
Some(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()
|
||||
@@ -274,7 +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()
|
||||
|
||||
@@ -1258,41 +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$60000".into(), ..Default::default() }).collect();
|
||||
let d = resolve(&wb, 0, &c);
|
||||
assert_eq!(d.series.len(), 40);
|
||||
assert_eq!(d.series[0].values.len(), 60_000);
|
||||
let total: usize = d.series.iter().map(|s| s.values.len()).sum();
|
||||
assert_eq!(total, crate::resolve::MAX_CHART_POINTS);
|
||||
assert!(d.series.last().is_some_and(|s| s.values.is_empty()));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn rejected_text_x_values_are_not_charged_to_the_point_budget() {
|
||||
// A text X range is ignored (the points are numbered), so repeating it in many series must not
|
||||
// use up the chart's budget and leave the later series without values.
|
||||
let mut wb = Workbook::new();
|
||||
let s = wb.sheet_mut(0).unwrap();
|
||||
for r in 0..100 {
|
||||
s.set_value(CellRef::new(r, 0), Value::Number(f64::from(r)));
|
||||
}
|
||||
s.set_value(CellRef::new(0, 1), Value::text("label"));
|
||||
s.set_value(CellRef::new(59_999, 1), Value::Number(2.0));
|
||||
let mut c = chart(ChartKind::Scatter);
|
||||
c.series = (0..40)
|
||||
.map(|_| Series { values: "Sheet1!$A$1:$A$100".into(), categories: Some("Sheet1!$B$1:$B$60000".into()), ..Default::default() })
|
||||
.collect();
|
||||
let d = resolve(&wb, 0, &c);
|
||||
assert_eq!(d.series.len(), 40);
|
||||
assert!(d.series.iter().all(|s| s.values.len() == 100 && s.x.is_none()));
|
||||
}
|
||||
|
||||
@@ -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];
|
||||
|
||||
@@ -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);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -53,25 +53,41 @@ fn concat(a: &[Arg], _c: &mut dyn Ctx) -> R<Value> {
|
||||
check_len(out)
|
||||
}
|
||||
|
||||
/// The error `v.to_text()` would give, found without building any text.
|
||||
fn text_error(v: &Value) -> Option<CellError> {
|
||||
match v {
|
||||
Value::Error(e) => Some(*e),
|
||||
Value::Array(a) => a.data.first().and_then(text_error),
|
||||
_ => None,
|
||||
}
|
||||
}
|
||||
|
||||
fn textjoin(a: &[Arg], _c: &mut dyn Ctx) -> R<Value> {
|
||||
let delims: Vec<String> = as_array(&arg(a, 0)?.value).iter().map(|v| v.to_text()).collect::<R<_>>()?;
|
||||
// Errors are found first, in argument order, so they win as before; text is then built one
|
||||
// piece at a time into the capped output, so many references to one long text (or a long
|
||||
// delimiter list) stop at the cap instead of first copying every piece.
|
||||
let delims = as_array(&arg(a, 0)?.value);
|
||||
if let Some(e) = delims.iter().find_map(text_error) {
|
||||
return Err(e);
|
||||
}
|
||||
let ignore = boolean(a, 1)?;
|
||||
let mut items = Vec::new();
|
||||
for x in a.iter().skip(2) {
|
||||
for v in as_array(&x.value).iter() {
|
||||
let s = v.to_text()?;
|
||||
if ignore && s.is_empty() {
|
||||
continue;
|
||||
}
|
||||
items.push(s);
|
||||
}
|
||||
let items: Vec<_> = a.iter().skip(2).map(|x| as_array(&x.value)).collect();
|
||||
if let Some(e) = items.iter().flat_map(|x| x.iter()).find_map(text_error) {
|
||||
return Err(e);
|
||||
}
|
||||
let mut out = String::new();
|
||||
for (i, s) in items.iter().enumerate() {
|
||||
if i > 0 && !delims.is_empty() {
|
||||
push_capped(&mut out, &delims[(i - 1) % delims.len()])?;
|
||||
let mut joined = 0usize;
|
||||
for v in items.iter().flat_map(|x| x.iter()) {
|
||||
let s = v.to_text()?;
|
||||
if ignore && s.is_empty() {
|
||||
continue;
|
||||
}
|
||||
push_capped(&mut out, s)?;
|
||||
if joined > 0 && !delims.data.is_empty() {
|
||||
let d = delims.data.get((joined - 1) % delims.data.len()).map(Value::to_text).transpose()?.unwrap_or_default();
|
||||
push_capped(&mut out, &d)?;
|
||||
}
|
||||
push_capped(&mut out, &s)?;
|
||||
joined += 1;
|
||||
}
|
||||
check_len(out)
|
||||
}
|
||||
@@ -1131,6 +1147,22 @@ mod tests {
|
||||
assert_eq!(rows_of(&ev("CONCATENATE", vec![av(row_t(&["a", "b"])), t("x")])), vec![vec![tv("ax"), tv("bx")]]);
|
||||
}
|
||||
|
||||
/// Security: TEXTJOIN builds its result piece by piece and stops at the text cap, rather
|
||||
/// than first copying every input (100,000 references to one 32,767-character text were
|
||||
/// 3 GB of copies). Errors among the inputs still win, in argument order.
|
||||
#[test]
|
||||
fn textjoin_stops_at_the_text_cap() {
|
||||
use gridcraft_core::Array;
|
||||
let long = tv(&"x".repeat(32_767));
|
||||
let many = || av(Array::filled(100_000, 1, long.clone()).into());
|
||||
is_err(ev("TEXTJOIN", vec![t(","), b(true), many()]), CellError::Value);
|
||||
is_err(ev("TEXTJOIN", vec![av(Array::filled(1, 100_000, long.clone()).into()), b(false), t("a"), t("b")]), CellError::Value);
|
||||
is_err(ev("TEXTJOIN", vec![t(","), b(true), many(), e(CellError::Div0)]), CellError::Div0);
|
||||
is_err(ev("TEXTJOIN", vec![av(arr(vec![vec![tv(","), Value::Error(CellError::NA)]])), b(true), t("a")]), CellError::NA);
|
||||
// Delimiters cycle and are converted only where used.
|
||||
is_text(ev("TEXTJOIN", vec![av(arr(vec![vec![nv(1.0), tv(";")]])), b(false), t("a"), t("b"), t("c"), t("d")]), "a1b;c1d");
|
||||
}
|
||||
|
||||
fn row_t(xs: &[&str]) -> Value {
|
||||
arr(vec![xs.iter().map(|x| tv(x)).collect()])
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user