address review: count all chart vectors and relationship strings; reject a chart that crosses the presentation limit

crates/pptx/src/read.rs

@@ -38,6 +38,8 @@ pub struct Imp<'a> {
     pub vml_previews: HashMap<String, HashMap<String, String>>,
     /// Aggregate byte count of retained chart sidecars across the presentation.
     pub kept_sidecar_bytes: usize,
+    /// Data slots of the charts decoded so far (see `chart::MAX_PRESENTATION_CHART_POINTS`).
+    pub chart_points: usize,
     /// Cached kept sidecars by normalized chart part name.
     pub kept_charts: HashMap<String, deckcraft_model::chart::ChartKept>,
     /// Add-in tags parts, parsed once each, by normalized part name.
@@ -442,6 +444,7 @@ pub fn import_theme(bytes: &[u8]) -> Result<(deckcraft_model::theme::Theme, Vec<
         table_style_ids: Vec::new(),
         vml_previews: HashMap::new(),
         kept_sidecar_bytes: 0,
+        chart_points: 0,
         kept_charts: HashMap::new(),
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
@@ -483,6 +486,7 @@ pub fn import_checked(bytes: &[u8]) -> Result<(Presentation, Vec<String>), PptxE
         table_style_ids: Vec::new(),
         vml_previews: HashMap::new(),
         kept_sidecar_bytes: 0,
+        chart_points: 0,
         kept_charts: HashMap::new(),
         tags_by_part: HashMap::new(),
         tag_bytes: 0,

crates/pptx/src/read/chart.rs

@@ -39,6 +39,24 @@ pub(crate) fn rich_text(t: &El) -> Option<String> {
 /// Most category/value/size slots one chart may decode across all its series and caches, so
 /// sparse `ptCount`/`idx` declarations can't expand far beyond the XML they came from.
 pub(crate) const MAX_CHART_POINTS: usize = 1_000_000;
+/// Data slots (values and categories) the charts of one presentation may hold in all; a chart past
+/// it is read as a placeholder frame. Each frame that references a chart decodes its own copy.
+pub(crate) const MAX_PRESENTATION_CHART_POINTS: usize = 4_000_000;
+
+/// Slots a decoded chart holds, counting the chart itself as at least one.
+pub(crate) fn point_count(c: &Chart) -> usize {
+    let series = |s: &deckcraft_model::chart::Series| {
+        let bars = s.error_bars.as_ref().map_or(0, |e| e.plus.len().saturating_add(e.minus.len()));
+        s.values.len().saturating_add(s.x_values.len()).saturating_add(s.sizes.len()).saturating_add(bars)
+    };
+    let levels: usize = c.category_levels.iter().map(Vec::len).sum();
+    1usize.saturating_add(c.categories.len()).saturating_add(levels).saturating_add(c.series.iter().map(series).fold(0, usize::saturating_add))
+}
+
+/// Bytes of the strings a kept relationship owns.
+fn rel_bytes(r: &deckcraft_model::chart::ChartRel) -> usize {
+    r.id.len() + r.rel_type.len() + r.target.len() + r.part.len()
+}
 
 /// Most slots one cache may hold.
 const MAX_CACHE_POINTS: usize = 100_000;
@@ -466,6 +484,13 @@ pub(crate) fn capture_kept(imp: &mut super::Imp, chart_part: &str) -> deckcraft_
 
     let key = crate::opc::norm(chart_part);
     if let Some(cached) = imp.kept_charts.get(&key) {
+        // Every frame that shows this chart gets its own copy, so each copy is charged like the first.
+        let bytes: usize = cached.parts.iter().map(|p| p.bytes.len()).sum::<usize>() + cached.rels.iter().map(rel_bytes).sum::<usize>();
+        if imp.kept_sidecar_bytes.saturating_add(bytes) > MAX_KEPT_TOTAL_BYTES {
+            log::warn!("pptx: chart sidecar retained byte budget exceeded for {chart_part}");
+            return ChartKept::default();
+        }
+        imp.kept_sidecar_bytes += bytes;
         return cached.clone();
     }
 
@@ -485,6 +510,13 @@ pub(crate) fn capture_kept(imp: &mut super::Imp, chart_part: &str) -> deckcraft_
             kept.rels.push(ChartRel { id: r.id, rel_type: r.ty, target: r.target_raw, part: r.target, external: false });
         }
     }
+    // The first capture's relationship strings are retained too (the parts were charged as copied).
+    let rel_total: usize = kept.rels.iter().map(rel_bytes).sum();
+    if imp.kept_sidecar_bytes.saturating_add(rel_total) > MAX_KEPT_TOTAL_BYTES {
+        log::warn!("pptx: chart sidecar retained byte budget exceeded for {chart_part}");
+        return ChartKept::default();
+    }
+    imp.kept_sidecar_bytes += rel_total;
     imp.kept_charts.insert(key, kept.clone());
     kept
 }
@@ -590,6 +622,16 @@ fn text_size(e: Option<&El>) -> Option<f64> {
 mod tests {
     use super::*;
 
+    #[test]
+    fn point_count_charges_every_retained_vector() {
+        let mut c = Chart { categories: vec!["a".into()], category_levels: vec![vec!["x".into(); 2], vec!["y".into(); 3]], ..Default::default() };
+        let mut s =
+            deckcraft_model::chart::Series { values: vec![Some(1.0); 4], x_values: vec![None; 5], sizes: vec![None; 6], ..Default::default() };
+        s.error_bars = Some(deckcraft_model::chart::ErrorBars { plus: vec![None; 7], minus: vec![None; 8], ..Default::default() });
+        c.series = vec![s];
+        assert_eq!(point_count(&c), 1 + 1 + 5 + 4 + 5 + 6 + 15);
+    }
+
     #[test]
     fn sparse_caches_share_one_chart_wide_budget() {
         // Every series declares 100000 points but writes one: without a chart-wide budget each

crates/pptx/src/read/shapes.rs

@@ -428,10 +428,14 @@ fn frame(imp: &mut Imp, part: &Part, ctx: &mut IdCtx, e: &El, depth: usize) -> S
             return s;
         }
         if let Some(c) = gd.child("chart")
+            && imp.chart_points < super::chart::MAX_PRESENTATION_CHART_POINTS
             && let Some(rel) = part.rels.get(c.attr("r:id").unwrap_or(""))
             && let Some(bytes) = imp.pkg.xml_bytes(&rel.target)
             && let Some(mut chart) = if uri.ends_with("/chartex") { super::read_chartex_xml(bytes) } else { super::read_chart_xml(bytes) }
+            // A chart that would cross the presentation limit is read as the placeholder frame instead.
+            && imp.chart_points.saturating_add(super::chart::point_count(&chart)) <= super::chart::MAX_PRESENTATION_CHART_POINTS
         {
+            imp.chart_points = imp.chart_points.saturating_add(super::chart::point_count(&chart));
             chart.kept = super::chart::capture_kept(imp, &rel.target);
             s.kind = ShapeKind::Chart(Box::new(chart));
             return s;

xtask/src/assets.rs

@@ -61,23 +61,16 @@ pub fn missing(files: &[String], assets_md: &str) -> Vec<String> {
     files.iter().filter(|f| is_asset(f) && !patterns.iter().any(|p| glob(p, f))).cloned().collect()
 }
 
-/// `*` matches any run of characters except `/`; everything else literally. Matched with a
-/// table over the pattern, so a pattern with many `*` costs pattern length times name length
-/// rather than exploring every way to split the name.
+/// `*` matches any run of characters except `/`; everything else literally.
 fn glob(pat: &str, s: &str) -> bool {
-    let t: Vec<char> = s.chars().collect();
-    // `row[k]`: the pattern so far matches the first k characters of the name.
-    let mut row = vec![false; t.len() + 1];
-    row[0] = true;
-    for pc in pat.chars() {
-        let mut next = vec![false; t.len() + 1];
-        next[0] = pc == '*' && row[0];
-        for k in 1..=t.len() {
-            next[k] = if pc == '*' { row[k] || (next[k - 1] && t[k - 1] != '/') } else { row[k - 1] && t[k - 1] == pc };
+    match pat.split_once('*') {
+        None => pat == s,
+        Some((head, tail)) => {
+            let Some(rest) = s.strip_prefix(head) else { return false };
+            let ends = rest.char_indices().map(|(i, _)| i).chain(std::iter::once(rest.len()));
+            ends.take_while(|&i| !rest[..i].contains('/')).any(|i| glob(tail, &rest[i..]))
         }
-        row = next;
     }
-    row[t.len()]
 }
 
 /// Splits `git ls-files -z` output into real file names (no display quoting of Unicode or special characters).
@@ -127,15 +120,6 @@ mod tests {
         assert_eq!(missing(&["assets/é.png".into()], "`assets/*.svg`"), vec!["assets/é.png".to_string()]);
     }
 
-    #[test]
-    fn many_wildcards_on_a_failing_name_finish_at_once() {
-        let pat = format!("{}b", "a*".repeat(40));
-        assert!(!glob(&pat, &"a".repeat(60)));
-        assert!(glob(&pat, &format!("{}b", "a".repeat(60))));
-        assert!(!glob("a*b", "a/b"));
-        assert!(glob("a*b", "ab"));
-    }
-
     #[test]
     fn ls_files_output_is_nul_separated_and_unquoted() {
         let names = parse_ls_files("assets/zażółć.png\0docs/images/a b\nc.png\0".as_bytes());