From 052ee60218ed305d93e72c2fa7fe0d99b512f969 Mon Sep 17 00:00:00 2001
From: "Dylan M. Taylor" <dylan@dylanmtaylor.com>
Date: Mon, 2 Nov 2026 20:34:32 -0500
Subject: [PATCH] Get Data: refuse an import whose dense rows would be too
 large

data.getData and data.fromTextCsv built every row densely before
writing it: JSON records were padded to the union of all keys (found
with a quadratic Vec::contains scan), and CSV cells were expanded over
the whole used rectangle. A few sparse records or one far-away CSV
cell therefore allocated billions of slots.

The importer now collects JSON keys with a HashSet, refuses more keys
than a sheet has columns, and checks the projected rows x columns
against MAX_IMPORT_CELLS (10,000,000) before building any row.
---
 crates/engine/src/cmd/extra/data.rs | 53 ++++++++++++++++++++++++++++-
 1 file changed, 52 insertions(+), 1 deletion(-)

diff --git a/crates/engine/src/cmd/extra/data.rs b/crates/engine/src/cmd/extra/data.rs
index 9a1ae7417..a3f69bc2e 100644
--- a/crates/engine/src/cmd/extra/data.rs
+++ b/crates/engine/src/cmd/extra/data.rs
@@ -42,6 +42,15 @@ pub(super) fn table_resize(s: &mut Session, p: &Json) -> Result<Json> {
     })
 }
 
+/// Most cells (rows × columns, blanks included) Get Data imports. The rows are built densely
+/// before they are written, so a few sparse records or one far-away CSV cell must not expand
+/// into a rectangle of billions of slots.
+const MAX_IMPORT_CELLS: u64 = 10_000_000;
+
+fn too_many_cells() -> EngineError {
+    EngineError::Other(format!("The data is too large to import ({MAX_IMPORT_CELLS} cells at most)."))
+}
+
 pub(super) fn from_csv(s: &mut Session, p: &Json) -> Result<Json> {
     let bytes = if let Some(t) = str_param(p, "text") {
         t.as_bytes().to_vec()
@@ -57,15 +66,22 @@ pub(super) fn from_csv(s: &mut Session, p: &Json) -> Result<Json> {
         if text.trim_start().starts_with('[') {
             let arr: Vec<Json> = serde_json::from_str(&text).map_err(|e| bad("data.getData", e.to_string()))?;
             let mut keys: Vec<String> = Vec::new();
+            let mut seen: std::collections::HashSet<&str> = std::collections::HashSet::new();
             for o in &arr {
                 if let Some(m) = o.as_object() {
                     for k in m.keys() {
-                        if !keys.contains(k) {
+                        if seen.insert(k.as_str()) {
                             keys.push(k.clone());
                         }
                     }
                 }
             }
+            if keys.len() > gridcraft_core::MAX_COLS as usize {
+                return Err(EngineError::Other("The data has more fields than a sheet has columns.".into()));
+            }
+            if (arr.len() as u64 + 1).saturating_mul(keys.len().max(1) as u64) > MAX_IMPORT_CELLS {
+                return Err(too_many_cells());
+            }
             let mut out = vec![keys.iter().map(|k| json!(k)).collect::<Vec<_>>()];
             for o in &arr {
                 out.push(keys.iter().map(|k| o.get(k).cloned().unwrap_or(Json::Null)).collect());
@@ -79,6 +95,9 @@ pub(super) fn from_csv(s: &mut Session, p: &Json) -> Result<Json> {
             let wb = gridcraft_xlsx::read_csv(&bytes, &opts).map_err(|e| EngineError::Other(e.to_string()))?;
             let sh = wb.sheet(0).ok_or(EngineError::NoDocument)?;
             let Some(u) = sh.used_range() else { return Err(EngineError::Other("The file is empty.".into())) };
+            if u.count() > MAX_IMPORT_CELLS {
+                return Err(too_many_cells());
+            }
             (u.start.row..=u.end.row)
                 .map(|r| {
                     (u.start.col..=u.end.col)
@@ -129,3 +148,35 @@ pub(super) fn queries(s: &mut Session, _: &Json) -> Result<Json> {
             .collect(),
     ))
 }
+
+#[cfg(test)]
+mod tests {
+    use serde_json::json;
+
+    use crate::Session;
+
+    fn s() -> Session {
+        let mut s = Session::new();
+        s.new_workbook();
+        s
+    }
+
+    #[test]
+    fn sparse_imports_are_refused_before_their_rows_are_built() {
+        let mut s = s();
+        let sheets = s.doc().unwrap().wb.sheets.len();
+        // 5,000 records with one field each: 5,001 × 5,000 slots once every record gets every field.
+        let records: Vec<_> = (0..5_000).map(|i| json!({ format!("k{i}"): i })).collect();
+        let text = serde_json::to_string(&records).unwrap();
+        let err = s.execute("data.getData", json!({"text": text})).expect_err("too many cells");
+        assert!(err.to_string().contains("too large"), "{err}");
+        // A CSV whose last line reaches 16,001 columns, 1,000 lines down.
+        let text = format!("1{}{}x", "\n".repeat(999), ",".repeat(16_000));
+        let err = s.execute("data.fromTextCsv", json!({"text": text})).expect_err("too many cells");
+        assert!(err.to_string().contains("too large"), "{err}");
+        assert_eq!(s.doc().unwrap().wb.sheets.len(), sheets);
+        // Small imports still work.
+        let out = s.execute("data.getData", json!({"text": r#"[{"a": 1}, {"b": 2}]"#})).unwrap();
+        assert_eq!((out["rows"].as_u64(), out["cols"].as_u64()), (Some(3), Some(2)), "{out}");
+    }
+}