XLS: a FORMAT record shorter than its character count keeps the characters it holds (fixes #731)

crates/xlsx/src/tests/xls_tests.rs

@@ -476,3 +476,39 @@ fn sheets_share_one_record_budget() {
     let err = read_xls(&compound("Workbook", &dup)).unwrap_err().to_string();
     assert!(err.contains("same substream"), "{err}");
 }
+
+#[test]
+fn a_format_shorter_than_its_count_keeps_its_characters() {
+    // A FORMAT whose cch is one more than the characters in the record (seen in real files):
+    // Excel opens them, the record's characters are the format code.
+    let short = rec(0x041E, &cat(&[&165u16.to_le_bytes(), &6u16.to_le_bytes(), &[0], b"0.000"]));
+    let xf = |ifmt: u16| rec(0x00E0, &cat(&[&u16s(&[0, ifmt, 0]), &[0; 14]]));
+    let globals = |pos: u32| {
+        cat(&[
+            &bof(0x0005),
+            &short,
+            &rec(0x041E, &cat(&[&164u16.to_le_bytes(), &xlstr("yyyy-mm-dd")])),
+            &xf(0),
+            &xf(164),
+            &xf(165),
+            &rec(0x0085, &cat(&[&pos.to_le_bytes(), &[0, 0, 6, 0], b"Sheet1"])),
+            &eof(),
+        ])
+    };
+    let sheet = cat(&[
+        &bof(0x0010),
+        &rec(0x0203, &cat(&[&cell(0, 0, 0), &7.0f64.to_le_bytes()])),
+        &rec(0x0203, &cat(&[&cell(1, 0, 1), &0.0f64.to_le_bytes()])),
+        &rec(0x0203, &cat(&[&cell(2, 0, 2), &1.25f64.to_le_bytes()])),
+        &eof(),
+    ]);
+    let stream = cat(&[&globals(globals(0).len() as u32), &sheet]);
+    let (wb, report) = read_xls(&compound("Workbook", &stream)).unwrap();
+    let sh = &wb.sheets[0];
+    let fmt = |a1: &str| wb.styles.get(sh.style_id(CellRef::parse(a1).unwrap())).num_fmt.as_str().to_string();
+    assert_eq!(v(&wb, 0, "A1"), Value::Number(7.0));
+    assert_eq!(v(&wb, 0, "A3"), Value::Number(1.25));
+    assert_eq!(fmt("A2"), "yyyy-mm-dd");
+    assert_eq!(fmt("A3"), "0.000");
+    assert!(report.warnings.iter().any(|w| w.contains("declared length")), "{:?}", report.warnings);
+}

crates/xlsx/src/xls/mod.rs

@@ -143,7 +143,7 @@ pub(crate) fn read_xls_with(bytes: &[u8], mut budget: Budget) -> Result<(Workboo
         }
     };
     let mut report = ReadReport::default();
-    let g = globals(&stream, &mut budget)?;
+    let g = globals(&stream, &mut budget, &mut report)?;
 
     let mut wb = Workbook::new();
     wb.sheets.clear();
@@ -205,7 +205,7 @@ fn bof(rec: &Record<'_>) -> Result<(u16, u16), IoError> {
     Ok((c.u16()?, c.u16()?))
 }
 
-fn globals(stream: &[u8], budget: &mut Budget) -> Result<Globals, IoError> {
+fn globals(stream: &[u8], budget: &mut Budget, report: &mut ReadReport) -> Result<Globals, IoError> {
     let mut recs = Records::new(stream, 0);
     let first = recs.next_record()?.ok_or_else(|| invalid("empty Workbook stream"))?;
     let (vers, dt) = bof(&first)?;
@@ -249,7 +249,15 @@ fn globals(stream: &[u8], budget: &mut Budget) -> Result<Globals, IoError> {
             0x0022 => g.date1904 = c.u16()? == 1,
             0x041E => {
                 let ifmt = c.u16()?;
-                let code = c.string()?;
+                let cch = usize::from(c.u16()?);
+                let high = c.u8()? & 1 != 0;
+                // Some writers count one character more than the record holds; Excel opens
+                // those files, so keep the characters that are there.
+                let fit = c.remaining() / if high { 2 } else { 1 };
+                if cch > fit {
+                    report.warn("Some number formats were shorter than their declared length; the characters present were kept.");
+                }
+                let code = c.chars(cch.min(fit), high)?;
                 budget.text(code.len())?;
                 g.formats.insert(ifmt, code);
             }