Spreadsheets put an HTML table next to the tab-separated text; Excel
(x:num) and LibreOffice Calc (sdval) record each cell's number there.
Paste now reads the HTML flavour (new paste_html host service, arboard on
desktop) and edit.paste takes an optional html param: a field that would
paste as text becomes the recorded number when the HTML cell shows the
same text, with an inferred format ("Rp"#,##0.00) that renders it back
identically.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Enter that commits the Name Box stayed in the frame's input; the grid,
drawn after the formula bar and focused again, read it as its own Enter
and moved the active cell down. The Name Box now consumes it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
PERCENTOF = SUM(subset) / SUM(all), written as _xlfn.PERCENTOF.
EUROCONVERT uses the fixed EU conversion rates for the currencies the
function documents, with currency rounding and triangulation precision.
New Add-in category for EUROCONVERT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A plain drag copies a lone number and extends every other series; Ctrl+drag
(Cmd on macOS) now does the opposite, as in Excel: a lone number counts up
by 1, and numbers, dates, lists and `Item 1`-style text are copied.
edit.autoFill gains mode "alternate" for this.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
epaint panics in set_fonts on a face skrifa can't open. Every system font
file is now checked with the same skrifa call (plus a readable cmap)
before it reaches FontDefinitions; a bad candidate is skipped with a
warning and the next one is tried, and a final pass drops any remaining
unparseable face from the definitions and every family.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Odd first and last coupon periods via quasi-coupon periods, from the
formulas on Microsoft's public function pages. ODDLYIELD is closed form;
ODDFYIELD bisects ODDFPRICE (bounded). Financial is now 55 / 55.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The file format has no `@` operator, but formulas with an explicit
implicit intersection were written with a literal `@`
(`<f>@A1:A3*2</f>`). Excel writes `@x` as `_xlfn.SINGLE(x)`, which only
dynamic-array formulas carry. The writer now does the same and marks such
formulas cm="1"; the parser reads `SINGLE(x)` back as `@x`, as it already
does for `ANCHORARRAY` and `#`.
Two cases in the #83 metadata test expected a redundant `@` (`SQRT(@A1:A3)`,
`@A1:A3*2`) to be written as an ordinary formula. Dropping the `@` there is
only lossless if the reader puts it back on legacy formulas, which it
doesn't yet (the read side of #64), so they now expect cm.
generated with Claude Code
Turning the filter on with a multi-cell selection used the selection as the
filter range. With only the header row selected, the filter had no data rows:
the dropdown list was empty and filtering hid nothing. A one-row selection
now extends down over the current region below it. An explicit `range`
parameter is still taken as given.
The filter menu gets a search box, as asked for in the issue. It narrows
the value list; Select All then reads "(Select All Search Results)" and
covers only the matches, and Apply keeps only the checked matches. The box
takes focus when the menu opens, so typing no longer lands in the active
cell.
generated with Claude Code
An icon-set rule without stored thresholds (every rule made by
home.conditionalFormat, and the sales sample) was written as an empty
<iconSet>, which CT_IconSet forbids (cfvo minOccurs 2); openpyxl drops it.
The writer now always emits one cfvo per icon: the stored thresholds when
there is one per icon, the lower bound prepended when it is missing, else
Excel's defaults (percent 0/33/67, 0/25/50/75, 0/20/40/60/80).
Evaluation read the thresholds as one fewer than the icons, but files
carry one per icon (the first is the lower bound), so thresholds read from
a file were ignored. Both now share icon_count and default_icon_thresholds
in gridcraft-model; icon_count is clamped to 3..=5 so a set name like
"0Foo" can't underflow k - 1.
generated with Claude Code
Typing 25:00 or 36:30:15 gave the right value but the h:mm format, so
the cell showed 1:00 and 12:30:15. Excel formats such entries as
[h]:mm:ss (25:00:00).
Excel decides duplicates by what a cell displays, not the value stored:
the same date shown as 3/8/2006 and as Mar 8, 2006 is two values, while
1.01 and 1.02 both shown as 1.0 are duplicates. Remove Duplicates
compared stored values, so it did the opposite in both cases.
7/4/1776 was entered as 7/4/3676 and DATEVALUE("12/31/1899") gave
693962: the text parser built dates with DATE()'s rule that adds 1900 to
years below 1900. Excel keeps such entries as text, and DATEVALUE gives
#VALUE!.
With Case sensitive checked, Sort compared text by code point, so every
uppercase word came before every lowercase one (A, B, a, b). Excel
compares ignoring case first and only then puts lowercase before
uppercase: a, A, b, B.
home.conditionalFormat took any operator it didn't know as "greater",
so a Cell Value rule given Excel's own names (lessThan, greaterThan,
lessThanOrEqual, greaterThanOrEqual, as in the file format and VBA)
silently became "greater than". Those names are now accepted, as data
validation already does, and anything else is an error.
The evaluator took a single index, so CHOOSE({1,2},"a","b") gave "a"
and VLOOKUP(x,CHOOSE({1,2},B1:B3,A1:A3),2,FALSE) gave #REF!. An array
index now goes through the library CHOOSE, which lifts over its
arguments as Excel does.
COUNTIF(range,"*") counted empty cells and COUNTIF(range,"<>*") skipped
them. Excel matches text patterns against text only: an empty string is
text, a blank cell is not.
Excel compares a blank cell as 0 (or empty text) in Cell Value rules,
so "less than 5" or "equal to 0" also formats blank cells, and a
cell holding an error never matches (the comparison is an error).
GridCraft skipped blank cells and matched errors as if they sorted
above every number ("greater than 5" formatted #N/A).
The No Blanks rule also counted a cell holding only spaces as not
blank, while the Blanks rule (like Excel's LEN(TRIM())) counted it as
blank, so such a cell matched both.
Only the first matching rule's format was used, so with a light red
fill for >300 and a red border for >500, 510 showed only the border.
Excel applies every matching rule; where two rules set the same part
(fill, font colour, a border edge…), the rule with the higher priority
wins, and Stop If True still ends the list.
Language::Es was registered and switches formulas to Spanish (SUMA, decimal comma, semicolon
arguments), but every interface label still fell back to English. This adds es.rs with all 266
labels in Excel's Spanish wording (Combinar y centrar, Inmovilizar paneles, Análisis de
hipótesis, Buscar objetivo, Rastrear precedentes…) and covers Spanish in the i18n tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012JC2QZ1Qc5KQMBswS38GKy
Excel shows 15 significant digits. When a number is an exact tie at the 16th
(its exact decimal expansion is 15 digits, a 5, then only zeros), Excel
rounds it toward zero: TEXT(1234567890123455,"0") is 1234567890123450.
GridCraft took Rust's `{:.14e}`, which rounds that tie to even, so any tie
with an odd 15th digit came out one unit too high (…460).
decimal_digits now formats 16 digits (correct unless the 16th is a 5),
and only on a 5 checks the exact expansion (a 30-digit screen, then the
full one) and truncates a true tie. Non-ties, including values just above
or below a tie, round as before. One format serves almost every number.
Tests: ties of every 15th-digit parity, negatives, fractional ties, the
neighbours just above and below a tie, the carry after fifteen 9s
(999999999999999.5 shows 999999999999999), and unchanged ordinary cases.
Evidence: Windows Excel 16.0.20430 TEXT() over a 174-row probe and a 24-row
carry probe agree on every row except those under a percent format and
General over negatives (the two bugs named below). The suite's number-format
golden goes from 183,843 to 184,900 of 187,810 passing with no row that
passed before failing now.
modified: crates/numfmt/src/decimal.rs
modified: crates/numfmt/src/tests.rs
The ODS and XLSB readers parse untrusted package metadata and content through
`quick_xml::NsReader`. In 0.38.4 that is RUSTSEC-2026-0195 (CVSS 7.5): the namespace
resolver pushes a binding for every `xmlns` / `xmlns:*` declaration on a start tag, with
no cap on how many there are, and `NsReader::read_event` reaches that path - not only
`read_resolved_event`.
None of our caps bound it. MAX_PART (512 MiB), MAX_TOTAL (1 GiB), MAX_ENTRIES,
MAX_CELLS, MAX_TEXT and MAX_DEPTH bound part size, entry count, cells, text and nesting,
never declarations per tag.
Measured with `gridcraft-cli info` on crafted .ods files, before this change:
754 B, 0 declarations -> 8.6 MB, exit 0
24 KB, 1,000,000 -> 52.6 MB, exit 0
187 KB, 8,000,000 -> 367.8 MB, exit 0
770 KB, 33,000,000 -> 1.46 GB, exit 0
About 45 bytes of resident memory per declaration, roughly 1,900x the file size, and no
error at all. At MAX_PART a ~1 MB file reaches ~1.9 GB, and every part is held in memory
while the package is read. Both readers share the path, so a downloaded .ods or .xlsb
was enough. Nothing panicked, so `Session::execute` never saw it.
0.41.0 caps declarations per start tag at 256 (DEFAULT_MAX_DECLARATIONS_PER_ELEMENT,
raisable through NamespaceResolver::set_max_declarations_per_element) and returns
`NamespaceError::TooManyDeclarations` beyond it. The same fixtures after:
754 B -> 8.6 MB, ok 24 KB -> 19 MB, error
187 KB -> 101.8 MB, error 770 KB -> 397 MB, error
with "start tag declares more than 256 namespace bindings". The residual peak is the
part itself: MAX_PART still lets one part decompress to 256 MiB, the existing intended
bound.
API changes for 0.41 (0.x bumps may break; all four are in the xlsx crate):
- `resolve_element` and `resolve_attribute` moved off `NsReader` on to the resolver:
`reader.resolver_mut().resolve_element(..)`. `attr` and `tag` in ods.rs take
`&mut NsReader` for that.
- `Attribute::unescape_value` is deprecated in favour of `normalized_value`, which also
applies the attribute-value normalization the XML specification requires.
- `BytesText::xml_content` now takes the document's XML version.
This also clears RUSTSEC-2026-0194, the quadratic duplicate-attribute check. xml.rs
already passed `with_checks(false)`, but ods.rs and xlsb.rs used the checked iterator, so
they were on the quadratic path as well.
Regression test: `ods_tests::ods_namespace_declaration_bomb_is_rejected` asserts that a
tag with thousands of declarations is an error and that a document with a handful still
reads. It fails on 0.38.4 (the tag is accepted) and passes on 0.41.0.
Verified: `cargo xtask ci`; the full xlsx suite (80 tests) plus the malformed-input
tests; and the four measurements above, before and after, with no panics either way.
Brandon Thomas