Every edit copies the workbook for its undo step (copy-on-write: sheets
and 64-row bands of cells are shared). A floating picture's bytes
(Image.data) and the VBA project (Workbook.vba_project) were plain
Vec<u8>, so each edit on a sheet with a picture, and every edit in a
workbook with macros, kept another full copy of them, up to the 100
steps of the history. They are Arc<Vec<u8>> now, as cell pictures
already are, with base64 helpers for serde (b64::shared, b64::opt). JSON
and XLSX are unchanged.
Measured on macOS 27 (Apple M3 Max), gridcraft-cli, kernel peak
footprint: a 19 MB picture and 50 cell edits on its sheet, 992 -> 22 MB;
10 edits, 215 -> 21 MB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A formula read with `cm` (Excel's XLDAPR dynamic-array record in
xl/metadata.xml) now carries `Formula::dynamic`, and the writer keeps the
`cm` and the metadata part for it even when the formula returns one value
and the export shape analysis would not need array evaluation. Before, such
formulas were saved as legacy formulas, so Excel recalculated some of them
with implicit intersection (SUMPRODUCT(IFERROR(range/range,0)) changed
value), and workbooks whose every dynamic formula was scalar lost
xl/metadata.xml altogether.
The mark survives moves, row and column insertion or deletion, sheet and
name renames and LAMBDA call-case fixes; retyping a formula drops it and
the writer decides from the formula's shape, as before. It is serialized
only when set, so saved documents without it are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
home.rowHeight and home.columnWidth take pixels at 96 dpi, the model's unit
(header drags and the Row Height dialog send pixels), but their parameter docs
said points. An agent asking for a 30 pt row got `ht="22.5"` in the file, and
the range check accepted 500 while its message said "between 0 and 409 points".
- The params docs say pixels: `height: 20 (pixels) | points: 15`,
`width: 64 (pixels) | chars: 8.43`, and `{width: 64 (pixels)}` for
home.defaultWidth. Pixel values behave as before.
- home.rowHeight takes `points`, Excel's unit (Row Height dialog, `ht` in the
file), as home.columnWidth already takes `chars`.
- Range errors give the limit in the unit passed: 546 pixels (409.5 points),
409.5 points, 1800 pixels, 255 characters. `chars` is checked against 255,
Excel's limit and the widest column a file holds (it went to 256.4 through
the pixel limit).
- The model's size comments say pixels.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Chart sheets (a sheet that holds one chart; Move Chart > New sheet, F11) were
skipped on open, and their chart sheet, drawing, chart, style and colour parts
were dropped with them because the pass-through treats those kinds as content
GridCraft rewrites. Saving a workbook lost every chart sheet.
They are now kept as they were: the <sheet> entry (name, visibility and its
place among the worksheets), the chartsheet part and everything behind it,
byte for byte, written back in place. The workbook's other sheet positions
(active tab, print areas, scoped names) are computed with the chart sheets
counted. A chart sheet is left out on save when the chart is no longer
valid for the workbook: a sheet it plots was deleted or renamed, or a
worksheet took its name.
Excel stores the path of a linked workbook relative to the file that uses it, and rewrites it when the file is saved into another folder. The path was kept as text, so after Save As it named a file next to the new copy: a missing link, or another workbook with the same name. Relative externalLinkPath targets are now resolved against the folder the workbook was read from and written relative to the folder it is saved into (absolute when there is no relative path, on another drive).
Cell format 0 always mapped to the model's built-in Calibri 11 default, so the file's Normal font was replaced on save and every default-formatted cell changed size. The Normal style now takes the font of cell format 0, and a Normal font in the theme's body face is written with scheme="minor" again.
Excel uses the cached cell values of a saved file only when workbook.xml has both <fileVersion> and the calcFeatures extension; without them it recalculates everything on open, and lookups into closed workbooks come back as #CALC!. Both are now read into the workbook's kept content and written back in schema order: fileVersion first, the calcFeatures ext in <extLst> last. A workbook that had neither gets neither.
With values on rows and no column field, the header cell above the data
column reads "Total", as Excel saves it (tabular checked against a real
file).
Σ Values in Columns always comes after the last column field: the column
layout doesn't place it anywhere else, so `pivot.moveField` no longer
stores a position there, the reader ignores where `-2` sits in colFields
and the writer puts it last.
`pivot.changeSource` removes dropped row and column fields one by one so
Σ Values keeps its place ahead of the field it preceded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With two or more value fields the Σ Values pseudo-field can sit in Rows
as well as Columns, at any position among the row fields. The model
records it (`PivotTable.values_on_rows`, `values_position`) and the
engine lays it out like Excel in compact, outline and tabular form: one
row item per value field at the Σ Values level, per-value subtotals
("East Sum of Sales") and grand totals ("Total Sum of Sales").
XLSX reads `dataOnRows` and the `-2` field's place in rowFields or
colFields and writes them back; the "values on rows are shown on
columns" warning is gone. `pivot.moveField` moves "Σ Values" between
Rows and Columns or within one, axis field moves keep it in place, and
the PivotTable Fields pane lists it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
cargo fmt reformatted the new example's while-loop condition onto its
own line; clippy's nonminimal_bool caught a negated is_some_and that
reads more plainly as is_none_or.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Follow-up to #210, which the reviewer correctly rejected: the stated
cause (a sparse column with a stray distant value) doesn't hold, since
the scan already stops at the first empty cell. This is a different,
narrower fix that doesn't change behaviour for any input — it never
truncates a long fill, unlike #210's cap — only how fast a genuinely
long, densely-populated run is scanned.
grid.rs's fill-handle double-click called Sheet::value_ref once per
row to find where the neighbouring column's data ends. Each call does
a band lookup + row lookup + column lookup (three BTreeMap probes) via
CellStore::get. For a real dense run (e.g. a filled column near the
1,048,576-row sheet limit) that's a lot of point lookups on the UI
thread in one frame: ~59ms for a 1,000,000-row run in a release build
(measured with the new fill_handle_perf example), likely worse under
wasm where GridCraft actually runs in the browser.
- crates/model/src/store.rs: CellStore::contiguous_value_run walks
each band's row BTreeMap with `range` (one in-order pass per band)
instead of one `get` per row, so a dense run costs roughly one
tree-traversal step per row instead of one binary search per row.
Deliberately narrower than `has`/`get`: a stored cell always decides
the row on its own (even with an empty value), matching what
`value_ref`'s callers actually test for, not `has`'s "formula counts
regardless of its cached value" (used for Ctrl+arrow navigation).
- crates/model/src/sheet.rs: Sheet::last_contiguous_nonempty layers
`spill` back in wherever contiguous_value_run finds no stored cell,
exactly where plain `value_ref` would fall through to it.
- crates/ui-egui/src/grid.rs: the fill-handle double-click now calls
this once instead of looping over value_ref.
Result: 999,999-row dense run, 59.2ms -> 24.7ms (2.40x), same answer
(asserted in the example). Four new tests in sheet.rs, including a
300-trial property test against a naive value_ref-loop reference
(spills, formula-with-empty-value, gaps, blanks) — all pass, plus the
full existing suite (`cargo test -p gridcraft-model -p gridcraft-ui-egui
-p gridcraft-engine`; the only failures are #210-era gridcraft-engine
CSV tests already failing on main, unrelated to this change).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
A `;`-separated CSV (the usual CSV where `,` is the decimal separator) was
rewritten with `,` on save: read_csv sniffed the delimiter but the result was
dropped, and save_bytes passed a hard-coded b','. file.save, Save As .csv,
file.exportCsv, file.saveBytes and gridcraft-cli convert all went through it.
The workbook now remembers the separator it was read with
(Workbook::csv_delimiter, set by read_csv) and a CSV save writes it back.
Workbooks not read from CSV, and .tsv/.txt sources saved as .csv, keep `,`.
Regression tests: `;`, `,` and tab .csv round trips byte for byte, new and
.xlsx-origin workbooks save with `,`, and open-by-path + edit + file.save on a
`;` file keeps `;` (also through exportCsv and saveBytes).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Model: Series.trendlines (Vec<Trendline>, serde default) with kind
(linear, exponential, logarithmic, polynomial 2..6, power, moving
average 2..255), name, forward/backward forecast, set intercept,
equation / R² display and line colour, width and dash.
XLSX: read and write <c:trendline> in CT_Trendline child order, placed
after dLbls and before the data references in line, bar, area, scatter,
bubble and stock series; unknown types are skipped.
Chart: least-squares fits (transformed for exp/log/power, skipping
non-positive values; polynomials via normal equations on centred,
scaled x with partial pivoting), moving averages, forecasts that widen
the category or x axis, clipping to the plot, equation and R² labels and
legend entries ("Linear (Sales)"). Degenerate data draws nothing.
Engine: chart.addElement element="trendline" adds (type, series, order,
period, forward, backward, displayEquation, displayRSquared) or removes
trendlines; document.inspect lists each chart's trendlines.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
XLSX headerFooter kept only oddHeader/oddFooter: a workbook whose header is on a
different first page (differentFirst + firstHeader) or on even pages showed no
header, and saving dropped it. The model now keeps the first/even headers and
footers and the differentFirst / differentOddEven / scaleWithDoc / alignWithMargins
flags; XLSX reads and writes them; the PDF and the Page Layout view use the right
pair per page; in-place editing edits the page's own pair, and the Header & Footer
tab has Different First Page and Different Odd & Even Pages. pageLayout.headerFooter
takes the variants and a page: odd|even|first for one-section edits.
Insert › Header & Footer now switches to Page Layout view and edits the first
page's centre header, as in Excel, instead of opening the codes dialog.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Text boxes ignored a:rPr (size, bold, italic, typeface, fill, outline),
a:pPr algn, a:bodyPr anchor and spPr/a:ln/a:noFill. They are now read into
Shape::format, drawn on screen and in PDF, and written back.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sheet protection stored an internal FNV hash that was never written to
XLSX, and an Excel sheet password was dropped on import, so imported
protected sheets could be unprotected without it. Protect Sheet now
stores a salted SHA-512 hash (spin count 100000) via PasswordHash; the
XLSX sheetProtection password (iterated or legacy) is read and written,
and Unprotect Sheet really verifies it. Hashes from older GridCraft JSON
workbooks are still read and verified.
Refs #346
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The ribbon button toggled structure protection silently. It now opens a
Protect Structure and Windows dialog with an optional password; when a
password was set, unprotecting asks for it and really verifies it
(SHA-512, spin count 100000, salted). The password hash is read from and
written to XLSX workbookProtection (workbookAlgorithmName/HashValue/
SaltValue/SpinCount or the legacy workbookPassword); an unreadable one
stays locked. File › Info uses the same flow.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
