chore(release): version packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@office-kit/xlsx@0.10.0

Minor Changes

• #124 3a8db65 Thanks @baseballyama! - fix: an invalid hyperlink target is now rejected instead of producing a file Excel refuses (#117)

  setHyperlink accepted any string as target and the writer emitted it
  verbatim into the worksheet rels, so { target: 'https:// www.example.com' }
  saved without complaint and Excel then reported "We found a problem with some
  content in …" and dropped content on repair. One bad row poisoned an entire
  export.

  The rels Target attribute is xsd:anyURI. Authoring a hyperlink now throws
  an OpenXmlSchemaError when the target is empty, contains whitespace or a
  control character, or has a non-ASCII host (percent-encoding does not rescue
  that one — the host has to be punycode). The check runs in makeHyperlink, so
  setHyperlink, addUrlHyperlink and addMailtoHyperlink are all covered.

  Reading is deliberately unaffected: a workbook whose rels already carry a
  broken target still loads, so it can be inspected and repaired.

  Behaviour change: makeHyperlink, setHyperlink, addUrlHyperlink and
  addMailtoHyperlink now throw on a target they previously accepted. That is
  the point of the change — the alternative was a workbook Excel refuses — but
  it can surface at a call site that used to succeed.

• #122 3bd3940 Thanks @baseballyama! - fix: rich text was flattened to plain text on save (#114)

  A shared string built from <r> runs came back as a single plain <t>: the
  loader collapsed rich-text <si> entries into their concatenated text before
  the worksheet reader saw them, so a cell with a bold first half round-tripped
  uniformly unformatted. Inline strings (<c t="inlineStr"><is>) lost their runs
  the same way on read.

  Rich-text <si> and <is> bodies now go through one CT_Rst parser and become
  { kind: 'rich-text' } cell values, so per-run fonts survive. Cells holding
  rich text are written into xl/sharedStrings.xml as <si><r>…</r></si> —
  where Excel itself stores them — instead of being inlined per cell, so a
  formatted string repeated across cells costs one entry rather than one copy
  each. <r> runs with no <rPr> stay separate runs, since that is how Excel
  writes the unformatted half of a rich string.

  Behaviour change: a cell whose shared string is rich text now reads back
  as { kind: 'rich-text', runs } where it used to read back as the concatenated
  plain string, and rich-text cells serialise into xl/sharedStrings.xml rather
  than as t="inlineStr". Code that assumed getCell(...).value was a string
  for those cells needs to handle the union.

Patch Changes

• #125 c5d706f Thanks @baseballyama! - fix: drawing shapes were rewritten as chart graphicFrames with an empty r:id (#110)

  An anchor holding anything the model doesn't cover — a shape, a group, a
  connector — was written back as a chart <xdr:graphicFrame> carrying
  <c:chart r:id=""/>, so a rectangle came back as "Chart 1" with a dangling
  reference, its geometry gone and no drawing rels part written at all. Anchors
  sitting inside Excel's <mc:AlternateContent> wrapper were dropped outright,
  leaving an empty <xdr:wsDr/>. Together those are what still broke worksheets
  carrying form controls after #105: the worksheet side was fixed, but the shapes
  that draw the controls live in drawing1.xml.

  Unmodeled drawing content is now kept as the verbatim source XML and written
  back untouched — the whole anchor element, so editAs and the
  <xdr:clientData fLocksWithSheet="0"> flags form controls rely on survive too
  — and <mc:AlternateContent> wrappers keep their place in document order,
  which is z-order. Relationships those nodes reference keep their original ids
  and their target parts, and the writer allocates ids for modeled charts and
  pictures around them.

• #121 2f7e6c4 Thanks @baseballyama! - fix: an empty cached formula value and its t="str" were dropped on save (#115)

  <c r="A1" t="str"><f>[1]Extern!$A$1</f><v/></c> came back as
  <c r="A1"><f>[1]Extern!$A$1</f></c>: the reader collapsed an empty <v/>
  and an absent <v> into the same "no cached value" state, so the writer had
  nothing left to emit the type from.

  It matters for formulas that reference another workbook — Excel cannot
  recalculate those without opening the other file, so the cached value is the
  only thing it has to display. An empty <v/> is now read as a cached empty
  string and written back with its t="str"; a genuinely absent <v> still
  means "no cached value", and an empty <v/> on a numeric cell still carries
  no number.

• #123 b3dd9c0 Thanks @baseballyama! - fix: cells without a value were dropped on save, leaving defined names dangling (#111)

  <c r="B1"/> and <c r="C1" s="0"/> disappeared from the saved sheet: the
  writer skipped any cell whose value was null unless it also carried a
  non-default style. Dropping a genuinely unreferenced empty cell is harmless,
  but an empty cell can still be the target of a definedName, a form control's
  fmlaLink, or a drawing or comment anchor — and those were left pointing at
  nothing.

  A valueless cell now always emits. deleteCell — not setCell(…, null) — is
  how a caller says the cell is gone; cells that were never in the model are
  still not written.

• #119 48b0d4d Thanks @baseballyama! - fix: styles.xml lost dxfs, tableStyles, colors and extLst on save (#113)

  Everything after <cellStyles> in xl/styles.xml was dropped when a loaded
  workbook was saved again. <dxfs> was only written when the pool held
  entries, and <tableStyles>, <colors> and <extLst> — where Excel keeps
  custom table styles, the MRU colour palette and the x14/x15 slicer and
  timeline styles — were never read in the first place. Every workbook Excel
  writes carries at least <dxfs> and <tableStyles>, so this hit essentially
  any real file.

  Stylesheet now carries the unmodeled tail (stylesXmlTail) verbatim and the
  writer re-emits it after <dxfs>, where CT_Stylesheet (ECMA-376 §18.8.39)
  puts it. <dxfs count="0"/> is now always written, matching Excel — a
  conditional-formatting rule's dxfId is an index into that list.

• #118 c97258e Thanks @baseballyama! - fix: workbook.xml was written out of CT_Workbook order (#112)

  CT_Workbook (ECMA-376 §18.2.27) is an xsd:sequence, so the children of
  <workbook> have a normative order. Saving a workbook wrote <definedNames>
  straight after <sheets> — ahead of <functionGroups> and
  <externalReferences> — and <pivotCaches> ahead of <calcPr>. A workbook
  carrying both an external reference and a defined name therefore came back as
  a package Excel refuses to open.

  workbook.xml is now emitted in schema order: functionGroups,
  externalReferences, definedNames, calcPr, oleSize,
  customWorkbookViews, pivotCaches, smartTagPr, smartTagTypes,
  fileRecoveryPr, then any unmodeled tail (webPublishing,
  webPublishObjects, extLst) that was carried over from the loaded file.

• #126 21c4a26 Thanks @baseballyama! - fix: workbook.xml lost mc:Ignorable and its namespace declarations on save (#116)

  Saving a workbook whose xl/workbook.xml carries the markup-compatibility
  block Excel writes into essentially every file stripped mc:Ignorable and the
  namespace declarations it names off the root, renamed the child prefixes
  (x15ac: → a generated ns0:, xr: → x16:), dropped the xr2:uid on
  <workbookView>, and moved <workbookPr> behind the mc:AlternateContent
  block. Excel reported the result as corrupt. This is the workbook-level sibling
  of #105, which fixed the same rewrite on the worksheet.

  The root's namespace declarations and mc:Ignorable are now carried through
  verbatim, and every captured child is written with the prefixes the root
  declares — mc:Ignorable names them by prefix, so the two have to agree.
  <workbookView> keeps namespaced attributes it doesn't model (xr2:uid),
  and the unmodeled head is emitted after <workbookPr>, where Excel puts it.