@textcortex/slidewise@1.21.1

Patch Changes

• 94e348b: fix(render): emit a valid SVG <image> for image-fill backgrounds

  renderDeckToSvg rendered a slide's image-fill background as a CSS background
  shorthand inside an SVG fill="…" attribute — fill="center / cover no-repeat url("data:image…")". That is not valid SVG (a non-paint value plus nested
  unescaped quotes); browsers tolerate it, but strict rasterisers
  (@resvg/resvg-js, librsvg, batik) reject it, blocking a Chromium-free
  parsePptx → renderDeckToSvg → resvg → PNG path.

  The pptx importer stores image backgrounds as a CSS shorthand, but the
  renderer's image-ref detection only inspected the start of the value, so the
  shorthand fell through to the fill path. The renderer now recognises a
  url(...) anywhere in the value and emits a real <image> element
  (preserveAspectRatio = slice for cover, meet for contain). A lock-in
  test asserts every rendered slide is valid SVG a strict XML parser accepts,
  with an image-background slide as the regression case.

@textcortex/slidewise@1.21.0

Minor Changes

• 652f804: feat(pptx): layout-instantiation in applyEdits (lossless scale-with-variety)

  applyEdits now supports source: { layoutId, fills? } in a PlannedSlide —
  instantiating a fresh slide from one of the template's own layouts inside
  the lossless byte-patch path. Because the layout is already a part of source,
  the new slide binds to ppt/slideLayouts/<layoutId>.xml (inheriting theme /
  master / background chrome) while every other part stays byte-identical. This
  unlocks lossless and scale-with-variety in one deck: clone slides where you
  want the exact thing, instantiate from layouts where you want variety.

  Each layout placeholder is materialised as an addressable, positioned element
  with a deterministic id — layoutSlotElementId(layoutId, key) (exported) where
  key is the placeholderKey / summarizeLayouts slot key. Text/obj slots
  are populated from fills and editable via setText; picture slots become a
  <p:pic> with a transparent placeholder blip so setImage can repoint them;
  chart/table/other slots expose their geometry so the host fills them with
  addChart / addDiagram. Placeholder geometry is read EMU-native from the
  layout (falling back to the matching master slot), so it stays correct without a
  canvas-px round-trip. An unresolvable layoutId is surfaced via onWarning and
  the slide is skipped rather than shipped wrong.

• 652f804: feat(render): headless renderDeckToImages + deck.fontUsage font transparency

  Headless render-to-image (visual-QA loop). New browser-free renderer that
  draws what the editor draws — native charts (buildChartOption + ECharts
  SSR), diagrams (layoutDiagram), text/shapes/images/backgrounds in z-order —
  not the OOXML raster fallbacks. No Playwright/Chromium/DOM.

  • renderDeckToSvg(deck, opts?) → one composed SVG per slide (ECharts is
    loaded on demand, so it never bloats the editor bundle).
  • renderDeckToImages(deck, opts?) / renderSlideToImage(deck, i, opts?) /
    renderPptxToImages(bytes, opts?) → raster bytes. Rasterisation is an
    injected hook (opts.rasterizeSvg, e.g. @resvg/resvg-js); when omitted the
    default tries a dynamic @resvg/resvg-js import and throws a clear error if
    it isn't installed — so there's no hard native dependency.
  • opts: slides (1-based subset), dpi (canvas scales by dpi/96),
    format, maxWidth (thumbnail cap). Deterministic (no animation).

  Enables the host's render → fresh-eyes inspect → targeted applyEdits fix →
  re-render cycle, rendering a final applyEdits output directly.

  Font transparency. parsePptx now stamps deck.fontUsage: { family, embedded }[] — every font family the deck's text uses, flagged
  whether the source PPTX actually embeds it (<p:embeddedFontLst> → a real
  ppt/fonts/* part) or merely references it (system-fallback risk on
  viewers that don't ship the brand font). Hosts use it to warn at generation
  time ("missing fonts for some ppts"). It's a read-only diagnostic, distinct from
  deck.fonts (the embeddable payloads).

@textcortex/slidewise@1.20.0

Minor Changes

• e56ddd2: feat(pptx): applyEdits — lossless surgical-edit API

  Add applyEdits(source, plan, options?): a patch on the original .pptx bytes
  rather than a full re-serialize. The create flow can now emit an EditPlan
  (subset/reorder/repeat of template slides, each with edits) and get back a valid
  package where everything not named by an edit is byte-identical to the source —
  masters, layouts, theme, embedded fonts, ppt/tags/*, notes, embeddings, and
  any untouched element. This removes the lossy round-trip that produced the
  custGeom/SVG-fallback/dangling-rel fidelity bugs and lets hosts drop their
  defensive cleanup. serializeDeck stays for the live editor / from-scratch decks.

  Edits address elements by the same stable ids parsePptx returns; slides by
  1-based template index. Supported ops: setText/clearText (preserve the
  template box + run styling, or rebuild from supplied runs), setChartData
  (repopulate a native chart in place — type/colours kept, caches and the
  embedded xlsx workbook updated so Edit-Data still works), setTableData,
  setImage, removeElement, addChart, addDiagram, plus per-slide
  background and deck title. Removed slides and any parts that become
  exclusive to them are reclaimed by a package-wide reachability sweep, then
  dangling relationships and content-types are reconciled. Unresolved element ids
  and unsupported layout-instantiation are surfaced via onWarning instead of
  throwing.

@textcortex/slidewise@1.19.1

Patch Changes

• 305dc0f: fix(pptx): emit a structurally valid package on serialize

  Three serializeDeck bugs corrupted the generated .pptx (missing parts /
  invalid image bytes) even from clean source templates, triggering a PowerPoint
  repair prompt and outright rejection by stricter consumers (Google Slides,
  LibreOffice, OOXML validators):

  • Dangling tags relationships: the chrome-preserve path re-pointed a
    slide's tag rel at a slidewise_preserved_* name, then clobbered that part by
    re-copying the source tags under their original names. The rel now resolves
    to the de-prefixed part it should always have pointed at.
  • Dangling notesMaster relationships: pptxgenjs writes a notesSlide per
    slide linked to a notes master, which chrome preservation removed without a
    source replacement. The orphaned (implicit, non-body-referenced) relationship
    is now dropped.
  • SVG markup in .png raster fallbacks: dual SVG images (<a:blip> raster
    • <asvg:svgBlip> vector) had the SVG source written into the .png
      fallback. The fallback is now a real rasterized PNG (browser) or a valid
      transparent PNG (SSR/Node); the vector svgBlip part is untouched.

  Adds a final reconcileDanglingRels invariant guard — every internal
  relationship target must resolve to a shipped part — that backstops both
  dangling-rel shapes (repairing recoverable targets, dropping only
  safe-to-remove optional ones, and leaving critical rels untouched). Also runs
  pruneDanglingContentTypes on the source-preservation path so stale
  [Content_Types] overrides (pptxgenjs's slideMaster1..N, leftover notes
  overrides) can't invalidate the package either.

@textcortex/slidewise@1.19.0

Minor Changes

• 3e7c3f1: First-class diagram element (P3 / F3). A new DiagramElement models
  process / timeline / funnel / matrix / cycle / list visuals as an ordered set
  of labelled nodes instead of a hand-placed cluster of shapes and lines. The
  renderer and the PPTX writer share one layout function (layoutDiagram, also
  exported), so the on-canvas preview and the saved file can't drift. On export a
  diagram serialises to a single labelled <p:grpSp> of real shapes + connectors
  — so it stays grouped and editable in PowerPoint (move/resize as one unit)
  rather than collapsing to anonymous floating shapes. Exposed via the
  DiagramElement / DiagramNode / DiagramKind types and the layoutDiagram
  helper.

• 3e7c3f1: Host deck-generation ergonomics (review follow-ups for layout instantiation):

  • summarizeLayouts(deck, options) — { compact: true } returns the minimal
    { id, name?, role, fillable } shape (no geometry) for a tight model-context
    budget; { dedupe: true } collapses layouts that share a role + full
    placeholder-slot inventory (text and non-text chart/picture/table slots)
    into one representative carrying the rest as aliases, so an 85-layout
    template surfaces as its handful of distinct kinds — and a chart-bearing
    layout never collapses into a text-only twin. Composable.
  • Robust sourceLayoutId resolution — a host-authored slide's
    sourceLayoutId now resolves from deck.layouts or, when the host didn't
    carry the layouts array, by the ppt/slideLayouts/<id>.xml id convention
    against the { source } archive. When it resolves to nothing, serializeDeck
    emits a structured { code: "layout-unresolved", slideIndex, layoutId }
    warning instead of silently falling back.
  • Richer chrome-skipped warning — now carries the detected sourceAspect
    and outputAspect so a host can explain why chrome was dropped.

  Plus README: the host "author-a-slide" contract ({ sourceLayoutId, background: "transparent", elements } without a JS call), the recipe for
  placing non-text slots from summarizeLayouts geometry, and the server-side
  layoutDiagram render recipe with its DOM-free guarantee.

• 3e7c3f1: Round-trip fidelity fixes (P4):

  • Image crop / radius now round-trip. Previously an image's crop
    (<a:srcRect>) was read on import but silently dropped on export, and corner
    radius was neither parsed nor written. serializeDeck now routes any image
    carrying a crop or radius through a dedicated <p:pic> writer (emitting
    <a:srcRect> and roundRect geometry) instead of pptxgenjs — whose
    cover/contain sizing emits its own <a:srcRect> and would fight a user crop —
    and parsePptx reads a rounded picture's corner radius back. Plain
    (uncropped, square-cornered) images keep the existing path unchanged.
  • Text-run letter-case (cap) now round-trips. A run's cap
    ("all" / "small", OOXML <a:rPr cap>) was parsed on import but dropped on
    export (pptxgenjs has no cap option). It's now re-applied per run in
    post-process, so all-caps / small-caps styling survives a save.

• 3e7c3f1: Harden layout instantiation for AI deck generation (P1 / F1):

  • Layout-instantiated slides now inherit their layout's background. A slide minted by addSlideFromLayout with the default transparent background no longer serialises an explicit <a:noFill/> <p:bg> — that empty background was overriding the layout/master/theme inheritance, so instantiated slides lost their on-brand background. They now stay <p:bg>-less and paint from their sourceLayoutId layout's chrome (matching the source-slide guarantee for cloned/reordered slides).
  • Layout-selection metadata. DeckLayout.type now carries the raw OOXML <p:sldLayout type> role, and the new summarizeLayouts(deck) returns a compact, model-context-friendly layout menu (friendly role label, fillable fills keys, per-placeholder kind/category/geometry) so a host can have a model pick a layout per slide. placeholderKey(ph) exposes the exact fills key for a placeholder.

• 3e7c3f1: Machine-readable serialization diagnostics (P2 / B3). serializeDeck now
  accepts SerializeOptions.onWarning, a callback invoked with a structured
  SerializeWarning when the output degrades. The key case is
  "chrome-skipped" — emitted when a source template's masters / layouts /
  theme / fonts can't be carried over because its slide size is unreadable, so
  the deck falls back to generic chrome. Hosts can now detect and surface the
  degradation instead of only seeing a console line. (Non-16:9 sizing for 4:3 /
  16:10 / custom templates already drives the output slide size; this adds the
  escape-hatch signal when it can't.)

@textcortex/slidewise@1.18.1

Patch Changes

• f16cfa5: fix(pptx): respect z-order for synthesised content (charts, custGeom shapes, connectors)

  Synthesised spTree content was always inserted at the back of the slide, so an
  in-app chart / custGeom "SVG" / connector with a higher z than its background
  card was buried behind that card (and its shadow) and rendered invisible. Each
  synth item now anchors directly on top of the pptxgenjs node it sits above
  (matched by shape name), preserving the deck's z-order, and only falls to the
  back when it is genuinely below every model element.

@textcortex/slidewise@1.18.0

Minor Changes

• 17c069f: Template-faithful serialization fixes and new authoring primitives for AI deck generation:

  • Per-slide source mapping — Slide.sourceSlideIndex lets a host that clones / reorders / subsets imported slides declare which source slide each output slide replays its background and layout from, instead of mapping by output position (fixes blank/white-on-white slides on reordered decks).
  • Deep chart preservation — preserved charts now carry their full dependency tree (embedded workbook, colors/style parts, rels, content types), so they no longer trigger a PowerPoint repair and keep "Edit Data" + custom styling.
  • Non-16:9 templates — a source's real slide size now drives the output (4:3, 16:10, custom), preserving its masters / layouts / theme / fonts and inverting the authoring-canvas fit for emitted elements, instead of silently falling back to a generic 16:9 deck.
  • Instantiable layouts — parsePptx exposes master layouts on Deck.layouts, and the new addSlideFromLayout(deck, layoutId, opts) mints a fresh slide bound to a layout with its text placeholders ready to fill.
  • Chart-option helpers — buildChartOption, defaultPaletteColor, and makeValueFormatter are exported so hosts can build the exact ECharts options Slidewise renders (e.g. for server-side previews).
  • Connector primitive — a first-class connector element (straight / bent / curved, arrowheads, flips) with renderer + <p:cxnSp> writer support.

@textcortex/slidewise@1.17.1

Patch Changes

• 83683e3: fix(pptx): render <a:grpFill/> shapes by inheriting the group's fill

  Decorative line-art (e.g. the swoosh graphic on a title slide) is often
  authored as many <a:custGeom> segments inside a single <p:grpSp>, with every
  segment declaring <a:grpFill/> so they share the group's one translucent
  colour. parsePptx had no grpFill branch, so each segment fell through to
  transparent and the entire graphic disappeared on import — present in the
  source but invisible in the deck.

  The group's resolved fill is now threaded down to descendants, and a shape with
  <a:grpFill/> paints with it (walking up the group chain when an inner group
  defines no fill of its own). Geometry was already preserved; only the paint was
  being dropped.

@textcortex/slidewise@1.17.0

Minor Changes

• 019e000: feat(pptx): support PowerPoint templates (.potx)

  .potx and .pptx share an identical OOXML package; only the main part's
  content type in [Content_Types].xml differs. This adds first-class template
  support across import and export:

  • parsePptx already parsed .potx transparently (it reads parts by path, not
    by content type) — now the rest of the pipeline preserves template-ness.
  • New exported isPptxTemplate(blob) detects a template by inspecting the
    package content type rather than trusting a filename extension (a mis-named
    .pptx that is really a template is detected correctly).
  • serializeDeck gains an asTemplate?: boolean option. When omitted,
    template-ness is inherited from the source archive, so a parsed .potx
    round-trips back to a .potx; pass true/false to force the output kind.
    Templates are emitted with the …presentationml.template.main+xml main-part
    content type and the .potx MIME type.

@textcortex/slidewise@1.16.4

Patch Changes

• a085c8d: fix(pptx): import-fidelity fixes for think-cell / brand-template decks

  • Skip shapes flagged hidden="1" (e.g. think-cell "do not delete" data objects)
  • Render run-level text highlight (<a:rPr><a:highlight>) end to end
  • Apply cap="all"/"small" (including when inherited from a placeholder list style) as a render-time letter-case transform
  • Derive font weight from weight-named families ("Gilroy ExtraBold" → 800, "… Medium" → 500, …) so substitute fonts render at the right heaviness
  • Tables: per-cell fills, text colours, per-side borders, proportional column widths / row heights, cell spans (gridSpan/hMerge/rowSpan/vMerge), per-cell vertical anchor, and rich per-cell runs (highlight / bold / ✓ glyphs / bullet line breaks). Unfilled cells stay transparent instead of inheriting a sibling fill
  • Map Wingdings bullet glyphs to Unicode (ü→✓, q→☐, §→▪, …)
  • Bullets: repeat a character bullet across in-paragraph line breaks, suppress the glyph on empty paragraphs, and trim trailing empty paragraphs
  • Synthesise block-arrow paths (down/up/left/rightArrow) and resolve outline colour from <p:style><a:lnRef> so dashed/outlined shapes draw
  • Keep a text-bearing preset or custom-geometry shape's fill, border, and corner radius behind its text (roundRect callouts, outlined chevrons)
  • Honour <a:bodyPr><a:spAutoFit> no-wrap for short single-line labels; skip the arrow-tip text inset on no-fill label shapes
  • Render per-paragraph hanging-indent bullets as one block per line so multi-line items align correctly