PowerPoint greys Flip out for a SmartArt graphic (SmartArt Design ▸ Right to
Left is the way to mirror it), also inside a group, as ungrouping would hand
the flip to the graphic. Before, a flip turned it into a plain group that the
smartArt.* commands then refused. arrange.flipHorizontal and flipVertical
reach the check through shape.flip. Split out of the SmartArt PR at review.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An agent run exporting a deck as an outline got the footer text and the
slide-number field (‹#›) as body lines under each title. PowerPoint's
outline view shows title and body placeholders only, so the text outline,
the RTF outline (Save As ▸ Outline/RTF) and the outline view now skip the
footer-row placeholders as they already skipped the title. Split out of the
frame export PR at review, which asked for all three together.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An agent run found that connecting to a grouped shape answered ok but left
the connector's end unset: the sites of a group member were computed in the
group's child space, and rerouting only looked at the slide's top-level
shapes, so the glue was dropped at once. Both now locate the shape anywhere
in the tree with the affine onto the slide, as PowerPoint glues to grouped
shapes. shape.sites answers in the same slide space. A line inside a group is
refused as `id`: its ends would land in the group's space, and rerouting
does not follow grouped connectors.
Split out of the connector default sites PR at review.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
`cells` must be a list of exact [row, col] pairs inside the table; an entry
that is not one, a cell past the table, or more than 100000 entries is refused
and names the command, instead of being dropped or read as absent (which acted
on the selection or the whole table). Borders use checked arithmetic for the
neighbour cell. Split out of the 0-based index PR at review.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With rotateWithShape false, the renderer lays linear and radial gradients
out upright on the slide, while rectangular and path gradients still turn
with the shape. The shape.fill params doc now says so.
Review follow-up on #1658 (CodeRabbit), both from storing the slide's
mapped scheme in 30ccc38e:
- Copying in Slide Master view used the current slide's colour map. Use
`ctx_for`, so the theme is that of what is being edited.
- `ColorScheme` shares an entry between `dk1` and `tx1`, so a mapped scheme
changed direct `dk1` colours too: a dark fill pasted white. `SourceTheme`
keeps the theme's colours and the map apart, and `bake_theme` maps only
`tx1`, `bg1`, `tx2` and `bg2`, as `Ctx::color` does.
The browser image test now checks that non-picture pastes keep their text.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review follow-up on #1658 (CodeRabbit): `source_theme` stored the master's
scheme, so a slide or layout that overrides the colour map had its colours
fixed through the wrong map. Detaching made this reach a placeholder's
inherited colours: a title drawn white under an override pasted black.
Store the scheme the slide draws with. Paste as Picture reads the same
value.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review follow-up on #1658:
- The drag duplicate cleared `ph` without detaching, so a title or body
dragged off with Alt lost its size and bullets. Copy, Duplicate and the
drag now share `detach_on_slide`.
- Detaching at copy time turned a slide placeholder pasted in Slide Master
view into a regular shape. The clipboard keeps the originals, and a paste
there takes them, so it stays a placeholder as before this PR. Keep
Source Formatting bakes them too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>