RS-22447: Export grid line type to editable PPT chart settings

[chschan]

RS-22447 — Grid line type in PowerPoint/Excel export

RS-22447

Companion to flipStandardCharts#130, which made the Grid line type control (Solid/Dot/Dash) work on-screen. This fixes the export path so exported (PPT/Excel) charts reflect it too.

Problem

getPPTSettings() hardcoded the MajorGridLine Style for both axes to "Solid" (or "None" when the grid width is 0), ignoring the grid dash setting.

Fix

Derive the style from values.grid.dash / categories.grid.dash using the existing getLineStyle() helper. It already:

• returns "None" for a zero-width line, and
• falls back to "Solid" when no dash is supplied,

so behaviour is unchanged when the control isn't present. getPPTSettings() runs before the categories/values → x/y rename, so args$values.grid.dash / args$categories.grid.dash are the correct names here (same as the existing zero-line-dash handling nearby).

Tests

• test-chartsettings.R — new case asserts Dot/Dash reach ValueAxis/PrimaryAxis MajorGridLine$Style, that a zero-width grid stays "None", and that omitting the dash still yields "Solid". Full file: PASS 138 (with flipStandardCharts 1.32.12).

🤖 Generated with Claude Code

[JustinCCYap]

Review note: omitted grid.width now exports Style = "None" instead of "Solid"

getPPTSettings() receives user.args (the raw list(...)) at R/cchart.R:356, before the chart-function defaults are applied, so args$values.grid.width / args$categories.grid.width are NULL whenever the caller doesn't pass them explicitly.

• Old: if (isTRUE(args$...grid.width == 0)) "None" else "Solid" → NULL width yields isTRUE(NULL == 0) → FALSE → "Solid".
• New: getLineStyle(list(width = args$...grid.width, ...)) → is.null(line$width) → TRUE → "None".

So exporting e.g. a Column chart (default y.grid.width = 1, grid visible on-screen) without explicitly passing values.grid.width flips the exported MajorGridLine$Style from "Solid" to "None" — the grid disappears in the PPT/Excel chart and diverges from the on-screen render, the opposite of this PR's fidelity goal. (Masked in the Displayr plugin flow where width is passed explicitly; direct R/API callers are exposed.) The existing tests always pass grid.width explicitly, so this path is uncovered.

Related: getLineStyle also throws on an NA width (if (NA) → "missing value where TRUE/FALSE needed"), where the old isTRUE(...) was NA-safe.

Recommendation: guard the width branch so an absent/NA width preserves the previous "Solid" default. For example:

Style = if (isTRUE(args$values.grid.width == 0)) "None"
        else getLineStyle(list(width = args$values.grid.width %||% 1,
                               dash = args$values.grid.dash))

(or have getLineStyle treat NULL/NA width as unspecified → "Solid" at this call site), and add a test for the omitted-grid.width case asserting "Solid".

Minor: .gitignore adds .positai but not .claude, whereas .Rbuildignore ignores both — consider adding .claude to .gitignore for consistency.

[JustinCCYap]

Does this and the .Rbuildignore‎ changes need to be added?

[JustinCCYap]

LGTM