[Geliogabalus]

Description

paper.getContentArea({ useModelGeometry: true }) returns graph.getBBox(), which includes every cell, also the ones the paper hides through cellVisibility. Anything that sizes or fits the paper from the model geometry therefore makes room for content that is not shown:

• fitToContent();
• getFitToContentArea();
• transformToFitContent();
• components built on these methods.

This PR lets useModelGeometry take an options object with a filter. It is non-breaking: useModelGeometry: true and false work as before.

dia.Paper

interface Paper.ModelGeometryOptions {
    filter?: (cell: Cell) => boolean;
}

interface Paper.ContentAreaOptions {
    useModelGeometry?: boolean | Paper.ModelGeometryOptions;
}

paper.fitToContent({
    useModelGeometry: { filter: (cell) => !cell.get('collapsed') }
});

• Accepted by getContentArea() and getContentBBox(). FitToContentOptions and TransformToFitContentOptions now extend ContentAreaOptions, so getFitToContentArea(), fitToContent() and transformToFitContent() pass it through.
• The filter sits under useModelGeometry because it only applies there. On the view path, hidden cells are not mounted, so they are never measured anyway.
• The filter gets only the cell: no this binding and no paper argument, since the caller already holds the paper. It is the same (cell) => boolean signature as the filter of Element.fitToChildren().
• An object without a filter (useModelGeometry: {}) measures every cell, the same as true.

Why only on the paper

Code that measures content through these paper methods passes an options object, so an option is the only way to tell it which cells count. graph.getCellsBBox(cells) has no such need, because its caller already chooses the cells: graph.getCellsBBox(graph.getCells().filter(fn)).

Why cellVisibility is not applied automatically

cellVisibility is also used for viewport culling, where it hides cells outside the visible area. Applying it while measuring would shrink the content area to the viewport, and a paper resized to fit its content would shrink with it. The callback also takes an isMounted argument that has no meaning outside the update loop. Only the caller can tell hidden cells from culled ones, so the filter is explicit.

Tests

• QUnit:
  • getContentArea() / getContentBBox() / getFitToContentArea() / fitToContent() / transformToFitContent() with useModelGeometry: { filter };
  • the filter is called once per cell, with the cell only;
  • an empty area when nothing passes the filter;
  • useModelGeometry: {} measures every cell.
• TypeScript usage in test/ts/index.test.ts.
• One changeset, for dia.Paper.

🤖 Generated with Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

[kumilingus]

Could we nest the filter under useModelGeometry instead of making it a separate option?

paper.getContentArea({ useModelGeometry: { filter: (cell) => !cell.get('hidden') } });

interface ContentAreaOptions {
    useModelGeometry?: boolean | { filter?: (cell: Cell) => boolean };
}

• filter only works with useModelGeometry. As a separate option, { filter } alone is silently ignored. Nesting it rules that combination out.
• It follows the existing boolean | { ...options } pattern (e.g. snapLinks: boolean | { radius }).
• Code that only forwards useModelGeometry picks up the filter with no changes, and the core check stays truthy.