RenderTextSource class

A single piece of painted text discovered by findRenderTextSources, with enough of its own RenderObject exposed to compute word-level highlight geometry without this file needing to know whether the underlying widget was a Text, a RichText, or an EditableText.

Why this exists at all

A SemanticsNode (what walkSemantics in semantics_walker.dart finds matches against) has no public path back to the RenderObject that produced it — that mapping simply isn't exposed by the framework outside of a debug-only accessor (see the example find-in-page spike's _findBarSubtreeId doc, in example/lib/src/sections/find_in_page/find_spike.dart, for the one place that spike does rely on that debug accessor, and why it is an accepted spike-only limitation there).

Rather than trying to resolve "this semantics match's exact RenderObject", this walks the render tree directly — independently of semantics entirely — collecting every RenderParagraph (backing Text and RichText) and RenderEditable (backing EditableText) it finds, each already carrying both the text it painted and the geometry to compute per-character boxes from (RenderParagraph.getBoxesForSelection / RenderEditable.getBoxesForSelection). A caller then correlates one of these against a given FindMatch.globalRect by simple rect containment (see findAllTextOccurrences callers in the example find-in-page spike's find_spike.dart, at example/lib/src/sections/find_in_page/) — geometry is a far more stable correlation key here than any node-identity scheme would be, since a Text/RichText's own Semantics wrapper paints at exactly the same box as the RenderParagraph beneath it (no intervening padding), and an unscrolled, single-line EditableText's RenderEditable does the same relative to its own Semantics node.

This is intentionally not wired into every match — see the example find-in-page spike's build() (in example/lib/src/sections/find_in_page/find_spike.dart) for why only the currently-visible matches are ever resolved against this list: getBoxesForSelection is real per-glyph layout work, and paying it for dozens of scrolled-away matches on every frame would be wasted cost for boxes nobody can see anyway.

Annotations

• @immutable

Constructors

RenderTextSource({required String plainText, required Rect globalRect, required List<Rect> boxesForSelection(TextSelection selection), required RenderObject renderObject})

Creates a RenderTextSource.

const

Properties

boxesForSelection → List<Rect> Function(TextSelection selection)

Returns the global-space rectangles covering selection within this source's painted text — one per visually-distinct run (a wrapped word spans two boxes; a bidirectional boundary can too).

final

globalRect → Rect

This render object's bounding box in the same global (screen-origin) coordinate space walkSemantics computes FindMatch.globalRect in — see RenderObject.getTransformTo (called with a null target) for why the two coordinate spaces coincide.

final

hashCode → int

The hash code for this object.

no setter, inherited

plainText → String

The exact text this render object painted, suitable for locating query occurrences at the same character offsets boxesForSelection expects.

final

renderObject → RenderObject

The underlying RenderParagraph or RenderEditable this source was built from.

final

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

Methods

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

toString() → String

A string representation of this object.

inherited

Operators

operator ==(Object other) → bool

The equality operator.

inherited