Convert a bounded, contiguous window of an object-path Result into a
normative GeoArrow columnar batch.
This is the inverse of columnarBatchToResult and is bounded the same
way: maxFeatures is required and enforced before any feature is read. A
Result produced by columnarBatchToResult carries enough provenance
to default the batch identity, geometry layout, and attribute bindings; a
plain object-path Result must supply id, schemaId, and identity, and
any attribute it wants lifted into a typed column.
Null and missing attributes are treated identically for the nullable
timestamp and dictionary columns and become null column values. A feature-id
column is not nullable, so a missing or null id fails closed.
Unlike the forward direction this one can be asked to lose something: a
columnar batch has four column roles and an object Result may carry any
number of attributes. An attribute no binding covers therefore fails closed
by default, and unmappedAttributes: "drop" is the only way past it —
reporting every dropped name on
ResultToColumnarBatchResult.droppedAttributes so the loss is stated
rather than assumed.
Type Parameters
T extends Record<string, unknown> = Record<string, unknown>
HonuaGeoArrowError row-limit-exceeded when the window exceeds
options.maxFeatures, checked before any feature is read.
Throws
HonuaGeoArrowError unsupported-layout when the window mixes
geometry kinds, which a single columnar geometry column cannot carry, or
carries an attribute no column binding covers.
Convert a bounded, contiguous window of an object-path Result into a normative GeoArrow columnar batch.
This is the inverse of columnarBatchToResult and is bounded the same way:
maxFeaturesis required and enforced before any feature is read. AResultproduced by columnarBatchToResult carries enough provenance to default the batch identity, geometry layout, and attribute bindings; a plain object-pathResultmust supplyid,schemaId, andidentity, and any attribute it wants lifted into a typed column.Null and missing attributes are treated identically for the nullable timestamp and dictionary columns and become null column values. A feature-id column is not nullable, so a missing or null id fails closed.
Unlike the forward direction this one can be asked to lose something: a columnar batch has four column roles and an object
Resultmay carry any number of attributes. An attribute no binding covers therefore fails closed by default, andunmappedAttributes: "drop"is the only way past it — reporting every dropped name on ResultToColumnarBatchResult.droppedAttributes so the loss is stated rather than assumed.