ColumnLite

Capitalized list of supported PyArrow datatypes outlined in https://arrow.apache.org/docs/python/api/datatypes.html.

Timestamp datatype is expected to be used with unit measured in "us" and timezone set to "UTC" inside of parquet column header schema or it won't be accepted; an example for PyArrow binding would look like this: pa.timestamp("us", tz="UTC")

Qualified path of the group this column belongs to: a bare title for a top-level group, or segments joined by ▸ for a nested group. Null means ungrouped.

Publisher-supplied free-form metadata for the column, as a JSON object. Keys must be 1-100 chars and cannot contain leading whitespace or dot, or the characters / \ : > < | ? " *; keys are unique case-insensitively. Values may be any JSON-compatible type. A single tags object serialised as UTF-8 JSON must be at most 10 KiB.

The human-readable label used to identify the column

The ID of the column's unit