JSON as a table
An array of objects becomes rows and columns; an array of arrays becomes rows; a {"data": […]}
envelope is opened at its data. Nested objects become dotted columns — address.city — and lists are
joined into one cell. Numbers keep the digits they were written with.
The reader is forgiving about what people do to JSON by hand — // and /* */ comments,
trailing commas, keys without quotes, single-quoted strings — and what it writes back is always strict JSON.
JSON Lines, YAML, XML and property lists
- JSON Lines (
.jsonl,.ndjson): one object per line, one row per line. - YAML: a sequence of mappings reads as a table, and the app writes YAML back.
- XML: the repeated element becomes the row, child elements and attributes become columns.
- Property lists: XML or binary, as macOS and iOS write them.
JSON inside a cell
A cell often holds JSON of its own. ⌥⌘J opens it as a tree — objects and lists that open, keys you can rename,
values whose type you can change — with a text view a click away. Key order is kept, 1.50 stays
1.50, and a twenty-digit identifier does not become a float. Format and Minify work over a
whole column at once. In SQL, JSON_EXTRACT, -> and ->> reach inside.
Writing JSON back
Export as an array of objects or of arrays, as JSON Lines, or as nested JSON, which puts the nesting back
from dotted column names — address.city becomes {"address": {"city": …}}. YAML, XML and
property lists are in the same export sheet.
