← All examples

EDJAS spec for the ONS UK National Balance Sheet reference tables.

This is the demonstration of scale: a 500 KB workbook of 19 sheets, whose numbered sheets are ~50-column time-series matrices and whose 216 Excel Tables are auto-generated fragments (Table1, Table2, ...) — names too meaningless to reference. So the spec instead carves out the readable headline statement from 'Table A' by coordinates: net worth at the start of 2024, by institutional sector and asset class, ending in the "Total economy" roll-up. Column headings keep the workbook's glued footnote markers (e.g. "Dwellings2") — fidelity by default.

The specification

The whole of nbs.toml, applied to the unmodified nbs.xlsx.

[extract]
title    = "'Contents'!B5"
source   = "'Contents'!B6"
subtitle = "'Table A'!B1"
contents = "['Contents'!B9:C26 | records]"
header   = "['Table A'!B6:G6]"
rows     = "['Table A'!B8:G19]"

The JSON it produces

See the extracted JSON — exactly what edjas nbs.xlsx nbs.toml prints, which is the thing the report below is only one rendering of.

The report it produces

nbs.xlsx — a large multi-sheet reference workbook

The UK national balance sheet estimates, Office for National Statistics — the full reference-tables workbook: 19 sheets, including ~50-column time-series matrices per institutional sector, carrying 216 auto-generated Excel Table fragments. The spec deliberately extracts only the headline statement from Table A (net worth at the start of 2024, by sector and asset class, down to the Total economy roll-up), demonstrating how a small spec carves a readable brief out of a workbook built for expert reuse. Releases: ons.gov.uk.

Downloaded from ons.gov.uk.