JSON & Data

JSON is easy to write and awkward to read, especially when it arrives minified from an API or nested six levels deep. These tools format and validate it, show you where a parse error actually is, compare two documents so you can see what changed, and convert between JSON, YAML, CSV, XML and TOML. Every one of them parses in the browser, so you can paste a production response or a token payload without it leaving your machine.

Tabular data has its own kind of comparison. Two exports of the same table rarely differ in the way a text diff reports: rows move, and what you want to know is which rows are new, which are gone and exactly which cells changed — matched on a key column, so reordering means nothing.

A CSV can also go straight into a database. Each column’s type is worked out from every value in it — whole numbers, exact decimals, floating point, booleans, dates, timestamps or text, with a leading zero keeping a postcode as text — and the table comes out as CREATE TABLE and INSERT statements for PostgreSQL, MySQL, SQLite or SQL Server, each quoted and escaped by that database’s own rules.

Data that comes one record a line — NDJSON or JSON Lines, from a log or an export — is checked line by line, turned into a JSON array or back, and summarised field by field, with every number, string and key copied exactly as written rather than parsed and rewritten.