File format
JSON for tabular data: structure and conversion
Fileshape reads and writes JSON (.json, MIME type application/json). Conversion runs in your browser.
JSON is the native format of web APIs. Tabular data usually appears as an array of objects, one object per record, often wrapped inside an envelope with metadata like paging. Unlike CSV, JSON has types (strings, numbers, booleans, null) and nesting, so a single record can contain sub-objects and lists.
When reading, Fileshape locates the largest array of objects in the document and flattens nested objects into dot-named columns, joining simple arrays with semicolons. When writing, values are emitted as strings to avoid corrupting identifiers, and an option rebuilds nested objects from dot-named columns.
Quirks
- Key order isn't significant in JSON, but Fileshape preserves first-seen order for predictable columns.
- Arrays of objects are kept as JSON text in one cell.
- Large arrays are better exchanged as JSON Lines.