YAML to CSV

What this tool does

Turn a block of YAML into a CSV table you can open in a spreadsheet. Paste a top-level list of records, or a mapping whose values are records, and every record becomes one row. The header is the union of every record's keys, in the order they are first seen, so rows that are missing a field simply leave that cell blank instead of shifting columns. Nested mappings are flattened to dot-path columns (address.city), and arrays are rendered the way you choose. Everything runs in your browser — the YAML never leaves your device.

How the shape is read

• A list of records — the common case. Each list item is a row:

  - name: Ada
    age: 36
    address:
      city: London
  - name: Bo
    age: 40
  

  becomes name,address.city,age with one row per person.

• A mapping of records — when the top level is a mapping whose values are themselves records, each entry is a row and its key is kept in an extra column (named by Mapping key column name, key by default):

  alice:
    age: 30
  bob:
    age: 40
  

  becomes key,age / alice,30 / bob,40.

• A single record — a top-level mapping whose values are all scalars is treated as one row (no key column).

Options

• Delimiter — separate columns with a comma, semicolon, tab, or pipe.
• Include header row — turn off to emit data rows only.
• Array handling — json writes an array as one compact JSON string in a single cell (["a","b"]); joined joins a scalar array with , in one cell; columns expands each element into its own dot-indexed column (tags.0, tags.1).
• Quote every field — wrap all cells in double quotes, not just the ones that contain a delimiter, quote, or newline.
• Mapping key column name — the header for the key column when the input is a mapping of records; leave it blank to drop the key.

Limits

• The whole document is parsed in memory, so extremely large files (tens of MB) may be slow in a browser tab.
• Only the top level decides the shape: a list of records, a mapping of records, or a single scalar mapping. A top-level scalar, or a top-level list of scalars, has no columns and is reported as an unsupported shape.
• Deeply nested mappings flatten to dot-paths; if two different structures produce the same dot-path, the later value wins.
• YAML anchors/aliases are expanded by the parser; custom !tags are dropped and only their value is kept.

FAQ