TOML to CSV
Two ways to get a TOML file into a spreadsheet: a flat key,value sheet with one row per setting, or a proper table with one row per [[entry]] and a column per field. Pick the layout, pick the delimiter, download the CSV.
What this tool does
It converts a TOML document into comma-separated rows, in one of two shapes. Key / value flattens the whole document into dotted paths and writes a two-column sheet — every setting in the file, one per row, which is what you want for review or diffing in a spreadsheet. Table takes an array of tables and writes it as a real data table: a header row from the union of all keys, then one row per entry.
The intent it closes: "I need this config in a spreadsheet." Sometimes that is for a review meeting, sometimes to compare two environments side by side, sometimes because the list of servers or feature flags in the config is genuinely tabular data that lives in a config file.
Nested values are handled honestly. In table mode, sub-table fields become dotted column names, so a [[servers]] entry with tls.enabled gets a tls.enabled column. Arrays inside a cell are written as compact JSON, since CSV has no way to nest.
When you'd reach for it
- Review a config with people who do not read TOML. A two-column sheet is a format everyone can open.
- Compare environments. Export staging and production to CSV, paste both into one sheet, and let conditional formatting show the differences.
- Pull tabular data out of a config. Lists of servers, routes, rate limits, or feature flags stored as
[[arrays]]are already rows and columns. - Feed a bulk-edit workflow. Export, edit in a spreadsheet, then bring it back with CSV to TOML.
- Get a count. Sometimes you just want to know how many keys a config has, and a spreadsheet answers that in one click.
How it works
The two layouts share a parser and diverge after it.
1. Parse the TOML
The document is parsed into a value tree, so dotted keys are already nested and [[array]] blocks are already arrays before any flattening happens. A parse error stops the run and the status bar reports it.
2. Choose rows
In Key / value mode the whole tree is flattened to dotted paths and each becomes a row. In Table mode the tool finds the array of tables to export — either the path you typed, or the first one it finds walking the document — and takes each entry as a row. Column headers are the union of every entry's keys, in first-seen order, so a field missing from one entry leaves an empty cell rather than shifting the row.
3. Write CSV
Values are stringified: dates as their TOML text form, booleans as true/false, arrays as JSON. Any cell containing the delimiter, a quote, or a newline is wrapped in double quotes with internal quotes doubled, per RFC 4180. Rows are joined with LF, or CRLF if you picked the Excel option.
Options reference
Layout
Key / value gives you the whole document as key,value rows, with dotted paths for nested keys — complete, but not tabular. Table gives you one array of tables as a proper sheet with a header row. If the document has no array of tables, table mode reports that instead of guessing.
Array path
Which array to export in table mode, as a dotted path such as servers or tool.deploy.targets. Leave it blank to auto-detect the first array of tables in the document — enough for the common case where there is only one.
Delimiter
Comma is the default and the safest for tooling. Semicolon is what Excel expects in locales that use a comma as the decimal separator. Tab avoids quoting almost entirely — if that is your preference, tsvkit.org is the sibling site for TSV work.
Line endings
LF for anything Unix-shaped, CRLF when the file is going straight into Excel on Windows. Both are valid CSV; only Excel is picky.
Example
Input, in Table layout:
[[servers]]
name = "alpha"
ip = "10.0.0.1"
region = "us-east"
[[servers]]
name = "beta"
ip = "10.0.0.2"Output:
name,ip,region
alpha,10.0.0.1,us-east
beta,10.0.0.2,The header is the union of both entries' keys. beta has no region, so its cell is empty rather than the row being short. The same input in Key / value layout would produce five rows: servers.0.name, servers.0.ip, servers.0.region, servers.1.name, servers.1.ip.
FAQ
Which layout should I pick?
If the thing you want in the spreadsheet is a list — servers, routes, flags stored as [[entries]] — pick Table. If you want the whole config for review or diffing, pick Key / value. Table mode needs an array of tables to exist; key/value mode always works.
What happens to nested tables inside an array entry?
Their fields become dotted column names: an entry with tls = { enabled = true } contributes a tls.enabled column. That keeps every value addressable without inventing extra header rows.
How are arrays written into a cell?
As compact JSON — ["a","b"]. CSV cannot nest, and JSON at least round-trips: CSV to TOML reads it back as a string you can convert, and any spreadsheet shows it readably.
Will Excel open the output correctly?
Yes, with two caveats worth knowing: pick the semicolon delimiter if your Excel locale uses commas for decimals, and pick CRLF line endings. Quoting is already RFC 4180-compliant.
Can I get back to TOML afterwards?
Yes — that is what CSV to TOML is for. It reads both shapes: a two-column key/value sheet (with optional re-nesting of dotted keys) and a full data table exported as an array of tables.