CSV Formatter

Drop in a file or paste text, and the tool parses it according to RFC 4180 rules, then displays the result as a table with the first row treated as a header.

Key features:

• Delimiter override. Auto-detection covers the common cases, but European exports often use ; (because the comma is a decimal separator) and database dumps frequently use a tab. Switch the delimiter and the table re-parses instantly.
• Quote-aware parsing. A field like "Smith, John" stays as one cell, and "She said ""hi""" correctly becomes She said "hi".
• Column sorting. Click a header to sort ascending or descending. Numeric-looking columns sort numerically, not lexically, so 2 comes before 10.
• Row and column counts. A quick sanity check that the file parsed into the shape you expected.
• CSV to JSON. Export the parsed data as objects or as arrays.

Because parsing happens locally, large pastes work without a network round-trip, and nothing is logged or uploaded.

The classic reason a CSV looks broken is a comma inside a value. Consider this file:

id,name,city,amount
1,"Doe, Jane","Portland, OR",1250.00
2,Lee,Austin,980.50

A tool that splits on every comma turns row 1 into six columns and misaligns everything after it. Proper parsing respects the quotes and keeps Doe, Jane and Portland, OR as single cells, giving you a clean four-column table.

Convert that to JSON as objects and you get:

[
  { "id": "1", "name": "Doe, Jane", "city": "Portland, OR", "amount": "1250.00" },
  { "id": "2", "name": "Lee", "city": "Austin", "amount": "980.50" }
]

Note that values come through as strings by default. CSV has no type system, so 1250.00 is text until you decide to coerce it. If your downstream code expects numbers, cast them explicitly rather than assuming the parser guessed right.

• Eyeballing an export. You exported users, orders, or logs from some admin panel and just want to confirm the columns are what you think before writing code against them.
• Fixing a delimiter mismatch. A file opens as one giant column in your spreadsheet because it used ; and the importer assumed ,. Re-pick the delimiter here and read it correctly.
• Feeding an API or script. Turn a CSV into a JSON array of objects to paste into a request body, a test fixture, or a seed file.
• Spot-checking data quality. Sort a column to find blanks, outliers, or rows that drifted out of alignment because of an unescaped quote.
• Sharing a quick view. When a colleague sends a raw CSV, paste it in to make it legible without opening heavyweight spreadsheet software.

A few things that trip people up:

• The header row matters. The first row becomes your keys when exporting to JSON objects. If the file has no header, the first data row gets treated as one, so add a header line or export as arrays instead.
• Empty fields versus missing fields. a,,c is three fields with an empty middle value, which is different from a row that simply has fewer columns. Ragged rows usually signal a quoting bug earlier in the file.
• A stray quote poisons everything after it. One unbalanced " can make the parser swallow the rest of the file as a single field. If the row count looks far too low, search the source for a lone quote.
• Leading zeros and big numbers are text in CSV. A ZIP code 02139 or an ID like 100000000000000001 is safe as a string. Be careful before coercing to a number, which can drop the zero or lose precision.
• BOM and encoding. A file saved with a byte-order mark can make the first header key start with an invisible character. If a key like id will not match, the BOM is a likely culprit.

CSV is a convention more than a strict standard, but RFC 4180 describes the common rules: records separated by line breaks, fields separated by commas, and any field containing a comma, a quote, or a newline wrapped in double quotes with internal quotes doubled ("").

What CSV does not define is just as important:

• No types. Every value is text. true, 42, and 2026-01-01 carry no type information.
• No fixed delimiter. Despite the name, semicolons, tabs, and pipes are all common in the wild.
• No encoding declaration. The file does not tell you whether it is UTF-8, Latin-1, or something else. Mojibake usually means an encoding mismatch.
• No schema. Nothing enforces that every row has the same number of columns.

This is why a viewer that exposes the delimiter, respects quoting, and shows row and column counts saves real time: it makes the implicit choices visible so you can correct them instead of guessing.