Turn a CSV export into JSON without uploading the file
Spreadsheet exports arrive with inconsistent delimiters, stray quotes and columns that mix numbers and text. This converter lets you pick the separator, decide how headers are treated, and choose whether values are coerced into numbers, booleans and nulls.
- Delimiter
- Auto-detected
- Type coercion
- Optional
- Rows
- Any number
Measured on Apple M4 Max: 4.67 ms parse with delimiter detection and type inference (10,000 rows × 7 columns) See the benchmarks
How to use it
- Paste the CSV Or drop a file onto the input. The first few lines are analysed to detect the delimiter, whether the first row is a header, and whether any field contains a delimiter that must be quoted.
- Set the conversion rules Override the delimiter if the detection is wrong, choose between an array of objects and a single keyed object, and decide whether numbers, booleans and empty strings should be typed or kept as text.
- Copy the JSON The output is pretty-printed by default so it stays readable, and every row becomes an object keyed by the header names.
Questions people ask
How does it detect the delimiter?
It counts the occurrences of comma, semicolon, tab and pipe across the first lines and picks the most consistent one. You can always override the choice with the delimiter selector.
Should I enable type inference?
Only for genuinely numeric or boolean columns. Keep it off for postcodes, phone numbers, account numbers, product codes and any value that may start with a zero.
What happens to rows with the wrong number of columns?
They are counted and reported in the output summary, and the extra or missing fields are marked so you can see the problem instead of having it silently absorbed.
Can I convert a file with quotes inside fields?
Yes. A field wrapped in double quotes may contain commas, newlines and escaped double quotes, and the parser reads it as a single value.
Does it handle UTF-8 characters and accents?
Yes. UTF-8 is the expected encoding, and accented letters, emoji and non-Latin scripts survive the conversion unchanged.
Can I get a keyed object instead of an array?
Yes. Choose the object output and pick which column becomes the key — a common pattern is to key the records by an id column.
Where this earns its keep
Feeding an analytics export into a script
A CSV of product rows becomes a JSON array you can paste into a Node script, a fetch body or a fixture, without writing a parsing step first.
Migrating a spreadsheet into a database seed
Type inference turns an "amount" column of 12.50 into a number rather than a string, which is what a seed file for a numeric column expects.
Debugging a broken import
When a CSV import rejects row 47, the converter shows you how that row was parsed, so you can see whether the problem is a stray quote or a ragged column count.
Details that catch people out
A zip code of 01234 becomes the number 1234 if you enable inference, and a postal code is a string in every country on earth. Leave inference off for identifiers, phone numbers, postcodes and any column with leading zeros.
A row with more fields than the header, or fewer, is flagged. Padding it silently is the fastest way to hide a real data problem.
Last reviewed 2026-10-02. Every tool on DevSnap runs entirely in your browser, so whatever you paste here never reaches a server.