CSV to JSON Converter

What is JSON and Why Convert from CSV?

// Input CSV (comma + LF, header on, infer types on)
id,name,active,score
1,Alice,true,98.5
2,Bob,false,87
3,Carol,true,

// Output JSON
[
  { "id": 1, "name": "Alice", "active": true, "score": 98.5 },
  { "id": 2, "name": "Bob", "active": false, "score": 87 },
  { "id": 3, "name": "Carol", "active": true, "score": null }
]

// Same input with Header off (no first-row keys)
1,Alice,true,98.5
2,Bob,false,87

// Becomes
[
  { "col1": 1, "col2": "Alice", "col3": true, "col4": 98.5 },
  { "col1": 2, "col2": "Bob", "col3": false, "col4": 87 }
]

Key Features

Examples

id,name,email,role
1,Alice,alice@example.com,admin
2,Bob,bob@example.com,editor
3,Carol,carol@example.com,viewer
4,Dan,dan@example.com,viewer

ts	event	user	duration
2026-05-09T10:00:00Z	signup	alice	142
2026-05-09T10:01:00Z	login	alice	87
2026-05-09T10:02:00Z	checkout	alice	312
2026-05-09T10:03:00Z	logout	alice	44

id;name;price
1;Alice;1234,56
2;Bob;9876,54
3;Carol;42,00

name,role,note
"Smith, Jr.",admin,"He said ""hi"""
"Doe, Jane",editor,"Two
lines"

id,event,user
9007199254740993,signup,alice
9007199254740994,login,bob
9007199254740995,checkout,carol

1,Alice,admin
2,Bob,editor
3,Carol,viewer
4,Dan,viewer

How to Use

1. 1

   Paste your CSV

   Enter or paste your CSV into the input field above. The tool accepts comma, semicolon, tab, and pipe-delimited data. You can also click 'Load example' to try a sample like a spreadsheet export, TSV log, or Excel-EU CSV with semicolons.

2. 2

   Pick the delimiter (or Tab)

   Click `,` (default), `;` (Excel-EU semicolon), `\t` (TSV), or `|` (Pipe) to switch the delimiter in one click. Open the Options panel for fine control: Header on/off and Infer types on/off. Header off auto-names columns col1, col2, col3.

3. 3

   Copy or Download the JSON

   Click Copy to grab the JSON to your clipboard, or Download to save it as a .json file ready for your code, API, or pipeline. For round-trips, click Swap direction to convert JSON back to CSV in place.

Common Conversion Pitfalls

name,role
Smith, Jr.,admin
// Parser reads 3 columns: "Smith", " Jr.", "admin"

name,role
"Smith, Jr.",admin
// Parser reads 2 columns: "Smith, Jr.", "admin"

// Wrong delimiter (default comma) on Excel-EU file
id;name;price
1;Alice;1234,56
// Each row becomes one column: { col1: "1;Alice;1234,56" }

// Correct: pick `;` Delimiter chip
id;name;price
1;Alice;1234,56
// Output: { id: 1, name: "Alice", price: "1234,56" }

// Without big-int detection
{"id": 9007199254740993}
// JavaScript reads as 9007199254740992 (precision lost)

// With Infer types on, big integers stay as strings
{"id": "9007199254740993"}
// Use BigInt(value) in code to preserve precision

id, name, email
1, Alice, alice@example.com
// Output keys: "id", " name", " email" (with leading spaces)

id,name,email
1,Alice,alice@example.com
// Output keys: "id", "name", "email" (clean)

name,role,note
Alice,admin
Bob,editor,first day
// Row 1 is short by one cell

// Output (note empty/null cell in row 1)
[
  { "name": "Alice", "role": "admin", "note": null },
  { "name": "Bob", "role": "editor", "note": "first day" }
]

// Expecting a Date object in the output
ts,event
2026-05-09T10:00:00Z,signup
// Output ts is the string "2026-05-09T10:00:00Z", NOT a Date

// Correct: parse at the point of use in your code
const rows = JSON.parse(output);
const when = new Date(rows[0].ts);
// when is now a Date object

Common Use Cases

Spreadsheet Export to API Import

Paste a CSV exported from Excel, Google Sheets, or Numbers and get a JSON array of objects ready for POST to a REST API, GraphQL mutation, or bulk-import endpoint. The most common use case — analysts produce spreadsheet data, engineers need typed JSON to feed the backend.

Excel Export to Tooling

Convert Excel CSV exports (including Excel-EU semicolon-delimited files with the `;` chip) into JSON for processing with JavaScript tooling, jq scripts, or any system that reads JSON. The parser handles BOM stripping and CRLF line endings correctly so Excel exports don't break on the first row.

TSV Log to Analytics

Tab-separated logs from BigQuery exports, Snowflake unloads, Vector pipelines, or Unix tools (cut, awk) often arrive as .tsv. Pick the Tab Delimiter chip and get a typed JSON array ready for ad-hoc analysis, dashboard ingest, or pipeline-stage transformation.

Database CSV Dump to ETL

Convert PostgreSQL COPY TO CSV output, MySQL SELECT INTO OUTFILE, or any database CSV dump to JSON for loading into a NoSQL store, feeding into a JavaScript ETL pipeline, or shipping to BigQuery as line-delimited JSON. Big-integer detection preserves numeric IDs that exceed JavaScript's safe range.

Postman/Newman CSV Test Result Consumption

Postman test runs export CSV reports of pass/fail per request. Convert to JSON for programmatic consumption — feed into a status dashboard, alert pipeline, or test-result aggregator. Mixed-shape rows (failed tests have an extra error column) are handled with empty/null fills.

Small CSV to Quick JSON Config

Have a small CSV of constants — currency codes, country names, product SKUs — and need a JSON array for a config file or a JavaScript constant? Paste, copy, paste. With Infer types on, numbers and booleans are typed correctly; with Header on, you get an array of named-field objects ready to drop into a .json file.

Technical Details

RFC 4180 State-Machine Parser Internals

The parser is a proper finite-state-machine implementation following RFC 4180. States include UnquotedField, QuotedField, AfterQuote, RowEnd, and EndOfInput. The parser correctly handles quoted fields containing the delimiter, embedded CR/LF inside quoted fields, escaped double quotes (doubled, like ""), and trailing newlines. This produces output that round-trips losslessly through Python's csv module, PostgreSQL COPY, AWS S3 SELECT, and any compliant parser. The state machine is delimiter-aware, so switching from `,` to `;` or `\t` does not change the quoting semantics — only the field separator.

Type Inference Algorithm

With Infer types on, each cell runs through an ordered detection pipeline. First, an empty cell becomes JSON null. Second, the literal strings true and false become JSON booleans. Third, leading-zero strings (^0[0-9]+$) are kept as strings to preserve identifier semantics — converting to numbers would silently strip the leading zeros. Fourth, integer literals are tested against the safe-integer boundary (-2^53+1 to 2^53-1); values outside this range are kept as strings to avoid IEEE 754 precision loss. Fifth, ISO 8601 date strings are detected by regex and intentionally kept as strings — JSON has no native date type. Anything that survives all five guards is converted via Number() (numeric) or kept as a string (everything else).

BOM Stripping and Encoding Handling

All input is treated as UTF-8. The optional UTF-8 BOM (0xEF 0xBB 0xBF) is silently stripped from the first cell of the first row when present — this prevents BOM bytes from being included as a stray character at the start of the first column name (Excel on Windows commonly emits the BOM, breaking naive parsers). Other encodings (Windows-1252, ISO-8859-1) are not auto-detected; the browser File API would have already decoded the bytes as UTF-8 by the time the text reaches this tool. If you have non-UTF-8 input, convert it first with iconv or your editor's encoding-export option before pasting.

Best Practices

Pick the Delimiter Explicitly for Non-Comma Data

Don't rely on auto-detection. If your CSV uses semicolons (Excel-EU), tabs (TSV from BigQuery, Snowflake, or Unix tools), or pipes (high-comma fields), click the matching Delimiter chip before pasting. The parser is delimiter-aware: switching the chip immediately re-parses the input. This avoids the most common CSV-to-JSON failure mode where every row collapses into one cell because the parser used the wrong separator.

Keep Infer Types On for Typed JSON

With Infer types on (the default), you get typed JSON: numbers as numbers, booleans as booleans, null where empty cells appear. This is what most consumers want — APIs, frontends, JavaScript code. Toggle Infer types off only when you specifically need every cell as a string (downstream type-strict consumers, validation pipelines that compare exact source bytes). The detection pipeline has guards for leading-zero strings, big integers, and ISO dates, so identifiers and dates stay safe even with inference on.

Quote IDs as Strings in Upstream CSV

If your CSV is generated by a database or pipeline you control, emit large numeric IDs (Twitter snowflakes, Discord IDs, K8s resourceVersion) as quoted CSV strings ("9007199254740993") so they pass through Type Inference cleanly. The parser will keep them as strings either way (big-int detection catches values above 2^53 - 1), but explicit quoting is the most robust upstream contract and avoids any ambiguity about precision.

Header Row Should Be the First Line

Header on (the default) auto-detects the first row as column names. If your CSV has comments, blank lines, or metadata before the header, strip them before pasting — the parser does not skip leading non-data lines. For headerless CSVs (raw exports, machine-generated dumps), toggle Header off and the columns will be auto-named col1, col2, col3 in order. Don't try to fake a header by prepending one to a headerless file; either toggle Header off or fix the source.

Use Stringify Mode for CSV → JSON → CSV Round-Trips

If you plan to round-trip data through both directions (CSV → JSON → CSV), the reverse direction (JSON → CSV) needs Stringify mode for any nested arrays or objects to survive losslessly. Flatten mode in the reverse direction emits dotted keys (customer.address.city) that can't be perfectly reconstructed by the CSV parser. See our JSON to CSV converter for the full reverse-direction reference and round-trip testing notes.

Frequently Asked Questions