CSV parsing follows common quoted-field rules

Fields may contain delimiters, line breaks, or double quotes when enclosed in double quotes. A doubled quote inside a quoted field becomes one literal quote. The parser accepts LF or CRLF line endings and reports the row and column when a quoted field is unfinished or text appears after its closing quote.

Headers become JSON property names

CSV-to-JSON treats the first row as headers. Blank or duplicate headers are rejected because either would make the object structure ambiguous. Every subsequent row must have the same field count. An optional inference setting converts conservative decimal numbers, true, false, and null; empty fields remain empty strings.

JSON input must be tabular

JSON-to-CSV accepts an array of objects. It builds a stable union of keys in first-seen order. Nested arrays and objects are preserved as compact JSON text inside a CSV field rather than silently flattened. Missing properties become empty cells, while explicit null values are written as null.

Delimiter detection is evidence-based

Automatic detection compares comma, tab, semicolon, and pipe candidates across the first non-empty logical records and prefers a delimiter with a consistent multi-column shape. If the file is genuinely one column or ambiguous, choose the delimiter explicitly.

Local processing and practical limits

Conversion runs in this browser tab. Gypes does not intentionally upload the input or result. Text and imported files are limited to 2 MB, output is limited to 5 MB, and tables are capped at 50,000 rows and 1,000 columns to reduce accidental browser exhaustion.

CSV has no single universal dialect or data-type system. Check delimiters, encodings, date formats, leading zeros, large identifiers, formulas, and downstream import settings before replacing source data. Spreadsheet applications may interpret cells beginning with =, +, -, or @ as formulas.