CSV Validator

It checks that every row has as many columns as the first row, that every quoted field is closed and that quotes are only used where CSV allows them. It also looks for signs of a wrong encoding, NUL characters, empty or duplicate headers and mixed line endings. Errors make the file invalid, while warnings are listed for review.

With Detect automatically the validator tries comma, semicolon, tab and pipe and picks the one that splits the rows most consistently. If a file uses a delimiter that also appears in the values, choose it explicitly in the Delimiter field. The result always shows which delimiter was used.

Every row is compared with the first row, and each row that differs is listed with its number, its column count and the expected count. Rows are numbered from 1 including the header row, and the first 20 differing rows are listed with a count of the rest. Quoted line breaks inside a field do not start a new row.

A field that starts with a double quote must end with one, and a quote inside a quoted field must be doubled as "". The validator reports quoted fields that are never closed and quotes that appear in the middle of an unquoted field. Both are common when a value containing the delimiter was exported without proper quoting.

Text that was saved in another encoding and then read as UTF-8 contains replacement characters, which the validator counts and reports as an error. NUL characters, a sign of a binary or UTF-16 file, are reported too. A UTF-8 byte order mark is allowed and shown in the Encoding line.

With First row contains headers enabled, empty header cells and headers that appear more than once are listed as warnings, because many import tools reject them. Turn the option off for files without a header row. The column count check runs either way.

The result shows whether the file uses LF, CRLF or CR line endings. A file that mixes them gets a warning, since some importers count rows differently in that case. Line breaks inside quoted fields are part of the value and allowed.

Lines that are empty or contain only whitespace are skipped and do not count as rows. A row of empty fields such as ,,, is a real row and must have the right number of columns.

Errors are problems that make the data unreliable, such as ragged rows, broken quotes or a wrong encoding, and they mark the file as invalid. Warnings point at things that are legal CSV but often cause trouble, such as duplicate headers or mixed line endings. A file with only warnings is reported as valid.

Inputs can be up to 5 MiB and 100,000 rows. To turn a valid CSV into JSON, YAML, XLSX or another format, use the Data Format Converter.