Change CSV delimiter

Change CSV delimiter: uses and practical guidance

Changing a table delimiter helps transfer data between programs expecting different conventions. One export may use commas, another semicolons and another tabs. Replacing every punctuation mark in a text editor is insufficient because a delimiter can also appear inside a quoted field. This tool parses rows and fields before serializing them with the selected output delimiter. Values are retained and necessary quoting prevents the new separator from accidentally splitting a cell. The operation changes representation rather than selecting records, renaming columns or inferring a new table structure.

Choose the actual input delimiter rather than the one you want to produce. An incorrect choice can turn several columns into one header or cause a row-width error. The first row must contain unique, nonempty names, and every data row must have the same number of fields as that header. Quoted fields, doubled interior quotation marks and line breaks inside quoted fields are accepted. A newline inside a cell remains part of that cell instead of being assumed to begin another record. Inspect a representative source row to establish its separator before converting a large export.

Values remain text. Numbers, dates, booleans and identifiers are not inferred, so 001 retains its zeros. Field spaces remain part of the value too. Output normalizes record endings to CRLF and may remove unnecessary quotes or add quotes required by the new format. Consequently, output is not byte-for-byte identical to the original even when it represents the same table values. A header-only input can be converted without inventing data rows. Downloaded output uses .csv even when tabs are selected, so choose the matching separator explicitly when importing rather than relying entirely on filename-based detection.

Expressions that a spreadsheet may interpret as formulas are not modified. Use suitable text-import settings when automatic destination conversions are unwanted. Review a row containing an interior comma, another containing quotation marks and one with leading-zero codes before processing an extensive table. Limits are ten thousand data rows, two hundred columns and one million input characters. Retain the source so conversion can be repeated with another setting if the destination interprets fields incorrectly. Structural editing and column selection require a separate data-editing operation; this tool focuses on delimiters and serialization of the same table while preserving the supplied cell values.