Fill Incomplete CSV Records

Two distinct things, depending on mode. Ragged rows are rows with fewer cells than the header – common when an export script gave up early on certain rows. Empty cells are cells with no content or only whitespace, even though the row’s total cell count matches. The “both” mode handles both at once; the narrower modes let you target just one.

Quoted fields can contain the delimiter, line breaks, and quotes themselves (escaped by doubling – ""). A row like "He said ""hi""",foo parses to two cells: He said "hi" and foo. The output stringifier round-trips this – cells get quoted only when they contain the delimiter, a quote, or a line-break character.

For each empty cell in a column, copies the most recent non-empty value from that column. Common in time-series data where a missing value usually means “same as before” (e.g. inventory counts, sensor readings). If no previous non-empty value exists yet (you hit an empty in column N before any non-empty has appeared), the fallback fill value is used.

Type a 1-based, comma-separated list in “Columns to fill” – e.g. 2,3,5 fills only those columns and leaves columns 1 and 4 untouched. Blank field = all columns. Useful when some columns genuinely allow empties (notes, optional fields) and you only want to fill the required ones.

Depends on what your downstream consumer expects. N/A or null for text contexts; 0 for numeric (be careful – zero is a real value); empty string when you only want to pad ragged rows without inventing a sentinel; Unknown for categorical / dropdown contexts. Match whatever your import target treats as “no data”.

Mostly. Header order, delimiter, and quoting strategy match. Line endings default to auto (whatever your input used – LF or CRLF) but can be forced. The only deliberate changes are the cells you asked to be filled and the padding cells added to ragged rows. Cells that didn’t need quoting on input but do on output (because the fill value contained the delimiter, say) will get quoted.

The parser is single-pass O(chars) and uses a single output array. A few MB of CSV (tens of thousands of rows) processes in well under a second. Browsers cap text-area paste at around 1 MB on some platforms; for bigger files, the tool still runs but the text-area UI gets sluggish – you might want a streaming tool instead.

The sniffer reads the first non-empty line, counts occurrences of comma, tab, semicolon, and pipe, and picks the most frequent one. If none of the candidates appear, falls back to comma. Override with the dropdown if your first row has unusual structure (e.g. one column of CSV-shaped data).

No. The page is three static files (HTML, CSS, JS). All parsing, filling, and stringifying runs in the browser via plain string operations – no fetch, no XHR, no analytics, no cookies. You can disconnect the network after page load and the tool still works.