You typed SEPT1, 1-2, or 3/4 into a spreadsheet. Excel decided it was a date, converted it, and saved it that way — the original text is gone. This tool can't recover it with certainty, but it can turn the date back into a plausible best guess, fast, so you're not retyping hundreds of rows by hand. Nothing leaves your browser.
plainkit found columns that look like dates. For each one, say what the values were probably supposed to be — it guesses a starting point, but you know your data.
A CSV is plain text with no column types. When Excel opens or pastes into one, it guesses the type of every cell from its shape — and SEPT1 (a gene symbol; this exact bug forced the Human Genome Organisation to rename dozens of genes), 1-2 (a ratio, a version number, a fraction), and 3/4 (also a fraction) are shaped exactly like dates. Excel converts them, and once you save, the cell holds a date serial number with a display format — the string you typed is gone. There is no "undo" that gets it back from the file alone.
What this tool actually does: it reads the date Excel left behind, and re-renders it in the shape you say the original had (day-month, month/day, or the gene-symbol style). For unambiguous cases — a consistent format across the column — that reconstruction is usually exact. It cannot be exact where the original was genuinely ambiguous (is 3/4 "3 divided by 4" or "the 4th of March"?) or where Excel had already discarded distinguishing information, such as a leading zero on the day.
To stop it happening again: before opening or pasting, format the column as Text (select the column → Format Cells → Text), or use Data → From Text/CSV → set that column's type to Text in the preview before loading. Never double-click a CSV that has values like this in it.