Pikkola CSV Workbench, tool No. 26

A table, cleaned up before it reaches a spreadsheet.

Drop a CSV, a tab separated export, or something copied out of a chat window. The delimiter and the encoding are read for you, the rows are yours to sort, filter and inspect, and the same data leaves as JSON, Markdown, SQL or a smaller CSV. No upload, no account.

0
bytes uploaded
4
export formats
3
languages
Workbench

Drop a file, or paste the table

Drop a CSV or TSV here

Comma, tab and semicolon files. UTF-8 or EUC-KR. Up to 10 MB, read in this tab.

Choose a file or
No file handy
01

Every column, read at a glance.

Type, empty cells, distinct values and the range, taken from the rows that are on screen. A column that should be a number and is not will say so here, before it becomes a formula that returns an error.

Load a file and every column gets a card here.

02

Leaves in the shape the next tool wants.

The export follows the table as you have it: the search, the filters, the sort and the rows you kept. Four formats, copied or downloaded, built in the tab.

An array of objects, one per row, keyed by the header. Numeric and boolean columns keep their type, empty cells become null.


      
03

How a file is read.

01

Delimiter from the data

Commas, tabs, semicolons and pipes are counted outside quotes on the first lines. The one that splits every row into the same number of columns wins, so a European export and a tab separated paste both open correctly.

02

Encoding in three steps

A byte order mark is trusted first. Then the bytes are tried as UTF-8, which fails loudly on invalid sequences. Then they are read as EUC-KR, the character set behind most Korean Excel exports. The first two are fast and exact; the third is the safety net.

03

Quotes follow RFC 4180

A quoted field may contain the delimiter, a line break and doubled quotes, which is what an address column with commas inside looks like. Rows are rebuilt from that grammar rather than split on new lines, so nothing shifts.

04

Nothing leaves the tab

The file is read with FileReader, parsed in JavaScript and drawn from memory. Exports are Blob downloads written by the same script. There is no upload endpoint, no account and no analytics event that carries a row.

04

Questions that come up first.

Is my file uploaded anywhere?

No. The file is read with the browser's FileReader and parsed in the same tab. There is no server endpoint for a file, no database and no analytics event that carries a row. Closing the tab ends it.

How large a file can it open?

Up to 10 MB or 50,000 rows, whichever comes first. That is set so a phone does not freeze rather than by a limit of the format. Sorting and filtering work on the whole data set, while the table draws one page of it at a time.

My Korean text looks broken. What happened?

The bytes were decoded with the wrong character set. UTF-8 is tried first and EUC-KR second, which covers most Korean Excel exports, but a file can be something else. Change the encoding in the toolbar and the same bytes are read again, without reloading the file.

What counts as a duplicate row?

Two rows are duplicates when every cell matches, compared as text. The first row is kept and later copies are removed, and the removal can be undone from the message that appears. Removing rows that differ in one column is a job for a filter.

Why does Excel show broken characters in the CSV I downloaded?

Excel opens a CSV with the local character set unless the file starts with a UTF-8 byte order mark. Tick the Excel option in the CSV tab and the mark is added, which fixes Korean and Japanese text on Windows. Modern tools do not need it.

What about semicolon files from European Excel?

The delimiter is detected from the data, and semicolon is one of the candidates, so those files open as they are. The export has its own separator setting if the file has to go back to that machine.