CSV & TSV Viewer

Pick the .csv file with the file button, or paste its text into the page, and the table is drawn straight away from the first row down — nothing is installed and nothing is sent anywhere. The first row becomes the header, the delimiter is read from that same row, and a long file is shown 500 rows at a time so the tab stays responsive while you scroll.

Copy and download hand back the complete source text rather than the shortened preview, so even a 5 MB file comes back exactly as it went in. If the original began with a UTF-8 byte order mark it is put back on the way out, and the download keeps your own file name, adding .csv or .tsv only when the name had no extension at all.

No — the CSV is read by the page you already have open, so the bytes are parsed on your own machine and no copy of the file is transmitted anywhere. Nothing is stored either: closing the tab is enough to be rid of the table, and there is no account that could hold a history of what you opened.

Yes — once the page has loaded, opening a CSV needs no further network access, because parsing, the table, copy and download all run from code already in the browser. A connection that drops mid-session does not interrupt reading a file you have already chosen.

Open this page in Edge, Chrome or Firefox on Windows 11 and choose the file — no Office licence, no download and no change to file associations is required. It also sidesteps the familiar Windows annoyance of a long order number being rewritten in scientific notation, since every field is displayed exactly as stored.

Yes — the viewer runs in mobile Safari and mobile Chrome, so you can open a CSV straight from Files, Drive or a mail attachment on a phone without installing an app. The 500-row window keeps scrolling smooth on a small screen, and the same 5 MB ceiling applies as on a desktop.

That error means a double quote was opened and the file ran out before a matching quote closed it, which would otherwise swallow the rest of the file into a single cell. The message names the row and column where the quote opened; the usual causes are a stray quote left inside a text field and an export that was cut short partway through writing.

The bytes are not valid UTF-8, so a legacy export in Windows-1252, Shift_JIS or GBK is refused outright rather than shown with question marks where the accented letters were. Re-save it as CSV UTF-8 from the program that produced it and it opens; a UTF-8 byte order mark is fine and is quietly stripped so it never lands in the first header cell.

It means some rows of the CSV carry more or fewer fields than the header row, and those rows are listed by number instead of being padded out or thrown away. Up to 200 of them are listed individually while the total stays exact, so you can jump to the record where the export went wrong and still read every original value.

Up to 5 MB, and the size is checked before any byte is decoded, so an oversized file is refused instantly instead of freezing the tab. Inside that ceiling the parser also stops with a clear error at 2,000,000 rows, 4,096 fields in one row, or 1,000,000 characters in a single field.

Yes — every field is handled as text, so 007, a nineteen-digit account number and 0031234567 all survive verbatim, with no scientific notation and no dropped zeros. Values that look like dates, such as 03/04/2026, are left alone as well rather than being reinterpreted into another regional order.

A CSV separates fields with a comma and a TSV with a tab; both are plain text and both open here without any conversion step. Because the delimiter is decided by the header row rather than the extension, a tab-separated file saved as .csv, and a semicolon or pipe export, are all read correctly too.

Reading it here is safe in the way that matters: the file is treated as text only, so a formula such as =cmd|... is displayed as characters and never evaluated, and no macro inside the file can run. That is not a malware verdict — the very same file can still be dangerous in a spreadsheet program that does execute formulas.

Yes — no installation, no sign-up, no per-file quota: open the page, choose the file, read the table. It is a reader rather than an editor, so if you need to change cells and save the result you will still want a spreadsheet program afterwards.

No — a clean table only proves the file is well-formed UTF-8 text with a consistent delimiter and consistent quoting. Whether the figures are accurate, the export complete, or the sender trustworthy is beyond what any parser can judge; the ragged-row and blank-row warnings describe structure, not truth.