Katak

Open huge CSV files instantly

Katak is a free, open-source viewer and editor for CSV and TSV files of 1 GB, 10 GB, 50 GB and more. Search, filter, sort, edit and convert them without loading the file into memory, on Windows, macOS and Linux.

MIT license · one 12 MB portable binary · no install, no account, no telemetry

Katak showing a 10 GB CSV file with 119 million rows, opened instantly while it is indexed in the background
A 10 GB CSV file with 119 million rows: on screen 0.2 s after launch, indexed in the background at 1.6 GB/s.
0.2 sto show a 10 GB or 50 GB file
1 GB/ssearch, regex and text filters
127 MBRAM with a 50 GB file open
0 byteschanged in rows you did not edit

CSV file too large for Excel?

Excel and LibreOffice Calc stop at 1,048,576 rows. Text editors try to read the whole file into memory and freeze. Importing into a database takes minutes before you see a single row.

Katak memory-maps the file, reads only what is on screen and builds a sparse index in a background thread, so a file with hundreds of millions of rows opens as fast as a small one and the window never freezes. Filter or select the rows you need and export them as a CSV that Excel can open, or work with the whole file directly.

Features

Opens any CSV, TSV or delimited text file

Delimiter, quotes, header, encoding and line endings are detected from the first megabyte. Change anything in the parsing dialog with a live preview: comma, semicolon, tab, pipe or custom multi-character delimiters, whitespace-aligned columns, quote and escape styles, skipped and comment lines, and a raw-text view.

Parsing options dialog with a live preview of a semicolon-separated file

Search and filter at gigabytes per second

Plain text or regex search with match case, whole word, whole cell and column scope. Per-column filters (contains, equals, starts with, regex, empty, numeric range) combined with AND. Results appear while the scan runs, and both see your unsaved edits.

Filtering a 10 GB CSV file: 1.5 million matching rows shown while the filter runs

Sorts millions of rows

Sort by any column as numbers, dates or text, ascending or descending. 12 million rows sort in 1.2 seconds within a configurable memory budget. The sort is a view: the file only changes if you export or save in sorted order.

12 million rows sorted by amount in descending order

Edits like a spreadsheet, saves without surprises

Edit cells, insert and delete rows and columns, copy and paste with Excel, find and replace with regex capture groups, and undo anything. Saving keeps the original delimiter, quoting, encoding and line endings, copies untouched rows byte for byte and replaces the file atomically, with an optional backup.

Editing cells and rows in a large CSV file

Converts and exports

Export all rows, the filtered or sorted rows, or a selection, with another delimiter, quoting, encoding or line ending: CSV to TSV, semicolons to commas, Windows-1251 to UTF-8. Exports stream, so a 10 GB conversion uses about 50 MB of RAM.

Export dialog converting a CSV file to TSV

Awkward files welcome

Multi-line quoted fields, ragged rows, unterminated quotes, NUL bytes, invalid UTF-8, a single 500 MB cell.

Wide files

Only visible rows and columns are parsed and drawn, so 10,000 or 1,000,000 columns scroll smoothly.

Every encoding

UTF-8, UTF-16, Windows code pages, ISO-8859, KOI8, Shift_JIS, GBK, Big5 and more, auto-detected.

Network drives

Slow or disconnected shares are read in the background and never freeze the window.

Changes on disk

If another program appends to or replaces the open file, Katak offers to reload it.

Keyboard friendly

Go to row or byte offset, F3 to the next match, Ctrl+Home/End, tabs, zoom, light and dark themes.

Benchmarks

Measured on an Intel Core Ultra 9 285 with Windows 11, files in the OS cache, using the generator and benchmark tools in the repository. All measurements.

TaskResult
Launch to first screen, 10 GB / 50 GB file198 ms / 231 ms
Background indexing (quote-aware)1.4–1.7 GB/s
Literal search, 1 GB / 10 GB file1.3 GB/s / 1.0 GB/s
Regex search, 1 GB file1.1–1.3 GB/s
Filter, 1 GB file (text / numeric range)1.3 GB/s / 792 MB/s
Sort 12 million rows by a number column1.2 s
Convert 10 GB to TSV350–400 MB/s, 53 MB RAM
Memory with a 50 GB file open and indexed127 MB
Index size0.03 % of the file
Longest UI frame while indexing a 50 GB file4.7 ms

Download

Windows 10 / 11

Installer with Start menu entry and "Open with" for .csv, .tsv and .txt, a portable .exe, or an MSI for IT deployment.

Installer (x64) Portable katak.exe (x64) MSI package (Group Policy, Intune)

macOS

Not notarized yet: run xattr -d com.apple.quarantine katak once after extracting.

Apple Silicon (arm64) Intel (x64)

Linux

x64, glibc 2.35 or newer (Ubuntu 22.04+, Debian 12+, Fedora 36+), X11 or Wayland.

katak-linux-x64.tar.gz Build from source

All versions and SHA-256 checksums: GitHub releases. With Rust installed: cargo install --git https://github.com/Erkinbek/katak --locked katak

Frequently asked questions

How do I open a CSV file that is too large for Excel?

Open it in Katak with File → Open, drag and drop, or katak file.csv. There is no row limit: the first rows appear immediately and the rest of the file is indexed in the background. To use the data in Excel, filter it or select the part you need and export it as a smaller CSV.

How large a CSV file can Katak open?

Files are memory-mapped, so the limit is your disk, not your RAM. Katak is tested with 50 GB files, 119 million rows, 100 million one-character rows and 1,000,000 columns. The index takes about 0.03% of the file size.

Does Katak load the whole file into memory?

No. Only the visible rows and columns are read and parsed. Search, filter, sort and save stream through the file on background threads with bounded memory; a 50 GB file open and indexed uses about 127 MB of RAM.

Will saving change data I did not edit?

No. Untouched rows are copied byte for byte; Katak never reformats numbers or dates, strips leading zeros or trims whitespace. Saves go to a temp file that atomically replaces the original, so a crash never leaves a half-written file.

Can Katak convert a huge CSV file to another delimiter or encoding?

Yes. Export writes all rows, the filtered rows, the sorted order or a selection with another delimiter, quoting policy, encoding or line ending. It streams, so converting a 10 GB file needs about 50 MB of RAM.

Is Katak free?

Yes. Katak is open source under the MIT license, with no account, no telemetry and no network access. The source code is on GitHub.