How to use
- Drop an
.xlsx,.csvor.tsvfile, or paste cells copied from Excel. This page opens with the Sort, filter & columns step already added. - Add sort keys, one per column. Keys higher in the list apply first; ties fall through to the next key.
- Add filter conditions and choose “all must match” or “any may match”.
- If needed, set the columns to keep and their order, transpose, or random sampling.
- Check the row count (
Filter 5 → 2 rows) in the result preview and download as xlsx, csv or json.
Order of processing
Inside this one step the order is filter → sort → random sample → column selection → transpose. So you can filter or sort on a column you do not export (column selection comes last), and sample randomly from the rows left after filtering.
Sorting rules
| Column type (auto-detected) | Compared by |
|---|---|
| All numbers | Numeric value |
| All dates | Time order |
| Anything else | Text order with Korean collation: Hangul before Latin letters, digit runs compared by value |
- You can also force number, date or text.
- Empty values always go last, in both ascending and descending order. The sort is stable, so rows with equal keys keep their original order.
- CSV values are all read as text, but a column that looks entirely numeric is compared as numbers. In the example table (Region, Product, Sales), Sales
9, 100, 10, 2, (empty)sorted descending gives100, 10, 9, 2, (empty). - Multi-key example: Region ascending then Sales descending gives
Daegu (empty),Busan 9,Busan 2,Seoul 100,Seoul 10(Korean order: Daegu < Busan < Seoul in the original Hangul names).
Filter conditions
There are 13: equals, not equals, contains, does not contain, starts with, ends with, greater than, at least, less than, at most, is empty, is not empty and regex.
- Equals, contains, starts/ends with and regex are case-insensitive.
- Greater/less comparisons are numeric when both sides are numbers, by date when both are dates, otherwise by text; empty cells never match. Sales
> 9keeps only100and10; the empty cell,9and2are excluded. - “Equals” compares numerically when both sides are numbers, so
1equals1.0. - In the same example, Region = Seoul and Product = B keeps 1 row, while Region = Seoul or Product = B keeps 3.
- To see rows with no Sales value use “is empty” (1 row). The regex
^(Seoul|Busan)$, written with the Hangul names, keeps 4 rows. Regex is limited to 200 characters and an invalid pattern is reported as an error.
Columns, transpose and random sample
- Select columns: the columns you tick appear in the order you tick them. Choosing
Sales, Regionoutputs just those two, in that order. Keep at least one column. - Transpose: flips the table including the header row. Values in the first column become the new headers; duplicates get a number, like
Seoul,Seoul (2). The example table gets the headersRegion, Busan, Seoul, Seoul (2), Busan (2), Daegu. - Random sample: set the row count n and a seed. The same seed always draws the same rows, so you can reproduce a result. From 10 rows numbered 1 to 10, drawing 3 gives
4, 8, 9with seed 1 and1, 5, 10with seed 2. Drawn rows keep their original order. An n at or above the row count leaves everything.
Related tools
- Several steps at once: Excel & CSV Workbench
- Sorting plain lines of text: Sort Lines
- Dedupe after sorting: Remove Duplicate Rows
- Per-column statistics: Column Statistics
Your file is processed only in your browser and is never sent to a server.