How to convert a TXT file to CSV: tab, pipe or semicolon

A .txt extension says nothing about the data inside. Open the file in a text editor and look at the first few lines: values are either separated by a character, lined up by character position, or not a table at all.

Most data exports saved as .txt are delimited by tabs, pipes, semicolons or commas. Fixed-width files pad each field with spaces so the columns line up, with no separator between them. Free text such as logs or notes has no consistent columns.

1. Drop the .txt file on rows.page, choose it from your computer, or click Paste data, paste the text and click Open.
2. rows.page reads .txt as CSV, and DuckDB's sniffer detects the delimiter (comma, semicolon, tab or pipe) and the header row. There is no manual delimiter setting.
3. Check the Columns panel to confirm the fields split cleanly: each column shows its type, distinct count, share of empty values, min and max for numbers and dates, and a histogram or the top 10 values.
4. Optionally clean the rows with a filter such as country:NL, price:>20 or -name:* (is empty), or with SQL against the table data, for example SELECT DISTINCT * EXCLUDE (__rid) FROM data to drop duplicate rows.
5. Click Export and choose CSV.

DuckDB-WASM reads the file inside your browser, so it is not uploaded. A .txt file over 1.5 GB loads partially, with the note Loaded the first N rows: the file is X, more than the browser can hold.

Export writes your current view, including the filter, sort, hidden columns or SQL result. The file is always comma-delimited with a header row, in UTF-8 without a BOM, and text cells starting with =, +, -, @, a tab or a carriage return get a leading apostrophe so spreadsheets don't run them as formulas.

You can't pick a different delimiter, encoding or BOM for the export. Filtered views export every matching row, SQL results stop at 1,000,000 rows, and you can't edit cells or merge two files in rows.page.

Text files from older systems are often not UTF-8, which garbles accented letters and currency signs. rows.page checks the bytes of every .txt, CSV and TSV file it opens and always exports UTF-8:

• UTF-8 files open as they are.
• UTF-16 files with a byte order mark (BOM) are decoded with the note Read as UTF-16 text.
• Other files that are not valid UTF-8 are read as Latin-1 (ISO-8859-1) with the note Read as Latin-1 text.
• Any byte from 0x80 to 0x9F makes the Latin-1 read fail, and rows.page shows Couldn't open this data. In Windows-1252 files those bytes are the euro sign, curly quotes, en and em dashes and the trademark sign.

For a Windows-1252 file, convert it to UTF-8 first and open the new file:

iconv -f WINDOWS-1252 -t UTF-8 in.txt > out.txt

rows.page does not parse fixed-width text, and it does not read or write .xlsx files. If your columns are lined up by character position, split them in Excel and save the sheet as CSV:

• Data > From Text/CSV reads delimited text, and its Delimiter list also has a fixed width option.
• In the Power Query editor, Split Column > By Positions breaks each line at the character offsets you enter.
• The legacy Text Import Wizard has a Fixed width mode where you place the column breaks by hand. Turn it on under File > Options > Data > Show legacy data import wizards > From Text (Legacy).
• Save with File > Save As and pick CSV UTF-8 (Comma delimited) so accented characters survive.

To convert files in a script, or to handle a file over 1.5 GB in full, run DuckDB in your terminal. Like rows.page, it detects the delimiter and header row of a .txt file:

COPY (SELECT * FROM read_csv('in.txt')) TO 'out.csv' (HEADER);

This writes a comma-separated out.csv with a header row. DuckDB rewrites values from their detected types, so 42.50 comes out as 42.5; add all_varchar = true inside read_csv(...) to keep every value exactly as written.

To share the result, push it with curl -T out.csv https://rows.page. Without an account the link lasts 7 days.