CSV Desk — offline CSV batch cleanup with a change report

CSV Desk cleans up to ten CSV files at a time in your browser, offline, and downloads one ZIP with a cleaned copy of each file, a JSON report of every change and, if you want them, byte-for-byte backups of the originals. Open the downloaded folder in desktop Chrome, choose your files and review each one before you export.

You pick each change: trim surrounding whitespace, give repeated or blank headers unique names, remove blank records, remove exact duplicate records, or put an apostrophe in front of formula-like cells. With every option off, values are kept exactly as they are and only the separator, quoting and encoding are normalised. Quoted separators and multiline fields are handled. Parsing is strict: a stray quote inside an unquoted field (for example 5" screen) blocks that file with a clear message instead of guessing. Files are read in the browser tab and never uploaded; the app has no remote dependencies.

Includes: the batch app and its source, a quick-start guide, a fictional CSV example, a single-file demo and developer tests. One-time purchase. One person may use the batch edition for personal and internal business work; see the included licence. The free demo is MIT licensed.

Try it first: https://ntoledo319.github.io/many-routes-csv-desk-demo/

Limits: tested in Chrome 150 on Linux only. Ten files, 5 MiB per file, 20 MiB per batch, and 100,000 records, 200 columns and 500,000 cells per file. The preview shows the first 25 records. No XLSX, merging or fuzzy matching. CSV has no column types, so import IDs as Text in your spreadsheet. Formula prefixing changes values (negative numbers included) and is not a security guarantee.

Support: https://payhip.com/TheLanternLibrary/contact for reproducible defects in the delivered version; please send a small fictional sample, never real data. No promised response time or data-recovery service. Refunds: contact the seller within 14 days if the download doesn't work as described; your statutory rights still apply.

Made with AI assistance: the code, documentation and store images were produced with AI tools, then tested as described above.