TidyCSV

Clean messy CSV files in your browser. Duplicates, whitespace, empty rows, broken headers, suspicious emails and mixed date formats - fixed locally. Your file never leaves your device.

Live: https://uniquepratham.github.io/tidycsv/ · Done-for-you variant: https://uniquepratham.github.io/csv-fix-service/ · Repository: https://github.com/UniquePratham/tidycsv

Why it exists

CSV exports from Shopify, Airtable, HubSpot, Salesforce and Google Sheets often arrive broken (bad emails, duplicates, mixed dates, lost headers) and cleaning them by hand costs 60+ minutes per file. Six paid products charge $9-29/month or $8.99-49.99 per job for exactly this. TidyCSV does the common fixes client-side for free and sells a $9 one-time unlock for the full clean export.

Use it

Open index.html in any browser (double-click, or serve statically):

python -m http.server 8000
# then visit http://localhost:8000/

1. Drop a .csv file (parsing happens in this tab - no upload).
2. Review the issue report (rule-by-rule counts + details).
3. Check the cleaned preview and clean-up notes.
4. Click Unlock clean export - $9 one-time, then Download cleaned CSV.

Pricing honesty

Checkout activates as soon as payments are connected (first demand signal). Until then the download stays open during launch - the $9 unlock is how you pay if the tool saved you time. No account, no subscription, no dark patterns.

Run the core locally

The same rules ship as a Python standard-library CLI:

python tidycsv.py input.csv                 # print issue report + clean-up steps
python tidycsv.py input.csv -o clean.csv    # also write the cleaned file
python tidycsv.py input.csv --report-only   # report only

Rules implemented

Rule	Detects	Auto-fix
empty_rows	entirely empty rows	drop
duplicate_rows	repeated rows (whitespace-insensitive)	keep first
whitespace	leading/trailing spaces in cells	strip
header_noise	blank or duplicate column names	name / suffix
invalid_email	bad addresses in *email* columns or cells with @	report only
mixed_dates	columns mixing YYYY-MM-DD, DD/MM/YYYY, DD.MM.YYYY	normalize to ISO only when unambiguous

Ambiguous dates (03/04/2024) are never silently rewritten - we do not guess whether you meant March 4 or April 3.

Tests

python test_tidycsv.py     # 28 tests: parsing, detection, fixes, ISO dates, CLI
python test_site.py        # 22 tests: 9 pages, canonical/sitemap/robots,
                           #           local-only assets, event allowlist,
                           #           Node syntax + browser/Python core parity

Pages in this site

File	What it is
index.html	the full tool (drop a file, issue report, cleaned preview, unlock)
LANDING.html	pricing / launch page
clean-csv-online.html	primary free cleaner page (functional)
remove-duplicate-csv-rows.html	dedupe page (functional)
fix-malformed-csv.html	structure repair page (functional)
csv-column-cleanup.html	whitespace + header cleanup page (functional)
csv-delimiter-conversion.html	delimiter converter (functional, mode delimiter)
csv-encoding-problems.html	BOM / mojibake repair (functional, mode encoding)
clean-csv-data.html, csv-formatting-tool.html, fix-broken-csv-file.html	how-to guides (article + links into the tools)
sitemap.xml, robots.txt	crawl configuration for the 9 pages above

Every functional page really runs: tidycsv.js (shared core) + seo-tool.js (widget) + site.css. Guides are articles only.

Measurement (no third-party analytics)

Static GitHub Pages have no access logs we can reach and no server to receive events, and the product tests forbid remote resources. Measurement therefore is:

1. GitHub traffic API for the repository (views/clones, fetched with gh).
2. zf-track.js - a client-side session log in the visitor's browser (allowlisted funnel events only, no cookies, nothing sent anywhere). The visitor (or owner) exports the JSON from the badge at the bottom right and imports it with:

   python -m zeroforge track-session --experiment exp_006 --file session.json
   

   Imported numbers are labelled client-side session log, never server-measured.

Design notes

• Zero dependencies: Python standard library + static HTML/JS/CSS files.
• Zero serving cost: static hosting only, no backend, no database.
• No tracking pixels, no CDNs, no third-party scripts (asserted by the tests).
• Honorable gating: the export button unlocks in the UI; there is no server to enforce payment yet (documented deliberately - enforcement comes with checkout).

Part of the ZeroForge zero-cash product factory. See LANDING.html for the launch page and ../csv-fix-service/ for the done-for-you variant.