Ask HN: CSV/Excel Spreadsheet Upload Service | Hacker News Reader

Ask HN: CSV/Excel Spreadsheet Upload Service

11 points1474 comments

I'm feeling a bit inspired from the discussion here: https://news.ycombinator.com/item?id=8961149 so I figured we should have a new thread.

I'm thinking about tackling this problem, that is, making the end user experience for structured data uploads really, really, good.

But like the previous discussion, it's a bit of a tough problem.

You have to deal with different spread sheet formats, malformed files, bad input, etc.

One possible solution that was offered was giving the users a template file to download and match their data to it. But do you really want to give your users more friction to using your software? Especially if it's something they have to do off of your web application, they're likely to just leave and never come back.

Thoughts? Anybody interested in using an external service to handle and validate CSV uploads?

View on HN

My conclusion is that meeting user expectations with regard to spreadsheets is impossibly hard. Excel, for all its warts, just sets the bar too high. That "bad csv" that causes problems is handled by Excel - people aren't uploading broken spreadsheets where broken is what the user deems broken not a programmer on the other end.

Spreadsheets are so hard that the resources of Google couldn't unseat Excel despite Google providing web access, simplifying collaboration, and giving it away for free.

It's not getting numbers in boxes that matters, it's the long tail of functionality. Microsoft has been dogfooding its spreadsheets since MultiPlan in 1982. They have more than 30 years of robustness in their sausage.

This means you'll always be getting bad inputs from workable spreadsheets. Excel deals with hand built data better than any likely automated external method.

Cleaning Excel data is a great idea for a consultancy not a startup.

Good Luck.

I can absolutely recommend sheet.js. It opens an xlsx file in a browser and dumps the data out as json that you can chuck in to something like handsontable.js for the user to view or modify data, then just take the handsontable data and pass it to your backend. If you want to do automated checks you can very easily do that to the json that sheet.js outputs. It's exceptional.

There's an xsl parser too, but if you want to handle csv you'll need something like papaparse.js (which is sort of the grandparent of harb, which is the node.js csv parser sheet.js uses).

Yes, interested. Curious to know your thoughts on user experience.

I've been involved in this space quite a bit back in 2012/2013. I founded Spreadsheet.io to address this problem as well.

I wrote the xls/xlsx/csv/tsv pipeline parser that converts to JSON. Also wrote a native Excel add-in the embeds a JavaScript runtime / REPL for applying JS scripts against local files. Using scripts to extract, clean up and integrate data, etc.

Anyway, thinking about open sourcing it and/or finding a partner of sorts, etc.

My first question is what are the use cases are. Do you know if it's just like Patrick Mckenzie describes?: A SaaS company looking to import user data but the user data can be structured oddly.

I'd need to know how people would use such a service in order to start figuring out the UX.

Reply on news.ycombinator.com