Parse a CSV in the Browser: Column Type Badges, Parsing Issues and the Bytes You Get

I work on OnePassApps, and the Data Studio's Viewer tab is the one screen in our tooling that tells you what a parser actually made of your file before you convert it.

It is not a converter and it is not a spreadsheet. It is the diagnostics view: it parses your CSV, badges every column with the type it guessed, shows you the rows it could not read, and hands you back the exact bytes it parsed. That last part matters more than it sounds, because the badges are a display hint, not a conversion.

You can try it on the Data Studio page, or read the OnePassApps guides page first.

The panel, before you run it

Open the Data Studio. The first tab radio reads Viewer (the page calls it "Data Studio: CSV viewer" and the preview stage is headed CSV PREVIEW), and the tab row carries eleven tools: Viewer, Sort / filter, Columns, Duplicates, Merge, Split, To JSON, Formatter, Validator, To CSV and YAML.

The three steps read:

• 01 Source - Clear, Paste local data, Browse files, Input limits and privacy
• 02 Options - a Delimiter picker shipping on Auto detect · comma, with one sentence that explains it: "Auto detect reads the first lines to find the separator."
• 03 Result - "Rows, columns and any parsing issues appear here after the tool runs."

The action button reads Parse CSV, and the source hint is "Drop a data file, or browse CSV, TSV, JSON or YAML · up to 20 MiB · or paste rows in". The privacy line is the usual one for the studios: "Processing stays in this browser." and "Nothing leaves this tab."

A clean file, and what the badges say

I parsed a six-column CSV with three rows, deliberately mixing numeric-looking values, booleans, a date, a currency string and an empty cell:

id,name,score,active,note,price 1,Ada,42.5,true,2026-11-03,$42 2,Grace,7,false,hello,19.99 3,Mika,007,true,,0 

The result line reads "Loaded 3 rows and 6 columns.", then "3 rows · 6 columns · no parsing issues", then "Delimiter found" / "Comma", then "Parsed in 649 ms" and the button becomes Parse again.

The preview badged each column with a type:

Read that as a guess about the column, not a promise about your data. score is badged NUMBER although one of its values is 007, active is TEXT although every value is true or false, and note is TEXT although it holds a date and an empty cell. price is TEXT because $42 is not a number. The badge is decided per column, not per cell.

The empty cell in row 3 renders as an em dash - — - in the preview table.

And here is the edge worth knowing: the badge does not change what you get back. I clicked Download CSV and captured the file. It is a real text/csv;charset=utf-8 blob of 160 bytes, every field quoted, CRLF line endings, and it preserves the characters exactly as they were:

"id","name","score","active","note","price" "1","Ada","42.5","true","2026-11-03","$42" "2","Grace","7","false","hello","19.99" "3","Mika","007","true","","0" 

So 007 is still 007 in the file - the badge never cost you a leading zero - and the empty cell is an empty quoted field, not the — you saw on screen. The badge is a label; the bytes are the truth.

The preview also has a Table / Raw toggle. Table shows the parsed grid. Raw swaps it for the file's own lines - for that four-line file the raw view listed four numbered lines - but the text of those lines did not render in my run, so I could not read the raw source back out of the preview.

Now break it on purpose

The interesting part of this tab is what it does with a file that is genuinely broken. I fed it a header with a repeated name, rows of different lengths, and an unterminated quote:

id,name,score,id 1,Ada,42.5,9 2,Grace 3,"Mika, Jr 4,"Line1 Line2",5,10 5,Mika,1,11, 

The panel does not refuse the file. It parses it, labels what went wrong, and shows you a table you can reason about:

• The summary reads "Loaded 4 rows and 5 columns." and "4 rows · 5 columns · 5 parsing issues".
• The step's state line reads 1 issue, and the counters below the result separate the two kinds: CSV parsing issues (1) and CSV warnings (4). One error plus four warnings is the five in the summary line.
• Opening the issues row gives the sentence that names the fault: "Row 1: a quote inside a quoted field is not doubled".
• The preview carries its own banner: "Couldn’t read line 1", "A quote inside a quoted field is not doubled", and a link to Open as text.

Two structural repairs happen quietly, and both are visible in the header row of the preview:

• The repeated header is renamed. My second id column comes back as id_2, so you never have two columns with the same name in the table.
• The extra field becomes a named column. Rows longer than the header grow an Extra 1 column, badged TEXT, and the short rows get an em dash in it.

The short rows are padded with — rather than rejected: row 2 has only two values, so its score, id_2 and Extra 1 cells are all —.

And the broken quote does what broken quotes do - it swallows the rest of the file. Row 3's name cell reads Mika, Jr 4,"Line1 Line2, with three physical lines collapsed into one field, and the values that follow shift along. That is why the file reports four rows when six data lines follow the header.

So: a CSV that is not well-formed does not stop you here. It gets parsed as far as it can, the fault is named in words you can act on, and the table still shows you what the parser believes. That is a different contract from "your file is invalid" - and it is the more useful one, because you can see the damage before you convert anything.

What it costs

Standard is free right now - $0 / month, 100,000 credits every 30 days, no end date set, no card required, and no automatic paid subscription. Parsing a CSV in the tab is local work and does not touch that allowance.

If you want to see the same numbers for your own file, the Data Studio is at https://app.onepassapps.com/studios/data. I work on OnePassApps, so if the badges ever disagree with the bytes you downloaded, tell me which one you trusted and I will go and look at why.