Parsing CSV properly is harder than it looks
The instinct to read a CSV by splitting each line on commas fails on real files, and it fails silently. A field may be quoted, and a quoted field may contain commas, line breaks and escaped quotes — so rows do not correspond to lines and columns do not correspond to positions between commas.
RFC 4180 sets out the rules: quote a field if it contains a comma, a quote or a newline, and escape an internal quote by doubling it. That last detail is the one hand-rolled exporters get wrong most often, writing a backslash escape that conforming parsers do not recognise.
This viewer parses to the specification, which is why a file containing free-text notes or multi-line addresses displays correctly here while appearing shifted or truncated in tools that split on delimiters. If your data looks misaligned elsewhere but correct here, an embedded newline or quote is almost certainly the reason.
Delimiters, encodings and headers
Comma is not universal. In locales where the comma is the decimal separator, Excel writes and expects semicolons, so a comma-delimited file from a British or American colleague opens as a single column for a German or French one. Tab-separated files avoid the ambiguity entirely and are worth preferring for interchange.
Encoding is the other frequent breakage. A UTF-8 file opened as Windows-1252 turns every accented character into a pair of symbols. Excel in particular needs a byte-order mark to recognise UTF-8 reliably on double-click, which is why exports intended for it often carry one — and why that same mark can appear as stray characters at the start of the first header when read by something else.
There is also no way to tell from the file whether the first row is a header. It is a convention, not a specification, and a parser must be told or must guess — which is why a data row occasionally ends up as column names when the first row happens to be all text.
Inspecting without a spreadsheet
The main reason to view a CSV in a browser rather than opening it in Excel is that a spreadsheet reinterprets data on open, and does so destructively. Leading zeros vanish from postcodes and zero-padded identifiers. Long digit strings are converted to scientific notation and rounded past 15 significant figures. Anything resembling a date becomes one — the behaviour that famously forced a genomics naming convention to change because gene symbols kept turning into dates.
Viewing the parsed file as text shows you what is actually in it, which matters when you are diagnosing a data problem rather than analysing it. It is also faster for large files and avoids the risk of accidentally saving the mangled version back over the original.
For converting between formats rather than inspecting, the JSON to CSV converter handles the structural mapping, including flattening nested objects into dotted column names.