# CSV structure inspection

Useful Horizons · https://usefulhorizons.com/guides/read-csv-without-losing-the-record
Published September 8, 2026

This is a blank working sheet with a labeled fictional reference example. It does not upload or save your observations. Keep the completed file under your control.

## Input description
- Source file or extract label:
- Exporting application and version:
- Delimiter (CSV Table Lab supports comma only):
- Encoding:
- Record breaks (CRLF / LF / CR / mixed):
- Heading record present:
- Expected data records and columns:
- Meaning of an empty value:

## Cell checks
| Check | Expected value | Observed value | Pass or follow-up |
|---|---|---|---|
| Identifier keeps leading zeros | | | |
| Quoted comma stays in one cell | | | |
| Embedded quote remains one quote | | | |
| Multiline note remains one cell | | | |
| Empty value differs from zero | | | |

## Fictional reference
The built-in inventory has 4 data records and 4 columns with headings on. `0017` remains text, `Pencils, blue` is one cell, the paper note contains two lines, and the folder quantity is empty. These are expected sample observations, not evidence about your input.

## Round trip
- Export mode used (labeled CSV / raw CSV text / JSON):
- Intentional value changes:
- Parsed record count after re-import:
- Exact-string differences found:
- Next check in receiving application:

References: RFC 4180 https://www.rfc-editor.org/rfc/rfc4180 and W3C tabular model https://www.w3.org/TR/tabular-data-model/

Reuse permission: https://usefulhorizons.com/terms