CSV Validator
Check a local CSV for structural issues and find the records that need attention.
Local file
Select one file up to 1 MiB.
Parsing settings
Validation report
Structural check
No issues found
Records
3
Expected columns
3
Fictional brokerage example
Comma detected · UTF-8 · First non-blank record is the header · CRLF
Issues
Read-only preview
Record 1 of 3 · Header
| Field | Source value |
|---|---|
| 1 | Loan ID |
| 2 | Client |
| 3 | Amount |
What CSV Validation Checks
The CSV validator checks whether the file can be decoded and split into records with a consistent field count. It reads the first non-blank record to set the expected width. It accepts Windows, Unix and older Mac line endings (CRLF, LF and CR). Mixed record endings, blank records and duplicate header names appear as warnings.
Quoting follows the conventional CSV format described in RFC 4180. A comma inside double quotes belongs to the field. Two double quotes inside a quoted field mean one literal double quote. The tool also accepts semicolon and tab delimiters.
Choose the File’s Parsing Settings
Select a CSV or tab-separated file up to 1 MiB. Auto detect samples the first records and proposes the most consistent delimiter. The report names the delimiter it used. If the choice is ambiguous, pick the export’s delimiter, such as Semicolon for a file that uses commas as decimal marks.
Auto detect under Character encoding uses a recognised byte-order mark (BOM) at the start of the file. Without a mark, it reads the file as UTF-8. For an older export, choose its known encoding, such as Windows-1252. Choose Header if the first non-blank record contains column names, or Data if it contains values.
The preview shows leading zeros, quoted field contents and embedded line breaks as text. A value beginning with an equals sign stays text. The preview shows up to eight records, 12 fields per record and 500 characters per field.
Read the Issue Report
Record numbers count logical records from one, including the header and any blank records. A quoted field can span two or more physical lines, so the report gives the line number as well when it differs. A width error shows the expected and found field counts. If there are more than 20 issues, the report gives the full count and shows the first 20.
The fictional example has three columns. Its second record, 0042,"Sample, Pty Ltd",150.00, is valid. Without the quotes, 0042,Sample, Pty Ltd,150.00 has four fields. Restore the quotes around the company name in the source export or in a copy of the CSV.
An unmatched quote can consume following lines as part of one field. Close the quote in the source and check where the next record starts. For text after a closing quote, check whether the quote should be doubled or the next delimiter is missing. Rerun the report on the corrected file.
What a Clean Structural Result Means
No issues found means the file decoded under the chosen encoding, the quotes parsed and the field counts matched. The report shows the delimiter, encoding, header choice and line endings used. If the bytes cannot be read under the chosen encoding, the report asks you to select the export’s encoding or re-export as UTF-8.
A structurally readable file can still contain the wrong amounts or fail your destination’s required-column and data-type rules. Compare the preview with the export and check those rules before importing.
Validation runs on your device and leaves the source bytes unchanged. The file stays in this browser and is never uploaded. Remove file clears the selected file, report and preview.
