CSV Preflight Checker for Classification Inputs
Before sending a file into Jev or another classification flow, check that the right rows and columns were read. This page performs deterministic input checks. It does not guess missing values or remove records. Load the fictional sample to see the issue report first.
Files are read in this browser only. Nothing is uploaded to JevLog or sent to a model. Refreshing clears the current result.
Choose a file or load the demo sample.
The spreadsheet CSV adds a prefix to formula-like cells and may change report values. Download the JSON to process exact report values in another program.
Reproduce the sample report
Section titled “Reproduce the sample report”Download the input with issues and select id and feedback. Expect 7 records, 5 issue entries and 1 duplicate ID. Record 4 has a duplicate ID and text, record 5 has blank text, record 6 has formula-like text, and record 8 has the wrong width. Record numbers include the header as 1; one record can have several issues.
Compare your download with the expected JSON report. The separately authored repaired teaching fixture should return 7 records and 0 issues. Its fictional edits are described in the worked tutorial. Correct real data only from verified source records.
Frequently asked questions
Section titled “Frequently asked questions”Does this automatically clean the input?
Section titled “Does this automatically clean the input?”No. It reports issues and does not delete or rewrite records. Repeated text may belong to different customers, so inspect the source before editing.
Do I need an account? Are results saved after refresh?
Section titled “Do I need an account? Are results saved after refresh?”No account is required. Files and results stay in the current page’s memory. Download the report before refreshing and select your file again afterward. Loading the sample fetches a public fixture; selected business files are not uploaded.
Why do invalid quotes or blank headers block the check?
Section titled “Why do invalid quotes or blank headers block the check?”The parser cannot reliably interpret the input structure. Export UTF-8 CSV or TSV and fix a copy first; renaming an XLSX file is insufficient.
Use the report
Section titled “Use the report”- Fix repeated headers, uneven rows, and duplicate IDs first. These can break the link between a result and its original record.
- Review blank or repeated text in context. Two identical complaints from different customers are still two events.
- If a cell is flagged as a spreadsheet formula risk, check the source before opening a CSV export in a spreadsheet. JSON keeps exact values; the spreadsheet CSV adds a prefix to formula-like cells.
- Fix a copy of the source file and run the check again. This page never rewrites your file.
The tool accepts UTF-8 CSV, tab-separated, and semicolon-separated files up to 5 MB. Quoted commas and line breaks are parsed as one field. Record numbers count data records rather than physical lines. Passing these checks says nothing about classification accuracy. Parsing follows the conventions documented in RFC 4180; export risks are described by OWASP CSV Injection.
Continue with the CSV preparation guide, fictional support-ticket classification, or run comparison tool.