Implementation notes

Parsing

The parser keeps quoted delimiters and line breaks within a field. Empty records are discarded. A UTF-8 byte-order mark is removed before parsing. Detection compares the field count of the first non-empty header under comma, semicolon and tab separators; explicit selection is available when detection is ambiguous.

Validation

Rows with the wrong field count are excluded from the profile and preview, with their count shown in the summary. Missing values do not count as numeric zero. A column is numeric only when all its non-empty values convert to finite numbers.

Timestamps

The range uses the browser's date parser and is displayed in UTC. Locale-specific date strings can produce inconsistent results. Use ISO 8601 with Z or an explicit offset; numeric epoch timestamps are not converted automatically.

Privacy and persistence

Opening a file reads it into memory in the current tab. No user file is sent to a server or saved to browser storage. Example buttons request static fixture files from this site. Reloading clears the loaded data.

Known limitations

• The preview is limited to the first 100 matching rows.
• Header detection does not support a quoted header spanning multiple lines.
• Leading zeroes remain visible in the preview but numeric profiling interprets them as numbers.
• There is no chart renderer, live device connection or binary-format decoder.