CSV Date Format Converter

A CSV Date Format Converter is useful when a receiving system requires one spelling of a date but your sources use several. A shipping date written as 03/04/2026 might mean April third or March fourth. Replacing slashes with hyphens cannot answer that question. First establish the source convention, then transform the representation. This tool creates a new result column and a status column while retaining the original date text and every other field.

A date can also represent a calendar day, a local clock reading, or an instant with an explicit offset. These meanings require different settings. A birthday or accounting day generally belongs to a calendar rather than a time zone. An appointment at a local time needs a location before it becomes an instant. An event timestamp carrying +08:00 already specifies its relation to UTC. Choose the meaning that your source actually supplies instead of adding information that was never recorded.

Use Choose files for UTF-8 CSV or TSV, or Paste a table for copied spreadsheet cells. Confirm the delimiter and header in the import dialog. Quotes, embedded delimiters, and line breaks are parsed as CSV structure. A logical record containing a multiline note still represents one record, even if it occupies several lines in a text editor. The preview helps identify columns; confirming the import reads the complete input under the selected parsing settings.

Select Add conversion column, choose Column to convert, and enter a distinct New result column name. No column is selected automatically. Repeat this for another date field if necessary: an order date and a dispatch timestamp can have separate source patterns and targets. Every configured column receives its own interpretation. An identifier containing digits is left alone unless you deliberately select it. Keep the default All records scope for a complete run; First 20 records only is a labeled sample.

Allowed source formats is a multiple-choice list. Calendar patterns include year-month-day, day/month/year, month/day/year, year/month/day, day.month.year, compact yearmonthday, and the displayed Chinese year-month-day pattern. All components use the widths shown in the control. Years have four digits, and month and day have two. The supported Chinese pattern therefore expects 2026年04月03日, with zero padding. Unlisted textual month names, natural language, spreadsheet serial numbers, and guessed timestamp units are not interpreted.

Each selected pattern is tested against the complete cell. If several patterns produce the same date, that agreement is acceptable. If they produce different dates, the result is ambiguous and remains blank. With both slash patterns selected, 03/04/2026 cannot be decided automatically. A later unambiguous value does not authorize guessing the earlier one. Separate columns or source files by a confirmed convention when their meanings differ, or narrow the allowed source formats using reliable information from the data owner.

The converter verifies actual calendar components, including leap years, before rendering a target. February thirtieth is invalid, and February twenty-ninth is accepted only in a leap year. Invalid month numbers and overflowing clock components are rejected. The engine does not use permissive date rollover to turn an impossible input into a different valid day. Such rollover would conceal a source error while producing a plausible-looking result.

For a complete example, import two records: identifier 001 with 03/04/2026, and identifier 002 with 30/02/2026. Select the date column, choose only DD/MM/YYYY, and keep YYYY-MM-DD as the target. The first result is 2026-04-03 with success status. The second result is blank with invalid status, while 30/02/2026 remains in its source column. Changing the interface language preserves both interpretations; it does not switch the source convention or recalculate the calendar day.

Choose Date and time for clock-bearing text. Supported patterns include second-resolution local timestamps with a space or T separator, offset timestamps with T, and offset timestamps containing three fractional-second digits. Offset syntax is Z or a signed hours-and-minutes pair. An offset-bearing source must retain an offset-bearing target. Conversely, a source without a zone cannot silently acquire Z. A target that would discard nonzero fractional seconds is reported as unsupported instead of rounding away that information.

Convert time zone explicitly lets you supply a source IANA zone for unzoned input and a target IANA zone. For example, a timestamp at 10:00:00 with +08:00 becomes 02:00:00Z when converted to UTC. Its instant is unchanged even though its clock representation changes. An existing source offset determines the instant directly; the source-zone setting is needed for a local clock reading without an offset. Named-zone conversions are restricted to years 1970 through 2100 and use the browser's supported zone data.

A local clock can skip forward or repeat during a zone transition. In America/New_York, 2026-03-08T02:30:00 is a nonexistent local time. The converter leaves it unresolved and explains the gap; it does not shift the event to a convenient nearby hour. Correct the source meaning before treating that event as a resolved instant. Choosing a different output format cannot repair information missing from the input.

The local time 2026-11-01T01:30:00 in that zone occurs twice. Leave unresolved is the default repeated-time policy. If supporting information identifies the intended occurrence, choose the earlier or later instant explicitly. The earlier occurrence converts to 05:30:00Z and the later one to 06:30:00Z. These are different instants, so selecting a policy is a semantic decision. Save rules preserves that choice and makes the interpretation repeatable, but reuse on a new source still requires reviewing field bindings and conventions.

Run and review produces an interpretation report with the source record, original text, interpreted value, target output, status, and reason. Separate result views expose successful, empty, invalid, ambiguous, and unsupported values. Empty means the configured empty definition matched; it is not a successful conversion. Missing cells and empty strings are empty by default, while whitespace and literal markers require explicit choices. Every unresolved date keeps its original text in the source table.

Export new copy offers the result or a selected report, and the complete bundle includes the configuration summary. The exporter reads generated text back before enabling Confirm download. A sample download has SAMPLE in its filename and must not be described as a full-file date audit. Changing a source or rule makes earlier output stale. Run again before export, and inspect the unresolved count even if most dates succeeded. A receiving spreadsheet may still reinterpret CSV text when opened; use an explicit text import if the literal spelling must stay visible.

Apply to workflow makes a completed full result available to number normalization and validation. The source columns remain present, allowing a validation rule to inspect a normalized date while an investigator checks the original spelling. Undo and Redo recompute ordered steps from retained inputs. Language switching preserves this session. Refreshing or closing the page clears the in-memory data, so export useful copies and save reusable rules before leaving.

Input, output, report size, and execution-time guards can stop a large job. A conversion across many columns produces more interpretation records than the input has rows, so a file within the import limit may still exceed the audit limit. Reduce the selected fields or split the file if that happens. Cancel task stops the active worker while preserving imported data and applied steps. File contents are processed in the browser; loading the website still requests static resources, and this tool does not claim that every possible historical calendar or time-zone convention is supported.