Schema Column Mapper
Prepare a table for a specific import contract. Define target fields, connect them to named source columns, and choose explicit transformations and types. Run the plan to separate valid output rows from rows that need repair while keeping the original parsed cells in a complete report. No database connection or arbitrary script execution is involved.
Key features
- Four mapping operations: copy one column, insert a literal, join ordered columns, or choose the first nonempty column
- Named target fields with string, safe integer, exact decimal text, boolean or Gregorian date types
- Explicit trimming, letter case, empty-value defaults and required-field checks
- Logical CSV record numbers and field-specific errors; original values retained in the full JSON report
- Reusable strict mapping JSON plus separate valid-output and rejected-source CSV downloads
How to use
- Paste CSV or select a UTF-8 CSV/TSV file, choose its delimiter and read the table.
- Use the initial source-column fields or add target fields; set their names, types and mapping operations.
- Order source references and choose trimming, letter case, required values and optional defaults.
- Run the mapping and inspect valid and invalid records. Changing input or rules clears the previous result.
- Download the mapping plan, all valid output rows, all rejected source rows or the complete JSON report.
Use cases
- Combine given and family names into a destination full_name column while preserving text IDs
- Use a backup contact column only when the preferred source is empty under the selected trim rule
- Reject invalid dates and numbers before preparing a CSV import for another system
- Save one named-column mapping and reuse it when the source columns appear in a different order
Frequently asked questions
How is this different from the JSON CSV converter or cleaning workbench?
The JSON CSV converter changes serialization, flattening and inferred types. The cleaning workbench applies a sequence of operations to an existing table. This tool defines a separate ordered target schema, maps named inputs into each target field, and reports required-value and type failures before exporting only valid rows.
What order do transformations and defaults follow?
The selected expression reads only original source cells. Then optional trim and letter-case conversion run. A default is used only if that result is empty and receives the same transformations. Required and type checks run last. An invalid nonempty value is reported rather than silently replaced. Coalesce checks candidate emptiness using the trim setting and uses the first available column.
How are numbers, booleans and dates validated?
Integer accepts signed whole digits within the JSON safe-integer range. Decimal accepts up to 18 integer digits and 12 fractional digits, with no exponent or thousands separator, and returns canonical exact text. Boolean requires true or false after selected transformations. Dates require real Gregorian YYYY-MM-DD dates from year 0001 through 9999.
Which rows are downloaded, and can values be lost?
The valid-output CSV contains every fully valid row in target-field order. The rejected CSV contains every invalid row’s original parsed cells and issue details. The JSON report contains both groups, their converted values, originals and rules. Preview filters and 25-row pages never limit downloads. Optional empty output values become null in JSON and blank CSV cells; keep the report to retain that distinction and the original cells.
Can the saved mapping be reused, and what JSON is accepted?
The column-mapping-v1 plan binds exact header names, not positions. Missing referenced columns stop the run. It allows 1–32 unique target fields and 1–8 ordered references for join/coalesce, including repeated references. Unknown properties, decoded duplicate JSON keys, unsupported operations and plans over 64 KiB are rejected. The plan is declarative; it does not execute JavaScript.
What are the limits and CSV export conventions?
Input is limited to 1 MiB,5,000 rows,32 columns and 100,000 data cells. The complete report is limited to 8 MiB and 9,000 issues; exceeding a limit stops the entire run. CSV exports use a UTF-8 BOM, quoted cells and CRLF. Formula-like string cells receive an apostrophe prefix; JSON preserves raw text. Spreadsheet software may still infer types, so import identifier and exact-decimal columns as text.
Privacy
Source cells, mapping rules and reports stay in this page’s memory. This tool does not upload them or save them to browser storage or URLs. Download before leaving if you want to keep a plan. Reports and rejected-row files contain original input values; a plan may also contain literal values or defaults. Cancelling or changing input prevents an older result from being applied, although an operating-system file read already in progress may finish.
Comments & questions