Dataset Merger & Joiner

1. Drop files into the file area or click browse files. For pasted data, use Or paste a table, choose Pasted format, and click Add pasted table.
2. Choose Merge mode, then select Append / Union (stack rows) or Keyed join (match rows by key).
3. Set the mappings and options. A join needs different Left dataset and Right dataset selections plus at least one key pair.
4. Read Result summary and the review tables, then choose an export button or Export audit report (JSON).

How Append / Union (stack rows) combines datasets

Append / Union stacks rows into target columns. Datasets stay in the order they were added, and rows keep their order within each dataset. Automatic column matching trims and lowercases names. Change a mapping or use Add a target column and Add column for a new output field.

Dataset A: id,name
1,Ada

Dataset B: id,name
2,Lin

Output:
id,name
1,Ada
2,Lin

For each target field, Type can be auto, text, number, boolean, or date. Trim & collapse whitespace, Case normalization, and Null / empty value handling change the output values. Drop exact duplicate rows checks rows after mapping. Add source file/row columns adds the source name and source row number.

How to join datasets by key

Keyed join uses exactly two different datasets. Select Left dataset and Right dataset, add pairs with Add key pair, and use Add field mapping when shared field names differ. Same-named non-key columns are matched automatically. Ignore case in keys, Ignore whitespace in keys, and Allow empty keys to match change key comparison.

Left dataset:
id,name
1,Ada
2,Lin

Right dataset:
id,name
1,Ada

Inner (matched rows only):
id,name
1,Ada

Inner (matched rows only) keeps the matching row in this example. Left (all left rows) also retains Lin. Right (all right rows) keeps every right row, while Full (all rows either side) includes unmatched rows from both sides. Matched rows follow left-row order, with matches in right-row order.

What duplicate keys and conflicts do in a join

Duplicate-key policy can keep all rows, keep the first row, keep the last row, or exclude a duplicate group and report it. Many-to-many match policy chooses Cartesian (all combinations) or First match only.

When shared fields differ, Field conflict resolution can choose Prefer left value, Prefer right value, or Keep both (suffixed columns). The first two choices add differences to the conflict report. The third writes fields such as field_left and field_right.

What the dataset merge preview leaves out

The merged-data preview shows up to 200 rows, and each review table shows up to 300. Exports contain the full result or report. Delimited input uses its first row as headers. JSON must be a top-level array, and every non-empty JSONL or NDJSON line must contain valid JSON.

Nested JSON values are converted to strings, so objects are not deep-merged. Spreadsheet layout, formulas, and formatting are not preserved. The tool does not connect to a database. Dataset content stays in the browser while the page parses, maps, joins, and exports it.

Where dataset merging comes up

For CSV exports with the same headers, use Append / Union (stack rows), check the mapping, choose Drop exact duplicate rows if needed, and export CSV.

For tables with differently named keys, choose Keyed join (match rows by key), add a key pair, map the field, and inspect unmatched rows before exporting JSON.

Choose the right dataset merge task

Use Dataset Merge Tools when combining datasets. This page covers row stacking and keyed joins.

Frequently asked questions