CSV / TSV
Join two tables by one or more keys, or append their records by column name.
Your tables stay in this browser; no data is uploaded.
No generated rows yet.
Inputs or settings changed. Estimate again before generating or exporting.
A customer table and an order export describe different parts of the same relationship. CSV join connects their identifiers while keeping each source available for checking. Repeated identifiers may create several output rows for one customer.
To merge CSV files from consecutive reporting periods, select append instead. That operation adds records beneath existing records, even when the two files list their columns in a different order.
CSV join counts matches before constructing the output. Two rows sharing a key on the left and two on the right produce four combinations. Review that expansion and the unmatched counts before generating the file.
A regional export may put the amount before the account identifier. Append CSV records by matching header names, preserving the left table’s column order and adding names found only on the right. Missing fields become empty cells.
The left and right tables can use different delimiters. Both need one header record with unique, nonblank names. If automatic detection is uncertain, choose the separator explicitly before interpreting the column positions.
To join CSV by key on account and date together, enter the corresponding column positions on each side, such as 1,3. Positions start at 1, and both lists need the same number of columns. Every paired component must match.
Inspect the duplicate groups and unmatched source records in the estimate. Generate only after that review. Changing either table or any parameter expires both the estimate and result, so exports stay disabled until you run the stages again. The table shows 50 records at a time; copying and downloading include the complete result.
Inner keeps matching combinations; left also keeps every unmatched left record. Full outer join adds unmatched records from the right too. A missing counterpart contributes empty cells, while each source’s own values remain available.
Repeated keys are reported on both sides. Matching groups expand to all pairs rather than silently taking the first lookup row. Counts describe the relationship before rows are materialized, making accidental many-to-many joins easier to spot. Separate left and right unmatched CSV reports include original fields and a source-line column.
Whitespace trimming, case-insensitive comparison and empty-key matching are optional. They affect key comparison only. Output cells retain their original strings, including leading zeros and the spelling present in each source.
For a join, shared column names receive left and right suffixes. Generated names avoid existing headers and use numeric additions when needed. Append instead treats equal header names as the same column and adds right-only columns at the end.
In join modes, both versions receive suffixes such as _left and _right. Numeric additions resolve further collisions with existing names. Neither value overwrites its counterpart. In append mode, identical header names identify one output column.
Use append for records of the same kind, such as monthly exports. Use a join to attach fields from another table by a shared identifier. Append needs no key selection and keeps repeated records rather than removing duplicates.
Each operation accepts two inputs. You can download a result and paste it alongside another table, provided that result fits the smaller input limits: 2 MiB, 2,000 records including the header and 100 columns.
Every left record is paired with every right record carrying the same key. A group of two and a group of three therefore produce six rows. The estimate exposes this multiplication; clean unexpected duplicates before joining again.
Yes. A join preserves columns from both tables with unique names. Append aligns equal header names and introduces columns found only on the right, filling missing values with empty cells. Both inputs still require unique, nonblank headers.
Did the matching counts differ from your expectation?
Try the CSV join example and compare the projected count with the relationship you expect before generating your own combined table.