CSV comparison by key

How to use

Values compare as exact strings, preserving leading zeros, spaces and case. Row order does not affect matching. Any empty key component or duplicate composite key is rejected; records are never discarded.

Headers must be nonempty and unique. Only shared columns are offered for comparison; other columns remain in exported records. A changed key becomes one removal and one addition.

Example: reconcile two inventory exports

Load the example and select sku as the key. The price of 001 changes, 002 is removed and 003 is added. Reordering rows creates no differences.

Export format

Added CSV uses complete updated rows; removed CSV uses complete original rows. Each changed CSV row contains all before:original columns followed by all after:updated columns.

Exports use UTF-8 BOM, commas, quoted fields and CRLF. Formula-risk cells, including headers, receive an apostrophe. This may change exported values, but never comparison data.

Each input: 1 MiB, 20,000 data records and 500 columns maximum. Combined exports: 4 MiB. Processing times out after 8 seconds and can be cancelled. Blank records are not silently skipped; only the final line break is ignored.