Compare two versions of a table row by row and cell by cell

Compares two versions of a table row by row and cell by cell.

Input: multipart old_file and new_file (or old_file_url and new_file_url): CSV, TSV, XLSX or a text-based PDF, each up to 20 MiB and 250,000 rows. Optional options as a JSON string following /v1/schemas/table-diff-options: key (header(s) of the key column; detected when absent), column_map (old → new header), ignore_columns, numeric_tolerance, ignore_case, header_row_old/header_row_new, sheet_old/sheet_new, pdf_old/pdf_new, max_changes, output (json or csv).

Behaviour: the header row is found in each file, columns are paired by header (also when spelled differently, e.g. case or punctuation), renamed columns are paired by matching values or a contained header with evidence. The key is the given column(s), else a unique column (ID-like names first), else a unique column pair, else the row position (confidence low). Numbers are compared by value with the number format inferred per column and file; text after Unicode and whitespace normalisation. Duplicate keys are matched in order and reported.

Output: ApiResponse with DiffResult: both tables (header row, columns), key, column changes, rows[] (changed rows with every cell change and numeric delta, added and removed rows with their values, source rows), summary (matched, unchanged, changed, added, removed, cell changes per column). With output=csv the changes as semicolon CSV in long format.

Errors: 400 invalid file, 413 too large, 422 no header row, no common column or invalid options, 429 rate limit, 504 timeout.

Languages: responses are in English. Error, warning and rule codes are stable and language-neutral; agents may translate messages for the user. Column headers and common cell words are recognised in English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Czech, Turkish, Swedish, Danish, Norwegian, Finnish, Korean, Japanese and Chinese; numbers in both 1.234,56 and 1,234.56 style; any ISO 4217 currency (no conversion).

Access: open, rate-limited per IP and per MCP session; an X-API-Key header raises the limit. Uploaded files are processed in memory and never stored. Also available as MCP tool table_diff at /mcp (Streamable HTTP).