Start from documented encoding evidence
A DBF header can contain a language-driver or code-page mark. This tool maps a supported mark to an explicit decoder; it does not silently treat every old table as UTF-8 or Windows-1252. An unknown or unsupported mark asks for a deliberate encoding choice. Check the original application’s export settings or documentation when that information is available.
DOS and Windows encodings can assign different characters to the same byte. CP437, CP850, CP852 and CP857 have explicit mapping tables here; supported Windows and East Asian choices use the browser’s strict decoder. Selecting a familiar language name is a starting point for verification, not proof that the file was created using that code page.
Compare recognizable records
Run the table and inspect names, addresses and notes you can recognize. The record preview preserves leading spaces while removing fixed-width trailing space/NUL padding from character fields. Search filters the preview only. The complete downloads still contain all records selected by the active/deleted setting, so a search never silently changes the exported dataset.
For example, Turkish text can contain dotted and dotless I alongside Ş and Ğ. A wrong encoding may turn those bytes into different valid letters instead of producing an error. Change the encoding control and run again, then compare the same source record number. The full-record inspector shows unshortened text when a preview cell is abbreviated.
Keep assumptions and failures visible
The migration report records the selected encoding and whether it came from the header or an explicit user choice. A decoding failure creates a cell-specific issue with stored bytes rather than replacement characters. A successful decode can still be linguistically wrong, especially with single-byte encodings in which most byte values map to something.
Memo text uses the same chosen encoding as the table. Binary character and memo fields are not passed through a text decoder. Retain both original files, the encoding evidence and representative expected strings for your migration review. This tool does not automatically detect mixed encodings within one table, repair previously corrupted text or translate a name into another language.