SAV and ZSAV reading
SAV Inspector reads SPSS system files and compressed ZSAV files into a shared Rust model used by both browser and server transports.
- Numeric and string variables
- Variable and value labels
- Print/write formats and measurement levels
- User-defined missing values
- File label, notes, weight variable and multiple-response sets
- Cases and supported compression variants
Writing and verification
The writer creates a new SAV or ZSAV rather than modifying the source bytes in place. The round-trip check immediately reads the result and compares supported dictionary properties and cases.
Proprietary extensions, extremely unusual records or metadata outside the supported model may not be preserved. Inspect the compatibility preview and retain the source whenever a third-party workflow depends on specialised SPSS behaviour.
CSV and XLSX imports
CSV and Excel tables contain cells, not an SPSS dictionary. The local import preview therefore asks for delimiters, encoding, decimal conventions, worksheets and inferred types. It does not invent value labels, missing definitions, weights or multiple-response sets that are absent from the source.
- Open and inspect
Confirm format, encoding, compression, rows, columns and dictionary properties.
- Review compatibility findings
Pay attention to unsupported or ambiguous metadata before editing.
- Write a new copy
Choose SAV or ZSAV and never discard the source during validation.
- Run the round-trip
Compare the reopened result and test it in the downstream application when that workflow is critical.
Frequently asked questions
Are SAV and ZSAV the same?
They represent SPSS system data, but ZSAV uses additional compression and requires newer SPSS versions. Conversion changes representation, not automatically the meaning.
Does XLSX preserve SPSS metadata?
No. A spreadsheet does not natively carry the full SPSS dictionary. Missing rules, value labels and response sets must be supplied or recreated deliberately.
Can every historical SAV file be guaranteed to open?
No parser can honestly guarantee every damaged, proprietary or unusual file. Safety limits and clear errors take precedence over attempting unbounded decoding.
How should an unsupported file be reported?
Keep the original, record the producing application/version and provide a non-sensitive reproduction if possible. Do not send confidential data through an ordinary support message.