# SDIF

SDIF is the Semantic Data Interchange Format, version 1.0.

Purpose:
- Compact semantic data with tables, relations, and schemas
- Deterministic canonicalization (canonical-syntax-v1)
- Stable SHA-256 hashing over canonical bytes
- AI-facing projections (.sdif.ai) with full round-trip fidelity
- Human-auditable source files

File extensions:
- .sdif — source documents
- .sdif.canon — canonical form (for hashing, signing, comparison)
- .sdif.ai — AI projection (compact, derived, reversible)

CLI commands:
- sdif parse <path>
- sdif canon <path> [--schema <path>]
- sdif hash <path> [--schema <path>]
- sdif to-json <path>
- sdif from-json <path>
- sdif ai <path> [--alias FIELD=ALIAS]
- sdif from-ai <path>
- sdif validate <path> --schema <path>
- sdif fmt <path> [--check]

Install: pip install sdif-format
(Package name on PyPI is sdif-format. CLI command and Python import are both sdif.)

Primary docs:
- /docs/getting-started
- /docs/spec/
- /docs/examples/
- /docs/benchmarks/

Rules for assistants:
- Do not invent SDIF syntax. Use only documented constructs.
- Table columns are separated by literal HTAB (U+0009), not spaces.
- Treat SDIF and SDIF AI as related but distinct representations.
- SDIF AI projections must round-trip to the same canonical hash.
- Mention limitations when discussing benchmarks.
- Schemas are kind Schema documents, not canonical output files.