{
  "$version": "0.8",
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "Cipher Corpus Record (v0.8)",
  "type": "object",
  "required": [
    "id",
    "title",
    "cipher_family",
    "cipher_type",
    "plaintext",
    "ciphertext",
    "key",
    "language",
    "alphabet",
    "text_length",
    "normalized_text_length",
    "spacing",
    "punctuation",
    "casing",
    "difficulty",
    "source_type",
    "source",
    "license",
    "expected_attacks",
    "tags",
    "created_by",
    "verified",
    "dataset_version",
    "split",
    "transcription_quality",
    "source_provenance",
    "authenticity",
    "pair_status",
    "verification_method",
    "benchmark_eligible",
    "corpus_version",
    "source_batch"
  ],
  "properties": {
    "id": {
      "type": "string"
    },
    "title": {
      "type": "string"
    },
    "cipher_family": {
      "type": "string"
    },
    "cipher_type": {
      "type": "string"
    },
    "plaintext": {
      "type": "string"
    },
    "ciphertext": {
      "type": "string"
    },
    "key": {
      "type": "object"
    },
    "language": {
      "type": "string",
      "description": "ISO 639-1/2 code or BCP-47 language tag (e.g., 'en', 'fr', 'de', 'la', 'es')"
    },
    "split": {
      "type": "string",
      "enum": [
        "public",
        "blind"
      ],
      "description": "Benchmark split label. Both splits are published and downloadable: \"public\" is the default evaluation split and \"blind\" is a held-aside subset used to report a second score on records a tool was not tuned against. Neither label is secret, confidential, or withheld from the download. For a genuinely unseen evaluation, use data that has not been distributed. The immutable v0.6 schema described this label less accurately; its exact archived wording and the correction are recorded in docs/CORPUS-V06-ERRATA.md."
    },
    "transcription_quality": {
      "type": "string",
      "enum": [
        "clean",
        "noisy"
      ],
      "description": "Indicates whether the transcription is clean or contains realistic noise/errors"
    },
    "source_provenance": {
      "type": "object",
      "description": "Provenance and rights info for historical or synthetic records",
      "properties": {
        "url": {
          "type": [
            "string",
            "null"
          ],
          "format": "uri",
          "description": "A verified matching source URL when known; null for retired or unsupported citations. Read citation_erratum when present."
        },
        "archive": {
          "type": "string"
        },
        "publication_date": {
          "type": "string"
        },
        "license": {
          "type": "string"
        }
      },
      "required": [
        "url",
        "archive",
        "publication_date",
        "license"
      ]
    },
    "alphabet": {
      "type": "string"
    },
    "text_length": {
      "type": "integer",
      "minimum": 0,
      "description": "Unicode code points in the plaintext after NFC normalization, counting every character as written. Derived from plaintext only; see scripts/corpus-text-metrics.js."
    },
    "normalized_text_length": {
      "type": "integer",
      "minimum": 0,
      "description": "Cipher-relevant symbols in the plaintext: NFD normalize, drop combining marks, then count code points in Unicode categories L and N. Never exceeds text_length. Derived from plaintext only."
    },
    "spacing": {
      "type": "string",
      "enum": [
        "preserved",
        "removed",
        "grouped",
        "mixed"
      ]
    },
    "punctuation": {
      "type": "string",
      "enum": [
        "preserved",
        "removed",
        "mixed"
      ]
    },
    "casing": {
      "type": "string",
      "enum": [
        "uppercase",
        "lowercase",
        "mixed"
      ]
    },
    "difficulty": {
      "type": "string",
      "enum": [
        "beginner",
        "intermediate",
        "advanced",
        "expert"
      ]
    },
    "source_type": {
      "type": "string",
      "enum": [
        "synthetic",
        "historical",
        "public_domain",
        "user_contributed"
      ]
    },
    "source": {
      "type": "string"
    },
    "license": {
      "type": "string"
    },
    "notes": {
      "type": "string"
    },
    "expected_attacks": {
      "type": "array",
      "items": {
        "type": "string"
      }
    },
    "tags": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "uniqueItems": true,
      "description": "Topic tags, unique within a record and ordered by first use."
    },
    "created_by": {
      "type": "string"
    },
    "verified": {
      "type": "boolean",
      "description": "v0.8: whether this is a verified, benchmark-eligible pair. The v0.6 flag did not imply an authenticated historical source."
    },
    "dataset_version": {
      "type": "string"
    },
    "year": {
      "type": "string"
    },
    "historical_context": {
      "type": "string"
    },
    "known_solution_method": {
      "type": "string"
    },
    "hints": {
      "type": "array",
      "items": {
        "type": "string"
      }
    },
    "tool_notes": {
      "type": "string"
    },
    "related_exhibit_url": {
      "type": "string"
    },
    "try_in_cipher_detective_url": {
      "type": "string"
    },
    "benchmark_split": {
      "type": "string",
      "enum": [
        "beginner",
        "intermediate",
        "advanced",
        "expert",
        "historical",
        "blind"
      ]
    },
    "scoring_notes": {
      "type": "string"
    },
    "authenticity": {
      "type": "string",
      "enum": [
        "authentic_artifact",
        "authentic_excerpt",
        "reconstruction",
        "catalogued_unreviewed",
        "synthetic"
      ],
      "description": "What the material is. See data/corpus-v0.7-decisions.json for the definition of each value."
    },
    "pair_status": {
      "type": "string",
      "enum": [
        "verified_pair",
        "unverified_pair",
        "not_a_pair"
      ],
      "description": "Whether plaintext and ciphertext form a reproducible known-answer pair. Independent of authenticity: a teaching reconstruction can be a verified pair, and an authentic artifact may not be."
    },
    "verification_method": {
      "type": "string",
      "enum": [
        "engine_replay",
        "documented_computation",
        "none"
      ],
      "description": "How pair_status was established. engine_replay: the record round-trips through this repository's engines. documented_computation: verified against a primary source by a procedure recorded in data/corpus-v0.7-decisions.json, which the engines do not implement. none: not a verified pair. Tooling that drives the museum engines should filter on engine_replay."
    },
    "benchmark_eligible": {
      "type": "boolean",
      "description": "True only when pair_status is verified_pair. Benchmarks and any claim requiring a complete verified pair must filter on this field."
    },
    "source_batch": {
      "type": "string",
      "description": "The generator batch label this record came from (\"0.2\", \"0.5\", \"2.0\", \"0.6\"). Provenance only; it does not order releases."
    },
    "corpus_version": {
      "type": "string",
      "description": "The dataset release this record belongs to."
    },
    "citation_erratum": {
      "type": "object",
      "description": "A retired source URL is preserved here for audit, never presented as verified provenance for this record.",
      "properties": {
        "status": {
          "type": "string",
          "enum": [
            "wrong_work",
            "link_unavailable"
          ]
        },
        "previous_url": {
          "type": "string",
          "format": "uri"
        },
        "previous_archive": {
          "type": "string"
        },
        "reviewed_on": {
          "type": "string"
        },
        "note": {
          "type": "string"
        }
      },
      "required": [
        "status",
        "previous_url",
        "previous_archive",
        "reviewed_on",
        "note"
      ]
    }
  }
}