{
  "type": "object",
  "properties": {
    "schema": {
      "const": "okf-evidence-unit.v1"
    },
    "kind": {
      "enum": [
        "section",
        "paragraph",
        "table",
        "definition",
        "exception",
        "cross-reference",
        "compound",
        "unresolved-fragment",
        "page-fallback"
      ]
    },
    "boundary_status": {
      "enum": [
        "machine-detected",
        "author-declared"
      ]
    },
    "completeness": {
      "enum": [
        "complete-within-declared-boundary",
        "unresolved",
        "fallback"
      ]
    },
    "offset_unit": {
      "const": "utf-8-bytes"
    },
    "joiner": {
      "enum": [
        "",
        "\n",
        "\n\n"
      ]
    },
    "spans": {
      "type": "array",
      "minItems": 1,
      "maxItems": 32,
      "items": {
        "type": "object",
        "properties": {
          "source_url": {
            "type": "string",
            "minLength": 1,
            "maxLength": 8000
          },
          "extraction_url": {
            "type": "string",
            "minLength": 1,
            "maxLength": 8000
          },
          "locator": {
            "type": "string",
            "minLength": 1,
            "maxLength": 8000
          },
          "source_sha256": {
            "type": "string",
            "pattern": "^[a-f0-9]{64}$"
          },
          "extraction_sha256": {
            "type": "string",
            "pattern": "^[a-f0-9]{64}$"
          },
          "source_text_sha256": {
            "type": "string",
            "pattern": "^[a-f0-9]{64}$"
          },
          "literal_sha256": {
            "type": "string",
            "pattern": "^[a-f0-9]{64}$"
          },
          "source_text_bytes": {
            "type": "integer",
            "minimum": 0,
            "maximum": 16777216
          },
          "source_start": {
            "type": "integer",
            "minimum": 0,
            "maximum": 16777216
          },
          "source_end": {
            "type": "integer",
            "minimum": 0,
            "maximum": 16777216
          },
          "unit_start": {
            "type": "integer",
            "minimum": 0,
            "maximum": 16777216
          },
          "unit_end": {
            "type": "integer",
            "minimum": 0,
            "maximum": 16777216
          }
        },
        "required": [
          "source_url",
          "extraction_url",
          "locator",
          "source_sha256",
          "extraction_sha256",
          "source_text_sha256",
          "literal_sha256",
          "source_text_bytes",
          "source_start",
          "source_end",
          "unit_start",
          "unit_end"
        ],
        "additionalProperties": false
      }
    }
  },
  "required": [
    "schema",
    "kind",
    "boundary_status",
    "completeness",
    "offset_unit",
    "joiner",
    "spans"
  ],
  "additionalProperties": false,
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://chris-page-gov.github.io/okf-explorer/profiles/context-assembly/v1/evidence-unit.schema.json",
  "title": "Exact logical evidence-unit metadata v1",
  "description": "Shape only. Runtime validates whole-text coverage, UTF-8 offsets, joins, provenance and fragment hashes; producer verifies inclusion in original source text.",
  "allOf": [
    {
      "if": {
        "properties": {
          "kind": {
            "const": "unresolved-fragment"
          }
        }
      },
      "then": {
        "properties": {
          "completeness": {
            "const": "unresolved"
          }
        }
      }
    },
    {
      "if": {
        "properties": {
          "kind": {
            "const": "page-fallback"
          }
        }
      },
      "then": {
        "properties": {
          "completeness": {
            "const": "fallback"
          }
        }
      }
    }
  ]
}