Dataset preflight report

{
  "duplicates": {
    "exact_duplicate_excess_files": 1,
    "exact_groups": [
      {
        "paths": [
          "data/batch-a/exact-a.txt",
          "data/batch-a/exact-b.txt"
        ],
        "sha256": "7b90339159154b1f28a746a9bc24d3f2a139189b0ca33f53dde7d3ca4b1fb1dd"
      }
    ],
    "normalized_duplicate_excess_files": 1,
    "normalized_groups": [
      {
        "normalization": "text-nfc-newline-rstrip-v1",
        "normalized_sha256": "f32f72e792714002d0e3a8e6ee421ae94d13371a4cd0077fedcf991556df06ec",
        "paths": [
          "data/batch-b/normalized-a.txt",
          "data/batch-b/normalized-b.txt"
        ]
      }
    ]
  },
  "inventory": {
    "by_type": {
      "text": 5,
      "unknown": 1
    },
    "file_count": 6,
    "files": [
      {
        "extension": "",
        "normalized_sha256": "1ab49e53282e8c1ecf7ee5a9e1f6c52bc4e78323f0bcdfac5fdbcfc7dbd55169",
        "path": ".env",
        "secret_scan_status": "inspected",
        "sha256": "67c16b7762eed7e1529480e8ba1e3135bbf03159deca2b34eba665a87ace6707",
        "size_bytes": 20,
        "text_status": "utf8",
        "type": "unknown"
      },
      {
        "extension": ".txt",
        "normalized_sha256": "fba1a88a828b4e7c91f9f16e19b2ff0eb099c59dc9a8ad8dcaca9b73d96945b4",
        "path": "data/batch-a/exact-a.txt",
        "secret_scan_status": "inspected",
        "sha256": "7b90339159154b1f28a746a9bc24d3f2a139189b0ca33f53dde7d3ca4b1fb1dd",
        "size_bytes": 18,
        "text_status": "utf8",
        "type": "text"
      },
      {
        "extension": ".txt",
        "normalized_sha256": "fba1a88a828b4e7c91f9f16e19b2ff0eb099c59dc9a8ad8dcaca9b73d96945b4",
        "path": "data/batch-a/exact-b.txt",
        "secret_scan_status": "inspected",
        "sha256": "7b90339159154b1f28a746a9bc24d3f2a139189b0ca33f53dde7d3ca4b1fb1dd",
        "size_bytes": 18,
        "text_status": "utf8",
        "type": "text"
      },
      {
        "extension": ".txt",
        "normalized_sha256": "f32f72e792714002d0e3a8e6ee421ae94d13371a4cd0077fedcf991556df06ec",
        "path": "data/batch-b/normalized-a.txt",
        "secret_scan_status": "inspected",
        "sha256": "09db226c93082846718fc528c8953a43471380d0349e60abbab9af548247eace",
        "size_bytes": 20,
        "text_status": "utf8",
        "type": "text"
      },
      {
        "extension": ".txt",
        "normalized_sha256": "f32f72e792714002d0e3a8e6ee421ae94d13371a4cd0077fedcf991556df06ec",
        "path": "data/batch-b/normalized-b.txt",
        "secret_scan_status": "inspected",
        "sha256": "2a2eea2042704277d6aa82bfc94ac5d06cad1105547ba21e7afc7b74522ceaeb",
        "size_bytes": 25,
        "text_status": "utf8",
        "type": "text"
      },
      {
        "extension": ".txt",
        "normalized_sha256": "e0bbbf175a7559a5a051cbe0ba91444f1c43aa31df5a2a7817a8cd39a70d3929",
        "path": "evidence/source-batch-a.txt",
        "secret_scan_status": "inspected",
        "sha256": "2d7a4a7b0ac6d437ced06ed63e64a22bc2685b47ff89edc1a432e32c7bff5dc4",
        "size_bytes": 48,
        "text_status": "utf8",
        "type": "text"
      }
    ],
    "total_bytes": 149
  },
  "issues": [],
  "policy": {
    "configured": true,
    "results": [
      {
        "limit": 1,
        "metric": "exact_duplicate_excess_files",
        "observed": 1,
        "passed": true
      },
      {
        "limit": 1,
        "metric": "normalized_duplicate_excess_files",
        "observed": 1,
        "passed": true
      },
      {
        "limit": 0,
        "metric": "scan_issues",
        "observed": 0,
        "passed": true
      },
      {
        "limit": 1,
        "metric": "secret_risk_files",
        "observed": 1,
        "passed": true
      },
      {
        "limit": 4,
        "metric": "uncovered_files",
        "observed": 4,
        "passed": true
      }
    ],
    "status": "pass",
    "thresholds": {
      "max_exact_duplicate_excess_files": 1,
      "max_normalized_duplicate_excess_files": 1,
      "max_scan_issues": 0,
      "max_secret_risk_files": 1,
      "max_uncovered_files": 4
    }
  },
  "provenance": {
    "configured": true,
    "conflicting_file_count": 0,
    "coverage": [
      {
        "declaration_ids": [],
        "has_conflict": false,
        "path": ".env"
      },
      {
        "declaration_ids": [
          "fixture-source"
        ],
        "has_conflict": false,
        "path": "data/batch-a/exact-a.txt"
      },
      {
        "declaration_ids": [
          "fixture-source"
        ],
        "has_conflict": false,
        "path": "data/batch-a/exact-b.txt"
      },
      {
        "declaration_ids": [],
        "has_conflict": false,
        "path": "data/batch-b/normalized-a.txt"
      },
      {
        "declaration_ids": [],
        "has_conflict": false,
        "path": "data/batch-b/normalized-b.txt"
      },
      {
        "declaration_ids": [],
        "has_conflict": false,
        "path": "evidence/source-batch-a.txt"
      }
    ],
    "covered_file_count": 2,
    "declarations": [
      {
        "evidence_files": [
          {
            "exists": true,
            "path": "evidence/source-batch-a.txt"
          }
        ],
        "id": "fixture-source",
        "kind": "source",
        "matched_file_count": 2,
        "scope": {
          "path": "data/batch-a",
          "type": "prefix"
        }
      }
    ],
    "uncovered_file_count": 4,
    "uncovered_paths": [
      ".env",
      "data/batch-b/normalized-a.txt",
      "data/batch-b/normalized-b.txt",
      "evidence/source-batch-a.txt"
    ]
  },
  "scan": {
    "normalization": "text-nfc-newline-rstrip-v1",
    "secret_text_limit_bytes": 2097152,
    "source": ".",
    "text_inspection_limit_bytes": 8388608
  },
  "schema_version": "yy-dataset-preflight-report/v1",
  "secret_risk": {
    "file_count": 1,
    "files": [
      {
        "path": ".env",
        "signal_codes": [
          "sensitive-config-filename"
        ]
      }
    ]
  },
  "tool": {
    "name": "yy-dataset-preflight",
    "version": "0.1.0"
  }
}