autonomous intelligent information publishing - MVP

P1/schema_map.json epoch 1

{
  "sync_id": "sync-fbaf01faca",
  "source_kind": "gdrive",
  "tables": {
    "comments": {
      "row_count": 4,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "cmt_1",
            "cmt_2",
            "cmt_3",
            "cmt_4"
          ]
        },
        "file_id": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.75,
          "samples": [
            "doc_handbook",
            "doc_pricing",
            "doc_roadmap26"
          ]
        },
        "author": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "alice@acme.com",
            "bob@acme.com",
            "carol@acme.com",
            "dana@vendor.io"
          ]
        },
        "resolved": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.5,
          "samples": [
            "no",
            "yes"
          ]
        },
        "quoted_text": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "19 per seat",
            "EU market",
            "eight weeks ahead",
            "negotiated above that"
          ]
        },
        "text": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "Can we get the ladder in writing for the renewal?",
            "Confirmed with finance.",
            "Needs a legal gate before we commit publicly.",
            "Softened per HR feedback."
          ]
        },
        "created": {
          "inferred_type": "date",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "2026-02-11T09:30:00Z",
            "2026-07-15T14:20:00Z",
            "2026-08-20T10:00:00Z",
            "2026-08-29T11:00:00Z"
          ]
        }
      }
    },
    "drives": {
      "row_count": 1,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 1,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 1.0,
          "samples": [
            "drv_product"
          ]
        },
        "name": {
          "inferred_type": "string",
          "rows": 1,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 1.0,
          "samples": [
            "Product"
          ]
        },
        "drive_kind": {
          "inferred_type": "string",
          "rows": 1,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 1.0,
          "samples": [
            "shared_drive"
          ]
        },
        "created": {
          "inferred_type": "date",
          "rows": 1,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 1.0,
          "samples": [
            "2025-03-01T09:00:00Z"
          ]
        }
      }
    },
    "files": {
      "row_count": 20,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 20,
          "unique_ratio": 1.0,
          "samples": [
            "doc_api",
            "doc_handbook",
            "doc_legacy",
            "doc_pricing",
            "doc_pricing_v1"
          ]
        },
        "drive_id": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 0.05,
          "samples": [
            "drv_product"
          ]
        },
        "parent_id": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 1,
          "distinct": 7,
          "unique_ratio": 0.368,
          "samples": [
            "fld_archive",
            "fld_legal",
            "fld_metrics",
            "fld_product",
            "fld_roadmap"
          ]
        },
        "path": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 20,
          "unique_ratio": 1.0,
          "samples": [
            "/Product",
            "/Product/Legal",
            "/Product/Legal/DPA Template.pdf",
            "/Product/Legal/Vendor MSA.pdf",
            "/Product/Metrics"
          ]
        },
        "name": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 19,
          "unique_ratio": 0.95,
          "samples": [
            "API Spec",
            "Archive",
            "DPA Template.pdf",
            "Funnel Analysis",
            "KPI Dashboard"
          ]
        },
        "depth": {
          "inferred_type": "integer",
          "rows": 20,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.2,
          "samples": [
            "0",
            "1",
            "2",
            "3"
          ]
        },
        "is_folder": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "no",
            "yes"
          ]
        },
        "item_kind": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 5,
          "unique_ratio": 0.25,
          "samples": [
            "document",
            "folder",
            "image",
            "pdf",
            "spreadsheet"
          ]
        },
        "mime_type": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 5,
          "unique_ratio": 0.25,
          "samples": [
            "application/pdf",
            "application/vnd.google-apps.document",
            "application/vnd.google-apps.folder",
            "application/vnd.google-apps.spreadsheet",
            "image/png"
          ]
        },
        "size_bytes": {
          "inferred_type": "integer",
          "rows": 20,
          "nulls": 17,
          "distinct": 3,
          "unique_ratio": 1.0,
          "samples": [
            "119800",
            "284500",
            "48200"
          ]
        },
        "owner_email": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.15,
          "samples": [
            "alice@acme.com",
            "bob@acme.com",
            "carol@acme.com"
          ]
        },
        "created": {
          "inferred_type": "date",
          "rows": 20,
          "nulls": 0,
          "distinct": 20,
          "unique_ratio": 1.0,
          "samples": [
            "2025-01-05T09:00:00Z",
            "2025-01-05T09:10:00Z",
            "2025-03-01T09:00:00Z",
            "2025-03-02T09:00:00Z",
            "2025-03-02T09:05:00Z"
          ]
        },
        "modified": {
          "inferred_type": "date",
          "rows": 20,
          "nulls": 0,
          "distinct": 13,
          "unique_ratio": 0.65,
          "samples": [
            "2025-02-01T09:00:00Z",
            "2025-03-06T09:00:00Z",
            "2025-06-20T09:30:00Z",
            "2026-03-01T10:00:00Z",
            "2026-04-18T10:30:00Z"
          ]
        },
        "modified_by": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.2,
          "samples": [
            "alice@acme.com",
            "bob@acme.com",
            "carol@acme.com",
            "dana@vendor.io"
          ]
        },
        "trashed": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "no",
            "yes"
          ]
        },
        "starred": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "no",
            "yes"
          ]
        },
        "name_is_unique": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "no",
            "yes"
          ]
        },
        "word_count": {
          "inferred_type": "integer",
          "rows": 20,
          "nulls": 8,
          "distinct": 11,
          "unique_ratio": 0.917,
          "samples": [
            "12",
            "13",
            "18",
            "19",
            "28"
          ]
        },
        "section_count": {
          "inferred_type": "integer",
          "rows": 20,
          "nulls": 8,
          "distinct": 5,
          "unique_ratio": 0.417,
          "samples": [
            "1",
            "2",
            "3",
            "4",
            "5"
          ]
        },
        "content_hash": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 8,
          "distinct": 12,
          "unique_ratio": 1.0,
          "samples": [
            "03a6bce525626526",
            "0825400fabf9143c",
            "1242599fce644003",
            "594f908eefef444b",
            "6f55c3605dc4142c"
          ]
        },
        "extract_status": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.2,
          "samples": [
            "exported_no_size",
            "extracted",
            "not_applicable",
            "unsupported_type"
          ]
        },
        "web_view_link": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 20,
          "unique_ratio": 1.0,
          "samples": [
            "https://drive.google.com/file/d/doc_api/view",
            "https://drive.google.com/file/d/doc_handbook/view",
            "https://drive.google.com/file/d/doc_legacy/view",
            "https://drive.google.com/file/d/doc_pricing/view",
            "https://drive.google.com/file/d/doc_pricing_v1/view"
          ]
        },
        "link_visibility": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "anyone_with_link",
            "domain"
          ]
        },
        "shared_externally": {
          "inferred_type": "string",
          "rows": 20,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.1,
          "samples": [
            "no",
            "yes"
          ]
        }
      }
    },
    "labels": {
      "row_count": 4,
      "columns": {
        "file_id": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 1.0,
          "samples": [
            "doc_pricing",
            "doc_pricing_v1",
            "doc_roadmap26",
            "pdf_msa"
          ]
        },
        "label": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.75,
          "samples": [
            "confidential",
            "lifecycle",
            "needs-review"
          ]
        },
        "value": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.75,
          "samples": [
            "internal-only",
            "legal",
            "superseded"
          ]
        }
      }
    },
    "links": {
      "row_count": 4,
      "columns": {
        "src_file_id": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.5,
          "samples": [
            "doc_q3plan",
            "doc_roadmap26"
          ]
        },
        "dst_file_id": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.75,
          "samples": [
            "doc_api",
            "doc_pricing",
            "doc_roadmap26"
          ]
        },
        "link_type": {
          "inferred_type": "string",
          "rows": 4,
          "nulls": 0,
          "distinct": 1,
          "unique_ratio": 0.25,
          "samples": [
            "mentions"
          ]
        }
      }
    },
    "permissions": {
      "row_count": 31,
      "columns": {
        "file_id": {
          "inferred_type": "string",
          "rows": 31,
          "nulls": 0,
          "distinct": 20,
          "unique_ratio": 0.645,
          "samples": [
            "doc_api",
            "doc_handbook",
            "doc_legacy",
            "doc_pricing",
            "doc_pricing_v1"
          ]
        },
        "principal": {
          "inferred_type": "string",
          "rows": 31,
          "nulls": 0,
          "distinct": 7,
          "unique_ratio": 0.226,
          "samples": [
            "acme.com",
            "alice@acme.com",
            "anyone",
            "bob@acme.com",
            "carol@acme.com"
          ]
        },
        "principal_type": {
          "inferred_type": "string",
          "rows": 31,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.129,
          "samples": [
            "anyone",
            "domain",
            "group",
            "user"
          ]
        },
        "role": {
          "inferred_type": "string",
          "rows": 31,
          "nulls": 0,
          "distinct": 5,
          "unique_ratio": 0.161,
          "samples": [
            "commenter",
            "organizer",
            "owner",
            "reader",
            "writer"
          ]
        },
        "inherited": {
          "inferred_type": "string",
          "rows": 31,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.065,
          "samples": [
            "no",
            "yes"
          ]
        }
      }
    },
    "principals": {
      "row_count": 7,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 7,
          "nulls": 0,
          "distinct": 7,
          "unique_ratio": 1.0,
          "samples": [
            "acme.com",
            "alice@acme.com",
            "anyone",
            "bob@acme.com",
            "carol@acme.com"
          ]
        },
        "principal_type": {
          "inferred_type": "string",
          "rows": 7,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.571,
          "samples": [
            "anyone",
            "domain",
            "group",
            "user"
          ]
        },
        "domain": {
          "inferred_type": "string",
          "rows": 7,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.429,
          "samples": [
            "*",
            "acme.com",
            "vendor.io"
          ]
        }
      }
    },
    "revisions": {
      "row_count": 8,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 8,
          "nulls": 0,
          "distinct": 8,
          "unique_ratio": 1.0,
          "samples": [
            "doc_handbook@r1",
            "doc_handbook@r2",
            "doc_pricing@r1",
            "doc_pricing@r2",
            "doc_roadmap26@r1"
          ]
        },
        "file_id": {
          "inferred_type": "string",
          "rows": 8,
          "nulls": 0,
          "distinct": 4,
          "unique_ratio": 0.5,
          "samples": [
            "doc_handbook",
            "doc_pricing",
            "doc_roadmap26",
            "pdf_msa"
          ]
        },
        "revision_id": {
          "inferred_type": "string",
          "rows": 8,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.375,
          "samples": [
            "r1",
            "r2",
            "r3"
          ]
        },
        "modified": {
          "inferred_type": "date",
          "rows": 8,
          "nulls": 0,
          "distinct": 8,
          "unique_ratio": 1.0,
          "samples": [
            "2025-03-05T09:00:00Z",
            "2025-06-20T09:30:00Z",
            "2025-11-04T10:12:00Z",
            "2026-01-09T14:00:00Z",
            "2026-03-14T09:00:00Z"
          ]
        },
        "modified_by": {
          "inferred_type": "string",
          "rows": 8,
          "nulls": 0,
          "distinct": 3,
          "unique_ratio": 0.375,
          "samples": [
            "alice@acme.com",
            "bob@acme.com",
            "carol@acme.com"
          ]
        },
        "size_bytes": {
          "inferred_type": "integer",
          "rows": 8,
          "nulls": 7,
          "distinct": 1,
          "unique_ratio": 1.0,
          "samples": [
            "284500"
          ]
        }
      }
    },
    "sections": {
      "row_count": 36,
      "columns": {
        "id": {
          "inferred_type": "string",
          "rows": 36,
          "nulls": 0,
          "distinct": 36,
          "unique_ratio": 1.0,
          "samples": [
            "doc_api#s0",
            "doc_api#s1",
            "doc_api#s2",
            "doc_api#s3",
            "doc_handbook#s0"
          ]
        },
        "file_id": {
          "inferred_type": "string",
          "rows": 36,
          "nulls": 0,
          "distinct": 12,
          "unique_ratio": 0.333,
          "samples": [
            "doc_api",
            "doc_handbook",
            "doc_legacy",
            "doc_pricing",
            "doc_pricing_v1"
          ]
        },
        "section_index": {
          "inferred_type": "integer",
          "rows": 36,
          "nulls": 0,
          "distinct": 5,
          "unique_ratio": 0.139,
          "samples": [
            "0",
            "1",
            "2",
            "3",
            "4"
          ]
        },
        "heading": {
          "inferred_type": "string",
          "rows": 36,
          "nulls": 0,
          "distinct": 35,
          "unique_ratio": 0.972,
          "samples": [
            "API Spec",
            "Authentication",
            "Data Processing Addendum",
            "Dependencies",
            "Endpoints"
          ]
        },
        "level": {
          "inferred_type": "integer",
          "rows": 36,
          "nulls": 0,
          "distinct": 2,
          "unique_ratio": 0.056,
          "samples": [
            "1",
            "2"
          ]
        },
        "char_start": {
          "inferred_type": "integer",
          "rows": 36,
          "nulls": 0,
          "distinct": 22,
          "unique_ratio": 0.611,
          "samples": [
            "0",
            "104",
            "112",
            "12",
            "120"
          ]
        },
        "char_end": {
          "inferred_type": "integer",
          "rows": 36,
          "nulls": 0,
          "distinct": 33,
          "unique_ratio": 0.917,
          "samples": [
            "104",
            "111",
            "112",
            "12",
            "120"
          ]
        },
        "word_count": {
          "inferred_type": "integer",
          "rows": 36,
          "nulls": 0,
          "distinct": 15,
          "unique_ratio": 0.417,
          "samples": [
            "0",
            "10",
            "11",
            "12",
            "13"
          ]
        },
        "text": {
          "inferred_type": "string",
          "rows": 36,
          "nulls": 9,
          "distinct": 27,
          "unique_ratio": 1.0,
          "samples": [
            "600 requests per minute per workspace, burst to 1200 for 10 seconds.",
            "Anything under 200 needs no approval. Travel requires a manager sign-off.",
            "Bearer tokens scoped per workspace. Tokens rotate every 90 days.",
            "Blocked on the pricing tiers decision in https://docs.google.com/document/d/doc_pricing.\nRoadmap context: https://docs.google.com/document/d/doc_roadmap26.",
            "Capped at fees paid in the preceding 12 months. Excludes gross negligence."
          ]
        }
      }
    }
  }
}

Back to instance