Add networkRailCorpus schema

protos/rail_backend/v1/schedule_payload.proto

@@ -1,198 +0,0 @@
-syntax = "proto3";
-
-package rail_backend.v1;
-
-enum SchedulePayloadType {
-    VSTP_MESSAGE_TYPE_UNSPECIFIED = 0;
-    Create = 1;
-    Delete = 2;
-}
-
-message RunsOnDays {
-    bool sunday = 1;
-    bool monday = 2;
-    bool tuesday = 3;
-    bool wednesday = 4;
-    bool thursday = 5;
-    bool friday = 6;
-    bool saturday = 7;
-    bool bank_holidays = 8;
-}
-
-enum TrainStatus {
-    TRAIN_STATUS_UNSPECIFIED = 0;
-    PERMANENT_BUS = 1;
-    PERMANENT_FREIGHT = 2;
-    PERMANENT_PASSENGER_OR_PARCELS = 3;
-    PERMANENT_SHIP = 4;
-    PERMANENT_TRIP = 5;
-    STP_PASSENGER_ORPARCELS = 6;
-    STP_FREIGHT = 7;
-    STP_TRIP = 8;
-    STP_SHIP = 9;
-    STP_BUS = 10;
-}
-
-enum TrainCategory {
-    TRAIN_CATEGORY_UNSPECIFIED = 0;
-    METRO_SERVICE = 1;
-    UNADVERTISED_PASSENGER_TRAIN = 2;
-    ORDINARY_PASSENGER_TRAIN = 3;
-    STAFF_TRAIN = 4;
-    MIXED_TRAIN = 5;
-    CHANNEL_TUNNEL_TRAIN = 6;
-    EUROPEAN_SLEEPER_TRAIN = 7;
-    INTERNATIONAL_PASSENGER_TRAIN = 8;
-    MOTORAIL_SERVICE = 9;
-    UNADVERTISED_EXPRESS_TRAIN = 10;
-    EXPRESS_PASSENGER_TRAIN = 11;
-    SLEEPER_SERVICE = 12;
-    REPLACEMENT_BUS_SERVICE = 13;
-    BUS_SERVICE = 14;
-    SHIP = 15;
-    EMPTY_COACHING_STOCK = 16;
-    EMPTY_METRO_SERVICE = 17;
-    METRO_STAFF_SERVICE = 18;
-    POSTAL_TRAIN = 19;
-    POST_OFFICE_PARCELS_TRAIN = 20;
-    EMPTY_NON_PASSENGER_STOCK = 21;
-    DEPARTMENTAL_TRAIN = 22;
-    CIVIL_ENGINEER_TRAIN = 23;
-    MECHANICAL_AND_ELECTRICAL_ENGINEER_TRAIN = 24;
-    STORES_TRAIN = 25;
-    TEST_TRAIN = 26;
-    SIGNAL_AND_TELECOMMUNICATIONS_ENGINEER_TRAIN = 27;
-    LOCOMOTIVE_AND_BRAKE_VAN = 28;
-    LIGHT_LOCOMOTIVE = 29;
-    AUTOMOTIVE_COMPONENTS_TRAIN = 30;
-    AUTOMOTIVE_VEHICLE_TRAIN = 31;
-    EDIBLE_PRODUCTS_TRAIN = 32;
-    INDUSTRIAL_MINERALS_TRAIN = 33;
-    CHEMICAL_TRAIN = 34;
-    BUILDING_MATERIALS_TRAIN = 35;
-    GENERAL_MERCHANDISE_TRAIN = 36;
-    EUROPEAN_RAILFREIGHT = 37;
-    FREIGHTLINER_CONTRACTS = 38;
-    FREIGHTLINER_OTHER = 39;
-    COAL_TRAIN = 40;
-    COAL_POWER_STATION_TRAIN = 41;
-    COAL_OR_NUCLEAR_TRAIN = 42;
-    METALS_TRAIN = 43;
-    AGGREGATES_TRAIN = 44;
-    DOMESTIC_AND_INDUSTRIAL_WASTE_TRAIN = 45;
-    TRAINLOAD_BUILDING_MATERIALS_TRAIN = 46;
-    PETROLEUM_PRODUCTS_TRAIN = 47;
-    MIXED_FREIGHT_CHANNEL_TUNNEL_TRAIN = 48;
-    INTERMODAL_CHANNEL_TUNNEL_TRAIN = 49;
-    AUTOMOTIVE_FREIGHT_CHANNEL_TUNNEL_TRAIN = 50;
-    CONTRACT_FREIGHT_CHANNEL_TUNNEL_TRAIN = 51;
-    HAULMARK_FREIGHT_CHANNEL_TUNNEL_TRAIN = 52;
-    JOINT_VENTURE_CHANNEL_TUNNEL_FREIGHT_TRAIN = 53;
-}
-
-enum PowerType {
-    POWER_TYPE_UNSPECIFIED = 0;
-    DIESEL = 1;
-    DIESEL_MULTIPLE_UNIT = 2;
-    DISEL_MECHANICAL_MULTIPLE_UNIT = 3;
-    ELECTRIC = 4;
-    ELECTRO_DIESEL = 5;
-    EMU_PLUS_LOCOMOTIVE = 6;
-    ELECTRIC_MULTIPLE_UNIT = 7;
-    HIGH_SPEED_TRAIN = 8;
-}
-
-message OperatingCharacteristics {
-    bool vacuum_braked = 1;         // B
-    bool timed_at_100mph = 2;       // C
-    bool driver_only_operated = 3;  // D
-    bool conveys_mk4_coaches = 4;   // E
-    bool guard_required = 5;        // G
-    bool timed_at_110mph = 6;       // M
-    bool push_pull_train = 7;       // P
-    bool runs_as_required = 8;      // Q
-    bool air_conditioned_with_pa = 9;//R
-    bool steam_heated = 10;
-    bool runs_to_terminals_as_required = 11; // Y
-    bool may_convey_sb1c_gauge = 12; // Z
-}
-
-enum SleepingAccommodation {
-    NO_SLEEPING_ACCOMMODATION = 0;
-    FIRST_AND_STANDARD = 1;
-    FIRST_ONLY = 2;
-    STANDARD_ONLY = 3;
-}
-
-message Reservations {
-    bool reservations_available = 1;
-    bool reservations_compulsory = 2;
-    bool bike_reservations_essential = 3;
-    bool reservations_from_any_station = 4;
-}
-
-message Catering {
-    bool buffet_service = 1;
-    bool first_class_restaurant = 2;
-    bool hot_food_available = 3;
-    bool first_class_meal_included = 4;
-    bool restaurant_available = 5;
-    bool trolley_service = 6;
-}
-
-enum STPIndicator {
-    STP_INDICATOR_NOT_SPECIFIED = 0;
-    C = 1;
-    N = 2;
-    O = 3;
-    P = 4;
-}
-
-enum RecordIdentity {
-    RECORD_IDENTITY_NOT_SPECIFIED = 0;
-    ORIGIN = 1;
-    INTERMEDIATE = 2;
-    TERMINATING = 3;
-}
-
-message ScheduleLocation {
-    RecordIdentity record_identity = 1;
-    string tiploc = 2;
-    int32 tiploc_instance = 3;
-    int32 arrival_offset_seconds = 4; // Seconds after midnight
-    int32 departure_offset_seconds = 5;
-    int32 pass_offset_seconds = 6;
-    int32 public_arrival_offset_seconds = 7;
-    int32 public_departure_offset_seconds = 8;
-    string platform = 9;
-    string line = 10;
-    string path = 11;
-    int32 engineering_allowance_seconds = 12;
-    int32 performance_allowance_seconds = 13;
-}
-
-message SchedulePayload {
-    string train_uid = 1;
-    SchedulePayloadType transaction_type = 2;
-    int32 schedule_start_date = 3; // Literal int of date: eg. 20251216
-    int32 schedule_end_date = 4;
-    RunsOnDays runs_on_day_of_week = 5;
-    TrainStatus train_status = 6;
-    TrainCategory train_category = 7;
-    string headcode = 8; // signalling_id
-    string nrs_code = 9; // CIF_headcode
-    string train_service_code = 10;
-    string portion_id = 11; // CIF_business_sector
-    PowerType power_type = 12;
-    string timing_load = 13; // Pretty String in Ingress Service
-    OperatingCharacteristics operating_characteristics = 14;
-    bool first_class_available = 15;
-    SleepingAccommodation sleeping_accommodation = 16;
-    Reservations reservations = 17;
-    Catering catering_code = 18;
-    bool is_eurostar = 19;
-    STPIndicator stp_indicator = 20;
-    string uic_code = 21;
-    string toc_code = 22; // Maybe ENUM but subject to frequent-ish change
-    repeated ScheduleLocation schedule_location = 23;
-}

schemas/data-ingress/darwin-data/darwin-association.json

@@ -0,0 +1,118 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/association.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinAssocData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "darwin-association",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "additionalProperties": false,
+                "required": [
+                    "mr",
+                    "ar",
+                    "t",
+                    "c",
+                    "s",
+                    "mt",
+                    "at"
+                ],
+                "properties": {
+                    "mr": {
+                        "type": "string",
+                        "name": "Main RID",
+                        "description": "The main RID of this association object"
+                    },
+                    "ar": {
+                        "type": "string",
+                        "name": "Associated RID",
+                        "description": "The associated RID of this association object"
+                    },
+                    "t": {
+                        "type": "string",
+                        "name": "TIPLOC",
+                        "description": "The location at which this association applies"
+                    },
+                    "c": {
+                        "enum": ["JJ", "VV", "LK", "NP"],
+                        "name": "Category",
+                        "description": "The type of association"
+                    },
+                    "s": {
+                        "type": "string",
+                        "format": "date",
+                        "name": "SSD",
+                        "description": "SSD Generated from the Main RID"
+                    },
+                    "mt": {
+                        "name": "Main Times",
+                        "description": "Timings that apply to the main RID",
+                        "type": "object",
+                        "properties": {
+                            "wta": { "type": "string" },
+                            "wtd": { "type": "string" },
+                            "wtp": { "type": "string" },
+                            "pta": { "type": "string" },
+                            "ptd": { "type": "string" }
+                        }
+                    },
+                    "at": {
+                        "name": "Associated Times",
+                        "description": "Timings that apply to the associated RID",
+                        "type": "object",
+                        "properties": {
+                            "wta": { "type": "string" },
+                            "wtd": { "type": "string" },
+                            "wtp": { "type": "string" },
+                            "pta": { "type": "string" },
+                            "ptd": { "type": "string" }
+                        }
+                    },
+                    "can": {
+                        "name": "Cancelled",
+                        "type": "boolean",
+                        "description": "Whether this association is cancelled"
+                    },
+                    "del": {
+                        "name": "Deleted",
+                        "type": "boolean",
+                        "description": "Whether this association is deleted"
+                    }
+                }
+            }
+        }
+    },
+    "required": [
+        "service_name",
+        "service_id",
+        "data_type",
+        "data_kind",
+        "sent_timestamp",
+        "payload"
+    ],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-data/darwin-schedule.json

@@ -0,0 +1,209 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/schedule.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinScheduleData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "darwin-schedule",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "additionalProperties": false,
+                "required": [
+                    "rid",
+                    "uid",
+                    "ssd",
+                    "tid",
+                    "toc",
+                    "isp",
+                    "isc",
+                    "l"
+                ],
+                "properties": {
+                    "rid": {
+                        "type": "string",
+                        "description": "Service RID",
+                        "maxLength": 16
+                    },
+                    "uid": {
+                        "type": "string",
+                        "description": "Service UID",
+                        "maxLength": 10
+                    },
+                    "ssd": {
+                        "type": "string",
+                        "format": "date",
+                        "description": "Date on which the service commences"
+                    },
+                    "tid": {
+                        "name": "train_id",
+                        "type": "string",
+                        "minLength": 4,
+                        "maxLength": 4,
+                        "description": "The signalling headcode of the service"
+                    },
+                    "toc": {
+                        "type": "string",
+                        "minLength": 2,
+                        "maxLength": 2,
+                        "description": "The two character code for the operating TOC"
+                    },
+                    "sts": {
+                        "type": "string",
+                        "maxLength": 1,
+                        "name": "status",
+                        "description": "The 'status' of the service"
+                    },
+                    "qt": {
+                        "name": "qtrain",
+                        "description": "qtrain?",
+                        "type": "boolean",
+                        "default": false
+                    },
+                    "d": {
+                        "name": "deleted",
+                        "description": "Whether the schedule is deleted",
+                        "type": "boolean",
+                        "default": false
+                    },
+                    "isp": {
+                        "name": "is_passenger",
+                        "type": "boolean",
+                        "description": "Whether the service is a passenger service"
+                    },
+                    "isc": {
+                        "name": "is_charter",
+                        "type": "boolean",
+                        "description": "Whether the service is a charter service"
+                    },
+                    "sid": {
+                        "name": "rsid",
+                        "type": ["string", "null"],
+                        "maxLength": 10
+                    },
+                    "q": {
+                        "name": "qtrain",
+                        "description": "True if the service is scheduled but not activated (if not activated, presumed not running)",
+                        "type": "boolean",
+                        "default": false
+                    },
+                    "cr": {
+                        "name": "cancel_reason",
+                        "type": ["integer", "null"],
+                        "description": "The reason code for the services cancellation"
+                    },
+                    "cl": {
+                        "name": "cancel_location",
+                        "type": "string",
+                        "description": "The location at which the service was cancelled"
+                    },
+                    "l": {
+                        "type": "array",
+                        "minItems": 2,
+                        "items": {
+                            "type": "object",
+                            "required": ["tpl", "lr", "s"],
+                            "properties": {
+                                "tpl": {
+                                    "type": "string",
+                                    "description": "The locations TIPLOC",
+                                    "maxLength": 7,
+                                    "minLength": 4
+                                },
+                                "lr": {
+                                    "name": "loc_role",
+                                    "description": "The role of the location (Orig, Call, Pass, Dest)",
+                                    "enum": ["ORIG", "CALL", "PASS", "DEST"]
+                                },
+                                "s": {
+                                    "name": "seq",
+                                    "description": "The sequence in the service at which this location occours",
+                                    "type": "integer",
+                                    "minimum": 1,
+                                    "maximum": 32767
+                                },
+                                "pta": {
+                                    "description": "Public timetable arrival",
+                                    "type": "string",
+                                    "format": "date-time"
+                                },
+                                "ptd": {
+                                    "description": "Public timetable departure",
+                                    "type": "string",
+                                    "format": "date-time"
+                                },
+                                "wta": {
+                                    "description": "Working timetable arrival",
+                                    "type": "string",
+                                    "format": "date-time"
+                                },
+                                "wtd": {
+                                    "description": "Working timetable arrival",
+                                    "type": "string",
+                                    "format": "date-time"
+                                },
+                                "wtp": {
+                                    "description": "Working timetable pass",
+                                    "type": "string",
+                                    "format": "date-time"
+                                },
+                                "c": {
+                                    "name": "can",
+                                    "description": "Whether the service is cancelled at this location",
+                                    "default": false,
+                                    "type": "boolean"
+                                },
+                                "a": {
+                                    "name": "activities",
+                                    "type": "string",
+                                    "maxLength": 12
+                                },
+                                "fid": {
+                                    "description": "Unique formation indentifier (Maps to formation feed)",
+                                    "type": "string"
+                                },
+                                "al": {
+                                    "name": "avg_loading",
+                                    "description": "Average loading at this location",
+                                    "type": "integer"
+                                }
+                            }
+                        }
+                    }
+                }
+            }
+        }
+    },
+    "required": [
+        "service_name",
+        "service_id",
+        "data_type",
+        "data_kind",
+        "sent_timestamp",
+        "payload"
+    ],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/darwin-stations.json

@@ -0,0 +1,50 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/stations.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinStationsNames",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-station-names",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["c", "n"],
+                "properties": {
+                    "c": {
+                        "type": "string",
+                        "description": "CRS of the location"
+                    },
+                    "n": {
+                        "type": "string",
+                        "description": "The display name of this location"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/darwin-tiplocs.json

@@ -0,0 +1,58 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/tiplocs.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinTiplocData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-tiplocs",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["t", "n"],
+                "properties": {
+                    "c": {
+                        "type": "string",
+                        "description": "CRS of the location"
+                    },
+                    "t": {
+                        "type": "string",
+                        "description": "TIPLOC assigned to the location"
+                    },
+                    "n": {
+                        "type": "string",
+                        "description": "The display name of this location"
+                    },
+                    "l": {
+                        "type": "string",
+                        "description": "The NLC of this location"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/darwin-via-text.json

@@ -0,0 +1,67 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/via-text.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinViaText",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-via-text",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["a", "d", "l1", "v"],
+                "properties": {
+                    "a": {
+                        "name": "at",
+                        "type": "string",
+                        "description": "The CRS defining where the text should be displayed"
+                    },
+                    "d": {
+                        "name": "destination",
+                        "type": "string",
+                        "description": "The text should apply to trains to this destination"
+                    },
+                    "l1": {
+                        "name": "location-one",
+                        "type": "string",
+                        "description": "Message triggered if loc1 & loc2 (if present) are listed as stops"
+                    },
+                    "l2": {
+                        "name": "location-two",
+                        "type": "string",
+                        "description": "Message triggered if loc1 & loc2 (if present) are listed as stops"
+                    },
+                    "v": {
+                        "name": "viaText",
+                        "description": "The text that should be displayed",
+                        "type": "string"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/loading-category.json

@@ -0,0 +1,77 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/loading-category.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinLoadingCatData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-loading-category",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["c", "n"],
+                "properties": {
+                    "c": {
+                        "name": "code",
+                        "type": "string",
+                        "description": "The 'code' for the 'category'"
+                    },
+                    "t": {
+                        "name": "typicalDescription",
+                        "type": "string",
+                        "description": "The 'typicalDescription' of the category"
+                    },
+                    "e": {
+                        "name": "expectedDescription",
+                        "type": "string",
+                        "description": "The 'expectedDescription' of the category"
+                    },
+                    "d": {
+                        "name": "definition",
+                        "description": "The 'definition' of the category",
+                        "type": "string"
+                    },
+                    "n": {
+                        "name": "name",
+                        "description": "The 'name' of the category",
+                        "type": "string"
+                    },
+                    "i": {
+                        "name": "image",
+                        "description": "The 'image' for the category",
+                        "type": "string"
+                    },
+                    "o": {
+                        "name": "TOC",
+                        "description": "The 'toc' for the category",
+                        "type": "string"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/reason-code.json

@@ -0,0 +1,57 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/reasons.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinReasonCodeData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-reason-code",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["c"],
+                "properties": {
+                    "c": {
+                        "type": "integer",
+                        "name": "code",
+                        "description": "The 'code' that the late/cancel reason maps to"
+                    },
+                    "lr": {
+                        "type": "string",
+                        "name": "Late Reason",
+                        "description": "The lateReason string for the item"
+                    },
+                    "cr": {
+                        "type": "string",
+                        "name": "Cancellation Reason",
+                        "description": "The cancelReason string for the item"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/darwin-reference-data/toc.json

@@ -0,0 +1,57 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/darwin/ref/tocs.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "DarwinTocData",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "ref-darwin-toc",
+            "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": ["c", "n"],
+                "properties": {
+                    "c": {
+                        "name": "ATOC Code",
+                        "type": "string",
+                        "description": "The ATOC 'code' for the TOC"
+                    },
+                    "n": {
+                        "name": "TOC Name",
+                        "type": "string",
+                        "description": "The name of the TOC"
+                    },
+                    "u": {
+                        "name": "URL",
+                        "type": "string",
+                        "description": "URL for TOC Information"
+                    }
+                }
+            }
+        }
+    },
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
+    "additionalProperties": false
+}

schemas/data-ingress/knowledgebase/kb-stations.json

@@ -0,0 +1,119 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/knowledgebase/stations.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "KnowledgebaseStations",
+    "type": "object",
+    "properties": {
+        "service_name": {
+            "type": "string",
+            "description": "Name of the service submitting the update"
+        },
+        "service_id": {
+            "type": "string",
+            "description": "ID of the service sending the message - usually a hostname"
+        },
+        "data_type": {
+            "type": "string",
+            "const": "raw",
+            "description": "The type of data contained in this message object"
+        },
+        "sent_timestamp": {
+            "type": "integer",
+            "description": "Unix timestamp representing the time the message was sent"
+        },
+        "data_kind": {
+            "type": "string",
+            "const": "kb-stations"
+        },
+        "payload": {
+            "type": "array",
+            "additionalItems": false,
+            "items": {
+                "type": "object",
+                "required": [
+                    "c",
+                    "mc",
+                    "l",
+                    "o",
+                    "p",
+                    "a",
+                    "r",
+                    "d"
+                ],
+                "properties": {
+                    "c": {
+                        "type": "string",
+                        "name": "CRS Code",
+                        "description": "Station CRS (Primary Key)"
+                    },
+                    "n": {
+                        "type": "string",
+                        "name": "NLC Code",
+                        "description": "Station NLC Code"
+                    },
+                    "mc": {
+                        "type": "number",
+                        "name": "Minimum Connection Time",
+                        "description": "The minimum connection time at this station"
+                    },
+                    "a": {
+                        "type": "string",
+                        "name": "Address",
+                        "description": "The address of the station"
+                    },
+                    "p": {
+                        "type": "string",
+                        "name": "Postcode",
+                        "description": "The public postcode of the station"
+                    },
+                    "l": {
+                        "type": "object",
+                        "name": "Location",
+                        "description": "LatLon of the station",
+                        "required": ["lt", "ln"],
+                        "properties": {
+                            "lt": {
+                                "type": "number",
+                                "name": "Latitude",
+                                "description": "Stations Latitude"
+                            },
+                            "ln": {
+                                "type": "number",
+                                "name": "Longitude",
+                                "description": "Stations Longitude"
+                            }
+                        }
+                    },
+                    "o": {
+                        "type": "string",
+                        "name": "Operator",
+                        "description": "The two character code denoting the stations operator"
+                    },
+                    "m": {
+                        "type": "string",
+                        "name": "Map URL"
+                    },
+                    "r": {
+                        "name": "Request",
+                        "type": "boolean",
+                        "description": "Denotes whether station is Request Stop"
+                    },
+                    "d": {
+                        "name": "Defibrillator",
+                        "type": "boolean",
+                        "description": "Denotes whether defibrillator is available"
+                    }
+                }
+            }
+        }
+    },
+    "required": [
+        "service_name",
+        "service_id",
+        "data_type",
+        "data_kind",
+        "sent_timestamp",
+        "payload"
+    ],
+    "additionalProperties": false
+}

schemas/data-ingress/message-envelope.json

@@ -14,7 +14,7 @@
         },
         "data_type": {
             "type": "string",
-            "enum": ["file"],
+            "enum": ["file", "raw"],
             "description": "The type of data contained in this message object"
         },
         "sent_timestamp": {
@@ -23,10 +23,11 @@
         },
         "data_kind": {
             "type": "string",
-            "enum": ["pis", "timetable", "knowledgebase", "darwin_schedule", "darwin_ts", "darwin_ref"],
+            "enum": ["pis", "timetable", "kb-stations", "PushPort16Schedule", "darwin_ts", "darwin_ref"],
             "description": "The data type contained in the message.  Currently supported PIS: PIS Data, Timetable: CIF or VSTP Data, Knowledgebase: Station Data"
-        }
+        },
+        "payload": {}
     },
-    "required": ["service_name", "service_id", "data_type", "sent_timestamp"],
+    "required": ["service_name", "service_id", "data_type", "data_kind", "sent_timestamp"],
-    "additionalProperties": true
+    "additionalProperties": false
 }

schemas/data-ingress/network-rail/networkRailCorpus.json

@@ -0,0 +1,31 @@
+{
+    "$id": "https://schema.owlboard.info/data-ingress/network-rail/networkRailCorpus.schema.json",
+    "$schema": "https://json-schema.org/draft-07/schema#",
+    "title": "networkRailCorpus",
+    "type": "array",
+    "additionalItems": false,
+    "items": {
+        "type": "object",
+        "required": ["t", "n"],
+        "additionalProperties": false,
+        "properties": {
+            "t": {
+                "type": "string",
+                "name": "TIPLOC",
+                "description": "The TIPLOC Code of the location"
+            },
+            "n": {
+                "type": "string",
+                "name": "Name",
+                "description": "The human readable name of the location"
+            },
+            "nc": {
+                "type": "integer",
+                "name": "NLC",
+                "description": "The national location code of the location",
+                "minimum": 0,
+                "maximum": 999999
+            }
+        }
+    }
+}

schemas/data-ingress/pis-data.json

@@ -25,13 +25,6 @@
             },
             "description": "List of 3ALPHA/CRS Codes"
         },
-        "crsHash": {
-            "type": "string",
-            "minLength": 1,
-            "maxLength": 64,
-            "pattern": "^[0-9]+$",
-            "description": "Stringified 64-bit hash"
-        },
         "tiplocStops": {
             "type": "array",
             "items": {
@@ -41,14 +34,8 @@
                 "pattern": "^[a-zA-Z0-9]+$"
             },
             "description": "List of TIPLOC Codes"
-        },
-        "tiplocHash": {
-            "type": "string",
-            "minLength": 1,
-            "maxLength": 64,
-            "pattern": "^[0-9]+$"
         }
     },
-    "required": ["code", "toc", "crsStops", "crsHash", "tiplocStops", "tiplocHash"],
+    "required": ["code", "toc", "crsStops", "tiplocStops"],
     "additionalProperties": false
 }

schemas/data-ingress/pushport/schedule.json

@@ -1,58 +0,0 @@
-{
-    "$id": "https://schema.owlboard.info/data-ingress/pushport-schedule.schema.json",
-    "$schema": "https://json-schema.org/draft-07/schema#",
-    "title": "PushPortSchedule",
-    "type": "object",
-    "required": ["rid", "uid", "trainId", "ssd", "toc", "locations"],
-    "properties": {
-        "rid": {"type": "string", "description": "Unique service ID"},
-        "uid": {"type": "string", "description": "Unique schedule ID - maps to NR CIF"},
-        "trainId": {"type": "string", "description": "Train Headcode"},
-        "ssd": {"type": "string", "format": "date", "description": "Schedule Start Date (When the service commences)"},
-        "toc": {"type": "string", "minLength": 2, "maxLength": 2, "description": "TOC Code"},
-        "isPassenger": {"type": "boolean", "description": "If service is for passengers", "default": true},
-        "isCharter": {"type": "boolean", "description": "If train is charter service", "default": false},
-        "status": {"type": "string", "enum": ["train", "bus", "ship"], "default": "train"},
-        "isDelayedDelivery": {"type": "boolean", "description": "If message is sent as part of Darwin service recovery"},
-        "cancelReason": {"type": ["string", "null"], "description": "Reason code for cancellation"},
-        "isDeleted": {"type": "boolean", "description": "Whether schedule should be deleted", "default": false},
-        "locations": {
-            "type": "array",
-            "items": {
-                "type": "object",
-                "required": ["tpl", "seq"],
-                "properties": {
-                    "tpl": {"type": "string", "description": "TIPLOC of location"},
-                    "seq": {"type": "integer", "description": "The order of the stop for sorting and handling repeated TIPLOCs"},
-                    "wta": { "type": ["string", "null"] },
-                    "wtd": { "type": ["string", "null"] },
-                      "wtp": { "type": ["string", "null"] },
-                      "pta": { "type": ["string", "null"] },
-                      "ptd": { "type": ["string", "null"] },
-                      "plat": { "type": ["string", "null"] },
-                      "isCancelled": { "type": "boolean", "default": false },
-                      "isSupressed": { "type": "boolean", "default": false, "description": "Defines whether the service should be hidden to the public"},
-                      "activities": {
-                        "type": "object",
-                        "properties": {
-                            "pass": {"type": "boolean", "description": "Indicates that the train does not stop, it only passes this location"},
-                            "pickUpDropOff": {"type": "boolean", "description": "Indicates the train stops for passengers to board and alight"},
-                            "requestStop": {"type": "boolean", "description": "Indicates if the train stops only when required"},
-                            "pickUpOnly": {"type": "boolean", "description": "Indicated the service picks up passengers only"},
-                            "dropOffOnly": {"type": "boolean", "description": "Indicates the service drops off passengers only"},
-                            "tokenExchange": {"type": "boolean", "description": "Indicates if the train stops for exhanging a token/staff or similar"},
-                            "vehiclesAttached": {"type": "boolean", "description": "Indicates if vehicles are attached at this location"},
-                            "vehiclesDetached": {"type": "boolean", "description": "Indicates if vehicles are detached at this location"},
-                            "trainCrewChange": {"type": "boolean", "description": "Indicates if train stops to exchange crew at this location"},
-                            "changeLocomotive": {"type": "boolean", "description": "Indicates if train stops to exchange locomotive at this location"},
-                            "changeEnds": {"type": "boolean", "description": "Indicates if the train reverses or driver changes ends at this location"},
-                            "staffStop": {"type": "boolean", "description": "Indicates that the train stops for staff use only"},
-                            "unadvertised": {"type": "boolean", "description": "Indicates that the stop is not advertised to the public"},
-                            "passesAnotherTrain": {"type": "boolean", "description": "Indicates if the train stops to pass another train on a single line crossing point"}
-                        }
-                      }
-                }
-            }
-        }
-    }
-}

schemas/data-ingress/pushport/ts.json.txt

@@ -1,39 +0,0 @@
-{
-    "$id": "https://schema.owlboard.info/data-ingress/pushport-ts.schema.json",
-    "$schema": "https://json-schema.org/draft-07/schema#",
-    "title": "PushPortTS",
-    "type": "object",
-    "properties": {
-        "rid": {
-            "type": "string",
-            "description": "Unique ID for the train service (eg. 2026022217610103)",
-            "pattern": "^[0-9]{15}$"
-        },
-        "uid": {
-            "type": "string",
-            "description": "Schedule ID - The permanent ID for the schedule in the static timetable"
-        },
-        "ssd": {
-            "type": "string",
-            "format": "date",
-            "description": "Scheduled Start Data, the date the train service started its journey"
-        },
-        "isDelayed": {
-            "type": "boolean",
-            "description": "Indicated if the message is delayed due to PushPort service disruption"
-        },
-        "lateReason": {
-            "type": "integer",
-            "description": "Reason the train is late (Maps to textual reason)"
-        },
-        "cancelReason": {
-            "type": "integer",
-            "description": "Reason the train is cancelled (Maps to textual reason)"
-        },
-        "locations": {
-            "type": "array",
-            "description": "A list of one or more locations where the train has recorded a movement or forecast",
-            "items": { "$ref": "https://schema.owlboard.info/data-ingress/pushport-ts-location.schema.json" }
-        }
-    }
-}

schemas/data-ingress/pushport/ts_location.json.txt

@@ -1,51 +0,0 @@
-{
-    "$id": "https://schema.owlboard.info/data-ingress/pushport-ts-location.schema.json",
-    "$schema": "https://json-schema.org/draft-07/schema#",
-    "title": "PushPortTSLocation",
-    "type": "object",
-    "required": ["tpl"],
-    "definitions": {
-        "TSTime": {
-            "type": "object",
-            "properties": {
-                "at": {"type": "string", "description": "Actual time"},
-                "et": {"type": "string", "description": "Estimated time"},
-                "wet": {"type": "string", "description": "Working estimated time"},
-                "src": {"type": "string", "description": "Data source (TD/Darwin/etc)"},
-                "atClass": {"type": "string", "description": "Time source Automatic/Manual"},
-                "delayed": {"type": "boolean", "description": "is train delayed with no estimated time"},
-                "srcInst": {"type": "string", "description": "Source instance (what caused the update)"},
-                "atRemoved": {"type": "boolean", "description": "Whether a previous 'at' was reported in error and should be removed"}
-            }
-        }
-    },
-    "properties": {
-        "tpl": {"type": "string", "description": "TIPLOC for the location"},
-        "wta": {"type": "string", "description": "Scheduled working time arrival"},
-        "wtd": {"type": "string", "description": "Scheduled working time departure"},
-        "wtp": {"type": "string", "description": "Scheduled working time pass"},
-        "pta": {"type": "string", "description": "Scheduled public arrival time"},
-        "ptd": {"type": "string", "description": "Scheduled public departure time"},
-        "arr": {"$ref": "#/definitions/TSTime"},
-        "dep": {"$ref": "#/definitions/TSTime"},
-        "pass": {"$ref": "#/definitions/TSTime"},
-        "plat": {
-            "type": "object",
-            "description": "Platform data",
-            "properties": {
-                "value": {"type": "string", "description": "Platform number/letter"},
-                "platsrc": {"type": "string", "enum": ["A", "M", "P"], "description": "Source of data Automatic/Manual/Planned"},
-                "platsup": {"type": "boolean", "description": "Is platform suppressed from public display"},
-                "cisPlatsup": {"type": "boolean", "description": "Whether CIS or Darwin Workstation has supressed the platform"},
-                "conf": {"type": "boolean", "description": "Whether platform has been confirmed by the local CIS"}
-            }
-        },
-        "suppr": {"type": "boolean", "description": "Whether location is supressed"},
-        "length": {"type": "integer", "description": "Train length (no. of coaches)"},
-        "detachFront": {"type": "boolean","description": "Indicated if coaches will detatch from front"},
-        "detachRead": {"type": "boolean", "description": "Indicates if coaches will detatch from rear"},
-        "attachFront": {"type": "boolean", "description": "Indicates if coaches will attach to front"},
-        "attachRear": {"type": "boolean", "description": "Indicates if coaches will be attached to rear"},
-        "adhoc": {"type": "boolean", "description": "Whether this is an ad-hoc stop"}
-    }
-}