datahub/metadata-integration/java/datahub-protobuf-example/schema/protobuf/meta/meta.proto

syntax = "proto3";

package meta;

import "google/protobuf/descriptor.proto";

/*
   This is assigned to metadata fields. It describes how the metadata field should be represented
   in DataHub. This enum must be used in the `meta` package. Multiple can be used for the same
   metadata annotation. This allows a single piece of information to be captured in DataHub
   as a property, tag and/or term.

   Tags can be strings, enums, or booleans
   Terms can be strings or enums
   Properties should be strings

*/
enum DataHubMetadataType {
    PROPERTY = 0; // Datahub Custom Property
    TAG      = 1; // Datahub Tag
    TAG_LIST = 2; //    comma delimited string
    TERM     = 3; // Datahub Term
    OWNER    = 4; // Datahub Owner
    DOMAIN   = 5; // Datahub Domain
}

// Assuming Glossary Term defined from bootstrap example
enum Classification {
    HighlyConfidential = 0;
    Confidential = 1;
    Sensitive = 2;
}

message datahubField {
      extend google.protobuf.FieldOptions {
          // Required: Mark option field with how to export to DataHub in one or more places.
          repeated DataHubMetadataType type = 5000;

          // Set true if the field is a primary key. This works for any boolean with `primary_key` in it.
          bool is_primary_key = 5010;
      }
}

message securityField {
    extend google.protobuf.FieldOptions {

        // Extract classification field option as a Term, either works
        string classification = 5100 [(datahubField.type) = TERM];
        Classification classification_enum = 5101 [(datahubField.type) = TERM];
    }
}

message field {
    extend google.protobuf.FieldOptions {

        string tags = 5150 [(datahubField.type) = TAG_LIST];
    }
}

message ownership {
    extend google.protobuf.MessageOptions {

        repeated string team = 5200 [(datahubField.type) = OWNER, (datahubField.type) = PROPERTY];
        string data_steward = 5201 [(datahubField.type) = OWNER];
        string domain = 5202 [(datahubField.type) = DOMAIN, (datahubField.type) = PROPERTY];
    }
}

message security {
    extend google.protobuf.MessageOptions {

        // Place the classification term at the Message/Dataset level, either string or enum is supported
        string classification = 5300 [(datahubField.type) = TERM, (datahubField.type) = PROPERTY];
        Classification classification_enum = 5301 [(datahubField.type) = TERM, (datahubField.type) = PROPERTY];
    }
}

message kafka {
    extend google.protobuf.MessageOptions {

        repeated string topics = 5400 [(datahubField.type) = PROPERTY];
    }
}

enum Frequency {
    REALTIME = 0;
    DAILY = 1;
    WEEKLY = 2;
    MONTHLY = 3;
    YEARLY = 4;
}

message lifecycle {
    extend google.protobuf.MessageOptions {

        bool archived = 5500 [(datahubField.type) = TAG, (datahubField.type) = PROPERTY];
        Frequency frequency = 5510 [(datahubField.type) = TAG, (datahubField.type) = PROPERTY];
        string ttl = 5520 [(datahubField.type) = TAG];
    }
}

enum MessageType {
    ENTITY = 0;
    EVENT = 1;
    IMPRESSION = 2;
}

message message {
    extend google.protobuf.MessageOptions {

        string tags = 5600 [(datahubField.type) = TAG_LIST];
        MessageType type = 5610 [(datahubField.type) = TAG, (datahubField.type) = PROPERTY];
    }
}

message props {
    extend google.protobuf.MessageOptions {

        string prop1      = 5701 [(datahubField.type) = PROPERTY];
        bool prop2        = 5702 [(datahubField.type) = PROPERTY];
        MessageType prop3 = 5703 [(datahubField.type) = PROPERTY];

        repeated string prop4      = 5704 [(datahubField.type) = PROPERTY];
        repeated MessageType prop6 = 5706 [(datahubField.type) = PROPERTY];
    }
}

message tags {
    extend google.protobuf.MessageOptions {

        string tag_str      = 5801 [(datahubField.type) = TAG];
        bool tag_bool        = 5802 [(datahubField.type) = TAG];
        MessageType tag_enum = 5803 [(datahubField.type) = TAG];

        string tag_list      = 5804 [(datahubField.type) = TAG_LIST];
    }
}

message fieldTags { extend google.protobuf.FieldOptions {
    string tag_str                      = 5801 [(datahubField.type) = TAG];
    bool tag_bool                     = 5802 [(datahubField.type) = TAG];
    MessageType tag_enum = 5803 [(datahubField.type) = TAG];

    string tag_list                      = 5804 [(datahubField.type) = TAG_LIST];
}}