syntax = "proto3";

package pb.lyft.datacatalog;

option go_package = "datacatalog";
option py_generic_services = true;

message Parameter {

    string name = 1;
    string value = 2;
}

// Before jumping to message definition, lets go over the expected flow-
//   An Artifact represents an unit-of-work identified by (task, version, inputs). This is
//   encoded as unique hash for faster queries(called provenance). An artifact is persisted with some other
//   attributes (revision, createdAt, reference_id, outputs).
//   Only Discovery service knows about the hashing algorithm; one can use the closure (task, version, inputs)
//   to query an artifact if it doesnt have the provenance value.
//
//   Before starting the work on a task, programming-model first checks if the task has been done.
//     Request:   GET (task, version, inputs)
//     Response:  (Exists, Artifact) or (NotFound, nil)
//   if not found, Task executor goes ahead with the execution and at the end of execution creates a new entry in
//   the discovery service
//     Request:  CREATE (task, version, inputs) + (revision, reference_id, outputs)
//     Response: (Exists, Artifact) or (Created, Artifact)
//
//   One can also Query all the artifacts by querying any subset of properties.
// Message Artifact represents the complete information of an artifact- field that unique define the artifact +
// properties.
// Message ArtifactInternal is our storage model where we create an additional derived column for faster queries.
// Message ArtifactId only represents field that uniquely define the artifact.
message Artifact {
    string provenance = 1;
    string name = 2;
    string version = 3;
    int64 revision = 4;       // strictly increasing value set by users. users can do range query on this attribute.
    int64 created_at = 5;     // the time when discovery service received the request.
    string reference_id = 6;  // this could be a workflow runid or something that ties it data elsewhere
    repeated Parameter inputs = 7;
    repeated Parameter outputs = 8;
}

message ArtifactId {
    string name = 1;
    string version = 2;
    repeated Parameter inputs = 3;
}

message GetRequest {
    oneof id {
        string provenance = 1;
        ArtifactId artifact_id = 2;
    }
}

message GetResponse {
    Artifact artifact = 1;
}

enum QueryOperator {
    EQUAL = 0;
    GREATER_THAN = 1;
    LESSER_THAN = 2;
}

message IntFilter {
    int64 value = 1;
    QueryOperator operator = 2;
}

message IntRangeFilter {
    int64 min = 1;
    int64 max = 2;
}

message IntQueryKey {
    oneof filter {
        IntFilter val = 1;
        IntRangeFilter range = 2;
    }
}

// QueryRequest allows queries on a range of values for revision column and point queries on created_at
// and reference_id
message QueryRequest {
    string name = 1;
    string version = 2;
    IntQueryKey revision = 3;
    int64 created_at = 4;
    string reference_id = 5;
}

message QueryResponse {
    repeated Artifact artifact = 1;
}

message CreateRequest {
    ArtifactId ref = 1;
    string reference_id = 2;
    int64 revision = 3;
    repeated Parameter outputs = 4;
}

message CreateResponse {
    Artifact artifact = 1;
    enum Status {
        ALREADY_EXISTS = 0;
        CREATED        = 1;
    }
    Status status = 2;
}

message GenerateProvenanceRequest {
    ArtifactId id = 1;
}

message GenerateProvenanceResponse {
    string provenance = 1;
}

service Artifacts {
    rpc Get (GetRequest) returns (GetResponse) {}
    rpc Query (QueryRequest) returns (QueryResponse) {}
    rpc Create (CreateRequest) returns (CreateResponse) {}
    rpc GenerateProvenance (GenerateProvenanceRequest) returns (GenerateProvenanceResponse) {}
}