syntax = "proto3";

import "google/protobuf/timestamp.proto";
import "google/protobuf/struct.proto";

package databuild.v1;

message PartitionRef {
  string str = 1;
}

///////////////////////////////////////////////////////////////////////////////////////////////
// Jobs
///////////////////////////////////////////////////////////////////////////////////////////////

//
// Job Config
//

// The type of dependency
enum DepType {
  QUERY = 0; // Default
  MATERIALIZE = 1;
}

// Represents a data dependency
message DataDep {
  DepType dep_type = 1;
  PartitionRef partition_ref = 2;
}

// Configuration for a job
message JobConfig {
  // The partitions that this parameterization produces
  repeated PartitionRef outputs = 1;

  // Required data dependencies
  repeated DataDep inputs = 2;

  // Command line arguments
  repeated string args = 3;

  // Environment variables
  map<string, string> env = 4;
}

// Request message for job configuration service
message JobConfigureRequest { repeated PartitionRef outputs = 1; }

// Response message for job configuration service
message JobConfigureResponse { repeated JobConfig configs = 1; }

// Implemented by the job.cfg bazel rule
service JobConfigure {
  rpc Configure(JobConfigureRequest) returns (JobConfigureResponse);
}

//
// Job Exec
//

// Manifest that records the literal partitions consumed (and their manifests) in order to
// produce the specified partitions
message PartitionManifest {
  // The refs of the partitions produced by this job
  repeated PartitionRef outputs = 1;

  // Input partition manifests
  repeated PartitionManifest inputs = 2;

  // Start time of job execution (Unix timestamp)
  google.protobuf.Timestamp start_time = 3;

  // End time of job execution (Unix timestamp)
  google.protobuf.Timestamp end_time = 4;

  // The configuration used to run the job
  Task task = 5;
}

message JobExecuteRequest { repeated PartitionRef outputs = 1; }

// Metadata for the complete set of partitions produced by this job
message JobExecuteResponse { repeated PartitionManifest manifests = 1; }

// Implemented by the job.exec bazel rule
service JobExecute {
  rpc Execute(JobExecuteRequest) returns (JobExecuteResponse);
}

///////////////////////////////////////////////////////////////////////////////////////////////
// Graphs
///////////////////////////////////////////////////////////////////////////////////////////////

//
// GraphLookup
//

message JobLabel {
  // The bazel label the references the job_target
  string label = 1;
}

message GraphLookupRequest { repeated PartitionRef outputs = 1; }

// Represents a not-yet configured task
message TaskRef {
  // The job whose configure/exec targets will be used
  JobLabel job = 1;

  // The partition refs this task is responsible for producing, and with which the configure
  // target will be invoked
  repeated PartitionRef outputs = 2;
}

// Represents the complete set of tasks needed to produce the requested partitions
message GraphLookupResponse { repeated TaskRef task_refs = 1; }

// Implemented per graph
service GraphLookup {
  rpc Lookup(GraphLookupRequest) returns (GraphLookupResponse);
}

// Request message for graph analyze service
message GraphAnalyzeRequest { repeated PartitionRef outputs = 1; }

//
// JobGraph
//

message Task {
  // The bazel label uniquely identifying the job
  JobLabel job = 1;

  // The configuration for the job
  JobConfig config = 2;
}

// The bazel label referencing the graph
message GraphLabel { string label = 1; }

// Represents a job graph
message JobGraph {
  // The bazel label of the graph to be executed
  GraphLabel label = 1;

  // The output partitions to be produced by this graph
  repeated PartitionRef outputs = 2;

  // The job configurations that make up this graph
  repeated Task nodes = 3;
}

// Response message for graph analyze service
message GraphAnalyzeResponse { JobGraph graph = 1; }

message GraphExecuteResponse { repeated PartitionManifest manifests = 1; }
message GraphBuildRequest { repeated PartitionRef outputs = 1; }
message GraphBuildResponse { repeated PartitionManifest manifests = 1; }

// Service for job configuration and graph analysis
service DataBuildService {
//  // Get job configurations for the specified output references
//  rpc GetJobConfigs(JobConfigureRequest) returns (JobConfigureResponse) {}

  // Analyze and get the job graph for the specified output references
  rpc AnalyzeGraph(GraphAnalyzeRequest) returns (GraphAnalyzeResponse);

  // Execute the specified job graph (implemented by databuild)
  rpc Execute(JobGraph) returns (GraphExecuteResponse);

  // User-facing: build the desired partitions
  rpc Build(GraphBuildRequest) returns (GraphBuildResponse);
}

///////////////////////////////////////////////////////////////////////////////////////////////
// Catalog
///////////////////////////////////////////////////////////////////////////////////////////////

//
// Job Run Events
//

// The complete lifecycle that each job goes through
enum JobRunStatus {
  // Initial state
  NOT_SCHEDULED = 0;
  // Job preconditions met, job scheduled
  JOB_SCHEDULED = 1;
  // Job execution has begun
  JOB_RUNNING = 2;
  // Job execution finished successfully
  JOB_SUCCESS = 3;
  // Job execution failed
  JOB_FAILED = 4;
}

message JobConfigEvent { JobConfig config = 1; }
message JobLogsEvent { repeated string logs = 1; }

//
message JobEvent {
  // IDs
  // Unique ID for this job graph run
  string job_graph_run_id = 1;
  // Unique ID for this job run
  string job_run_id = 2;
  // Unique ID for this event
  string job_graph_run_event_id = 3;
  // Sequence number for this event, such that the ordering within job_graph_run_id is correct
  int64 sequence = 4;

  // Metadata
  // Status of the job run as of this event
  JobRunStatus status = 10;
  // Hash key of (label, outputs) to associate different runs over time
  string job_key = 11;
  JobLabel label = 12;
  repeated PartitionRef outputs = 13;

  // Sum type of the below events
  JobConfigEvent config = 20;
  JobLogsEvent logs = 21;
}

//
// Job Graph Run Events
//

message GraphAnalyzeEvent { JobGraph graph = 1; }

// Represents a change in status for a datadep
message GraphDataDepEvent {}

// The complete lifecycle that each job graph run goes through
enum JobGraphRunStatus {
  // Initial state - graph will be analyzed before executing
  GRAPH_STARTED = 0;
  // Analysis completed
  GRAPH_ANALYZED = 1;
  // Graph is in this state until deps are satisfied for at least 1 job
  AWAITING_DEPS = 2;
  // Graph is executing at least 1 job (if this drops to 0, state goes back to AWAITING_DEPS)
  GRAPH_RUNNING = 3;
  // Graph execution finished successfully
  GRAPH_SUCCESS = 4;
  // Graph execution failed
  GRAPH_FAILED = 5;
}

message GraphEvent {
  // IDs
  // Unique ID for this job graph run
  string job_graph_run_id = 1;
  // Unique ID for this event
  string job_graph_run_event_id = 2;
  // Sequence number for this event, such that the ordering within job_graph_run_id is correct
  int64 sequence = 3;

  // Metadata
  // Current status of the job graph run as of this event
  JobGraphRunStatus status = 10;
  GraphLabel label = 11;

  // Sum type of below events
  GraphAnalyzeEvent analysis = 20;
}

// The sequence of events that completely describes progress of the job graph build
message JobGraphRunEvent {
  string job_graph_run_event_id = 1;
  google.protobuf.Timestamp timestamp = 2;

  // Sum type for potential events
  JobEvent job_event = 10;
  GraphEvent graph_event = 11;
}

message JobGraphRun {
  string job_graph_run_id = 1;
  JobGraph graph = 2;
  repeated JobGraphRunEvent events = 3;
}

message JobGraphRunQuery {
  // TODO
}

message ListJobGraphRunsRequest { JobGraphRunQuery query = 1; }
message ListJobGraphRunsResponse { repeated JobGraphRun runs = 1; }

message PartitionManifestsQuery {
  // TODO
}

message ListPartitionManifestsRequest { PartitionManifestsQuery query = 1; }
message ListPartitionManifestsResponse { repeated PartitionManifest manifests = 1; }

service Catalog {
  // JTBDs
  // -

  // Enables lookup of job graph runs (current or past)
  rpc ListJobGraphRuns(ListJobGraphRunsRequest) returns (ListJobGraphRunsResponse);

  // Enables lookup of partition manifests produced as part of prior job runs
  rpc ListPartitions(ListPartitionManifestsRequest) returns (ListPartitionManifestsResponse);
}