datasynth-server 5.8.0

syntax = "proto3";

package synth;

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

// Main service for synthetic data generation
service SyntheticDataService {
    // Bulk generation - returns when complete
    rpc BulkGenerate(BulkGenerateRequest) returns (BulkGenerateResponse);

    // Streaming - continuous real-time data generation
    rpc StreamData(StreamDataRequest) returns (stream DataEvent);

    // Control commands (pause, resume, stop)
    rpc Control(ControlCommand) returns (ControlResponse);

    // Configuration management
    rpc GetConfig(google.protobuf.Empty) returns (ConfigResponse);
    rpc SetConfig(ConfigRequest) returns (ConfigResponse);

    // Server health and metrics
    rpc GetMetrics(google.protobuf.Empty) returns (MetricsResponse);
    rpc HealthCheck(google.protobuf.Empty) returns (HealthResponse);
}

// ============================================================================
// Generation Requests
// ============================================================================

message BulkGenerateRequest {
    // Optional configuration override (uses server default if not provided)
    optional GenerationConfig config = 1;
    // Number of journal entries to generate
    uint64 entry_count = 2;
    // Output format
    OutputFormat output_format = 3;
    // Whether to include master data in response
    bool include_master_data = 4;
    // Whether to inject anomalies
    bool inject_anomalies = 5;
}

message BulkGenerateResponse {
    // Total entries generated
    uint64 entries_generated = 1;
    // Total time in milliseconds
    uint64 duration_ms = 2;
    // Generated journal entries (serialized)
    repeated JournalEntryProto journal_entries = 3;
    // Anomaly labels (if injection enabled)
    repeated AnomalyLabelProto anomaly_labels = 4;
    // Statistics
    GenerationStats stats = 5;
}

message StreamDataRequest {
    // Optional configuration override
    optional GenerationConfig config = 1;
    // Target events per second (0 = unlimited)
    uint32 events_per_second = 2;
    // Maximum number of events (0 = unlimited)
    uint64 max_events = 3;
    // Whether to inject anomalies
    bool inject_anomalies = 4;
    // Anomaly injection rate (0.0 to 1.0)
    float anomaly_rate = 5;
}

// ============================================================================
// Data Events
// ============================================================================

message DataEvent {
    // Event sequence number
    uint64 sequence = 1;
    // Event timestamp
    google.protobuf.Timestamp timestamp = 2;
    // Event type
    oneof event {
        JournalEntryProto journal_entry = 3;
        MasterDataEvent master_data = 4;
        DocumentFlowEvent document_flow = 5;
        AnomalyEvent anomaly = 6;
        MetricsEvent metrics = 7;
    }
}

message JournalEntryProto {
    string document_id = 1;
    string company_code = 2;
    uint32 fiscal_year = 3;
    uint32 fiscal_period = 4;
    string posting_date = 5;
    string document_date = 6;
    string created_at = 7;
    string source = 8;
    optional string business_process = 9;
    repeated JournalLineProto lines = 10;
    bool is_anomaly = 11;
    optional string anomaly_type = 12;
}

message JournalLineProto {
    uint32 line_number = 1;
    string account_number = 2;
    string account_name = 3;
    string amount = 4;  // Decimal as string for precision
    bool is_debit = 5;
    optional string cost_center = 6;
    optional string profit_center = 7;
    optional string vendor_id = 8;
    optional string customer_id = 9;
    optional string material_id = 10;
    optional string text = 11;
}

message MasterDataEvent {
    oneof data {
        VendorProto vendor = 1;
        CustomerProto customer = 2;
        MaterialProto material = 3;
        FixedAssetProto fixed_asset = 4;
        EmployeeProto employee = 5;
    }
}

message VendorProto {
    string vendor_id = 1;
    string name = 2;
    string country = 3;
    string currency = 4;
    string payment_terms = 5;
    bool is_intercompany = 6;
}

message CustomerProto {
    string customer_id = 1;
    string name = 2;
    string country = 3;
    string currency = 4;
    string credit_rating = 5;
    string credit_limit = 6;
    bool is_intercompany = 7;
}

message MaterialProto {
    string material_id = 1;
    string description = 2;
    string material_type = 3;
    string unit_of_measure = 4;
    string standard_cost = 5;
}

message FixedAssetProto {
    string asset_id = 1;
    string description = 2;
    string asset_class = 3;
    string acquisition_value = 4;
    string acquisition_date = 5;
    uint32 useful_life_months = 6;
}

message EmployeeProto {
    string employee_id = 1;
    string name = 2;
    string department = 3;
    string job_title = 4;
    optional string manager_id = 5;
}

message DocumentFlowEvent {
    string document_type = 1;
    string document_id = 2;
    string company_code = 3;
    string document_date = 4;
    optional string reference_document = 5;
    string total_amount = 6;
    string currency = 7;
}

message AnomalyEvent {
    string anomaly_id = 1;
    string anomaly_type = 2;
    string document_id = 3;
    string description = 4;
    float severity_score = 5;
}

message MetricsEvent {
    uint64 total_entries = 1;
    uint64 entries_per_second = 2;
    float anomaly_rate = 3;
    uint64 uptime_seconds = 4;
}

// ============================================================================
// Control
// ============================================================================

message ControlCommand {
    ControlAction action = 1;
    // Optional pattern to trigger (for TriggerPattern action)
    optional string pattern_name = 2;
}

enum ControlAction {
    CONTROL_ACTION_UNSPECIFIED = 0;
    PAUSE = 1;
    RESUME = 2;
    STOP = 3;
    TRIGGER_PATTERN = 4;  // Trigger a specific anomaly pattern
}

message ControlResponse {
    bool success = 1;
    string message = 2;
    StreamStatus current_status = 3;
}

enum StreamStatus {
    STREAM_STATUS_UNSPECIFIED = 0;
    RUNNING = 1;
    PAUSED = 2;
    STOPPED = 3;
}

// ============================================================================
// Configuration
// ============================================================================

message GenerationConfig {
    // Industry sector
    string industry = 1;
    // Simulation start date (YYYY-MM-DD)
    string start_date = 2;
    // Simulation period in months
    uint32 period_months = 3;
    // Random seed (0 = random)
    uint64 seed = 4;
    // CoA complexity (small, medium, large)
    string coa_complexity = 5;
    // Companies configuration
    repeated CompanyConfigProto companies = 6;
    // Enable fraud injection
    bool fraud_enabled = 7;
    // Fraud rate (0.0 to 1.0)
    float fraud_rate = 8;
    // Enable master data generation
    bool generate_master_data = 9;
    // Enable document flows
    bool generate_document_flows = 10;
}

message CompanyConfigProto {
    string code = 1;
    string name = 2;
    string currency = 3;
    string country = 4;
    uint64 annual_transaction_volume = 5;
    float volume_weight = 6;
}

message ConfigRequest {
    GenerationConfig config = 1;
}

message ConfigResponse {
    bool success = 1;
    string message = 2;
    GenerationConfig current_config = 3;
}

// ============================================================================
// Metrics & Health
// ============================================================================

message MetricsResponse {
    // Generation metrics
    uint64 total_entries_generated = 1;
    uint64 total_anomalies_injected = 2;
    uint64 uptime_seconds = 3;
    // Current session metrics
    uint64 session_entries = 4;
    double session_entries_per_second = 5;
    // Stream metrics
    uint32 active_streams = 6;
    uint64 total_stream_events = 7;
}

message HealthResponse {
    bool healthy = 1;
    string version = 2;
    uint64 uptime_seconds = 3;
}

// ============================================================================
// Statistics
// ============================================================================

message GenerationStats {
    uint64 total_entries = 1;
    uint64 total_lines = 2;
    string total_debit_amount = 3;
    string total_credit_amount = 4;
    uint64 anomaly_count = 5;
    map<string, uint64> entries_by_company = 6;
    map<string, uint64> entries_by_source = 7;
}

message AnomalyLabelProto {
    string anomaly_id = 1;
    string document_id = 2;
    string anomaly_type = 3;
    string anomaly_category = 4;
    string description = 5;
    float severity_score = 6;
}

enum OutputFormat {
    OUTPUT_FORMAT_UNSPECIFIED = 0;
    JSON = 1;
    PROTOBUF = 2;
    CSV = 3;
    PARQUET = 4;
}