syntax = "proto3";

package langium_ai_tools;

// Position in a text document (line and character offsets)
message Position {
    // Line position (zero-based)
    uint32 line = 1;
    // Character position (zero-based)  
    uint32 character = 2;
}

// Text range with start and end positions
message Range {
    // Start position
    Position start = 1;
    // End position
    Position end = 2;
}

// Diagnostic severity levels
enum DiagnosticSeverity {
    DIAGNOSTIC_SEVERITY_UNSPECIFIED = 0;
    DIAGNOSTIC_SEVERITY_ERROR = 1;
    DIAGNOSTIC_SEVERITY_WARNING = 2;
    DIAGNOSTIC_SEVERITY_INFORMATION = 3;
    DIAGNOSTIC_SEVERITY_HINT = 4;
}

// VS Code diagnostic information
message Diagnostic {
    // Range where the diagnostic applies
    Range range = 1;
    // Human-readable message
    string message = 2;
    // Severity level
    optional DiagnosticSeverity severity = 3;
    // Optional diagnostic code
    optional string code = 4;
    // Optional source identifier (e.g., "langium")
    optional string source = 5;
}

// Langium-specific evaluator result data
message LangiumEvaluatorResultDataMsg {

    // Base EvaluatorResultData START

    // Optional runtime in milliseconds
    optional double runtime = 1;

    // Base EvaluatorResultData END

    // Number of validation failures
    uint32 failures = 2;
    // Number of errors
    uint32 errors = 3;
    // Number of warnings
    uint32 warnings = 4;
    // Number of infos
    uint32 infos = 5;
    // Number of hints
    uint32 hints = 6;
    // Number of unassigned diagnostics
    uint32 unassigned = 7;
    // Length of the response in characters
    uint32 response_length = 8;
    // Raw diagnostic data
    repeated Diagnostic diagnostics = 9;
}

// Diversity metrics for rule usage patterns
message DiversityMetrics {
    // Shannon entropy - information diversity measure
    // Range: 0 to log₂(n) where n = number of rules
    // Low (0-1): dominated by few rules
    // Medium (1-3): moderate diversity
    // High (>3): high diversity
    double entropy = 1;
    
    // Gini coefficient - inequality measure
    // Range: 0 to 1
    // Low (0-0.3): equal distribution
    // Medium (0.3-0.7): moderate inequality
    // High (0.7-1): high inequality
    double gini_coefficient = 2;
    
    // Simpson's diversity index - probability that two randomly selected items are different
    // Range: 0 to 1
    // Low (0-0.3): low diversity
    // Medium (0.3-0.7): moderate diversity
    // High (0.7-1): high diversity
    double simpson_index = 3;
}

// Syntax usage statistics for grammar analysis
message SyntaxStatistic {
    // Map of rule names to their usage counts
    map<string, uint32> rule_usage = 1;
    
    // Percentage of used rules compared to all available rules
    double coverage = 2;
    
    // Diversity metrics for rule usage patterns
    DiversityMetrics diversity = 3;
}

// Nested metadata container (workaround for map in oneof restriction)
message NestedMetadata {
    map<string, MetadataValue> values = 1;
}

// Metadata value wrapper to support different types of metadata
message MetadataValue {
    oneof value {
        // String value
        string string_value = 1;
        // Number value
        double number_value = 2;
        // Boolean value
        bool bool_value = 3;
        // Nested metadata (workaround for map limitation in oneof)
        NestedMetadata nested_metadata_value = 4;
        // Syntax statistics value
        SyntaxStatistic syntax_statistic_value = 5;
        // Can add more types as needed in the future
    }
}

// Generic evaluator result container
message EvaluatorResultMsg {
    // Name of this evaluation
    string name = 1;
    // Optional metadata as key-value pairs supporting different value types
    map<string, MetadataValue> metadata = 2;
    // Langium-specific data
    LangiumEvaluatorResultDataMsg data = 3;
}