tools/schema_test.go

package tools

import (
	"encoding/json"
	"fmt"
	"reflect"
	"testing"

	"github.com/korap/korap-mcp/service"
	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"
)

// TestAllToolsSchemaCompliance ensures all tools have properly structured schemas
func TestAllToolsSchemaCompliance(t *testing.T) {
	client := &service.Client{}

	tools := []Tool{
		NewSearchTool(client),
		NewMetadataTool(client),
	}

	for _, tool := range tools {
		t.Run(tool.Name(), func(t *testing.T) {
			schema := tool.InputSchema()

			// Verify basic schema structure
			validateBasicSchemaStructure(t, schema, tool.Name())

			// Verify schema is valid JSON
			validateSchemaIsValidJSON(t, schema, tool.Name())

			// Verify schema documentation completeness
			validateSchemaDocumentation(t, schema, tool.Name())
		})
	}
}

// validateBasicSchemaStructure checks that the schema has the required top-level properties
func validateBasicSchemaStructure(t *testing.T, schema map[string]any, toolName string) {
	// Must be object type
	assert.Equal(t, "object", schema["type"], "Tool %s schema must be object type", toolName)

	// Must have properties
	properties, exists := schema["properties"]
	assert.True(t, exists, "Tool %s schema must have properties", toolName)
	assert.IsType(t, map[string]any{}, properties, "Tool %s properties must be an object", toolName)

	// Must have required array
	required, exists := schema["required"]
	assert.True(t, exists, "Tool %s schema must have required array", toolName)
	assert.IsType(t, []string{}, required, "Tool %s required must be string array", toolName)

	// Should have title and description
	assert.Contains(t, schema, "title", "Tool %s schema should have title", toolName)
	assert.Contains(t, schema, "description", "Tool %s schema should have description", toolName)

	// Should prevent additional properties for strict validation
	assert.Contains(t, schema, "additionalProperties", "Tool %s schema should specify additionalProperties", toolName)
	assert.Equal(t, false, schema["additionalProperties"], "Tool %s should not allow additional properties", toolName)
}

// validateSchemaIsValidJSON ensures the schema can be properly serialized to JSON
func validateSchemaIsValidJSON(t *testing.T, schema map[string]any, toolName string) {
	jsonBytes, err := json.Marshal(schema)
	require.NoError(t, err, "Tool %s schema must be serializable to JSON", toolName)

	var unmarshalled map[string]any
	err = json.Unmarshal(jsonBytes, &unmarshalled)
	require.NoError(t, err, "Tool %s schema JSON must be valid", toolName)

	// Verify the unmarshalled schema has the same structure
	assert.Equal(t, schema["type"], unmarshalled["type"], "Tool %s schema type must survive JSON round-trip", toolName)
}

// validateSchemaDocumentation checks that all properties have proper documentation
func validateSchemaDocumentation(t *testing.T, schema map[string]any, toolName string) {
	properties, ok := schema["properties"].(map[string]any)
	require.True(t, ok, "Tool %s properties must be accessible", toolName)

	for propName, propSchema := range properties {
		propMap, ok := propSchema.(map[string]any)
		require.True(t, ok, "Tool %s property %s must be an object", toolName, propName)

		// Must have type
		assert.Contains(t, propMap, "type", "Tool %s property %s must have type", toolName, propName)

		// Must have description
		assert.Contains(t, propMap, "description", "Tool %s property %s must have description", toolName, propName)

		description, ok := propMap["description"].(string)
		assert.True(t, ok, "Tool %s property %s description must be string", toolName, propName)
		assert.NotEmpty(t, description, "Tool %s property %s description must not be empty", toolName, propName)
		assert.Greater(t, len(description), 10, "Tool %s property %s description should be descriptive", toolName, propName)

		// String properties should have examples
		if propType, ok := propMap["type"].(string); ok && propType == "string" {
			if enumValues, hasEnum := propMap["enum"]; hasEnum {
				// Enum properties should have examples matching enum values
				assert.Contains(t, propMap, "examples", "Tool %s property %s with enum should have examples", toolName, propName)
				examples, ok := propMap["examples"].([]string)
				assert.True(t, ok, "Tool %s property %s examples must be string array", toolName, propName)

				enumArray, ok := enumValues.([]string)
				assert.True(t, ok, "Tool %s property %s enum must be string array", toolName, propName)

				// All examples should be valid enum values
				for _, example := range examples {
					assert.Contains(t, enumArray, example, "Tool %s property %s example '%s' must be valid enum value", toolName, propName, example)
				}
			} else {
				// Regular string properties should have examples
				assert.Contains(t, propMap, "examples", "Tool %s property %s should have examples", toolName, propName)
			}
		}

		// Integer properties should have examples and constraints
		if propType, ok := propMap["type"].(string); ok && propType == "integer" {
			assert.Contains(t, propMap, "examples", "Tool %s property %s should have examples", toolName, propName)

			// Should have minimum/maximum constraints
			assert.Contains(t, propMap, "minimum", "Tool %s property %s should have minimum", toolName, propName)
			assert.Contains(t, propMap, "maximum", "Tool %s property %s should have maximum", toolName, propName)
		}
	}
}

// TestSchemaExamples verifies that schema examples are realistic and valid
func TestSchemaExamples(t *testing.T) {
	client := &service.Client{}

	t.Run("SearchTool", func(t *testing.T) {
		tool := NewSearchTool(client)
		schema := tool.InputSchema()

		properties := schema["properties"].(map[string]any)

		// Test query examples
		queryProp := properties["query"].(map[string]any)
		examples := queryProp["examples"].([]string)

		for _, example := range examples {
			assert.NotEmpty(t, example, "Query example should not be empty")
			assert.LessOrEqual(t, len(example), 100, "Query example should be reasonably short")
		}

		// Test corpus examples
		corpusProp := properties["corpus"].(map[string]any)
		corpusExamples := corpusProp["examples"].([]string)

		for _, example := range corpusExamples {
			assert.NotEmpty(t, example, "Corpus example should not be empty")
			// Should match the pattern defined in the schema
			pattern := corpusProp["pattern"].(string)
			assert.NotEmpty(t, pattern, "Corpus should have validation pattern")
		}
	})

	t.Run("MetadataTool", func(t *testing.T) {
		tool := NewMetadataTool(client)
		schema := tool.InputSchema()

		properties := schema["properties"].(map[string]any)

		// Test action examples
		actionProp := properties["action"].(map[string]any)
		examples := actionProp["examples"].([]string)
		enumValues := actionProp["enum"].([]string)

		for _, example := range examples {
			assert.Contains(t, enumValues, example, "Action example should be valid enum value")
		}
	})
}

// TestSchemaConstraints verifies that schema constraints are reasonable
func TestSchemaConstraints(t *testing.T) {
	client := &service.Client{}

	t.Run("SearchTool", func(t *testing.T) {
		tool := NewSearchTool(client)
		schema := tool.InputSchema()
		properties := schema["properties"].(map[string]any)

		// Test query constraints
		queryProp := properties["query"].(map[string]any)
		minLength := queryProp["minLength"].(int)
		maxLength := queryProp["maxLength"].(int)

		assert.Greater(t, minLength, 0, "Query minimum length should be positive")
		assert.Greater(t, maxLength, minLength, "Query maximum length should be greater than minimum")
		assert.LessOrEqual(t, maxLength, 10000, "Query maximum length should be reasonable")

		// Test count constraints
		countProp := properties["count"].(map[string]any)
		minimum := countProp["minimum"].(int)
		maximum := countProp["maximum"].(int)
		defaultValue := countProp["default"].(int)

		assert.GreaterOrEqual(t, minimum, 0, "Count minimum should be non-negative")
		assert.Greater(t, maximum, minimum, "Count maximum should be greater than minimum")
		assert.GreaterOrEqual(t, defaultValue, minimum, "Count default should be >= minimum")
		assert.LessOrEqual(t, defaultValue, maximum, "Count default should be <= maximum")
	})
}

// TestSchemaVersioning ensures schema structure is backwards compatible
func TestSchemaVersioning(t *testing.T) {
	client := &service.Client{}

	// This test would be extended when we add new versions
	// For now, we ensure the basic structure is stable

	tools := []Tool{
		NewSearchTool(client),
		NewMetadataTool(client),
	}

	for _, tool := range tools {
		t.Run(tool.Name(), func(t *testing.T) {
			schema := tool.InputSchema()

			// Core schema structure must remain stable
			assert.Equal(t, "object", schema["type"])
			assert.Contains(t, schema, "properties")
			assert.Contains(t, schema, "required")

			// Tool-specific stability checks
			switch tool.Name() {
			case "korap_search":
				properties := schema["properties"].(map[string]any)
				assert.Contains(t, properties, "query", "Search tool must always have query parameter")

				required := schema["required"].([]string)
				assert.Contains(t, required, "query", "Search tool must always require query parameter")

			case "korap_metadata":
				properties := schema["properties"].(map[string]any)
				assert.Contains(t, properties, "action", "Metadata tool must always have action parameter")

				required := schema["required"].([]string)
				assert.Contains(t, required, "action", "Metadata tool must always require action parameter")
			}
		})
	}
}

// BenchmarkSchemaGeneration measures schema generation performance
func BenchmarkSchemaGeneration(b *testing.B) {
	client := &service.Client{}

	tools := []Tool{
		NewSearchTool(client),
		NewMetadataTool(client),
	}

	for _, tool := range tools {
		b.Run(tool.Name(), func(b *testing.B) {
			for i := 0; i < b.N; i++ {
				schema := tool.InputSchema()
				_ = schema // Prevent optimization
			}
		})
	}
}

// TestSchemaDeepCopy ensures schemas are independent instances
func TestSchemaDeepCopy(t *testing.T) {
	client := &service.Client{}
	tool := NewSearchTool(client)

	// Get two schema instances
	schema1 := tool.InputSchema()
	schema2 := tool.InputSchema()

	// They should be equal in content but different instances
	assert.True(t, reflect.DeepEqual(schema1, schema2), "Schemas should have equal content")

	// Modifying one should not affect the other
	schema1["modified"] = true
	assert.NotContains(t, schema2, "modified", "Schema instances should be independent")
}

// TestSchemaHelpGeneration verifies schema can be used to generate help text
func TestSchemaHelpGeneration(t *testing.T) {
	client := &service.Client{}

	tools := []Tool{
		NewSearchTool(client),
		NewMetadataTool(client),
	}

	for _, tool := range tools {
		t.Run(tool.Name(), func(t *testing.T) {
			schema := tool.InputSchema()

			// Generate help text from schema
			helpText := generateHelpText(schema, tool.Name())

			// Verify help text contains essential information
			assert.Contains(t, helpText, tool.Name(), "Help text should contain tool name")
			assert.Contains(t, helpText, "Parameters:", "Help text should list parameters")

			// Should contain information about required parameters
			required := schema["required"].([]string)
			for _, param := range required {
				assert.Contains(t, helpText, param, "Help text should mention required parameter %s", param)
				assert.Contains(t, helpText, "(required)", "Help text should mark required parameters")
			}
		})
	}
}

// generateHelpText creates human-readable help text from a schema
func generateHelpText(schema map[string]any, toolName string) string {
	help := fmt.Sprintf("Tool: %s\n", toolName)

	if desc, ok := schema["description"].(string); ok {
		help += fmt.Sprintf("Description: %s\n\n", desc)
	}

	properties, ok := schema["properties"].(map[string]any)
	if !ok {
		return help
	}

	required, _ := schema["required"].([]string)
	requiredMap := make(map[string]bool)
	for _, req := range required {
		requiredMap[req] = true
	}

	help += "Parameters:\n"
	for paramName, paramSchema := range properties {
		paramMap, ok := paramSchema.(map[string]any)
		if !ok {
			continue
		}

		requiredText := ""
		if requiredMap[paramName] {
			requiredText = " (required)"
		}

		paramType, _ := paramMap["type"].(string)
		description, _ := paramMap["description"].(string)

		help += fmt.Sprintf("  %s (%s)%s: %s\n", paramName, paramType, requiredText, description)

		// Add enum values if present
		if enum, ok := paramMap["enum"].([]string); ok {
			help += fmt.Sprintf("    Allowed values: %v\n", enum)
		}

		// Add examples if present
		if examples, ok := paramMap["examples"]; ok {
			help += fmt.Sprintf("    Examples: %v\n", examples)
		}
	}

	return help
}