cache validation and avoid re-encoding

Author: daveshanley

schema_validation/validate_schema.go

@@ -8,14 +8,15 @@ import (
 	"errors"
 	"fmt"
 	"log/slog"
+	"math"
 	"os"
 	"reflect"
 	"regexp"
 	"strconv"
 	"sync"
 
+	"github.com/pb33f/libopenapi-validator/cache"
 	"github.com/pb33f/libopenapi/datamodel/high/base"
-	"github.com/pb33f/libopenapi/utils"
 	"github.com/santhosh-tekuri/jsonschema/v6"
 	"go.yaml.in/yaml/v4"
 	"golang.org/x/text/language"
@@ -123,49 +124,114 @@ func (s *schemaValidator) validateSchemaWithVersion(schema *base.Schema, payload
 	}
 
 	var renderedSchema []byte
-
-	// render the schema, to be used for validation, stop this from running concurrently, mutations are made to state
-	// and, it will cause async issues.
-	// Create isolated render context for this validation to prevent false positive cycle detection
-	// when multiple validations run concurrently.
-	// Use validation mode to force full inlining of discriminator refs - the JSON schema compiler
-	// needs a self-contained schema without unresolved $refs.
-	renderCtx := base.NewInlineRenderContextForValidation()
-	s.lock.Lock()
-	var e error
-	renderedSchema, e = schema.RenderInlineWithContext(renderCtx)
-	if e != nil {
-		// schema cannot be rendered, so it's not valid!
-		violation := &liberrors.SchemaValidationFailure{
-			Reason:          e.Error(),
-			Location:        "unavailable",
-			ReferenceSchema: string(renderedSchema),
-			ReferenceObject: string(payload),
+	var renderedNode *yaml.Node
+	var compiledSchema *jsonschema.Schema
+
+	// Check cache first — reuses existing SchemaCache (populated by NewValidationOptions).
+	var cacheKey uint64
+	canCache := s.options.SchemaCache != nil && schema.GoLow() != nil
+	if canCache {
+		// Include version in key so 3.0 (nullable) and 3.1 compile differently.
+		cacheKey = schema.GoLow().Hash() ^ uint64(math.Float32bits(version))
+		if cached, ok := s.options.SchemaCache.Load(cacheKey); ok &&
+			cached != nil && cached.CompiledSchema != nil {
+			renderedSchema = cached.RenderedInline
+			renderedNode = cached.RenderedNode
+			compiledSchema = cached.CompiledSchema
 		}
-		validationErrors = append(validationErrors, &liberrors.ValidationError{
-			ValidationType:         helpers.RequestBodyValidation,
-			ValidationSubType:      helpers.Schema,
-			Message:                "schema does not pass validation",
-			Reason:                 fmt.Sprintf("The schema cannot be decoded: %s", e.Error()),
-			SpecLine:               schema.GoLow().GetRootNode().Line,
-			SpecCol:                schema.GoLow().GetRootNode().Column,
-			SchemaValidationErrors: []*liberrors.SchemaValidationFailure{violation},
-			HowToFix:               liberrors.HowToFixInvalidSchema,
-			Context:                string(renderedSchema),
-		})
+	}
+
+	// Cache miss — render, convert to JSON, and compile.
+	if compiledSchema == nil {
+		renderCtx := base.NewInlineRenderContextForValidation()
+		s.lock.Lock()
+		nodeIface, renderErr := schema.MarshalYAMLInlineWithContext(renderCtx)
 		s.lock.Unlock()
-		return false, validationErrors
 
-	}
-	s.lock.Unlock()
+		if renderErr != nil {
+			violation := &liberrors.SchemaValidationFailure{
+				Reason:          renderErr.Error(),
+				Location:        "unavailable",
+				ReferenceSchema: string(renderedSchema),
+				ReferenceObject: string(payload),
+			}
+			validationErrors = append(validationErrors, &liberrors.ValidationError{
+				ValidationType:         helpers.RequestBodyValidation,
+				ValidationSubType:      helpers.Schema,
+				Message:                "schema does not pass validation",
+				Reason:                 fmt.Sprintf("The schema cannot be decoded: %s", renderErr.Error()),
+				SpecLine:               schema.GoLow().GetRootNode().Line,
+				SpecCol:                schema.GoLow().GetRootNode().Column,
+				SchemaValidationErrors: []*liberrors.SchemaValidationFailure{violation},
+				HowToFix:               liberrors.HowToFixInvalidSchema,
+				Context:                string(renderedSchema),
+			})
+			return false, validationErrors
+		}
+
+		// MarshalYAMLInlineWithContext returns *yaml.Node (from NodeBuilder.Render)
+		renderedNode, _ = nodeIface.(*yaml.Node)
 
-	jsonSchema, _ := utils.ConvertYAMLtoJSON(renderedSchema)
+		// yaml.Node → map → JSON bytes (skips yaml.Marshal + yaml.Unmarshal round-trip)
+		var jsonMap map[string]interface{}
+		if renderedNode != nil {
+			_ = renderedNode.Decode(&jsonMap)
+		}
+		jsonSchema, _ := json.Marshal(jsonMap)
+
+		// YAML bytes generated once for error messages / context strings
+		renderedSchema, _ = yaml.Marshal(renderedNode)
+
+		path := ""
+		if schema.GoLow().GetIndex() != nil {
+			path = schema.GoLow().GetIndex().GetSpecAbsolutePath()
+		}
+
+		var compileErr error
+		compiledSchema, compileErr = helpers.NewCompiledSchemaWithVersion(path, jsonSchema, s.options, version)
+		if compileErr != nil {
+			violation := &liberrors.SchemaValidationFailure{
+				Reason:          compileErr.Error(),
+				Location:        "schema compilation",
+				ReferenceSchema: string(renderedSchema),
+				ReferenceObject: string(payload),
+			}
+			line := 1
+			col := 0
+			if schema.GoLow().Type.KeyNode != nil {
+				line = schema.GoLow().Type.KeyNode.Line
+				col = schema.GoLow().Type.KeyNode.Column
+			}
+			validationErrors = append(validationErrors, &liberrors.ValidationError{
+				ValidationType:         helpers.Schema,
+				ValidationSubType:      helpers.Schema,
+				Message:                "schema compilation failed",
+				Reason:                 fmt.Sprintf("Schema compilation failed: %s", compileErr.Error()),
+				SpecLine:               line,
+				SpecCol:                col,
+				SchemaValidationErrors: []*liberrors.SchemaValidationFailure{violation},
+				HowToFix:               liberrors.HowToFixInvalidSchema,
+				Context:                string(renderedSchema),
+			})
+			return false, validationErrors
+		}
+
+		// Store in cache for subsequent validations of the same schema.
+		if canCache && compiledSchema != nil {
+			s.options.SchemaCache.Store(cacheKey, &cache.SchemaCacheEntry{
+				Schema:          schema,
+				RenderedInline:  renderedSchema,
+				ReferenceSchema: string(renderedSchema),
+				RenderedJSON:    jsonSchema,
+				CompiledSchema:  compiledSchema,
+				RenderedNode:    renderedNode,
+			})
+		}
+	}
 
 	if decodedObject == nil && len(payload) > 0 {
 		err := json.Unmarshal(payload, &decodedObject)
 		if err != nil {
-
-			// cannot decode the request body, so it's not valid
 			violation := &liberrors.SchemaValidationFailure{
 				Reason:          err.Error(),
 				Location:        "unavailable",
@@ -191,45 +257,12 @@ func (s *schemaValidator) validateSchemaWithVersion(schema *base.Schema, payload
 			})
 			return false, validationErrors
 		}
-
-	}
-
-	path := ""
-	if schema.GoLow().GetIndex() != nil {
-		path = schema.GoLow().GetIndex().GetSpecAbsolutePath()
 	}
-	jsch, err := helpers.NewCompiledSchemaWithVersion(path, jsonSchema, s.options, version)
 
 	var schemaValidationErrors []*liberrors.SchemaValidationFailure
-	if err != nil {
-		violation := &liberrors.SchemaValidationFailure{
-			Reason:          err.Error(),
-			Location:        "schema compilation",
-			ReferenceSchema: string(renderedSchema),
-			ReferenceObject: string(payload),
-		}
-		line := 1
-		col := 0
-		if schema.GoLow().Type.KeyNode != nil {
-			line = schema.GoLow().Type.KeyNode.Line
-			col = schema.GoLow().Type.KeyNode.Column
-		}
-		validationErrors = append(validationErrors, &liberrors.ValidationError{
-			ValidationType:         helpers.Schema,
-			ValidationSubType:      helpers.Schema,
-			Message:                "schema compilation failed",
-			Reason:                 fmt.Sprintf("Schema compilation failed: %s", err.Error()),
-			SpecLine:               line,
-			SpecCol:                col,
-			SchemaValidationErrors: []*liberrors.SchemaValidationFailure{violation},
-			HowToFix:               liberrors.HowToFixInvalidSchema,
-			Context:                string(renderedSchema),
-		})
-		return false, validationErrors
-	}
 
-	if jsch != nil && decodedObject != nil {
-		scErrs := jsch.Validate(decodedObject)
+	if compiledSchema != nil && decodedObject != nil {
+		scErrs := compiledSchema.Validate(decodedObject)
 		if scErrs != nil {
 
 			var jk *jsonschema.ValidationError
@@ -238,7 +271,7 @@ func (s *schemaValidator) validateSchemaWithVersion(schema *base.Schema, payload
 				// flatten the validationErrors
 				schFlatErr := jk.BasicOutput().Errors
 				schemaValidationErrors = extractBasicErrors(schFlatErr, renderedSchema,
-					decodedObject, payload, jk, schemaValidationErrors)
+					renderedNode, decodedObject, payload, jk, schemaValidationErrors)
 			}
 			line := 1
 			col := 0
@@ -266,13 +299,28 @@ func (s *schemaValidator) validateSchemaWithVersion(schema *base.Schema, payload
 }
 
 func extractBasicErrors(schFlatErrs []jsonschema.OutputUnit,
-	renderedSchema []byte, decodedObject interface{},
+	renderedSchema []byte, renderedNode *yaml.Node,
+	decodedObject interface{},
 	payload []byte, jk *jsonschema.ValidationError,
 	schemaValidationErrors []*liberrors.SchemaValidationFailure,
 ) []*liberrors.SchemaValidationFailure {
 	// Extract property name info once before processing errors (performance optimization)
 	propertyInfo := extractPropertyNameFromError(jk)
 
+	// Determine root content node ONCE (not per-error).
+	// NodeBuilder.Render() returns MappingNode directly, no DocumentNode unwrapping needed.
+	var rootNode *yaml.Node
+	if renderedNode != nil {
+		rootNode = renderedNode
+	} else if len(renderedSchema) > 0 {
+		// Fallback: parse bytes ONCE
+		var docNode yaml.Node
+		_ = yaml.Unmarshal(renderedSchema, &docNode)
+		if len(docNode.Content) > 0 {
+			rootNode = docNode.Content[0]
+		}
+	}
+
 	for q := range schFlatErrs {
 		er := schFlatErrs[q]
 
@@ -282,12 +330,11 @@ func extractBasicErrors(schFlatErrs []jsonschema.OutputUnit,
 		}
 		if er.Error != nil {
 
-			// re-encode the schema.
-			var renderedNode yaml.Node
-			_ = yaml.Unmarshal(renderedSchema, &renderedNode)
-
 			// locate the violated property in the schema
-			located := LocateSchemaPropertyNodeByJSONPath(renderedNode.Content[0], er.KeywordLocation)
+			var located *yaml.Node
+			if rootNode != nil {
+				located = LocateSchemaPropertyNodeByJSONPath(rootNode, er.KeywordLocation)
+			}
 
 			// extract the element specified by the instance
 			val := instanceLocationRegex.FindStringSubmatch(er.InstanceLocation)
@@ -331,9 +378,9 @@ func extractBasicErrors(schFlatErrs []jsonschema.OutputUnit,
 				// location of the violation within the rendered schema.
 				violation.Line = line
 				violation.Column = located.Column
-			} else {
+			} else if rootNode != nil {
 				// handles property name validation errors that don't provide useful InstanceLocation
-				applyPropertyNameFallback(propertyInfo, renderedNode.Content[0], violation)
+				applyPropertyNameFallback(propertyInfo, rootNode, violation)
 			}
 			schemaValidationErrors = append(schemaValidationErrors, violation)
 		}