refactor: improve unmarshalling performance and memory usage

Author: TristanSpeakEasy

.gitignore

@@ -1,2 +0,0 @@
-CLAUDE.local.md
-.claude/

arazzo/arazzo.go

@@ -16,7 +16,6 @@ import (
 	"github.com/speakeasy-api/openapi/internal/utils"
 	"github.com/speakeasy-api/openapi/marshaller"
 	"github.com/speakeasy-api/openapi/validation"
-	"github.com/speakeasy-api/openapi/yml"
 )
 
 // Version is the version of the Arazzo Specification that this package conforms to.
@@ -69,88 +68,66 @@ func Unmarshal(ctx context.Context, doc io.Reader, opts ...Option[unmarshalOptio
 		opt(&o)
 	}
 
-	c, err := core.Unmarshal(ctx, doc)
+	var arazzo Arazzo
+	validationErrs, err := marshaller.Unmarshal(ctx, doc, &arazzo)
 	if err != nil {
-		return nil, nil, err
+		return nil, nil, fmt.Errorf("failed to unmarshal Arazzo document: %w", err)
 	}
 
-	arazzo := &Arazzo{}
-	if err := marshaller.Populate(*c, arazzo); err != nil {
-		return nil, nil, err
+	if o.skipValidation {
+		return &arazzo, nil, nil
 	}
 
-	var validationErrs []error
-	if !o.skipValidation {
-		validationErrs = append(validationErrs, arazzo.Validate(ctx)...)
-		slices.SortFunc(validationErrs, func(a, b error) int {
-			var aValidationErr *validation.Error
-			var bValidationErr *validation.Error
-			aIsValidationErr := errors.As(a, &aValidationErr)
-			bIsValidationErr := errors.As(b, &bValidationErr)
-			if aIsValidationErr && bIsValidationErr {
-				if aValidationErr.Line == bValidationErr.Line {
-					return aValidationErr.Column - bValidationErr.Column
-				}
-				return aValidationErr.Line - bValidationErr.Line
-			} else if aIsValidationErr {
-				return -1
-			} else if bIsValidationErr {
-				return 1
+	validationErrs = append(validationErrs, arazzo.Validate(ctx)...)
+	slices.SortFunc(validationErrs, func(a, b error) int {
+		var aValidationErr *validation.Error
+		var bValidationErr *validation.Error
+		aIsValidationErr := errors.As(a, &aValidationErr)
+		bIsValidationErr := errors.As(b, &bValidationErr)
+		if aIsValidationErr && bIsValidationErr {
+			if aValidationErr.Line == bValidationErr.Line {
+				return aValidationErr.Column - bValidationErr.Column
 			}
+			return aValidationErr.Line - bValidationErr.Line
+		} else if aIsValidationErr {
+			return -1
+		} else if bIsValidationErr {
+			return 1
+		}
 
-			return 0
-		})
-	}
+		return 0
+	})
 
-	return arazzo, validationErrs, nil
+	return &arazzo, validationErrs, nil
 }
 
 // Marshal will marshal the provided Arazzo document to the provided io.Writer.
 func Marshal(ctx context.Context, arazzo *Arazzo, w io.Writer) error {
-	if arazzo == nil {
-		return errors.New("nil *Arazzo")
-	}
-
-	if err := arazzo.Marshal(ctx, w); err != nil {
-		return err
-	}
-
-	return nil
+	return marshaller.Marshal(ctx, arazzo, w)
 }
 
 // Sync will sync any changes made to the Arazzo document models back to the core models.
 func (a *Arazzo) Sync(ctx context.Context) error {
-	if _, err := marshaller.SyncValue(ctx, a, a.GetCore(), nil, false); err != nil {
+	if _, err := marshaller.SyncValue(ctx, a, a.GetCore(), a.GetRootNode(), false); err != nil {
 		return err
 	}
 	return nil
 }
 
-// Marshal will marshal the Arazzo document to the provided io.Writer.
-func (a *Arazzo) Marshal(ctx context.Context, w io.Writer) error {
-	ctx = yml.ContextWithConfig(ctx, a.GetCore().Config)
-
-	if _, err := marshaller.SyncValue(ctx, a, a.GetCore(), nil, false); err != nil {
-		return err
-	}
-
-	return a.GetCore().Marshal(ctx, w)
-}
-
 // Validate will validate the Arazzo document against the Arazzo Specification.
 func (a *Arazzo) Validate(ctx context.Context, opts ...validation.Option) []error {
 	opts = append(opts, validation.WithContextObject(a))
 
 	core := a.GetCore()
-	errs := core.GetValidationErrors()
+	errs := []error{}
 
 	arazzoMajor, arazzoMinor, arazzoPatch, err := utils.ParseVersion(a.Arazzo)
 	if err != nil {
-		errs = append(errs, validation.NewValueError(fmt.Sprintf("invalid Arazzo version in document %s: %s", a.Arazzo, err.Error()), core, core.Arazzo))
+		errs = append(errs, validation.NewValueError(validation.NewValueValidationError("invalid Arazzo version in document %s: %s", a.Arazzo, err.Error()), core, core.Arazzo))
 	}
 
 	if arazzoMajor != VersionMajor || arazzoMinor != VersionMinor || arazzoPatch > VersionPatch {
-		errs = append(errs, validation.NewValueError(fmt.Sprintf("only Arazzo version %s and below is supported", Version), core, core.Arazzo))
+		errs = append(errs, validation.NewValueError(validation.NewValueValidationError("only Arazzo version %s and below is supported", Version), core, core.Arazzo))
 	}
 
 	errs = append(errs, a.Info.Validate(ctx, opts...)...)
@@ -161,7 +138,7 @@ func (a *Arazzo) Validate(ctx context.Context, opts ...validation.Option) []erro
 		errs = append(errs, sourceDescription.Validate(ctx, opts...)...)
 
 		if _, ok := sourceDescriptionNames[sourceDescription.Name]; ok {
-			errs = append(errs, validation.NewSliceError(fmt.Sprintf("sourceDescription name %s is not unique", sourceDescription.Name), core, core.SourceDescriptions, i))
+			errs = append(errs, validation.NewSliceError(validation.NewValueValidationError("sourceDescription name %s is not unique", sourceDescription.Name), core, core.SourceDescriptions, i))
 		}
 
 		sourceDescriptionNames[sourceDescription.Name] = true
@@ -173,7 +150,7 @@ func (a *Arazzo) Validate(ctx context.Context, opts ...validation.Option) []erro
 		errs = append(errs, workflow.Validate(ctx, opts...)...)
 
 		if _, ok := workflowIds[workflow.WorkflowID]; ok {
-			errs = append(errs, validation.NewSliceError(fmt.Sprintf("workflowId %s is not unique", workflow.WorkflowID), core, core.Workflows, i))
+			errs = append(errs, validation.NewSliceError(validation.NewValueValidationError("workflowId %s is not unique", workflow.WorkflowID), core, core.Workflows, i))
 		}
 
 		workflowIds[workflow.WorkflowID] = true

arazzo/arazzo_examples_test.go

@@ -10,7 +10,7 @@ import (
 	"github.com/speakeasy-api/openapi/pointer"
 )
 
-// The below examples should be copied into the README.md file if every changed TODO: automate this
+// The below examples should be copied into the README.md file if ever changed TODO: automate this
 func Example_readAndMutate() {
 	ctx := context.Background()
 
@@ -48,7 +48,7 @@ func Example_readAndMutate() {
 func Example_creating() {
 	ctx := context.Background()
 
-	arazzo := &arazzo.Arazzo{
+	a := &arazzo.Arazzo{
 		Arazzo: arazzo.Version,
 		Info: arazzo.Info{
 			Title:   "My Workflow",
@@ -60,7 +60,7 @@ func Example_creating() {
 
 	buf := bytes.NewBuffer([]byte{})
 
-	err := arazzo.Marshal(ctx, buf)
+	err := arazzo.Marshal(ctx, a, buf)
 	if err != nil {
 		panic(err)
 	}
@@ -76,16 +76,16 @@ func Example_mutating() {
 		panic(err)
 	}
 
-	arazzo, _, err := arazzo.Unmarshal(ctx, f)
+	a, _, err := arazzo.Unmarshal(ctx, f)
 	if err != nil {
 		panic(err)
 	}
 
-	arazzo.Info.Title = "My updated workflow title"
+	a.Info.Title = "My updated workflow title"
 
 	buf := bytes.NewBuffer([]byte{})
 
-	if err := arazzo.Marshal(ctx, buf); err != nil {
+	if err := arazzo.Marshal(ctx, a, buf); err != nil {
 		panic(err)
 	}
 

arazzo/arazzo_order_test.go

@@ -8,7 +8,7 @@ import (
 	"testing"
 
 	"github.com/speakeasy-api/openapi/arazzo"
-	"github.com/speakeasy-api/openapi/arazzo/expression"
+	"github.com/speakeasy-api/openapi/expression"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"gopkg.in/yaml.v3"

arazzo/arazzo_test.go

@@ -15,7 +15,7 @@ import (
 	"github.com/speakeasy-api/openapi/arazzo"
 	"github.com/speakeasy-api/openapi/arazzo/core"
 	"github.com/speakeasy-api/openapi/arazzo/criterion"
-	"github.com/speakeasy-api/openapi/arazzo/expression"
+	"github.com/speakeasy-api/openapi/expression"
 	"github.com/speakeasy-api/openapi/extensions"
 	"github.com/speakeasy-api/openapi/jsonpointer"
 	"github.com/speakeasy-api/openapi/jsonschema/oas31"
@@ -292,11 +292,11 @@ sourceDescriptions:
 	require.NoError(t, err)
 
 	assert.Equal(t, []error{
-		&validation.Error{Line: 1, Column: 1, Message: "field workflows is missing"},
-		&validation.Error{Line: 1, Column: 9, Message: "only Arazzo version 1.0.1 and below is supported"},
-		&validation.Error{Line: 4, Column: 3, Message: "field version is missing"},
-		&validation.Error{Line: 6, Column: 5, Message: "field url is missing"},
-		&validation.Error{Line: 7, Column: 11, Message: "type must be one of [openapi, arazzo]"},
+		&validation.Error{Line: 1, Column: 1, UnderlyingError: validation.NewMissingFieldError("field workflows is missing")},
+		&validation.Error{Line: 1, Column: 9, UnderlyingError: validation.NewValueValidationError("only Arazzo version 1.0.1 and below is supported")},
+		&validation.Error{Line: 4, Column: 3, UnderlyingError: validation.NewMissingFieldError("field version is missing")},
+		&validation.Error{Line: 6, Column: 5, UnderlyingError: validation.NewMissingFieldError("field url is missing")},
+		&validation.Error{Line: 7, Column: 11, UnderlyingError: validation.NewValueValidationError("type must be one of [openapi, arazzo]")},
 	}, validationErrs)
 
 	expected := &arazzo.Arazzo{

arazzo/components.go

@@ -2,7 +2,6 @@ package arazzo
 
 import (
 	"context"
-	"fmt"
 	"regexp"
 
 	"github.com/speakeasy-api/openapi/arazzo/core"
@@ -41,11 +40,11 @@ type componentKey struct {
 // Validate validates the Components object.
 func (c *Components) Validate(ctx context.Context, opts ...validation.Option) []error {
 	core := c.GetCore()
-	errs := core.GetValidationErrors()
+	errs := []error{}
 
 	for key, input := range c.Inputs.All() {
 		if !componentNameRegex.MatchString(key) {
-			errs = append(errs, validation.NewMapKeyError(fmt.Sprintf("input key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.Inputs, key))
+			errs = append(errs, validation.NewMapKeyError(validation.NewValueValidationError("input key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.Inputs, key))
 		}
 
 		if input.IsLeft() {
@@ -57,7 +56,7 @@ func (c *Components) Validate(ctx context.Context, opts ...validation.Option) []
 
 	for key, parameter := range c.Parameters.All() {
 		if !componentNameRegex.MatchString(key) {
-			errs = append(errs, validation.NewMapKeyError(fmt.Sprintf("parameter key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.Parameters, key))
+			errs = append(errs, validation.NewMapKeyError(validation.NewValueValidationError("parameter key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.Parameters, key))
 		}
 
 		paramOps := append(opts, validation.WithContextObject(&componentKey{name: key}))
@@ -67,7 +66,7 @@ func (c *Components) Validate(ctx context.Context, opts ...validation.Option) []
 
 	for key, successAction := range c.SuccessActions.All() {
 		if !componentNameRegex.MatchString(key) {
-			errs = append(errs, validation.NewMapKeyError(fmt.Sprintf("successAction key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.SuccessActions, key))
+			errs = append(errs, validation.NewMapKeyError(validation.NewValueValidationError("successAction key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.SuccessActions, key))
 		}
 
 		successActionOps := append(opts, validation.WithContextObject(&componentKey{name: key}))
@@ -77,7 +76,7 @@ func (c *Components) Validate(ctx context.Context, opts ...validation.Option) []
 
 	for key, failureAction := range c.FailureActions.All() {
 		if !componentNameRegex.MatchString(key) {
-			errs = append(errs, validation.NewMapKeyError(fmt.Sprintf("failureAction key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.FailureActions, key))
+			errs = append(errs, validation.NewMapKeyError(validation.NewValueValidationError("failureAction key must be a valid key [%s]: %s", componentNameRegex.String(), key), core, core.FailureActions, key))
 		}
 
 		failureActionOps := append(opts, validation.WithContextObject(&componentKey{name: key}))

arazzo/core/arazzo.go

@@ -1,16 +1,8 @@
 package core
 
 import (
-	"context"
-	"errors"
-	"fmt"
-	"io"
-
 	"github.com/speakeasy-api/openapi/extensions/core"
-	"github.com/speakeasy-api/openapi/json"
 	"github.com/speakeasy-api/openapi/marshaller"
-	"github.com/speakeasy-api/openapi/yml"
-	"gopkg.in/yaml.v3"
 )
 
 type Arazzo struct {
@@ -22,51 +14,4 @@ type Arazzo struct {
 	Workflows          marshaller.Node[[]*Workflow]          `key:"workflows" required:"true"`
 	Components         marshaller.Node[*Components]          `key:"components"`
 	Extensions         core.Extensions                       `key:"extensions"`
-
-	Config *yml.Config
-}
-
-func Unmarshal(ctx context.Context, doc io.Reader) (*Arazzo, error) {
-	data, err := io.ReadAll(doc)
-	if err != nil {
-		return nil, fmt.Errorf("failed to read Arazzo document: %w", err)
-	}
-
-	if len(data) == 0 {
-		return nil, errors.New("empty document")
-	}
-
-	var root yaml.Node
-	if err := yaml.Unmarshal(data, &root); err != nil {
-		return nil, fmt.Errorf("failed to unmarshal Arazzo document: %w", err)
-	}
-
-	var arazzo Arazzo
-	if err := marshaller.Unmarshal(ctx, &root, &arazzo); err != nil {
-		return nil, err
-	}
-
-	arazzo.Config = yml.GetConfigFromDoc(data, &root)
-
-	return &arazzo, nil
-}
-
-func (a *Arazzo) Marshal(ctx context.Context, w io.Writer) error {
-	cfg := yml.GetConfigFromContext(ctx)
-
-	switch cfg.OutputFormat {
-	case yml.OutputFormatYAML:
-		enc := yaml.NewEncoder(w)
-
-		enc.SetIndent(cfg.Indentation)
-		if err := enc.Encode(a.RootNode); err != nil {
-			return err
-		}
-	case yml.OutputFormatJSON:
-		if err := json.YAMLToJSON(a.RootNode, cfg.Indentation, w); err != nil {
-			return err
-		}
-	}
-
-	return nil
 }