Added ‘bundler’ module

resolves a v3 document or a model into a single document.

Signed-off-by: quobix <dave@quobix.com>

bundler/bundler.go

@@ -0,0 +1,70 @@
+// Copyright 2023-2024 Princess Beef Heavy Industries, LLC / Dave Shanley
+// https://pb33f.io
+// SPDX-License-Identifier: MIT
+
+package bundler
+
+import (
+	"errors"
+	"github.com/pb33f/libopenapi"
+	"github.com/pb33f/libopenapi/datamodel"
+	"github.com/pb33f/libopenapi/datamodel/high/v3"
+	"github.com/pb33f/libopenapi/index"
+)
+
+// BundleBytes will take a byte slice of an OpenAPI specification and return a bundled version of it.
+// This is useful for when you want to take a specification with external references, and you want to bundle it
+// into a single document.
+//
+// This function will 'resolve' all references in the specification and return a single document. The resulting
+// document will be a valid OpenAPI specification, containing no references.
+//
+// Circular references will not be resolved and will be skipped.
+func BundleBytes(bytes []byte, configuration *datamodel.DocumentConfiguration) ([]byte, error) {
+	doc, err := libopenapi.NewDocumentWithConfiguration(bytes, configuration)
+	if err != nil {
+		return nil, err
+	}
+
+	v3Doc, errs := doc.BuildV3Model()
+	err = errors.Join(errs...)
+
+	bundledBytes, e := BundleDocument(&v3Doc.Model)
+	return bundledBytes, errors.Join(err, e)
+}
+
+// BundleDocument will take a v3.Document and return a bundled version of it.
+// This is useful for when you want to take a document that has been built
+// from a specification with external references, and you want to bundle it
+// into a single document.
+//
+// This function will 'resolve' all references in the specification and return a single document. The resulting
+// document will be a valid OpenAPI specification, containing no references.
+//
+// Circular references will not be resolved and will be skipped.
+func BundleDocument(model *v3.Document) ([]byte, error) {
+	rolodex := model.Rolodex
+	compress := func(idx *index.SpecIndex) {
+		mappedReferences := idx.GetMappedReferences()
+		sequencedReferences := idx.GetRawReferencesSequenced()
+		for _, sequenced := range sequencedReferences {
+			mappedReference := mappedReferences[sequenced.FullDefinition]
+			if mappedReference != nil && !mappedReference.Circular {
+				sequenced.Node.Content = mappedReference.Node.Content
+			}
+			if mappedReference != nil && mappedReference.Circular {
+				if idx.GetLogger() != nil {
+					idx.GetLogger().Warn("[bundler] skipping circular reference",
+						"ref", sequenced.FullDefinition)
+				}
+			}
+		}
+	}
+
+	indexes := rolodex.GetIndexes()
+	compress(rolodex.GetRootIndex())
+	for _, idx := range indexes {
+		compress(idx)
+	}
+	return model.Render()
+}

bundler/bundler_test.go

@@ -0,0 +1,133 @@
+// Copyright 2023-2024 Princess Beef Heavy Industries, LLC / Dave Shanley
+// https://pb33f.io
+
+package bundler
+
+import (
+	"bytes"
+	"encoding/json"
+	"github.com/pb33f/libopenapi"
+	"github.com/pb33f/libopenapi/datamodel"
+	"github.com/stretchr/testify/assert"
+	"log"
+	"log/slog"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestBundleDocument_DigitalOcean(t *testing.T) {
+
+	// test the mother of all exploded specs.
+	tmp, _ := os.MkdirTemp("", "openapi")
+	cmd := exec.Command("git", "clone", "https://github.com/digitalocean/openapi", tmp)
+	defer os.RemoveAll(filepath.Join(tmp, "openapi"))
+
+	err := cmd.Run()
+	if err != nil {
+		log.Fatalf("cmd.Run() failed with %s\n", err)
+	}
+
+	spec, _ := filepath.Abs(filepath.Join(tmp+"/specification", "DigitalOcean-public.v2.yaml"))
+	digi, _ := os.ReadFile(spec)
+
+	doc, err := libopenapi.NewDocumentWithConfiguration([]byte(digi), &datamodel.DocumentConfiguration{
+		BasePath:                tmp + "/specification",
+		ExtractRefsSequentially: true,
+		Logger: slog.New(slog.NewJSONHandler(os.Stdout, &slog.HandlerOptions{
+			Level: slog.LevelWarn,
+		})),
+	})
+	if err != nil {
+		panic(err)
+	}
+
+	v3Doc, errs := doc.BuildV3Model()
+	if len(errs) > 0 {
+		panic(errs)
+	}
+
+	bytes, e := BundleDocument(&v3Doc.Model)
+
+	assert.NoError(t, e)
+	assert.False(t, strings.Contains("$ref", string(bytes)), "should not contain $ref")
+
+}
+
+func TestBundleDocument_Circular(t *testing.T) {
+
+	digi, _ := os.ReadFile("../test_specs/circular-tests.yaml")
+
+	var logs []byte
+	byteBuf := bytes.NewBuffer(logs)
+
+	config := &datamodel.DocumentConfiguration{
+		ExtractRefsSequentially: true,
+		Logger: slog.New(slog.NewJSONHandler(byteBuf, &slog.HandlerOptions{
+			Level: slog.LevelWarn,
+		})),
+	}
+	doc, err := libopenapi.NewDocumentWithConfiguration(digi, config)
+	if err != nil {
+		panic(err)
+	}
+
+	v3Doc, errs := doc.BuildV3Model()
+
+	// three circular ref issues.
+	assert.Len(t, errs, 3)
+
+	bytes, e := BundleDocument(&v3Doc.Model)
+	assert.NoError(t, e)
+	assert.Len(t, bytes, 3069)
+
+	logEntries := strings.Split(byteBuf.String(), "\n")
+
+	assert.Len(t, logEntries, 5)
+	for _, entry := range logEntries {
+		items := make(map[string]any)
+		if entry != "" {
+			_ = json.Unmarshal([]byte(entry), &items)
+			assert.Equal(t, "[bundler] skipping circular reference", items["msg"])
+		}
+	}
+	assert.NoError(t, e)
+}
+
+func TestBundleBytes(t *testing.T) {
+
+	digi, _ := os.ReadFile("../test_specs/circular-tests.yaml")
+
+	var logs []byte
+	byteBuf := bytes.NewBuffer(logs)
+
+	config := &datamodel.DocumentConfiguration{
+		ExtractRefsSequentially: true,
+		Logger: slog.New(slog.NewJSONHandler(byteBuf, &slog.HandlerOptions{
+			Level: slog.LevelWarn,
+		})),
+	}
+
+	bytes, e := BundleBytes(digi, config)
+	assert.Error(t, e)
+	assert.Len(t, bytes, 3069)
+
+	logEntries := strings.Split(byteBuf.String(), "\n")
+
+	assert.Len(t, logEntries, 5)
+	for _, entry := range logEntries {
+		items := make(map[string]any)
+		if entry != "" {
+			_ = json.Unmarshal([]byte(entry), &items)
+			assert.Equal(t, "[bundler] skipping circular reference", items["msg"])
+		}
+	}
+}
+
+func TestBundleBytes_Bad(t *testing.T) {
+	bytes, e := BundleBytes(nil, nil)
+	assert.Error(t, e)
+	assert.Nil(t, bytes)
+}

datamodel/document_config.go

@@ -101,6 +101,12 @@ type DocumentConfiguration struct {
 	// Logger is a structured logger that will be used for logging errors and warnings. If not set, a default logger
 	// will be used, set to the Error level.
 	Logger *slog.Logger
+
+	// ExtractRefsSequentially will extract all references sequentially, which means the index will look up references
+	// as it finds them, vs looking up everything asynchronously.
+	// This is a more thorough way of building the index, but it's slower. It's required building a document
+	// to be bundled.
+	ExtractRefsSequentially bool
 }
 
 func NewDocumentConfiguration() *DocumentConfiguration {

datamodel/low/v3/create_document.go

@@ -46,8 +46,9 @@ func createDocument(info *datamodel.SpecInfo, config *datamodel.DocumentConfigur
 	idxConfig.BaseURL = config.BaseURL
 	idxConfig.BasePath = config.BasePath
 	idxConfig.Logger = config.Logger
+	extract := config.ExtractRefsSequentially
+	idxConfig.ExtractRefsSequentially = extract
 	rolodex := index.NewRolodex(idxConfig)
-	//<-info.GetJSONParsingChannel() // Need to wait for JSON parsing to complete before we can index.
 	rolodex.SetRootNode(info.RootNode)
 	doc.Rolodex = rolodex
 

document.go

@@ -332,6 +332,7 @@ func (d *document) BuildV3Model() (*DocumentModel[v3high.Document], []error) {
 	}
 
 	highDoc := v3high.NewDocument(lowDoc)
+	highDoc.Rolodex = lowDoc.Index.GetRolodex()
 
 	d.highOpenAPI3Model = &DocumentModel[v3high.Document]{
 		Model: *highDoc,

index/extract_refs.go

@@ -571,7 +571,10 @@ func (index *SpecIndex) ExtractComponentsFromRefs(refs []*Reference) []*Referenc
 	var found []*Reference
 
 	// run this async because when things get recursive, it can take a while
-	c := make(chan bool)
+	var c chan bool
+	if !index.config.ExtractRefsSequentially {
+		c = make(chan bool)
+	}
 
 	locate := func(ref *Reference, refIndex int, sequence []*ReferenceMapped) {
 		located := index.FindComponent(ref.FullDefinition)
@@ -604,7 +607,9 @@ func (index *SpecIndex) ExtractComponentsFromRefs(refs []*Reference) []*Referenc
 			index.refErrors = append(index.refErrors, indexError)
 			index.errorLock.Unlock()
 		}
-		c <- true
+		if !index.config.ExtractRefsSequentially {
+			c <- true
+		}
 	}
 
 	var refsToCheck []*Reference
@@ -615,14 +620,20 @@ func (index *SpecIndex) ExtractComponentsFromRefs(refs []*Reference) []*Referenc
 
 	for r := range refsToCheck {
 		// expand our index of all mapped refs
-		go locate(refsToCheck[r], r, mappedRefsInSequence)
-		//locate(refsToCheck[r], r, mappedRefsInSequence) // used for sync testing.
+		if !index.config.ExtractRefsSequentially {
+			go locate(refsToCheck[r], r, mappedRefsInSequence) // run async
+		} else {
+			locate(refsToCheck[r], r, mappedRefsInSequence) // run synchronously
+		}
+
 	}
 
-	completedRefs := 0
-	for completedRefs < len(refsToCheck) {
-		<-c
-		completedRefs++
+	if !index.config.ExtractRefsSequentially {
+		completedRefs := 0
+		for completedRefs < len(refsToCheck) {
+			<-c
+			completedRefs++
+		}
 	}
 	for m := range mappedRefsInSequence {
 		if mappedRefsInSequence[m] != nil {

index/index_model.go

@@ -148,6 +148,12 @@ type SpecIndexConfig struct {
 	// the file is a JSON Schema. To allow JSON Schema files to be included set this to true.
 	SkipDocumentCheck bool
 
+	// ExtractRefsSequentially will extract all references sequentially, which means the index will look up references
+	// as it finds them, vs looking up everything asynchronously.
+	// This is a more thorough way of building the index, but it's slower. It's required building a document
+	// to be bundled.
+	ExtractRefsSequentially bool
+
 	// private fields
 	uri []string
 }

index/spec_index.go

@@ -230,6 +230,12 @@ func (index *SpecIndex) GetMappedReferences() map[string]*Reference {
 	return index.allMappedRefs
 }
 
+// GetRawReferencesSequenced returns a slice of every single reference found in the document, extracted raw from the doc
+// returned in the exact order they were found in the document.
+func (index *SpecIndex) GetRawReferencesSequenced() []*Reference {
+	return index.rawSequencedRefs
+}
+
 // GetMappedReferencesSequenced will return all references that were mapped successfully to nodes, performed in sequence
 // as they were read in from the document.
 func (index *SpecIndex) GetMappedReferencesSequenced() []*ReferenceMapped {