Another massive surgical strike with the rolodex and index reshuffle.

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

index/extract_refs.go

@@ -186,25 +186,48 @@ func (index *SpecIndex) ExtractRefs(node, parent *yaml.Node, seenPath []string,
 
 				segs := strings.Split(value, "/")
 				name := segs[len(segs)-1]
-				_, p := utils.ConvertComponentIdIntoFriendlyPathSearch(value)
+
+				var p string
+				uri := strings.Split(value, "#/")
+				if strings.HasPrefix(value, "http") || filepath.IsAbs(value) {
+					if len(uri) == 2 {
+						_, p = utils.ConvertComponentIdIntoFriendlyPathSearch(fmt.Sprintf("#/%s", uri[1]))
+					} else {
+						_, p = utils.ConvertComponentIdIntoFriendlyPathSearch(uri[0])
+					}
+				} else {
+					if len(uri) == 2 {
+						_, p = utils.ConvertComponentIdIntoFriendlyPathSearch(fmt.Sprintf("#/%s", uri[1]))
+					} else {
+						_, p = utils.ConvertComponentIdIntoFriendlyPathSearch(value)
+					}
+				}
 
 				// determine absolute path to this definition
 				iroot := filepath.Dir(index.specAbsolutePath)
-				uri := strings.Split(value, "#/")
 				var componentName string
 				var fullDefinitionPath string
 				if len(uri) == 2 {
 					if uri[0] == "" {
 						fullDefinitionPath = fmt.Sprintf("%s#/%s", index.specAbsolutePath, uri[1])
 					} else {
+
+						if strings.HasPrefix(uri[0], "http") {
+							fullDefinitionPath = value
+						} else {
 							abs, _ := filepath.Abs(filepath.Join(iroot, uri[0]))
 							fullDefinitionPath = fmt.Sprintf("%s#/%s", abs, uri[1])
 						}
+					}
 					componentName = fmt.Sprintf("#/%s", uri[1])
+				} else {
+					if strings.HasPrefix(uri[0], "http") {
+						fullDefinitionPath = value
 					} else {
 						fullDefinitionPath = fmt.Sprintf("%s#/%s", iroot, uri[0])
 						componentName = fmt.Sprintf("#/%s", uri[0])
 					}
+				}
 
 				ref := &Reference{
 					FullDefinition: fullDefinitionPath,
@@ -470,6 +493,7 @@ func (index *SpecIndex) ExtractComponentsFromRefs(refs []*Reference) []*Referenc
 		located := index.FindComponent(ref.FullDefinition, ref.Node)
 		if located != nil {
 			index.refLock.Lock()
+			// have we already mapped this?
 			if index.allMappedRefs[ref.Definition] == nil {
 				found = append(found, located)
 				index.allMappedRefs[ref.Definition] = located
@@ -478,8 +502,22 @@ func (index *SpecIndex) ExtractComponentsFromRefs(refs []*Reference) []*Referenc
 					Definition:     ref.Definition,
 					FullDefinition: ref.FullDefinition,
 				}
-
 				sequence[refIndex] = rm
+			} else {
+				// it exists, but is it a component with the same ID?
+				d := index.allMappedRefs[ref.Definition]
+
+				// if the full definition matches, we're good and can skip this.
+				if d.FullDefinition != ref.FullDefinition {
+					found = append(found, located)
+					index.allMappedRefs[ref.FullDefinition] = located
+					rm := &ReferenceMapped{
+						Reference:      located,
+						Definition:     ref.Definition,
+						FullDefinition: ref.FullDefinition,
+					}
+					sequence[refIndex] = rm
+				}
 			}
 			index.refLock.Unlock()
 		} else {

index/find_component.go

@@ -45,10 +45,24 @@ func (index *SpecIndex) FindComponent(componentId string, parent *yaml.Node) *Re
 
 	//witch DetermineReferenceResolveType(componentId) {
 	//case LocalResolve: // ideally, every single ref in every single spec is local. however, this is not the case.
-	//	return index.FindComponentInRoot(componentId)
+	//return index.FindComponentInRoot(componentId)
 
 	//case HttpResolve, FileResolve:
-	return index.performExternalLookup(strings.Split(componentId, "#/"))
+
+	uri := strings.Split(componentId, "#/")
+	if len(uri) == 2 {
+		if uri[0] != "" {
+			if index.specAbsolutePath == uri[0] {
+				return index.FindComponentInRoot(fmt.Sprintf("#/%s", uri[1]))
+			} else {
+				return index.lookupRolodex(uri)
+			}
+		} else {
+			return index.FindComponentInRoot(fmt.Sprintf("#/%s", uri[1]))
+		}
+	} else {
+		return index.FindComponentInRoot(fmt.Sprintf("#/%s", uri[0]))
+	}
 
 	//}
 	//return nil
@@ -326,23 +340,26 @@ func FindComponent(root *yaml.Node, componentId, absoluteFilePath string) *Refer
 	return nil
 }
 
-//func (index *SpecIndex) FindComponentInRoot(componentId string) *Reference {
-//	if index.root != nil {
-//		return FindComponent(index.root, componentId, )
-//	}
-//	return nil
-//}
+func (index *SpecIndex) FindComponentInRoot(componentId string) *Reference {
+	if index.root != nil {
+		return FindComponent(index.root, componentId, index.specAbsolutePath)
+	}
+	return nil
+}
 
-func (index *SpecIndex) performExternalLookup(uri []string) *Reference {
+func (index *SpecIndex) lookupRolodex(uri []string) *Reference {
 
 	if len(uri) > 0 {
 
 		// split string to remove file reference
 		file := strings.ReplaceAll(uri[0], "file:", "")
-		fileName := filepath.Base(file)
 
-		var absoluteFileLocation string
-		if filepath.IsAbs(file) {
+		var absoluteFileLocation, fileName string
+
+		// is this a local or a remote file?
+
+		fileName = filepath.Base(file)
+		if filepath.IsAbs(file) || strings.HasPrefix(file, "http") {
 			absoluteFileLocation = file
 		} else {
 			absoluteFileLocation, _ = filepath.Abs(filepath.Join(filepath.Dir(index.specAbsolutePath), file))
@@ -363,6 +380,9 @@ func (index *SpecIndex) performExternalLookup(uri []string) *Reference {
 				return nil
 			}
 
+			if rFile == nil {
+				panic("FUCK")
+			}
 			parsedDocument, err = rFile.GetContentAsYAMLNode()
 			if err != nil {
 				logger.Error("unable to parse rolodex file", "file", absoluteFileLocation, "error", err)

index/find_component_test.go

@@ -50,8 +50,6 @@ func TestSpecIndex_CheckCircularIndex(t *testing.T) {
 	rolo.SetRootNode(&rootNode)
 	cf.Rolodex = rolo
 
-	// TODO: pick up here.
-
 	fsCfg := LocalFSConfig{
 		BaseDirectory: cf.BasePath,
 		FileFilters:   []string{"first.yaml", "second.yaml", "third.yaml", "fourth.yaml"},

index/index_model.go

@@ -96,8 +96,8 @@ type SpecIndexConfig struct {
 	// exploits, but it's better to be safe than sorry.
 	//
 	// To read more about this, you can find a discussion here: https://github.com/pb33f/libopenapi/pull/64
-	//AllowRemoteLookup bool // Allow remote lookups for references. Defaults to false
-	//AllowFileLookup   bool // Allow file lookups for references. Defaults to false
+	AllowRemoteLookup bool // Allow remote lookups for references. Defaults to false
+	AllowFileLookup   bool // Allow file lookups for references. Defaults to false
 
 	// ParentIndex allows the index to be created with knowledge of a parent, before being parsed. This allows
 	// a breakglass to be used to prevent loops, checking the tree before cursing down.
@@ -280,6 +280,8 @@ type SpecIndex struct {
 	specAbsolutePath string
 	resolver         *Resolver
 
+	built bool
+
 	//parentIndex *SpecIndex
 	uri []string
 	//children    []*SpecIndex

index/resolver.go

@@ -269,7 +269,7 @@ func (resolver *Resolver) VisitReference(ref *Reference, seen map[string]bool, j
 			if j.Definition == r.Definition {
 
 				var foundDup *Reference
-				foundRefs := resolver.specIndex.SearchIndexForReference(r.Definition)
+				foundRefs := resolver.specIndex.SearchIndexForReferenceByReference(r)
 				if len(foundRefs) > 0 {
 					foundDup = foundRefs[0]
 				}
@@ -311,7 +311,7 @@ func (resolver *Resolver) VisitReference(ref *Reference, seen map[string]bool, j
 
 		if !skip {
 			var original *Reference
-			foundRefs := resolver.specIndex.SearchIndexForReference(r.Definition)
+			foundRefs := resolver.specIndex.SearchIndexForReferenceByReference(r)
 			if len(foundRefs) > 0 {
 				original = foundRefs[0]
 			}
@@ -408,8 +408,27 @@ func (resolver *Resolver) extractRelatives(ref *Reference, node, parent *yaml.No
 				}
 
 				value := node.Content[i+1].Value
+				var locatedRef []*Reference
+				searchRef := &Reference{
+					Definition:     value,
+					FullDefinition: ref.FullDefinition,
+					RemoteLocation: ref.RemoteLocation,
+					IsRemote:       true,
+				}
 
-				locatedRef := resolver.specIndex.SearchIndexForReference(value)
+				// we're searching a remote document, we need to build a full path to the reference
+				if ref.IsRemote {
+					if ref.RemoteLocation != "" {
+						searchRef = &Reference{
+							Definition:     value,
+							FullDefinition: fmt.Sprintf("%s%s", ref.RemoteLocation, value),
+							RemoteLocation: ref.RemoteLocation,
+							IsRemote:       true,
+						}
+					}
+				}
+
+				locatedRef = resolver.specIndex.SearchIndexForReferenceByReference(searchRef)
 
 				if locatedRef == nil {
 					_, path := utils.ConvertComponentIdIntoFriendlyPathSearch(value)

index/resolver_test.go

@@ -407,7 +407,7 @@ func TestResolver_ResolveComponents_Stripe(t *testing.T) {
 
 	resolveFile, _ := os.ReadFile(baseDir)
 
-	info, err := datamodel.ExtractSpecInfoWithDocumentCheck(resolveFile, true)
+	info, _ := datamodel.ExtractSpecInfoWithDocumentCheck(resolveFile, true)
 
 	fileFS, err := NewLocalFS(baseDir, os.DirFS(filepath.Dir(baseDir)))
 	if err != nil {

index/rolodex.go

@@ -5,6 +5,7 @@ package index
 
 import (
 	"errors"
+	"fmt"
 	"github.com/pb33f/libopenapi/datamodel"
 	"gopkg.in/yaml.v3"
 	"io"
@@ -17,9 +18,12 @@ import (
 	"time"
 )
 
+type HasIndex interface {
+	GetIndex() *SpecIndex
+}
+
 type CanBeIndexed interface {
 	Index(config *SpecIndexConfig) (*SpecIndex, error)
-	GetIndex() *SpecIndex
 }
 
 type RolodexFile interface {
@@ -86,10 +90,10 @@ func (rf *rolodexFile) Name() string {
 
 func (rf *rolodexFile) GetIndex() *SpecIndex {
 	if rf.localFile != nil {
-		return rf.localFile.index
+		return rf.localFile.GetIndex()
 	}
 	if rf.remoteFile != nil {
-		// TODO: remote file index
+		return rf.remoteFile.GetIndex()
 	}
 	return nil
 }
@@ -207,7 +211,6 @@ func (rf *rolodexFile) GetErrors() []error {
 }
 
 func NewRolodex(indexConfig *SpecIndexConfig) *Rolodex {
-
 	r := &Rolodex{
 		indexConfig: indexConfig,
 		localFS:     make(map[string]fs.FS),
@@ -304,16 +307,23 @@ func (r *Rolodex) IndexTheRolodex() error {
 			indexChan <- idx
 		}
 
-		if lfs, ok := fs.(*LocalFS); ok {
-			for _, f := range lfs.Files {
+		if lfs, ok := fs.(RolodexFS); ok {
+			wait := false
+			for _, f := range lfs.GetFiles() {
 				if idxFile, ko := f.(CanBeIndexed); ko {
 					wg.Add(1)
+					wait = true
 					go indexFileFunc(idxFile, f.GetFullPath())
 				}
 			}
+			if wait {
 				wg.Wait()
+			}
 			doneChan <- true
 			return
+		} else {
+			errChan <- errors.New("rolodex file system is not a RolodexFS")
+			doneChan <- true
 		}
 	}
 
@@ -440,15 +450,12 @@ func (r *Rolodex) Open(location string) (RolodexFile, error) {
 	var errorStack []error
 
 	var localFile *LocalFile
-	//var remoteFile *RemoteFile
+	var remoteFile *RemoteFile
 
 	if r == nil || r.localFS == nil && r.remoteFS == nil {
 		panic("WHAT NO....")
 	}
 
-	for k, v := range r.localFS {
-
-		// check if this is a URL or an abs/rel reference.
 	fileLookup := location
 	isUrl := false
 	u, _ := url.Parse(location)
@@ -456,8 +463,12 @@ func (r *Rolodex) Open(location string) (RolodexFile, error) {
 		isUrl = true
 	}
 
-		// TODO handle URLs.
 	if !isUrl {
+
+		for k, v := range r.localFS {
+
+			// check if this is a URL or an abs/rel reference.
+
 			if !filepath.IsAbs(location) {
 				fileLookup, _ = filepath.Abs(filepath.Join(k, location))
 			}
@@ -504,8 +515,52 @@ func (r *Rolodex) Open(location string) (RolodexFile, error) {
 					break
 				}
 			}
+
+		}
+	} else {
+
+		if !r.indexConfig.AllowRemoteLookup {
+			return nil, fmt.Errorf("remote lookup for '%s' not allowed, please set the index configuration to "+
+				"AllowRemoteLookup to true", fileLookup)
+		}
+
+		for _, v := range r.remoteFS {
+			f, err := v.Open(fileLookup)
+			if err == nil {
+
+				if rf, ok := interface{}(f).(*RemoteFile); ok {
+					remoteFile = rf
+					break
+				} else {
+
+					bytes, rErr := io.ReadAll(f)
+					if rErr != nil {
+						errorStack = append(errorStack, rErr)
+						continue
+					}
+					s, sErr := f.Stat()
+					if sErr != nil {
+						errorStack = append(errorStack, sErr)
+						continue
+					}
+					if len(bytes) > 0 {
+						remoteFile = &RemoteFile{
+							filename:     filepath.Base(fileLookup),
+							name:         filepath.Base(fileLookup),
+							extension:    ExtractFileType(fileLookup),
+							data:         bytes,
+							fullPath:     fileLookup,
+							lastModified: s.ModTime(),
+							index:        r.rootIndex,
+						}
+						break
 					}
 				}
+			}
+
+		}
+	}
+
 	if localFile != nil {
 		return &rolodexFile{
 			rolodex:   r,
@@ -514,5 +569,13 @@ func (r *Rolodex) Open(location string) (RolodexFile, error) {
 		}, errors.Join(errorStack...)
 	}
 
+	if remoteFile != nil {
+		return &rolodexFile{
+			rolodex:    r,
+			location:   remoteFile.fullPath,
+			remoteFile: remoteFile,
+		}, errors.Join(errorStack...)
+	}
+
 	return nil, errors.Join(errorStack...)
 }

index/rolodex_file_loader.go

@@ -20,10 +20,8 @@ type LocalFS struct {
 	entryPointDirectory string
 	baseDirectory       string
 	Files               map[string]RolodexFile
-	parseTime           int64
 	logger              *slog.Logger
 	readingErrors       []error
-	filters             []string
 }
 
 func (l *LocalFS) GetFiles() map[string]RolodexFile {
@@ -180,26 +178,23 @@ func NewLocalFSWithConfig(config *LocalFSConfig) (*LocalFS, error) {
 		switch extension {
 		case YAML, JSON:
 
-			file, readErr := config.DirFS.Open(p)
+			dirFile, readErr := config.DirFS.Open(p)
 			modTime := time.Now()
 			if readErr != nil {
-				readingErrors = append(readingErrors, readErr)
 				allErrors = append(allErrors, readErr)
 				logger.Error("[rolodex] cannot open file: ", "file", abs, "error", readErr.Error())
 				return nil
 			}
-			stat, statErr := file.Stat()
+			stat, statErr := dirFile.Stat()
 			if statErr != nil {
-				readingErrors = append(readingErrors, statErr)
 				allErrors = append(allErrors, statErr)
 				logger.Error("[rolodex] cannot stat file: ", "file", abs, "error", statErr.Error())
 			}
 			if stat != nil {
 				modTime = stat.ModTime()
 			}
-			fileData, readErr = io.ReadAll(file)
+			fileData, readErr = io.ReadAll(dirFile)
 			if readErr != nil {
-				readingErrors = append(readingErrors, readErr)
 				allErrors = append(allErrors, readErr)
 				logger.Error("cannot read file data: ", "file", abs, "error", readErr.Error())
 				return nil

index/rolodex_ref_extractor.go

@@ -12,17 +12,6 @@ import (
 // var refRegex = regexp.MustCompile(`['"]?\$ref['"]?\s*:\s*['"]?([^'"]*?)['"]`)
 var refRegex = regexp.MustCompile(`('\$ref'|"\$ref"|\$ref)\s*:\s*('[^']*'|"[^"]*"|\S*)`)
 
-/*
-r := regexp.MustCompile(`('\$ref'|"\$ref"|\$ref)\s*:\s*('[^']*'|"[^"]*"|\S*)`)
-	matches := r.FindAllStringSubmatch(text, -1)
-	for _, submatches := range matches {
-		if len(submatches) > 2 {
-			fmt.Println("Full match:", submatches[0])
-			fmt.Println("JSON Schema reference: ", submatches[2])
-		}
-	}
-*/
-
 type RefType int
 
 const (
@@ -108,12 +97,12 @@ func ExtractRefType(ref string) RefType {
 
 func ExtractRefs(content string) [][]string {
 
-	res := refRegex.FindAllStringSubmatch(content, -1)
+	return refRegex.FindAllStringSubmatch(content, -1)
+
+	//var results []*ExtractedRef
+	//for _, r := range res {
+	//	results = append(results, &ExtractedRef{Location: r[1], Type: ExtractRefType(r[1])})
+	//}
 
-	var results []*ExtractedRef
-	for _, r := range res {
-		results = append(results, &ExtractedRef{Location: r[1], Type: ExtractRefType(r[1])})
-	}
 
-	return res
 }

index/rolodex_remote_loader.go

@@ -6,6 +6,7 @@ package index
 import (
 	"errors"
 	"fmt"
+	"github.com/pb33f/libopenapi/datamodel"
 	"golang.org/x/exp/slog"
 	"golang.org/x/sync/syncmap"
 	"gopkg.in/yaml.v3"
@@ -19,6 +20,7 @@ import (
 )
 
 type RemoteFS struct {
+	indexConfig       *SpecIndexConfig
 	rootURL           string
 	rootURLParsed     *url.URL
 	RemoteHandlerFunc RemoteURLHandler
@@ -41,6 +43,9 @@ type RemoteFile struct {
 	URL           *url.URL
 	lastModified  time.Time
 	seekingErrors []error
+	index         *SpecIndex
+	parsed        *yaml.Node
+	offset        int64
 }
 
 func (f *RemoteFile) GetFileName() string {
@@ -52,7 +57,25 @@ func (f *RemoteFile) GetContent() string {
 }
 
 func (f *RemoteFile) GetContentAsYAMLNode() (*yaml.Node, error) {
-	return nil, errors.New("not implemented")
+	if f.parsed != nil {
+		return f.parsed, nil
+	}
+	if f.index != nil && f.index.root != nil {
+		return f.index.root, nil
+	}
+	if f.data == nil {
+		return nil, fmt.Errorf("no data to parse for file: %s", f.fullPath)
+	}
+	var root yaml.Node
+	err := yaml.Unmarshal(f.data, &root)
+	if err != nil {
+		return nil, err
+	}
+	if f.index != nil && f.index.root == nil {
+		f.index.root = &root
+	}
+	f.parsed = &root
+	return &root, nil
 }
 
 func (f *RemoteFile) GetFileExtension() FileExtension {
@@ -71,6 +94,8 @@ func (f *RemoteFile) GetFullPath() string {
 	return f.fullPath
 }
 
+// fs.FileInfo interfaces
+
 func (f *RemoteFile) Name() string {
 	return f.name
 }
@@ -91,40 +116,52 @@ func (f *RemoteFile) IsDir() bool {
 	return false
 }
 
+// fs.File interfaces
+
 func (f *RemoteFile) Sys() interface{} {
 	return nil
 }
 
-func (f *RemoteFile) Index(config *SpecIndexConfig) (*SpecIndex, error) {
-
-	// TODO
-	return nil, nil
-}
-func (f *RemoteFile) GetIndex() *SpecIndex {
-
-	// TODO
+func (f *RemoteFile) Close() error {
 	return nil
 }
-
-type remoteRolodexFile struct {
-	f      *RemoteFile
-	offset int64
+func (f *RemoteFile) Stat() (fs.FileInfo, error) {
+	return f, nil
 }
-
-func (f *remoteRolodexFile) Close() error               { return nil }
-func (f *remoteRolodexFile) Stat() (fs.FileInfo, error) { return f.f, nil }
-func (f *remoteRolodexFile) Read(b []byte) (int, error) {
-	if f.offset >= int64(len(f.f.data)) {
+func (f *RemoteFile) Read(b []byte) (int, error) {
+	if f.offset >= int64(len(f.data)) {
 		return 0, io.EOF
 	}
 	if f.offset < 0 {
-		return 0, &fs.PathError{Op: "read", Path: f.f.name, Err: fs.ErrInvalid}
+		return 0, &fs.PathError{Op: "read", Path: f.name, Err: fs.ErrInvalid}
 	}
-	n := copy(b, f.f.data[f.offset:])
+	n := copy(b, f.data[f.offset:])
 	f.offset += int64(n)
 	return n, nil
 }
 
+func (f *RemoteFile) Index(config *SpecIndexConfig) (*SpecIndex, error) {
+
+	if f.index != nil {
+		return f.index, nil
+	}
+	content := f.data
+
+	// first, we must parse the content of the file
+	info, err := datamodel.ExtractSpecInfoWithDocumentCheck(content, true)
+	if err != nil {
+		return nil, err
+	}
+
+	index := NewSpecIndexWithConfig(info.RootNode, config)
+	index.specAbsolutePath = f.fullPath
+	f.index = index
+	return index, nil
+}
+func (f *RemoteFile) GetIndex() *SpecIndex {
+	return f.index
+}
+
 type FileExtension int
 
 const (
@@ -133,19 +170,39 @@ const (
 	UNSUPPORTED
 )
 
-func NewRemoteFS(rootURL string) (*RemoteFS, error) {
+func NewRemoteFSWithConfig(specIndexConfig *SpecIndexConfig) (*RemoteFS, error) {
+	remoteRootURL := specIndexConfig.BaseURL
+	rfs :=  &RemoteFS{
+		logger: slog.New(slog.NewJSONHandler(os.Stdout, &slog.HandlerOptions{
+			Level: slog.LevelDebug,
+		})),
+
+		rootURLParsed: remoteRootURL,
+		FetchChannel:  make(chan *RemoteFile),
+	}
+	if remoteRootURL != nil {
+	  rfs.rootURL = remoteRootURL.String()
+	}
+	return rfs, nil
+}
+
+func NewRemoteFS() (*RemoteFS, error) {
+	config := CreateOpenAPIIndexConfig()
+	return NewRemoteFSWithConfig(config)
+}
+
+func NewRemoteFSWithRootURL(rootURL string) (*RemoteFS, error) {
 	remoteRootURL, err := url.Parse(rootURL)
 	if err != nil {
 		return nil, err
 	}
-	return &RemoteFS{
-		logger: slog.New(slog.NewJSONHandler(os.Stdout, &slog.HandlerOptions{
-			Level: slog.LevelDebug,
-		})),
-		rootURL:       rootURL,
-		rootURLParsed: remoteRootURL,
-		FetchChannel:  make(chan *RemoteFile),
-	}, nil
+	config := CreateOpenAPIIndexConfig()
+	config.BaseURL = remoteRootURL
+	return NewRemoteFSWithConfig(config)
+}
+
+func (i *RemoteFS) SetIndexConfig(config *SpecIndexConfig) {
+	i.indexConfig = config
 }
 
 func (i *RemoteFS) GetFiles() map[string]RolodexFile {
@@ -200,7 +257,7 @@ func (i *RemoteFS) seekRelatives(file *RemoteFile) {
 			fmt.Printf("Found relative HTTP reference: %s\n", ref[1])
 		}
 	}
-	if i.remoteRunning == false {
+	if !i.remoteRunning {
 		i.remoteRunning = true
 		i.remoteWg.Wait()
 		i.remoteRunning = false
@@ -215,15 +272,29 @@ func (i *RemoteFS) Open(remoteURL string) (fs.File, error) {
 		return nil, err
 	}
 
+	remoteParsedOrig, _ := url.Parse(remoteURL)
+
+	// try path first
+	if r, ok := i.Files.Load(remoteParsedURL.Path); ok {
+		return r.(*RemoteFile), nil
+	}
+
 	fileExt := ExtractFileType(remoteParsedURL.Path)
 
 	if fileExt == UNSUPPORTED {
 		return nil, &fs.PathError{Op: "open", Path: remoteURL, Err: fs.ErrInvalid}
 	}
 
-	i.logger.Debug("Loading remote file", "file", remoteParsedURL.Path)
+	// if the remote URL is absolute (http:// or https://), and we have a rootURL defined, we need to override
+	// the host being defined by this URL, and use the rootURL instead, but keep the path.
+	if i.rootURLParsed != nil && remoteParsedURL.Host != "" {
+		remoteParsedURL.Host = i.rootURLParsed.Host
+		remoteParsedURL.Scheme = i.rootURLParsed.Scheme
+	}
 
-	response, clientErr := i.RemoteHandlerFunc(i.rootURL + remoteURL)
+	i.logger.Debug("Loading remote file", "file", remoteURL, "remoteURL", remoteParsedURL.String())
+
+	response, clientErr := i.RemoteHandlerFunc(remoteParsedURL.String())
 	if clientErr != nil {
 		i.logger.Error("client error", "error", response.StatusCode)
 
@@ -238,7 +309,7 @@ func (i *RemoteFS) Open(remoteURL string) (fs.File, error) {
 	if response.StatusCode >= 400 {
 		i.logger.Error("Unable to fetch remote document %s",
 			"file", remoteParsedURL.Path, "status", response.StatusCode, "resp", string(responseBytes))
-		return nil, errors.New(fmt.Sprintf("Unable to fetch remote document: %s", string(responseBytes)))
+		return nil, fmt.Errorf("unable to fetch remote document: %s", string(responseBytes))
 	}
 
 	absolutePath, pathErr := filepath.Abs(remoteParsedURL.Path)
@@ -253,10 +324,12 @@ func (i *RemoteFS) Open(remoteURL string) (fs.File, error) {
 	lastModifiedTime, parseErr := time.Parse(time.RFC1123, lastModified)
 
 	if parseErr != nil {
-		return nil, parseErr
+		// can't extract last modified, so use now
+		lastModifiedTime = time.Now()
 	}
 
 	filename := filepath.Base(remoteParsedURL.Path)
+
 	remoteFile := &RemoteFile{
 		filename:     filename,
 		name:         remoteParsedURL.Path,
@@ -266,14 +339,31 @@ func (i *RemoteFS) Open(remoteURL string) (fs.File, error) {
 		URL:          remoteParsedURL,
 		lastModified: lastModifiedTime,
 	}
+
+	copiedCfg := *i.indexConfig
+
+	newBase := fmt.Sprintf("%s://%s%s", remoteParsedOrig.Scheme, remoteParsedOrig.Host,
+		filepath.Dir(remoteParsedOrig.Path))
+	newBaseURL, _ := url.Parse(newBase)
+
+	copiedCfg.BaseURL = newBaseURL
+	copiedCfg.SpecAbsolutePath = remoteURL
+	idx, _ := remoteFile.Index(&copiedCfg)
+
+	// for each index, we need a resolver
+	resolver := NewResolver(idx)
+	idx.resolver = resolver
+
 	i.Files.Store(absolutePath, remoteFile)
 
 	i.logger.Debug("successfully loaded file", "file", absolutePath)
 	i.seekRelatives(remoteFile)
 
-	if i.remoteRunning == false {
-		return &remoteRolodexFile{remoteFile, 0}, errors.Join(i.remoteErrors...)
+	idx.BuildIndex()
+
+	if !i.remoteRunning {
+		return remoteFile, errors.Join(i.remoteErrors...)
 	} else {
-		return &remoteRolodexFile{remoteFile, 0}, nil
+		return remoteFile, nil
 	}
 }

index/rolodex_remote_loader_test.go

@@ -86,7 +86,7 @@ func TestNewRemoteFS_BasicCheck(t *testing.T) {
     defer server.Close()
 
     //remoteFS := NewRemoteFS("https://raw.githubusercontent.com/digitalocean/openapi/main/specification/")
-	remoteFS, _ := NewRemoteFS(server.URL)
+    remoteFS, _ := NewRemoteFSWithRootURL(server.URL)
     remoteFS.RemoteHandlerFunc = test_httpClient.Get
 
     file, err := remoteFS.Open("/file1.yaml")
@@ -112,7 +112,7 @@ func TestNewRemoteFS_BasicCheck_Relative(t *testing.T) {
     server := test_buildServer()
     defer server.Close()
 
-	remoteFS, _ := NewRemoteFS(server.URL)
+    remoteFS, _ := NewRemoteFSWithRootURL(server.URL)
     remoteFS.RemoteHandlerFunc = test_httpClient.Get
 
     file, err := remoteFS.Open("/deeper/file2.yaml")
@@ -138,7 +138,7 @@ func TestNewRemoteFS_BasicCheck_Relative_Deeper(t *testing.T) {
     server := test_buildServer()
     defer server.Close()
 
-	remoteFS, _ := NewRemoteFS(server.URL)
+    remoteFS, _ := NewRemoteFSWithRootURL(server.URL)
     remoteFS.RemoteHandlerFunc = test_httpClient.Get
 
     file, err := remoteFS.Open("/deeper/even_deeper/file3.yaml")
@@ -164,7 +164,7 @@ func TestNewRemoteFS_BasicCheck_SeekRelatives(t *testing.T) {
     server := test_buildServer()
     defer server.Close()
 
-	remoteFS, _ := NewRemoteFS(server.URL)
+    remoteFS, _ := NewRemoteFSWithRootURL(server.URL)
     remoteFS.RemoteHandlerFunc = test_httpClient.Get
 
     file, err := remoteFS.Open("/bag/list.yaml")

index/search_index.go

@@ -9,10 +9,9 @@ import (
 	"strings"
 )
 
-// SearchIndexForReference searches the index for a reference, first looking through the mapped references
-// and then externalSpecIndex for a match. If no match is found, it will recursively search the child indexes
-// extracted when parsing the OpenAPI Spec.
-func (index *SpecIndex) SearchIndexForReference(ref string) []*Reference {
+func (index *SpecIndex) SearchIndexForReferenceByReference(fullRef *Reference) []*Reference {
+
+	ref := fullRef.FullDefinition
 
 	absPath := index.specAbsolutePath
 	if absPath == "" {
@@ -22,8 +21,12 @@ func (index *SpecIndex) SearchIndexForReference(ref string) []*Reference {
 	uri := strings.Split(ref, "#/")
 	if len(uri) == 2 {
 		if uri[0] != "" {
+			if strings.HasPrefix(uri[0], "http") {
+				roloLookup = fullRef.FullDefinition
+			} else {
 				roloLookup, _ = filepath.Abs(filepath.Join(absPath, uri[0]))
 			}
+		}
 		ref = fmt.Sprintf("#/%s", uri[1])
 	} else {
 		roloLookup, _ = filepath.Abs(filepath.Join(absPath, uri[0]))
@@ -38,7 +41,6 @@ func (index *SpecIndex) SearchIndexForReference(ref string) []*Reference {
 		return []*Reference{r}
 	}
 
-	// TODO: look in the rolodex.
 	if roloLookup != "" {
 		rFile, err := index.rolodex.Open(roloLookup)
 		if err != nil {
@@ -70,33 +72,17 @@ func (index *SpecIndex) SearchIndexForReference(ref string) []*Reference {
 		}
 	}
 
-	panic("should not be here")
-	fmt.Println(roloLookup)
-	return nil
-
-	//if r, ok := index.allMappedRefs[ref]; ok {
-	//	return []*Reference{r}jh
-	//}
-	//for c := range index.children {
-	//	found := goFindMeSomething(index.children[c], ref)
-	//	if found != nil {
-	//		return found
-	//	}
-	//}
-	//return nil
-}
-
-func (index *SpecIndex) SearchAncestryForSeenURI(uri string) *SpecIndex {
-	//if index.parentIndex == nil {
-	//	return nil
-	//}
-	//if index.uri[0] != uri {
-	//	return index.parentIndex.SearchAncestryForSeenURI(uri)
-	//}
-	//return index
+	fmt.Printf("unable to locate reference: %s, within index: %s\n", ref, index.specAbsolutePath)
 	return nil
 }
 
-func goFindMeSomething(i *SpecIndex, ref string) []*Reference {
-	return i.SearchIndexForReference(ref)
+// SearchIndexForReference searches the index for a reference, first looking through the mapped references
+// and then externalSpecIndex for a match. If no match is found, it will recursively search the child indexes
+// extracted when parsing the OpenAPI Spec.
+func (index *SpecIndex) SearchIndexForReference(ref string) []*Reference {
+	return index.SearchIndexForReferenceByReference(&Reference{FullDefinition: ref})
+}
+
+func (index *SpecIndex) SearchIndexForReferenceWithParent(ref string, reference *Reference) []*Reference {
+	return index.SearchIndexForReferenceByReference(&Reference{FullDefinition: ref})
 }

index/spec_index.go

@@ -103,6 +103,9 @@ func createNewIndex(rootNode *yaml.Node, index *SpecIndex, avoidBuildOut bool) *
 // useful for looking up things, the count operations are all run in parallel and then the final calculations are run
 // the index is ready.
 func (index *SpecIndex) BuildIndex() {
+	if index.built {
+		return
+	}
 	countFuncs := []func() int{
 		index.GetOperationCount,
 		index.GetComponentSchemaCount,
@@ -132,6 +135,7 @@ func (index *SpecIndex) BuildIndex() {
 	index.GetInlineDuplicateParamCount()
 	index.GetAllDescriptionsCount()
 	index.GetTotalTagsCount()
+	index.built = true
 }
 
 // GetRootNode returns document root node.
@@ -998,7 +1002,6 @@ func (index *SpecIndex) GetOperationCount() int {
 						}
 					}
 					if valid {
-						fmt.Sprint(p)
 						ref := &Reference{
 							Definition: m.Value,
 							Name:       m.Value,

index/spec_index_test.go

@@ -6,6 +6,8 @@ package index
 import (
     "fmt"
     "log"
+    "net/http"
+    "net/http/httptest"
     "net/url"
     "os"
     "os/exec"
@@ -433,18 +435,81 @@ func TestSpecIndex_NoRoot(t *testing.T) {
     assert.Equal(t, -1, index.GetGlobalLinksCount())
 }
 
+func test_buildMixedRefServer() *httptest.Server {
+
+    bs, _ := os.ReadFile("../test_specs/burgershop.openapi.yaml")
+    return httptest.NewServer(http.HandlerFunc(func(rw http.ResponseWriter, req *http.Request) {
+        if req.URL.String() == "/daveshanley/vacuum/main/model/test_files/burgershop.openapi.yaml" {
+            rw.Header().Set("Last-Modified", "Wed, 21 Oct 2015 07:28:00 GMT")
+            _, _ = rw.Write(bs)
+            return
+        }
+
+        _, _ = rw.Write([]byte(`OK`))
+    }))
+}
+
 func TestSpecIndex_BurgerShopMixedRef(t *testing.T) {
-	spec, _ := os.ReadFile("../test_specs/mixedref-burgershop.openapi.yaml")
+
+    // create a test server.
+    server := test_buildMixedRefServer()
+    defer server.Close()
+
+    // create a new config that allows local and remote to be mixed up.
+    cf := CreateOpenAPIIndexConfig()
+    cf.AvoidBuildIndex = true
+    cf.AllowRemoteLookup = true
+    cf.AvoidCircularReferenceCheck = true
+    cf.BasePath = "../test_specs"
+
+    // setting this baseURL will override the base
+    //cf.BaseURL, _ = url.Parse(server.URL)
+
+    cFile := "../test_specs/mixedref-burgershop.openapi.yaml"
+    yml, _ := os.ReadFile(cFile)
     var rootNode yaml.Node
-	_ = yaml.Unmarshal(spec, &rootNode)
+    _ = yaml.Unmarshal([]byte(yml), &rootNode)
 
-	cwd, _ := os.Getwd()
+    // create a new rolodex
+    rolo := NewRolodex(cf)
 
-	index := NewSpecIndexWithConfig(&rootNode, &SpecIndexConfig{
-		//AllowRemoteLookup: true,
-		//	AllowFileLookup:   true,
-		BasePath: cwd,
-	})
+    // set the rolodex root node to the root node of the spec.
+    rolo.SetRootNode(&rootNode)
+
+    // create a new remote fs and set the config for indexing.
+    //remoteFS, _ := NewRemoteFSWithRootURL(server.URL)
+    remoteFS, _ := NewRemoteFS()
+    remoteFS.SetIndexConfig(cf)
+
+    // set our remote handler func
+
+    c := http.Client{}
+
+    remoteFS.RemoteHandlerFunc = c.Get
+
+    // configure the local filesystem.
+    fsCfg := LocalFSConfig{
+        BaseDirectory: cf.BasePath,
+        FileFilters:   []string{"burgershop.openapi.yaml"},
+        DirFS:         os.DirFS(cf.BasePath),
+    }
+
+    // create a new local filesystem.
+    fileFS, err := NewLocalFSWithConfig(&fsCfg)
+    assert.NoError(t, err)
+
+    // add file systems to the rolodex
+    rolo.AddLocalFS(cf.BasePath, fileFS)
+    rolo.AddRemoteFS(server.URL, remoteFS)
+
+    // index the rolodex.
+    indexedErr := rolo.IndexTheRolodex()
+    rolo.BuildIndexes()
+
+    assert.NoError(t, indexedErr)
+
+    index := rolo.GetRootIndex()
+    rolo.CheckForCircularReferences()
 
     assert.Len(t, index.allRefs, 5)
     assert.Len(t, index.allMappedRefs, 5)
@@ -459,6 +524,8 @@ func TestSpecIndex_BurgerShopMixedRef(t *testing.T) {
     assert.Equal(t, 2, index.GetOperationsParameterCount())
     assert.Equal(t, 1, index.GetInlineDuplicateParamCount())
     assert.Equal(t, 1, index.GetInlineUniqueParamCount())
+    assert.Len(t, index.refErrors, 0)
+    assert.Len(t, index.GetCircularReferences(), 0)
 }
 
 func TestSpecIndex_TestEmptyBrokenReferences(t *testing.T) {
@@ -630,7 +697,7 @@ func TestSpecIndex_TestPathsNodeAsArray(t *testing.T) {
     _ = yaml.Unmarshal([]byte(yml), &rootNode)
 
     index := NewSpecIndexWithConfig(&rootNode, CreateOpenAPIIndexConfig())
-	assert.Nil(t, index.performExternalLookup(nil))
+    assert.Nil(t, index.lookupRolodex(nil))
 }
 
 func TestSpecIndex_lookupRemoteReference_SeenSourceSimulation_Error(t *testing.T) {

index/utility_methods.go

@@ -441,9 +441,5 @@ func GenerateCleanSpecConfigBaseURL(baseURL *url.URL, dir string, includeFile bo
 		}
 
 	}
-	if strings.HasSuffix(p, "/") {
-		p = p[:len(p)-1]
-	}
-	return p
-
+	return strings.TrimSuffix(p, "/")
 }

test_specs/mixedref-burgershop.openapi.yaml

@@ -234,7 +234,7 @@ paths:
           content:
             application/json:
               schema:
-                $ref: 'https://raw.githubusercontent.com/daveshanley/vacuum/main/model/test_files/burgershop.openapi.yaml'
+                $ref: 'https://raw.githubusercontent.com/daveshanley/vacuum/main/model/test_files/burgershop.openapi.yaml#/components/schemas/Error'
 components:
   schemas:
     Error: