feat: Add id for CodeGraphData (#63337)

This allows locating a CodeGraphData object later so that the occurrences
can be paginated over correctly.

Fixes GRAPH-692

cmd/frontend/graphqlbackend/codeintel.codenav.graphql

@@ -98,7 +98,13 @@ EXPERIMENTAL: This type may change in a backwards-incompatible way.
 TODO(issue: GRAPH-614): 'commit' field should have type GitCommit
 before stabilizing this API.
 """
-type CodeGraphData {
+type CodeGraphData implements Node {
+    """
+    ID of this object that can be used to identify the CodeGraphData
+    for pagination in the codeGraphData field.
+    """
+    id: ID!
+
     """
     Coarse description of the data source for this code graph data.
     """

cmd/frontend/graphqlbackend/node.go

@@ -381,3 +381,8 @@ func (r *NodeResolver) ToSearchJob() (SearchJobResolver, bool) {
 	n, ok := r.Node.(SearchJobResolver)
 	return n, ok
 }
+
+func (r *NodeResolver) ToCodeGraphData() (resolverstubs.CodeGraphDataResolver, bool) {
+	n, ok := r.Node.(resolverstubs.CodeGraphDataResolver)
+	return n, ok
+}

internal/codeintel/codenav/transport/graphql/BUILD.bazel

@@ -43,6 +43,7 @@ go_library(
         "//lib/errors",
         "//lib/pointers",
         "@com_github_graph_gophers_graphql_go//:graphql-go",
+        "@com_github_graph_gophers_graphql_go//relay",
         "@com_github_sourcegraph_go_lsp//:go-lsp",
         "@com_github_sourcegraph_log//:log",
         "@com_github_sourcegraph_scip//bindings/go/scip",

internal/codeintel/codenav/transport/graphql/root_resolver.go

@@ -5,11 +5,14 @@ import (
 	"fmt"
 	"sync"
 
+	"github.com/graph-gophers/graphql-go"
+	"github.com/graph-gophers/graphql-go/relay"
 	orderedmap "github.com/wk8/go-ordered-map/v2"
 	"go.opentelemetry.io/otel/attribute"
 
 	"github.com/sourcegraph/scip/bindings/go/scip"
 
+	"github.com/sourcegraph/sourcegraph/internal/api"
 	"github.com/sourcegraph/sourcegraph/internal/authz"
 	"github.com/sourcegraph/sourcegraph/internal/codeintel/codenav"
 	resolverstubs "github.com/sourcegraph/sourcegraph/internal/codeintel/resolvers"
@@ -159,6 +162,10 @@ func (r *rootResolver) CodeGraphData(ctx context.Context, opts *resolverstubs.Co
 	return &[]resolverstubs.CodeGraphDataResolver{}, nil
 }
 
+func (r *rootResolver) CodeGraphDataByID(ctx context.Context, id graphql.ID) (resolverstubs.CodeGraphDataResolver, error) {
+	return newCodeGraphDataResolverFromID(ctx, r.repoStore, r.svc, r.operations, id)
+}
+
 func preferUploadsWithLongestRoots(uploads []shared.CompletedUpload) []shared.CompletedUpload {
 	// Use orderedmap instead of a map to preserve the order of the uploads
 	// and to avoid introducing non-determinism.
@@ -317,7 +324,7 @@ type codeGraphDataResolver struct {
 
 	// Arguments
 	svc        CodeNavService
-	upload     shared.CompletedUpload
+	upload     UploadData
 	opts       *resolverstubs.CodeGraphDataOpts
 	provenance resolverstubs.CodeGraphDataProvenance
 
@@ -325,6 +332,26 @@ type codeGraphDataResolver struct {
 	operations *operations
 }
 
+// UploadData represents the subset of information of shared.CompletedUpload
+// that we actually care about for the purposes of the GraphQL API.
+//
+// All fields are left public for JSON marshaling/unmarshaling.
+type UploadData struct {
+	UploadID       int
+	Commit         string
+	Indexer        string
+	IndexerVersion string
+}
+
+func NewUploadData(upload shared.CompletedUpload) UploadData {
+	return UploadData{
+		UploadID:       upload.ID,
+		Commit:         upload.Commit,
+		Indexer:        upload.Indexer,
+		IndexerVersion: upload.IndexerVersion,
+	}
+}
+
 func newCodeGraphDataResolver(
 	svc CodeNavService,
 	upload shared.CompletedUpload,
@@ -337,23 +364,81 @@ func newCodeGraphDataResolver(
 		/*document*/ nil,
 		/*documentRetrievalError*/ nil,
 		svc,
-		upload,
+		NewUploadData(upload),
 		opts,
 		provenance,
 		operations,
 	}
 }
 
+// CodeGraphDataID represents the serializable state needed to materialize
+// a CodeGraphData value from an opaque GraphQL ID.
+//
+// All fields are left public for JSON marshaling/unmarshaling.
+type CodeGraphDataID struct {
+	UploadData
+	Args *resolverstubs.CodeGraphDataArgs
+	api.RepoID
+	Commit api.CommitID
+	Path   string
+	resolverstubs.CodeGraphDataProvenance
+}
+
+func newCodeGraphDataResolverFromID(
+	ctx context.Context,
+	repoStore database.RepoStore,
+	svc CodeNavService,
+	operations *operations,
+	rawID graphql.ID,
+) (resolverstubs.CodeGraphDataResolver, error) {
+	var id CodeGraphDataID
+	if err := relay.UnmarshalSpec(rawID, &id); err != nil {
+		return nil, errors.Wrap(err, "malformed ID")
+	}
+	repos, err := repoStore.GetByIDs(ctx, id.RepoID)
+	if err != nil {
+		return nil, errors.Wrap(err, "repo for CodeGraphData value no longer exists")
+	}
+	opts := resolverstubs.CodeGraphDataOpts{
+		Args:   id.Args,
+		Repo:   repos[0],
+		Commit: id.Commit,
+		Path:   id.Path,
+	}
+	return &codeGraphDataResolver{
+		sync.Once{},
+		/*document*/ nil,
+		/*documentRetrievalError*/ nil,
+		svc,
+		id.UploadData,
+		&opts,
+		id.CodeGraphDataProvenance,
+		operations,
+	}, nil
+}
+
 func (c *codeGraphDataResolver) tryRetrieveDocument(ctx context.Context) (*scip.Document, error) {
 	// NOTE(id: scip-doc-optimization): In the case of pagination, if we retrieve the document ID
 	// from the database, we can avoid performing a JOIN between codeintel_scip_document_lookup
 	// and codeintel_scip_documents
 	c.retrievedDocument.Do(func() {
-		c.document, c.documentRetrievalError = c.svc.SCIPDocument(ctx, c.upload.ID, c.opts.Path)
+		c.document, c.documentRetrievalError = c.svc.SCIPDocument(ctx, c.upload.UploadID, c.opts.Path)
 	})
 	return c.document, c.documentRetrievalError
 }
 
+func (c *codeGraphDataResolver) ID() graphql.ID {
+	dataID := CodeGraphDataID{
+		c.upload,
+		c.opts.Args,
+		c.opts.Repo.ID,
+		c.opts.Commit,
+		c.opts.Path,
+		c.provenance,
+	}
+	return relay.MarshalID(resolverstubs.CodeGraphDataIDKind, dataID)
+}
+
 func (c *codeGraphDataResolver) Provenance(_ context.Context) (resolverstubs.CodeGraphDataProvenance, error) {
 	return c.provenance, nil
 }

internal/codeintel/resolvers/codenav.go

@@ -31,9 +31,13 @@ type CodeNavServiceResolver interface {
 	// that it is not what is exactly provided as input from the GraphQL
 	// client.
 	CodeGraphData(ctx context.Context, opts *CodeGraphDataOpts) (*[]CodeGraphDataResolver, error)
+	// CodeGraphDataByID materializes a CodeGraphDataResolver purely from a graphql.ID.
+	CodeGraphDataByID(ctx context.Context, id graphql.ID) (CodeGraphDataResolver, error)
 	UsagesForSymbol(ctx context.Context, args *UsagesForSymbolArgs) (UsageConnectionResolver, error)
 }
 
+const CodeGraphDataIDKind = "CodeGraphData"
+
 type GitBlobLSIFDataArgs struct {
 	Repo      *types.Repo
 	Commit    api.CommitID
@@ -161,6 +165,8 @@ type DiagnosticResolver interface {
 }
 
 type CodeGraphDataResolver interface {
+	// ID satisfies the Node interface.
+	ID() graphql.ID
 	Provenance(ctx context.Context) (CodeGraphDataProvenance, error)
 	Commit(ctx context.Context) (string, error)
 	ToolInfo(ctx context.Context) (*CodeGraphToolInfo, error)
@@ -168,6 +174,10 @@ type CodeGraphDataResolver interface {
 	Occurrences(ctx context.Context, args *OccurrencesArgs) (SCIPOccurrenceConnectionResolver, error)
 }
 
+// CodeGraphDataProvenance corresponds to the matching type in the GraphQL API.
+//
+// Make sure this type maintains its marshaling/unmarshaling behavior in
+// case the type definition is changed.
 type CodeGraphDataProvenance string
 
 const (
@@ -192,6 +202,10 @@ func (f *CodeGraphDataFilter) String() string {
 	return ""
 }
 
+// CodeGraphDataArgs represents the arguments to the codeGraphData(...)
+// field on GitBlob in the GraphQL API.
+//
+// All fields are left public for JSON marshaling/unmarshaling.
 type CodeGraphDataArgs struct {
 	Filter *CodeGraphDataFilter
 }

internal/codeintel/resolvers/root_resolver.go

@@ -63,6 +63,9 @@ func (r *Resolver) NodeResolvers() map[string]NodeByIDFunc {
 		"PreciseIndex": func(ctx context.Context, id graphql.ID) (Node, error) {
 			return r.uploadsRootResolver.PreciseIndexByID(ctx, id)
 		},
+		CodeGraphDataIDKind: func(ctx context.Context, id graphql.ID) (Node, error) {
+			return r.codenavResolver.CodeGraphDataByID(ctx, id)
+		},
 	}
 }
 
@@ -129,6 +132,10 @@ func (r *Resolver) CodeGraphData(ctx context.Context, opts *CodeGraphDataOpts) (
 	return r.codenavResolver.CodeGraphData(ctx, opts)
 }
 
+func (r *Resolver) CodeGraphDataByID(ctx context.Context, id graphql.ID) (CodeGraphDataResolver, error) {
+	return r.codenavResolver.CodeGraphDataByID(ctx, id)
+}
+
 func (r *Resolver) UsagesForSymbol(ctx context.Context, args *UsagesForSymbolArgs) (UsageConnectionResolver, error) {
 	return r.codenavResolver.UsagesForSymbol(ctx, args)
 }