refactor(github): Have a dedicated PublicRepository type (#63607)

internal/extsvc/github/common.go

@@ -1811,6 +1811,19 @@ type Repository struct {
 	DiskUsageKibibytes int `json:"DiskUsage,omitempty"`
 }
 
+// PublicRepository is a reduced set of fields from a GitHub repository
+// that corresponds with the fields returned from the "List public repositories"
+// GitHub API endpoint: https://docs.github.com/en/rest/repos/repos?apiVersion=2022-11-28#list-public-repositories
+type PublicRepository struct {
+	ID            string // ID of repository (GitHub GraphQL ID, not GitHub database ID)
+	DatabaseID    int64  // The integer database id
+	NameWithOwner string // Full name of repository ("owner/name")
+	Description   string // Description of repository
+	URL           string // The web URL of this repository ("https://github.com/foo/bar")
+	IsPrivate     bool   // Whether the repository is private
+	IsFork        bool   // Whether the repository is a fork of another repository
+}
+
 func (r *Repository) SizeBytes() bytesize.Bytes {
 	return bytesize.Bytes(r.DiskUsageKibibytes) * bytesize.KiB
 }
@@ -1861,6 +1874,19 @@ type restRepository struct {
 	DiskUsageKibibytes int `json:"size"`
 }
 
+// restPublicRepository is a reduced set of fields from a GitHub repository
+// that corresponds with the fields returned from the "List public repositories"
+// GitHub API endpoint: https://docs.github.com/en/rest/repos/repos?apiVersion=2022-11-28#list-public-repositories
+type restPublicRepository struct {
+	ID          string `json:"node_id"` // GraphQL ID
+	DatabaseID  int64  `json:"id"`
+	FullName    string `json:"full_name"` // same as nameWithOwner
+	Description string `json:"description"`
+	HTMLURL     string `json:"html_url"` // web URL
+	Private     bool   `json:"private"`
+	Fork        bool   `json:"fork"`
+}
+
 // getRepositoryFromAPI attempts to fetch a repository from the GitHub API without use of the redis cache.
 func (c *V3Client) getRepositoryFromAPI(ctx context.Context, owner, name string) (*Repository, error) {
 	// If no token, we must use the older REST API, not the GraphQL API. See
@@ -1907,6 +1933,22 @@ func convertRestRepo(restRepo restRepository) *Repository {
 	return &repo
 }
 
+// convertRestPublicRepo converts public repo information returned by the rest API
+// to a standard format.
+func convertRestPublicRepo(restPublicRepo restPublicRepository) *PublicRepository {
+	repo := PublicRepository{
+		ID:            restPublicRepo.ID,
+		DatabaseID:    restPublicRepo.DatabaseID,
+		NameWithOwner: restPublicRepo.FullName,
+		Description:   restPublicRepo.Description,
+		URL:           restPublicRepo.HTMLURL,
+		IsPrivate:     restPublicRepo.Private,
+		IsFork:        restPublicRepo.Fork,
+	}
+
+	return &repo
+}
+
 // convertRestRepoPermissions converts repo information returned by the rest API
 // to a standard format.
 func convertRestRepoPermissions(restRepoPermissions restRepositoryPermissions) string {

internal/extsvc/github/v3.go

@@ -642,15 +642,15 @@ func (c *V3Client) ListTeamMembers(ctx context.Context, owner, team string, page
 // An empty sinceRepoID returns the first page of results.
 // This is only intended to be called for GitHub Enterprise, so no rate limit information is returned.
 // https://developer.github.com/v3/repos/#list-all-public-repositories
-func (c *V3Client) getPublicRepositories(ctx context.Context, sinceRepoID int64) ([]*Repository, bool, error) {
+func (c *V3Client) getPublicRepositories(ctx context.Context, sinceRepoID int64) ([]*PublicRepository, bool, error) {
 	path := "repositories"
 	if sinceRepoID > 0 {
 		path += "?per_page=100&since=" + strconv.FormatInt(sinceRepoID, 10)
 	}
-	return c.listRepositories(ctx, path)
+	return c.listPublicRepositories(ctx, path)
 }
 
-func (c *V3Client) ListPublicRepositories(ctx context.Context, sinceRepoID int64) ([]*Repository, bool, error) {
+func (c *V3Client) ListPublicRepositories(ctx context.Context, sinceRepoID int64) ([]*PublicRepository, bool, error) {
 	return c.getPublicRepositories(ctx, sinceRepoID)
 }
 
@@ -801,6 +801,27 @@ func (c *V3Client) listRepositories(ctx context.Context, requestURI string) ([]*
 	return repos, respState.hasNextPage(), nil
 }
 
+func (c *V3Client) listPublicRepositories(ctx context.Context, requestURI string) ([]*PublicRepository, bool, error) {
+	var restPublicRepos []restPublicRepository
+	respState, err := c.get(ctx, requestURI, &restPublicRepos)
+	if err != nil {
+		return nil, false, err
+	}
+	repos := make([]*PublicRepository, 0, len(restPublicRepos))
+	for _, r := range restPublicRepos {
+		// Sometimes GitHub API returns null JSON objects and JSON decoder unmarshalls
+		// them as a zero-valued `restRepository` objects.
+		//
+		// See https://github.com/sourcegraph/customer/issues/1688 for details.
+		if r.ID == "" {
+			c.log.Warn("GitHub returned a public repository without an ID", log.String("restPublicRepository", fmt.Sprintf("%#v", r)))
+			continue
+		}
+		repos = append(repos, convertRestPublicRepo(r))
+	}
+	return repos, respState.hasNextPage(), nil
+}
+
 func (c *V3Client) GetRepo(ctx context.Context, owner, repo string) (*Repository, error) {
 	var restRepo restRepository
 	if _, err := c.get(ctx, "repos/"+owner+"/"+repo, &restRepo); err != nil {

internal/repos/github.go

@@ -712,7 +712,7 @@ func (s *GitHubSource) listRepos(ctx context.Context, repos []string, results ch
 	}
 }
 
-func batchRepos(repos []*github.Repository, batchSize int) (batches [][]*github.Repository) {
+func batchPublicRepos(repos []*github.PublicRepository, batchSize int) (batches [][]*github.PublicRepository) {
 	for i := 0; i < len(repos); i += batchSize {
 		end := i + batchSize
 		if end > len(repos) {
@@ -753,7 +753,7 @@ func (s *GitHubSource) listPublic(ctx context.Context, results chan *githubResul
 		// The ListPublicRepositories endpoint returns incomplete information,
 		// so we make additional calls to get the full information of each repo.
 
-		batchedRepos := batchRepos(repos, 30)
+		batchedRepos := batchPublicRepos(repos, 30)
 		for _, batch := range batchedRepos {
 			namesWithOwners := make([]string, 0, len(batch))