feat: filter repositories using topics

cmd/platform.go

@@ -27,6 +27,7 @@ func configurePlatform(cmd *cobra.Command) {
 	flags.StringSliceP("group", "G", nil, "The name of a GitLab organization. All repositories in that group will be used.")
 	flags.StringSliceP("user", "U", nil, "The name of a user. All repositories owned by that user will be used.")
 	flags.StringSliceP("repo", "R", nil, "The name, including owner of a GitHub repository in the format \"ownerName/repoName\".")
+	flags.StringSliceP("topic", "", nil, "The topic of a GitHub/GitLab/Gitea repository. All repositories having at least one matching topic are targeted.")
 	flags.StringSliceP("project", "P", nil, "The name, including owner of a GitLab project in the format \"ownerName/repoName\".")
 	flags.BoolP("include-subgroups", "", false, "Include GitLab subgroups when using the --group flag.")
 	flags.BoolP("ssh-auth", "", false, `Use SSH cloning URL instead of HTTPS + token. This requires that a setup with ssh keys that have access to all repos and that the server is already in known_hosts.`)
@@ -119,6 +120,7 @@ func createGithubClient(flag *flag.FlagSet, verifyFlags bool, readOnly bool) (mu
 	orgs, _ := flag.GetStringSlice("org")
 	users, _ := flag.GetStringSlice("user")
 	repos, _ := flag.GetStringSlice("repo")
+	topics, _ := flag.GetStringSlice("topic")
 	forkMode, _ := flag.GetBool("fork")
 	forkOwner, _ := flag.GetString("fork-owner")
 	sshAuth, _ := flag.GetBool("ssh-auth")
@@ -149,6 +151,7 @@ func createGithubClient(flag *flag.FlagSet, verifyFlags bool, readOnly bool) (mu
 		Organizations: orgs,
 		Users:         users,
 		Repositories:  repoRefs,
+		Topics:        topics,
 	}, mergeTypes, forkMode, forkOwner, sshAuth, readOnly)
 	if err != nil {
 		return nil, err
@@ -162,6 +165,7 @@ func createGitlabClient(flag *flag.FlagSet, verifyFlags bool) (multigitter.Versi
 	groups, _ := flag.GetStringSlice("group")
 	users, _ := flag.GetStringSlice("user")
 	projects, _ := flag.GetStringSlice("project")
+	topics, _ := flag.GetStringSlice("topic")
 	includeSubgroups, _ := flag.GetBool("include-subgroups")
 	sshAuth, _ := flag.GetBool("ssh-auth")
 
@@ -186,6 +190,7 @@ func createGitlabClient(flag *flag.FlagSet, verifyFlags bool) (multigitter.Versi
 		Groups:   groups,
 		Users:    users,
 		Projects: projRefs,
+		Topics:   topics,
 	}, gitlab.Config{
 		IncludeSubgroups: includeSubgroups,
 		SSHAuth:          sshAuth,
@@ -202,6 +207,7 @@ func createGiteaClient(flag *flag.FlagSet, verifyFlags bool) (multigitter.Versio
 	orgs, _ := flag.GetStringSlice("org")
 	users, _ := flag.GetStringSlice("user")
 	repos, _ := flag.GetStringSlice("repo")
+	topics, _ := flag.GetStringSlice("topic")
 	sshAuth, _ := flag.GetBool("ssh-auth")
 
 	if verifyFlags && len(orgs) == 0 && len(users) == 0 && len(repos) == 0 {
@@ -234,6 +240,7 @@ func createGiteaClient(flag *flag.FlagSet, verifyFlags bool) (multigitter.Versio
 		Organizations: orgs,
 		Users:         users,
 		Repositories:  repoRefs,
+		Topics:        topics,
 	}, mergeTypes, sshAuth)
 	if err != nil {
 		return nil, err

internal/scm/gitea/gitea.go

@@ -10,6 +10,7 @@ import (
 	"code.gitea.io/sdk/gitea"
 	"github.com/lindell/multi-gitter/internal/scm"
 	"github.com/pkg/errors"
+	log "github.com/sirupsen/logrus"
 
 	internalHTTP "github.com/lindell/multi-gitter/internal/http"
 )
@@ -67,6 +68,7 @@ type RepositoryListing struct {
 	Organizations []string
 	Users         []string
 	Repositories  []RepositoryReference
+	Topics        []string
 }
 
 // RepositoryReference contains information to be able to reference a repository
@@ -96,6 +98,20 @@ func (g *Gitea) GetRepositories(ctx context.Context) ([]scm.Repository, error) {
 
 	repos := make([]scm.Repository, 0, len(allRepos))
 	for _, repo := range allRepos {
+		log := log.WithField("repo", repo.FullName)
+
+		if len(g.Topics) != 0 {
+			topics, err := g.getRepoTopics(ctx, repo)
+			if err != nil {
+				return repos, fmt.Errorf("could not fetch repository topics: %w", err)
+			}
+
+			if !scm.RepoContainsTopic(topics, g.Topics) {
+				log.Debug("Skipping repository since it does not match repository topics")
+				continue
+			}
+		}
+
 		convertedRepo, err := g.convertRepository(repo)
 		if err != nil {
 			return nil, err
@@ -106,6 +122,11 @@ func (g *Gitea) GetRepositories(ctx context.Context) ([]scm.Repository, error) {
 	return repos, nil
 }
 
+func (g *Gitea) getRepoTopics(ctx context.Context, repo *gitea.Repository) ([]string, error) {
+	topics, _, err := g.giteaClient(ctx).ListRepoTopics(repo.Owner.UserName, repo.Name, gitea.ListRepoTopicsOptions{})
+	return topics, err
+}
+
 func (g *Gitea) getRepositories(ctx context.Context) ([]*gitea.Repository, error) {
 	allRepos := []*gitea.Repository{}
 

internal/scm/github/github.go

@@ -99,6 +99,7 @@ type RepositoryListing struct {
 	Organizations []string
 	Users         []string
 	Repositories  []RepositoryReference
+	Topics        []string
 }
 
 // RepositoryReference contains information to be able to reference a repository
@@ -151,6 +152,9 @@ func (g *Github) GetRepositories(ctx context.Context) ([]scm.Repository, error)
 		case !g.Fork && !g.ReadOnly && !permissions["push"]:
 			log.Debug("Skipping repository since the token does not have push permissions and the run will not fork")
 			continue
+		case len(g.Topics) != 0 && !scm.RepoContainsTopic(r.Topics, g.Topics):
+			log.Debug("Skipping repository since it does not match repository topics")
+			continue
 		}
 
 		newRepo, err := g.convertRepo(r)

internal/scm/github/github_test.go

@@ -49,6 +49,9 @@ func Test_GetRepositories(t *testing.T) {
 					"name": "test1",
 					"full_name": "test-org/test1",
 					"private": false,
+					"topics": [
+						"frontend"
+					],
 					"owner": {
 						"login": "test-org",
 						"type": "Organization",
@@ -93,6 +96,10 @@ func Test_GetRepositories(t *testing.T) {
 					"name": "test2",
 					"full_name": "lindell/test2",
 					"private": false,
+					"topics": [
+						"backend",
+						"go"
+					],
 					"owner": {
 						"login": "lindell",
 						"type": "User",
@@ -163,6 +170,22 @@ func Test_GetRepositories(t *testing.T) {
 		}
 	}
 
+	// Topics
+	{
+		gh, err := github.New("", "", transport.Wrapper, github.RepositoryListing{
+			Organizations: []string{"test-org"},
+			Topics:        []string{"frontend", "backend"},
+		}, []scm.MergeType{scm.MergeTypeMerge}, false, "", false, false)
+		require.NoError(t, err)
+
+		repos, err := gh.GetRepositories(context.Background())
+		assert.NoError(t, err)
+		if assert.Len(t, repos, 1) {
+			assert.Equal(t, "master", repos[0].DefaultBranch())
+			assert.Equal(t, "test-org/test1", repos[0].FullName())
+		}
+	}
+
 	// Multiple
 	{
 		gh, err := github.New("", "", transport.Wrapper, github.RepositoryListing{

internal/scm/gitlab/gitlab.go

@@ -11,6 +11,7 @@ import (
 
 	internalHTTP "github.com/lindell/multi-gitter/internal/http"
 	"github.com/lindell/multi-gitter/internal/scm"
+	log "github.com/sirupsen/logrus"
 	"github.com/xanzy/go-gitlab"
 )
 
@@ -54,6 +55,7 @@ type RepositoryListing struct {
 	Groups   []string
 	Users    []string
 	Projects []ProjectReference
+	Topics   []string
 }
 
 // Config includes extra config parameters for the GitLab client
@@ -89,6 +91,12 @@ func (g *Gitlab) GetRepositories(ctx context.Context) ([]scm.Repository, error)
 
 	repos := make([]scm.Repository, 0, len(allProjects))
 	for _, project := range allProjects {
+		log := log.WithField("repo", project.NameWithNamespace)
+		if len(g.Topics) != 0 && !scm.RepoContainsTopic(project.Topics, g.Topics) {
+			log.Debug("Skipping repository since it does not match repository topics")
+			continue
+		}
+
 		p, err := g.convertProject(project)
 		if err != nil {
 			return nil, err
@@ -124,6 +132,7 @@ func (g *Gitlab) getProjects(ctx context.Context) ([]*gitlab.Project, error) {
 		if err != nil {
 			return nil, err
 		}
+
 		allProjects = append(allProjects, project)
 	}
 

internal/scm/repository.go

@@ -9,3 +9,18 @@ type Repository interface {
 	// FullName returns the full id of the repository, usually ownerName/repoName
 	FullName() string
 }
+
+func RepoContainsTopic(repoTopics []string, filterTopics []string) bool {
+	repoTopicsMap := map[string]struct{}{}
+	for _, v := range repoTopics {
+		repoTopicsMap[v] = struct{}{}
+	}
+
+	for _, v := range filterTopics {
+		if _, ok := repoTopicsMap[v]; ok {
+			return true
+		}
+	}
+
+	return false
+}