From dbc948a453f8cbdf11c8c595299b114ea4b4a814 Mon Sep 17 00:00:00 2001
From: Mustafa Gezen <mustafa@rockylinux.org>
Date: Tue, 22 Dec 2020 06:13:27 +0100
Subject: [PATCH] add no-dup-mode to skip already imported tags

---
 cmd/srpmproc/main.go |  3 +++
 internal/git.go      |  7 ++++---
 internal/process.go  | 48 +++++++++++++++++++++++++++++++++++++++++++-
 3 files changed, 54 insertions(+), 4 deletions(-)

diff --git a/cmd/srpmproc/main.go b/cmd/srpmproc/main.go
index 114a1b2..ef1f603 100644
--- a/cmd/srpmproc/main.go
+++ b/cmd/srpmproc/main.go
@@ -26,6 +26,7 @@ var (
 	gitCommitterEmail string
 	modulePrefix      string
 	rpmPrefix         string
+	noDupMode         bool
 )
 
 var root = &cobra.Command{
@@ -87,6 +88,7 @@ func mn(_ *cobra.Command, _ []string) {
 		GitCommitterEmail: gitCommitterEmail,
 		ModulePrefix:      modulePrefix,
 		Authenticator:     authenticator,
+		NoDupMode:         noDupMode,
 	})
 }
 
@@ -106,6 +108,7 @@ func main() {
 	root.Flags().StringVar(&gitCommitterEmail, "git-committer-email", "mustafa+distrobuild@bycrates.com", "Email of committer")
 	root.Flags().StringVar(&modulePrefix, "module-prefix", "https://git.centos.org/modules", "Where to retrieve modules if exists. Only used when source-rpm is a git repo")
 	root.Flags().StringVar(&rpmPrefix, "rpm-prefix", "https://git.centos.org/rpms", "Where to retrieve SRPM content. Only used when source-rpm is not a local file")
+	root.Flags().BoolVar(&noDupMode, "no-dup-mode", false, "If enabled, skips already imported tags")
 
 	if err := root.Execute(); err != nil {
 		log.Fatal(err)
diff --git a/internal/git.go b/internal/git.go
index 8f2739a..e1c4c7e 100644
--- a/internal/git.go
+++ b/internal/git.go
@@ -71,6 +71,7 @@ func (g *GitMode) WriteSource(md *modeData) {
 
 	match := tagImportRegex.FindStringSubmatch(md.tagBranch)
 	refspec := config.RefSpec(fmt.Sprintf("+refs/heads/%s:%s", match[2], md.tagBranch))
+	log.Printf("checking out upstream refspec %s", refspec)
 	err = remote.Fetch(&git.FetchOptions{
 		RemoteName: "upstream",
 		RefSpecs:   []config.RefSpec{refspec},
@@ -115,9 +116,9 @@ func (g *GitMode) WriteSource(md *modeData) {
 			continue
 		}
 
-		lineInfo := strings.Split(line, " ")
-		hash := lineInfo[0]
-		path := lineInfo[1]
+		lineInfo := strings.SplitN(line, " ", 2)
+		hash := strings.TrimSpace(lineInfo[0])
+		path := strings.TrimSpace(lineInfo[1])
 
 		url := fmt.Sprintf("https://git.centos.org/sources/%s/%s/%s", md.rpmFile.Name(), match[2], hash)
 		log.Printf("downloading %s", url)
diff --git a/internal/process.go b/internal/process.go
index 9f45af2..1bf071b 100644
--- a/internal/process.go
+++ b/internal/process.go
@@ -35,6 +35,7 @@ type ProcessData struct {
 	Authenticator     *ssh.PublicKeys
 	Importer          ImportMode
 	BlobStorage       blob.Storage
+	NoDupMode         bool
 }
 
 type ignoredSource struct {
@@ -63,6 +64,40 @@ type modeData struct {
 func ProcessRPM(pd *ProcessData) {
 	md := pd.Importer.RetrieveSource(pd)
 
+	// if no-dup-mode is enabled then skip already imported versions
+	var tagIgnoreList []string
+	if pd.NoDupMode {
+		repo, err := git.Init(memory.NewStorage(), memfs.New())
+		if err != nil {
+			log.Fatalf("could not init git repo: %v", err)
+		}
+		remoteUrl := fmt.Sprintf("%s/dist/%s.git", pd.UpstreamPrefix, md.rpmFile.Name())
+		refspec := config.RefSpec("+refs/heads/*:refs/remotes/origin/*")
+
+		remote, err := repo.CreateRemote(&config.RemoteConfig{
+			Name:  "origin",
+			URLs:  []string{remoteUrl},
+			Fetch: []config.RefSpec{refspec},
+		})
+		if err != nil {
+			log.Fatalf("could not create remote: %v", err)
+		}
+
+		list, err := remote.List(&git.ListOptions{
+			Auth: pd.Authenticator,
+		})
+		if err != nil {
+			log.Fatalf("could not list remote: %v", err)
+		}
+
+		for _, ref := range list {
+			if !strings.HasPrefix(string(ref.Name()), "refs/tags/imports") {
+				continue
+			}
+			tagIgnoreList = append(tagIgnoreList, string(ref.Name()))
+		}
+	}
+
 	sourceRepo := *md.repo
 	sourceWorktree := *md.worktree
 
@@ -89,6 +124,18 @@ func ProcessRPM(pd *ProcessData) {
 
 		match := tagImportRegex.FindStringSubmatch(md.tagBranch)
 		md.pushBranch = "rocky" + strings.TrimPrefix(match[2], "c")
+		newTag := "imports/rocky" + strings.TrimPrefix(match[1], "imports/c")
+
+		shouldContinue := true
+		for _, ignoredTag := range tagIgnoreList {
+			if ignoredTag == "refs/tags/"+newTag {
+				log.Printf("skipping %s", ignoredTag)
+				shouldContinue = false
+			}
+		}
+		if !shouldContinue {
+			continue
+		}
 
 		// create a new remote
 		remoteUrl := fmt.Sprintf("%s/dist/%s.git", pd.UpstreamPrefix, rpmFile.Name())
@@ -225,7 +272,6 @@ func ProcessRPM(pd *ProcessData) {
 
 		log.Printf("committed:\n%s", obj.String())
 
-		newTag := "imports/rocky" + strings.TrimPrefix(match[1], "imports/c")
 		_, err = repo.CreateTag(newTag, commit, &git.CreateTagOptions{
 			Tagger: &object.Signature{
 				Name:  pd.GitCommitterName,