autogits/workflow-pr/pr_processor.go

package main

//go:generate mockgen -source=pr_processor.go -destination=mock/pr_processor.go -typed

import (
	"encoding/json"
	"fmt"
	"path"
	"runtime/debug"
	"slices"
	"strings"

	"github.com/opentracing/opentracing-go/log"
	"src.opensuse.org/autogits/common"
	"src.opensuse.org/autogits/common/gitea-generated/client/repository"
	"src.opensuse.org/autogits/common/gitea-generated/models"
)

func prGitBranchNameForPR(repo string, prNo int64) string {
	return fmt.Sprintf("PR_%s#%d", repo, prNo)
}

func PrjGitDescription(prset *common.PRSet) (title string, desc string) {
	title_refs := make([]string, 0, len(prset.PRs)-1)
	refs := make([]string, 0, len(prset.PRs)-1)

	for _, pr := range prset.PRs {
		if prset.IsPrjGitPR(pr.PR) {
			continue
		}
		if pr.PR.State != "open" {
			// remove PRs that are not open from description
			continue
		}
		org, repo, idx := pr.PRComponents()

		title_refs = append(title_refs, repo)
		ref := fmt.Sprintf(common.PrPattern, org, repo, idx)
		refs = append(refs, ref)
	}

	title = "Forwarded PRs: " + strings.Join(title_refs, ", ")
	desc = fmt.Sprintf("This is a forwarded pull request by %s\nreferencing the following pull request(s):\n\n", GitAuthor) + strings.Join(refs, "\n") + "\n"

	if prset.Config.ManualMergeOnly {
		desc = desc + "\n###  ManualMergeOnly enabled. To merge, 'merge ok' is required in either the project PR or every package PR."
	}
	if prset.Config.ManualMergeProject {
		desc = desc + "\n###  ManualMergeProject enabled. To merge, 'merge ok' is required by project maintainer in the project PR."
	}
	if !prset.Config.ManualMergeOnly && !prset.Config.ManualMergeProject {
		desc = desc + "\n###  Automatic merge enabled. This will merge when all review requirements are satisfied."
	}
	return
}

func verifyRepositoryConfiguration(repo *models.Repository) error {
	if repo.AutodetectManualMerge && repo.AllowManualMerge {
		return nil
	}

	// modify repo to allow above
	common.LogDebug("Adjusting repo to accept manual merges:", repo.Owner.UserName+"/"+repo.Name)
	_, err := Gitea.SetRepoOptions(repo.Owner.UserName, repo.Name, true)
	if err != nil {
		common.LogError("Failed to set repo to manual merges:", err)
	}
	return err
}

func updateSubmoduleInPR(submodule, headSha string, git common.Git) {
	common.LogDebug("updating submodule", submodule, "to HEAD", headSha)
	// NOTE: this can fail if current PrjGit is pointing to outdated, GC'ed commit
	// as long as we can update to newer one later, we are still OK
	git.GitExec(common.DefaultGitPrj, "submodule", "update", "--init", "--checkout", "--depth", "1", submodule)
	common.PanicOnError(git.GitExec(path.Join(common.DefaultGitPrj, submodule), "fetch", "--depth", "1", "origin", headSha))
	common.PanicOnError(git.GitExec(path.Join(common.DefaultGitPrj, submodule), "checkout", "-f", headSha))
}

type PRProcessor struct {
	config *common.AutogitConfig
	git    common.Git
}

func AllocatePRProcessor(req *models.PullRequest, configs common.AutogitConfigs) (*PRProcessor, error) {
	org := req.Base.Repo.Owner.UserName
	repo := req.Base.Repo.Name
	id := req.Index

	branch := req.Base.Ref

	PRstr := fmt.Sprintf("%s/%s!%d", org, repo, id)
	common.LogInfo("*** Starting processing PR:", PRstr, "branch:", branch)

	config := configs.GetPrjGitConfig(org, repo, branch)
	if config == nil {
		if req.Base.Repo.DefaultBranch == branch {
			common.LogDebug("Default branch submission...", org, repo)
			config = configs.GetPrjGitConfig(org, repo, "")
		}
	}
	if config == nil {
		common.LogError("Cannot find config for PR.")
		return nil, fmt.Errorf("Cannot find config for PR")
	}

	if common.GetLoggingLevel() >= common.LogLevelDebug {
		cjson, _ := json.Marshal(config)
		common.LogDebug("found config:", string(cjson))
	}
	if config == nil {
		common.LogError("Cannot find config for branch '%s'", req.Base.Ref)
		return nil, fmt.Errorf("Cannot find config for branch '%s'", req.Base.Ref)
	}

	git, err := GitHandler.CreateGitHandler(config.Organization)
	if err != nil {
		common.LogError("Cannot allocate GitHandler:", err)
		return nil, fmt.Errorf("Error allocating GitHandler. Err: %w", err)
	}
	common.LogDebug("git path:", git.GetPath())

	//	git.GitExecOrPanic("", "config", "set", "--global", "advice.submoduleMergeConflict", "false")
	//	git.GitExecOrPanic("", "config", "set", "--global", "advice.mergeConflict", "false")

	return &PRProcessor{
		config: config,
		git:    git,
	}, nil
}

func (pr *PRProcessor) SetSubmodulesToMatchPRSet(prset *common.PRSet) error {
	git := pr.git
	subList, err := git.GitSubmoduleList(common.DefaultGitPrj, "HEAD")
	if err != nil {
		common.LogError("Error fetching submodule list for PrjGit", err)
		return err
	}

	for _, pr := range prset.PRs {
		if prset.IsPrjGitPR(pr.PR) {
			continue
		}

		org, repo, idx := pr.PRComponents()
		prHead := pr.PR.Head.Sha
		revert := false

		if pr.PR.State != "open" {
			prjGitPR, err := prset.GetPrjGitPR()
			if prjGitPR != nil {
				// remove PR from PrjGit
				var valid bool
				if prHead, valid = git.GitSubmoduleCommitId(common.DefaultGitPrj, repo, prjGitPR.PR.MergeBase); !valid {
					common.LogError("Failed fetching original submodule commit id for repo")
					return err
				}
			}
			revert = true
		}

		// find 'repo' in the submodule list
		submodule_found := false
		for submodulePath, id := range subList {
			if path.Base(submodulePath) == repo {
				submodule_found = true
				if id != prHead {
					ref := fmt.Sprintf(common.PrPattern, org, repo, idx)
					commitMsg := fmt.Sprintln("auto-created for", repo, "\n\nThis commit was autocreated by", GitAuthor, "\n\nreferencing PRs:\n", ref)

					if revert {
						commitMsg = fmt.Sprintln("auto-created for", repo, "\n\nThis commit was autocreated by", GitAuthor, "\n\nremoving PRs:\n", ref)
					}

					updateSubmoduleInPR(submodulePath, prHead, git)
					status, err := git.GitStatus(common.DefaultGitPrj)
					common.LogDebug("status:", status)
					common.LogDebug("submodule", repo, " hash:", id, " -> ", prHead)
					common.PanicOnError(err)
					common.PanicOnError(git.GitExec(common.DefaultGitPrj, "commit", "-a", "-m", commitMsg))

					pr.PR.Head.Sha = id // update the prset
				}
				submodule_found = true
				break
			}
		}

		if !submodule_found {
			common.LogError("Failed to find expected repo:", repo)
		}
	}
	return nil
}

func (pr *PRProcessor) CreatePRjGitPR(prjGitPRbranch string, prset *common.PRSet) error {
	git := pr.git
	PrjGitOrg, PrjGitRepo, PrjGitBranch := prset.Config.GetPrjGit()
	PrjGit, err := Gitea.GetRepository(PrjGitOrg, PrjGitRepo)
	if err != nil {
		common.LogError("Failed to fetch PrjGit repository data.", PrjGitOrg, PrjGitRepo, err)
		return err
	}
	RemoteName, err := git.GitClone(common.DefaultGitPrj, PrjGitBranch, PrjGit.SSHURL)
	common.PanicOnError(err)
	git.GitExecOrPanic(common.DefaultGitPrj, "checkout", "-B", prjGitPRbranch, RemoteName+"/"+PrjGitBranch)

	headCommit, err := git.GitBranchHead(common.DefaultGitPrj, prjGitPRbranch)
	if err != nil {
		common.LogError("Failed to fetch PrjGit branch", prjGitPRbranch, err)
		return err
	}
	if err := pr.SetSubmodulesToMatchPRSet(prset); err != nil {
		return err
	}
	newHeadCommit, err := git.GitBranchHead(common.DefaultGitPrj, prjGitPRbranch)
	if err != nil {
		common.LogError("Failed to fetch updated PrjGit branch", prjGitPRbranch, err)
		return err
	}

	if !common.IsDryRun && !pr.config.NoProjectGitPR {
		if headCommit != newHeadCommit {
			common.PanicOnError(git.GitExec(common.DefaultGitPrj, "push", RemoteName, "+HEAD:"+prjGitPRbranch))
		}

		title, desc := PrjGitDescription(prset)
		pr, err := Gitea.CreatePullRequestIfNotExist(PrjGit, prjGitPRbranch, PrjGitBranch, title, desc)
		if err != nil {
			common.LogError("Error creating PrjGit PR:", err)
			return err
		}
		Gitea.UpdatePullRequest(PrjGit.Owner.UserName, PrjGit.Name, pr.Index, &models.EditPullRequestOption{
			RemoveDeadline: true,
		})

		prinfo := prset.AddPR(pr)
		prinfo.RemoteName = RemoteName
	}
	return nil
}

func (pr *PRProcessor) RebaseAndSkipSubmoduleCommits(prset *common.PRSet, branch string) error {
	git := pr.git
	PrjGitPR, err := prset.GetPrjGitPR()
	common.PanicOnError(err)

	remoteBranch := PrjGitPR.RemoteName + "/" + branch

	common.LogDebug("Rebasing on top of", remoteBranch)
	for conflict := git.GitExec(common.DefaultGitPrj, "rebase", remoteBranch); conflict != nil; {
		statuses, err := git.GitStatus(common.DefaultGitPrj)
		if err != nil {
			git.GitExecOrPanic(common.DefaultGitPrj, "rebase", "--abort")
			common.PanicOnError(err)
		}
		for _, s := range statuses {
			if s.SubmoduleChanges != "S..." {
				git.GitExecOrPanic(common.DefaultGitPrj, "rebase", "--abort")
				return fmt.Errorf("Unexpected conflict in rebase. %v", s)
			}
		}
		conflict = git.GitExec(common.DefaultGitPrj, "rebase", "--skip")
	}

	return nil
}

func (pr *PRProcessor) UpdatePrjGitPR(prset *common.PRSet) error {
	_, _, PrjGitBranch := prset.Config.GetPrjGit()
	PrjGitPR, err := prset.GetPrjGitPR()
	if err != nil {
		common.LogError("Updating PrjGitPR but not found?", err)
		return err
	}

	git := pr.git
	if len(prset.PRs) == 1 {
		git.GitExecOrPanic(common.DefaultGitPrj, "fetch", PrjGitPR.RemoteName, PrjGitPR.PR.Head.Sha)
		common.LogDebug("Only project git in PR. Nothing to update.")
		return nil
	}

	PrjGit := PrjGitPR.PR.Base.Repo
	prjGitPRbranch := PrjGitPR.PR.Head.Name
	if strings.Contains(prjGitPRbranch, "/") {
		PrjGitPR.RemoteName, err = git.GitClone(common.DefaultGitPrj, "", PrjGit.SSHURL)
		git.GitExecOrPanic(common.DefaultGitPrj, "fetch", PrjGitPR.RemoteName, PrjGitPR.PR.Head.Sha)
		git.GitExecOrPanic(common.DefaultGitPrj, "checkout", PrjGitPR.PR.Head.Sha)
		common.LogInfo("Cannot update this PR as it's on another remote, not branch:", prjGitPRbranch, "Assuming this is by-design. (eg. project git PR only)")
		return nil
	}

	PrjGitPR.RemoteName, err = git.GitClone(common.DefaultGitPrj, prjGitPRbranch, PrjGit.SSHURL)
	common.PanicOnError(err)
	git.GitExecOrPanic(common.DefaultGitPrj, "fetch", PrjGitPR.RemoteName, PrjGitBranch)

	forcePush := false
	// trust Gitea here on mergeability
	if !PrjGitPR.PR.Mergeable {
		common.PanicOnError(pr.RebaseAndSkipSubmoduleCommits(prset, PrjGitBranch))
		forcePush = true
	}

	headCommit, err := git.GitBranchHead(common.DefaultGitPrj, prjGitPRbranch)
	if err != nil {
		common.LogError("Failed to fetch PrjGit branch", prjGitPRbranch, err)
		return err
	}
	if err := pr.SetSubmodulesToMatchPRSet(prset); err != nil {
		return err
	}
	newHeadCommit, err := git.GitBranchHead(common.DefaultGitPrj, prjGitPRbranch)
	if err != nil {
		common.LogError("Failed to fetch updated PrjGit branch", prjGitPRbranch, err)
		return err
	}

	PrjGitTitle, PrjGitBody := PrjGitDescription(prset)
	if PrjGitPR.PR.Title != PrjGitTitle || PrjGitPR.PR.Body != PrjGitBody {
		common.LogDebug("New title:", PrjGitTitle)
		common.LogDebug(PrjGitBody)
	}

	if !common.IsDryRun {
		if headCommit != newHeadCommit {
			params := []string{"push", PrjGitPR.RemoteName, "+HEAD:" + prjGitPRbranch}
			if forcePush {
				params = slices.Insert(params, 1, "-f")
			}
			common.PanicOnError(git.GitExec(common.DefaultGitPrj, params...))
		}

		// update PR
		if PrjGitPR.PR.Body != PrjGitBody || PrjGitPR.PR.Title != PrjGitTitle {
			Gitea.UpdatePullRequest(PrjGit.Owner.UserName, PrjGit.Name, PrjGitPR.PR.Index, &models.EditPullRequestOption{
				RemoveDeadline: true,
				Title:          PrjGitTitle,
				Body:           PrjGitBody,
			})
		}
	}

	// remove closed PRs from prset
	prset.RemoveClosedPRs()
	return nil
}

func (pr *PRProcessor) Process(req *models.PullRequest) error {
	config := pr.config
	git := pr.git

	// requests against project are not handled here
	common.LogInfo("processing opened PR:", req.URL)
	prOrg := req.Base.Repo.Owner.UserName
	prRepo := req.Base.Repo.Name
	prNo := req.Index

	common.LogError(req)

	prset, err := common.FetchPRSet(CurrentUser.UserName, Gitea, prOrg, prRepo, prNo, config)
	if err != nil {
		common.LogError("Cannot fetch PRSet:", err)
		return err
	}
	common.LogInfo("fetched PRSet of size:", len(prset.PRs))

	prjGitPRbranch := prGitBranchNameForPR(prRepo, prNo)
	prjGitPR, err := prset.GetPrjGitPR()
	if err == common.PRSet_PrjGitMissing {
		common.LogDebug("Missing PrjGit. Need to create one under branch", prjGitPRbranch)

		if err = pr.CreatePRjGitPR(prjGitPRbranch, prset); err != nil {
			return err
		}
	} else if err == nil {
		common.LogDebug("Found PrjGit PR:", common.PRtoString(prjGitPR.PR))
		prjGitPRbranch = prjGitPR.PR.Head.Name

		if prjGitPR.PR.State != "open" {
			if prjGitPR.PR.HasMerged {
				// update branches in project
				prjGitPR.RemoteName, err = git.GitClone(common.DefaultGitPrj, prjGitPRbranch, prjGitPR.PR.Base.Repo.SSHURL)
				common.PanicOnError(err)

				old_pkgs, err := git.GitSubmoduleList(common.DefaultGitPrj, prjGitPR.PR.MergeBase)
				common.PanicOnError(err)
				new_pkgs, err := git.GitSubmoduleList(common.DefaultGitPrj, prjGitPRbranch)
				common.PanicOnError(err)

				pkgs := make(map[string]string)
				for pkg, old_commit := range old_pkgs {
					if new_commit, found := new_pkgs[pkg]; found {
						// pkg modified
						if new_commit != old_commit {
							pkgs[pkg] = new_commit
						}
					} else { // not found, pkg removed
						pkgs[pkg] = ""
					}
				}
				for pkg, commit := range new_pkgs {
					if _, found := old_pkgs[pkg]; !found {
						// pkg added
						pkgs[pkg] = commit
					}
				}

				PrjGitSubmoduleCheck(config, git, common.DefaultGitPrj, pkgs)
			}

			// manually merge or close entire prset that is still open
			for _, pr := range prset.PRs {
				if pr.PR.State == "open" {
					org, repo, idx := pr.PRComponents()
					if prjGitPR.PR.HasMerged {
						Gitea.AddComment(pr.PR, "This PR is merged via the associated Project PR.")
						err = Gitea.ManualMergePR(org, repo, idx, pr.PR.Head.Sha, false)
						if _, ok := err.(*repository.RepoMergePullRequestConflict); !ok {
							common.PanicOnError(err)
						}
					} else {
						Gitea.AddComment(pr.PR, "Closing here because the associated Project PR has been closed.")
						Gitea.UpdatePullRequest(org, repo, idx, &models.EditPullRequestOption{
							State: "closed",
						})
					}
				}
			}
			return nil
		}

		if len(prset.PRs) > 1 {
			for _, pr := range prset.PRs {
				if prset.IsPrjGitPR(pr.PR) {
					continue
				}
			}
		}

		if err = pr.UpdatePrjGitPR(prset); err != nil {
			return err
		}
	}

	if prjGitPR == nil {
		prjGitPR, err = prset.GetPrjGitPR()
		if err == common.PRSet_PrjGitMissing && config.NoProjectGitPR {
			// we could be waiting for other tooling to create the
			// project git PR. In meantime, we can assign some
			// reviewers here.
		} else if err != nil {
			common.LogError("Error fetching PrjGitPR:", err)
			return nil
		}
	}

	common.LogDebug("Updated PR")

	// make sure that prjgit is consistent and only submodules that are to be *updated*
	// reset anything that changed that is not part of the prset
	// package removals/additions are *not* counted here
	org, repo, branch := config.GetPrjGit()
	// TODO: this is broken...
	if pr, err := prset.GetPrjGitPR(); err == nil && false {
		common.LogDebug("Submodule parse begin")
		orig_subs, err := git.GitSubmoduleList(common.DefaultGitPrj, pr.PR.MergeBase)
		common.PanicOnError(err)
		new_subs, err := git.GitSubmoduleList(common.DefaultGitPrj, "HEAD")
		common.PanicOnError(err)
		common.LogDebug("Submodule parse done")

		reset_submodule := func(submodule, sha string) {
			updateSubmoduleInPR(submodule, sha, git)
		}

		common.LogDebug("Checking we only change linked commits")
		for path, commit := range new_subs {
			if old, ok := orig_subs[path]; ok && old != commit {
				found := false
				for _, pr := range prset.PRs {
					if pr.PR.Base.Repo.Name == path && commit == pr.PR.Head.Sha {
						found = true
						break
					} else if pr.PR.Base.Repo.Name == path {
						common.LogError(path, "-- commits not match", commit, pr.PR.Head.Sha)
					}
				}
				if !found {
					reset_submodule(path, old)
				}
			}
		}

		stats, err := git.GitStatus(common.DefaultGitPrj)
		common.LogDebug("Check Done", len(stats), "changes")
		common.PanicOnError(err)
		if len(stats) > 0 {
			git.GitExecOrPanic(common.DefaultGitPrj, "commit", "-a", "-m", "Sync submodule updates with PR-set")
			git.GitExecQuietOrPanic(common.DefaultGitPrj, "submodule", "deinit", "--all", "--force")
			if !common.IsDryRun {
				git.GitExecOrPanic(common.DefaultGitPrj, "push")
			}
		}
	}

	if prjGitPR != nil {
		common.LogDebug("  num of reviewers:", len(prjGitPR.PR.RequestedReviewers))
	} else {
		common.LogInfo("* No prjgit")
	}
	maintainers, err := common.FetchProjectMaintainershipData(Gitea, org, repo, branch)
	if err != nil {
		return err
	}

	// handle case where PrjGit PR is only one left and there are no changes, then we can just close the PR
	if len(prset.PRs) == 1 && prjGitPR != nil && prset.PRs[0] == prjGitPR && prjGitPR.PR.User.UserName == prset.BotUser {
		common.LogDebug("  --> checking if superflous PR")
		diff, err := git.GitDiff(common.DefaultGitPrj, prjGitPR.PR.MergeBase, prjGitPR.PR.Head.Sha)
		if err != nil {
			return err
		}
		if len(diff) == 0 {
			common.LogInfo("PR is no-op and can be closed. Closing.")
			if !common.IsDryRun {
				Gitea.AddComment(prjGitPR.PR, "Pull request no longer contains any changes. Closing.")
				_, err = Gitea.UpdatePullRequest(prjGitPR.PR.Base.Repo.Owner.UserName, prjGitPR.PR.Base.Repo.Name, prjGitPR.PR.Index, &models.EditPullRequestOption{
					State: "closed",
				})
				if err != nil {
					return err
				}
			}
			return nil
		}
		common.LogDebug("  --> NOT superflous PR")
	}

	for _, pr := range prset.PRs {
		if err := verifyRepositoryConfiguration(pr.PR.Base.Repo); err != nil {
			common.LogError("Cannot set manual merge... aborting processing")
			return err
		}
	}

	common.LogInfo("Consistent PRSet:", prset.IsConsistent())
	common.LogInfo("Reviewed?", prset.IsApproved(Gitea, maintainers))
	if prset.IsConsistent() && prset.IsApproved(Gitea, maintainers) {
		common.LogInfo("Merging...")
		if err = prset.Merge(Gitea, git); err != nil {
			common.LogError("merge error:", err)
		}
	} else {
		prset.AssignReviewers(Gitea, maintainers)
	}
	return err
}

type RequestProcessor struct {
	configuredRepos map[string][]*common.AutogitConfig
}

func ProcesPullRequest(pr *models.PullRequest, configs []*common.AutogitConfig) error {
	if len(configs) < 1 {
		// ignoring pull request against unconfigured project (could be just regular sources?)
		return nil
	}

	PRProcessor, err := AllocatePRProcessor(pr, configs)
	if err != nil {
		log.Error(err)
		return err
	}
	defer PRProcessor.git.Close()

	return PRProcessor.Process(pr)
}

func (w *RequestProcessor) ProcessFunc(request *common.Request) error {
	defer func() {
		if r := recover(); r != nil {
			common.LogInfo("panic cought --- recovered")
			common.LogError(string(debug.Stack()))
		}
	}()

	var pr *models.PullRequest
	var err error
	if req, ok := request.Data.(*common.PullRequestWebhookEvent); ok {
		pr, err = Gitea.GetPullRequest(req.Pull_Request.Base.Repo.Owner.Username, req.Pull_Request.Base.Repo.Name, req.Pull_Request.Number)
		if err != nil {
			common.LogError("Cannot find PR for issue:", req.Pull_Request.Base.Repo.Owner.Username, req.Pull_Request.Base.Repo.Name, req.Pull_Request.Number)
			return err
		}
	} else if req, ok := request.Data.(*common.IssueWebhookEvent); ok {
		pr, err = Gitea.GetPullRequest(req.Repository.Owner.Username, req.Repository.Name, int64(req.Issue.Number))
		if err != nil {
			common.LogError("Cannot find PR for issue:", req.Repository.Owner.Username, req.Repository.Name, int64(req.Issue.Number))
			return err
		}
	} else {
		common.LogError("*** Invalid data format for PR processing.")
		return fmt.Errorf("*** Invalid data format for PR processing.")
	}

	configs, ok := w.configuredRepos[pr.Base.Repo.Owner.UserName]
	if !ok {
		common.LogError("*** Cannot find config for org:", pr.Base.Repo.Owner.UserName)
	}
	return ProcesPullRequest(pr, configs)
}