Files
gitea/services/migrations/gitea_downloader.go
T
silverwindandbircni fc49368845 feat: Improve Gitea-compatible forge migrations, fix many migration bugs (#39424)
Supports migrating from Gitea-compatible forges like Forgejo and fixes
many migration bugs.

1. Pre-release and `+gitea-` suffixed server versions no longer drop
reviews, reactions and pagination.
2. Review states, dismissals, scoped and archived labels and merge bases
are migrated.
3. Disabled units, external trackers, unavailable release assets, and
GitLab confidential issues and internal notes are skipped.
4. Credentials are only sent to the source host, and `Retry-After` rate
limits are waited out.
5. GitHub migrations of only issues or pull requests, basic auth clones
and dismissed reviews work again.
6. The API returns 404 for releases of a disabled unit and shows
migrated authors as mannequins, replacing the never populated
`original_author` and `original_author_id` fields.

Fixes: https://github.com/go-gitea/gitea/issues/19533
Fixes: https://github.com/go-gitea/gitea/issues/35885
Fixes: https://github.com/go-gitea/gitea/issues/18275
Fixes: https://github.com/go-gitea/gitea/issues/35069
Fixes: https://github.com/go-gitea/gitea/issues/17763
Fixes: https://github.com/go-gitea/gitea/issues/39548
Co-authored-by: bircni <bircni@icloud.com>
2026-10-24 01:15:31 +00:00

748 lines
21 KiB
Go

// Copyright 2020 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package migrations
import (
"cmp"
"context"
"errors"
"fmt"
"io"
"net/http"
"net/url"
"strings"
"time"
"gitea.dev/modules/container"
"gitea.dev/modules/log"
base "gitea.dev/modules/migration"
"gitea.dev/modules/structs"
"gitea.dev/modules/util"
gitea_sdk "gitea.dev/sdk"
"github.com/hashicorp/go-version"
)
var (
_ base.Downloader = &GiteaDownloader{}
_ base.DownloaderFactory = &GiteaDownloaderFactory{}
)
func init() {
RegisterDownloaderFactory(&GiteaDownloaderFactory{})
}
// GiteaDownloaderFactory defines a gitea downloader factory
type GiteaDownloaderFactory struct{}
// New returns a Downloader related to this factory according MigrateOptions
func (f *GiteaDownloaderFactory) New(ctx context.Context, opts base.MigrateOptions) (base.Downloader, error) {
u, err := url.Parse(opts.CloneAddr)
if err != nil {
return nil, err
}
baseURL := u.Scheme + "://" + u.Host
repoNameSpace := strings.TrimPrefix(u.Path, "/")
repoNameSpace = strings.TrimSuffix(repoNameSpace, ".git")
path := strings.Split(repoNameSpace, "/")
if len(path) < 2 {
return nil, fmt.Errorf("invalid path: %s", repoNameSpace)
}
repoPath := strings.Join(path[len(path)-2:], "/")
if len(path) > 2 {
subPath := strings.Join(path[:len(path)-2], "/")
baseURL += "/" + subPath
}
log.Trace("Create gitea downloader. BaseURL: %s RepoName: %s", baseURL, repoNameSpace)
return NewGiteaDownloader(ctx, baseURL, repoPath, opts.AuthUsername, opts.AuthPassword, opts.AuthToken)
}
// GitServiceType returns the type of git service
func (f *GiteaDownloaderFactory) GitServiceType() structs.GitServiceType {
return structs.GiteaService
}
// GiteaDownloader implements a Downloader interface to get repository information's
type GiteaDownloader struct {
base.NullDownloader
ctx context.Context
client *gitea_sdk.Client
baseURL string
repoOwner string
repoName string
httpClient *http.Client
pagination bool
maxPerPage int
}
// NewGiteaDownloader creates a gitea Downloader via gitea API
//
// Use either a username/password or personal token. token is preferred
// Note: Public access only allows very basic access
func NewGiteaDownloader(ctx context.Context, baseURL, repoPath, username, password, token string) (*GiteaDownloader, error) {
httpClient := newMigrationHTTPClient(baseURL, util.Iif(token != "", "token "+token, basicAuthorization(username, password)))
giteaClient, err := gitea_sdk.NewClient(baseURL, gitea_sdk.SetHTTPClient(httpClient))
if err != nil {
log.Error(fmt.Sprintf("Failed to create NewGiteaDownloader for: %s. Error: %v", baseURL, err))
return nil, err
}
if err = pinGiteaVersion(ctx, giteaClient); err != nil {
return nil, err
}
path := strings.Split(repoPath, "/")
paginationSupport := true
if err = giteaClient.CheckServerVersionConstraint(ctx, ">=1.12"); err != nil {
paginationSupport = false
}
// set small maxPerPage since we can only guess
// (default would be 50 but this can differ)
maxPerPage := 10
// gitea instances >=1.13 can tell us what maximum they have
apiConf, _, err := giteaClient.Settings.GetGlobalAPISettings(ctx)
if err != nil {
log.Info("Unable to get global API settings. Ignoring these.")
log.Debug("giteaClient.GetGlobalAPISettings. Error: %v", err)
} else if apiConf != nil && apiConf.MaxResponseItems > 0 {
maxPerPage = min(apiConf.MaxResponseItems, 100)
}
return &GiteaDownloader{
ctx: ctx,
client: giteaClient,
baseURL: baseURL,
repoOwner: path[0],
repoName: path[1],
httpClient: httpClient,
pagination: paginationSupport,
maxPerPage: maxPerPage,
}, nil
}
func pinGiteaVersion(ctx context.Context, client *gitea_sdk.Client) error { // the SDK's checks reject pre-releases
rawVersion, _, err := client.Meta.ServerVersion(ctx)
if err != nil {
return err
}
if serverVersion, err := version.NewVersion(rawVersion); err == nil {
if giteaVersion, ok := strings.CutPrefix(serverVersion.Metadata(), "gitea-"); ok { // Forgejo
if serverVersion, err = version.NewVersion(giteaVersion); err != nil {
return err
}
}
return gitea_sdk.SetGiteaVersion(serverVersion.Core().String())(client)
}
return nil
}
func (g *GiteaDownloader) unitDisabled(resp *gitea_sdk.Response, hasUnit func(*gitea_sdk.Repository) bool) bool { // Gitea answers a disabled unit and missing access alike with 403 or 404, Forgejo with 404
if resp == nil || (resp.StatusCode != http.StatusNotFound && resp.StatusCode != http.StatusForbidden) {
return false
}
repo, _, err := g.client.Repositories.GetRepo(g.ctx, g.repoOwner, g.repoName)
return err == nil && !hasUnit(repo)
}
// String implements Stringer
func (g *GiteaDownloader) String() string {
return fmt.Sprintf("migration from gitea server %s %s/%s", g.baseURL, g.repoOwner, g.repoName)
}
func (g *GiteaDownloader) LogString() string {
if g == nil {
return "<GiteaDownloader nil>"
}
return fmt.Sprintf("<GiteaDownloader %s %s/%s>", g.baseURL, g.repoOwner, g.repoName)
}
// GetRepoInfo returns a repository information
func (g *GiteaDownloader) GetRepoInfo(_ context.Context) (*base.Repository, error) {
if g == nil {
return nil, errors.New("error: GiteaDownloader is nil")
}
repo, _, err := g.client.Repositories.GetRepo(g.ctx, g.repoOwner, g.repoName)
if err != nil {
return nil, err
}
return &base.Repository{
Name: repo.Name,
Owner: repo.Owner.UserName,
IsPrivate: repo.Private,
Description: repo.Description,
Website: repo.Website,
CloneURL: repo.CloneURL,
OriginalURL: repo.HTMLURL,
DefaultBranch: repo.DefaultBranch,
}, nil
}
// GetTopics return gitea topics
func (g *GiteaDownloader) GetTopics(_ context.Context) ([]string, error) {
topics, _, err := g.client.Repositories.ListRepoTopics(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListRepoTopicsOptions{})
return topics, err
}
// GetMilestones returns milestones
func (g *GiteaDownloader) GetMilestones(ctx context.Context) ([]*base.Milestone, error) {
milestones := make([]*base.Milestone, 0, g.maxPerPage)
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, nil
default:
}
ms, _, err := g.client.Repositories.ListMilestones(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListMilestoneOption{
ListOptions: gitea_sdk.ListOptions{
PageSize: g.maxPerPage,
Page: i,
},
State: gitea_sdk.StateAll,
})
if err != nil {
return nil, err
}
for i := range ms {
// old gitea instances don't have this information
createdAT := time.Time{}
var updatedAT *time.Time
if ms[i].Closed != nil {
createdAT = *ms[i].Closed
updatedAT = ms[i].Closed
}
// new gitea instances (>=1.13) do
if !ms[i].Created.IsZero() {
createdAT = ms[i].Created
}
if ms[i].Updated != nil && !ms[i].Updated.IsZero() {
updatedAT = ms[i].Updated
}
milestones = append(milestones, &base.Milestone{
Title: ms[i].Title,
Description: ms[i].Description,
Deadline: ms[i].Deadline,
Created: createdAT,
Updated: updatedAT,
Closed: ms[i].Closed,
State: string(ms[i].State),
})
}
if !g.pagination || len(ms) < g.maxPerPage {
break
}
}
return milestones, nil
}
func (g *GiteaDownloader) convertGiteaLabel(label *gitea_sdk.Label) *base.Label {
return &base.Label{
Name: label.Name,
Color: label.Color,
Description: label.Description,
Exclusive: label.Exclusive,
Archived: label.IsArchived,
}
}
// GetLabels returns labels
func (g *GiteaDownloader) GetLabels(ctx context.Context) ([]*base.Label, error) {
labels := make([]*base.Label, 0, g.maxPerPage)
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, nil
default:
}
ls, _, err := g.client.Repositories.ListRepoLabels(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListLabelsOptions{ListOptions: gitea_sdk.ListOptions{
PageSize: g.maxPerPage,
Page: i,
}})
if err != nil {
return nil, err
}
for i := range ls {
labels = append(labels, g.convertGiteaLabel(ls[i]))
}
if !g.pagination || len(ls) < g.maxPerPage {
break
}
}
return labels, nil
}
func (g *GiteaDownloader) convertGiteaRelease(rel *gitea_sdk.Release) *base.Release {
r := &base.Release{
TagName: rel.TagName,
TargetCommitish: rel.Target,
Name: rel.Title,
Body: rel.Note,
Draft: rel.IsDraft,
Prerelease: rel.IsPrerelease,
PublisherID: rel.Publisher.ID,
PublisherName: rel.Publisher.UserName,
PublisherEmail: rel.Publisher.Email,
Published: rel.PublishedAt,
Created: rel.CreatedAt,
}
for _, asset := range rel.Attachments {
size := int(asset.Size)
dlCount := int(asset.DownloadCount)
r.Assets = append(r.Assets, &base.ReleaseAsset{
ID: asset.ID,
Name: asset.Name,
Size: &size,
DownloadCount: &dlCount,
Created: asset.Created,
DownloadFunc: func() (io.ReadCloser, error) {
return downloadAsset(g.ctx, g.httpClient, g.baseURL+"/attachments/"+url.PathEscape(asset.UUID))
},
})
}
return r
}
// GetReleases returns releases
func (g *GiteaDownloader) GetReleases(ctx context.Context) ([]*base.Release, error) {
releases := make([]*base.Release, 0, g.maxPerPage)
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, nil
default:
}
rl, resp, err := g.client.Releases.ListReleases(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListReleasesOptions{ListOptions: gitea_sdk.ListOptions{
PageSize: g.maxPerPage,
Page: i,
}})
if g.unitDisabled(resp, func(repo *gitea_sdk.Repository) bool { return repo.HasReleases }) {
return releases, nil
}
if err != nil {
return nil, err
}
for i := range rl {
releases = append(releases, g.convertGiteaRelease(rl[i]))
}
if !g.pagination || len(rl) < g.maxPerPage {
break
}
}
return releases, nil
}
func (g *GiteaDownloader) getIssueReactions(ctx context.Context, index int64) ([]*base.Reaction, error) {
if err := g.client.CheckServerVersionConstraint(g.ctx, ">=1.11"); err != nil {
log.Info("GiteaDownloader: instance to old, skip getIssueReactions")
return nil, nil
}
allReactions := make([]*base.Reaction, 0, g.maxPerPage)
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, nil
default:
}
reactions, _, err := g.client.Issues.ListIssueReactions(g.ctx, g.repoOwner, g.repoName, index, gitea_sdk.ListIssueReactionsOptions{ListOptions: gitea_sdk.ListOptions{
PageSize: g.maxPerPage,
Page: i,
}})
if err != nil {
return nil, err
}
for _, reaction := range reactions {
allReactions = append(allReactions, &base.Reaction{
UserID: reaction.User.ID,
UserName: reaction.User.UserName,
Content: reaction.Reaction,
})
}
if !g.pagination || len(reactions) < g.maxPerPage {
break
}
}
return allReactions, nil
}
func (g *GiteaDownloader) getCommentReactions(commentID int64) ([]*base.Reaction, error) {
var reactions []*base.Reaction
if err := g.client.CheckServerVersionConstraint(g.ctx, ">=1.11"); err != nil {
log.Info("GiteaDownloader: instance to old, skip getCommentReactions")
return reactions, nil
}
rl, _, err := g.client.Issues.GetIssueCommentReactions(g.ctx, g.repoOwner, g.repoName, commentID)
if err != nil {
return nil, err
}
for i := range rl {
reactions = append(reactions, &base.Reaction{
UserID: rl[i].User.ID,
UserName: rl[i].User.UserName,
Content: rl[i].Reaction,
})
}
return reactions, nil
}
// GetIssues returns issues according start and limit
func (g *GiteaDownloader) GetIssues(ctx context.Context, page, perPage int) ([]*base.Issue, bool, error) {
if perPage > g.maxPerPage {
perPage = g.maxPerPage
}
allIssues := make([]*base.Issue, 0, perPage)
issues, resp, err := g.client.Issues.ListRepoIssues(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListIssueOption{
ListOptions: gitea_sdk.ListOptions{Page: page, PageSize: perPage},
State: gitea_sdk.StateAll,
Type: gitea_sdk.IssueTypeIssue,
})
if g.unitDisabled(resp, func(repo *gitea_sdk.Repository) bool { return repo.HasIssues && repo.ExternalTracker == nil }) {
return nil, true, nil
}
if err != nil {
return nil, false, fmt.Errorf("error while listing issues: %w", err)
}
for _, issue := range issues {
labels := make([]*base.Label, 0, len(issue.Labels))
for i := range issue.Labels {
labels = append(labels, g.convertGiteaLabel(issue.Labels[i]))
}
var milestone string
if issue.Milestone != nil {
milestone = issue.Milestone.Title
}
reactions, err := g.getIssueReactions(ctx, issue.Index)
if err != nil {
WarnAndNotice("Unable to load reactions during migrating issue #%d in %s. Error: %v", issue.Index, g, err)
}
var assignees []string
for i := range issue.Assignees {
assignees = append(assignees, issue.Assignees[i].UserName)
}
allIssues = append(allIssues, &base.Issue{
Title: issue.Title,
Number: issue.Index,
PosterID: issue.Poster.ID,
PosterName: issue.Poster.UserName,
PosterEmail: issue.Poster.Email,
Content: issue.Body,
Milestone: milestone,
State: string(issue.State),
Created: issue.Created,
Updated: issue.Updated,
Closed: issue.Closed,
Reactions: reactions,
Labels: labels,
Assignees: assignees,
IsLocked: issue.IsLocked,
ForeignIndex: issue.Index,
Context: issue.Comments,
})
}
isEnd := len(issues) < perPage
if !g.pagination {
isEnd = len(issues) == 0
}
return allIssues, isEnd, nil
}
// GetComments returns comments according issueNumber
func (g *GiteaDownloader) GetComments(ctx context.Context, commentable base.Commentable) ([]*base.Comment, bool, error) {
if commentable.GetContext() == 0 {
return nil, true, nil
}
allComments := make([]*base.Comment, 0, g.maxPerPage)
seenIDs := container.Set[int64]{}
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, false, nil
default:
}
comments, _, err := g.client.Issues.ListIssueComments(g.ctx, g.repoOwner, g.repoName, commentable.GetForeignIndex(), gitea_sdk.ListIssueCommentOptions{ListOptions: gitea_sdk.ListOptions{
PageSize: g.maxPerPage,
Page: i,
}})
if err != nil {
return nil, false, fmt.Errorf("error while listing comments for issue #%d. Error: %w", commentable.GetForeignIndex(), err)
}
if len(comments) == 0 || seenIDs.Contains(comments[0].ID) {
break // the endpoint ignores page and limit, so page 2 repeats page 1
}
for _, comment := range comments {
seenIDs.Add(comment.ID)
reactions, err := g.getCommentReactions(comment.ID)
if err != nil {
WarnAndNotice("Unable to load comment reactions during migrating issue #%d for comment %d in %s. Error: %v", commentable.GetForeignIndex(), comment.ID, g, err)
}
allComments = append(allComments, &base.Comment{
IssueIndex: commentable.GetLocalIndex(),
Index: comment.ID,
PosterID: comment.Poster.ID,
PosterName: comment.Poster.UserName,
PosterEmail: comment.Poster.Email,
Content: comment.Body,
Created: comment.Created,
Updated: comment.Updated,
Reactions: reactions,
})
}
if !g.pagination || len(comments) != g.maxPerPage {
break
}
}
return allComments, true, nil
}
// GetPullRequests returns pull requests according page and perPage
func (g *GiteaDownloader) GetPullRequests(ctx context.Context, page, perPage int) ([]*base.PullRequest, bool, error) {
if perPage > g.maxPerPage {
perPage = g.maxPerPage
}
allPRs := make([]*base.PullRequest, 0, perPage)
prs, resp, err := g.client.PullRequests.ListRepoPullRequests(g.ctx, g.repoOwner, g.repoName, gitea_sdk.ListPullRequestsOptions{
ListOptions: gitea_sdk.ListOptions{
Page: page,
PageSize: perPage,
},
State: gitea_sdk.StateAll,
})
if g.unitDisabled(resp, func(repo *gitea_sdk.Repository) bool { return repo.HasPullRequests }) {
return nil, true, nil
}
if err != nil {
return nil, false, fmt.Errorf("error while listing pull requests (page: %d, pagesize: %d). Error: %w", page, perPage, err)
}
for _, pr := range prs {
var milestone string
if pr.Milestone != nil {
milestone = pr.Milestone.Title
}
labels := make([]*base.Label, 0, len(pr.Labels))
for i := range pr.Labels {
labels = append(labels, g.convertGiteaLabel(pr.Labels[i]))
}
var (
headUserName string
headRepoName string
headCloneURL string
headRef string
headSHA string
)
if pr.Head != nil {
if pr.Head.Repository != nil {
headUserName = pr.Head.Repository.Owner.UserName
headRepoName = pr.Head.Repository.Name
headCloneURL = pr.Head.Repository.CloneURL
}
headSHA = pr.Head.Sha
headRef = pr.Head.Ref
}
var mergeCommitSHA string
if pr.MergedCommitID != nil {
mergeCommitSHA = *pr.MergedCommitID
}
reactions, err := g.getIssueReactions(ctx, pr.Index)
if err != nil {
WarnAndNotice("Unable to load reactions during migrating pull #%d in %s. Error: %v", pr.Index, g, err)
}
var assignees []string
for i := range pr.Assignees {
assignees = append(assignees, pr.Assignees[i].UserName)
}
createdAt := time.Time{}
if pr.Created != nil {
createdAt = *pr.Created
}
updatedAt := time.Time{}
if pr.Updated != nil {
updatedAt = *pr.Updated
}
closedAt := pr.Closed
if pr.Merged != nil && closedAt == nil {
closedAt = pr.Merged
}
allPRs = append(allPRs, &base.PullRequest{
Title: pr.Title,
Number: pr.Index,
PosterID: pr.Poster.ID,
PosterName: pr.Poster.UserName,
PosterEmail: pr.Poster.Email,
Content: pr.Body,
State: string(pr.State),
Created: createdAt,
Updated: updatedAt,
Closed: closedAt,
Labels: labels,
Milestone: milestone,
Reactions: reactions,
Assignees: assignees,
Merged: pr.HasMerged,
MergedTime: pr.Merged,
MergeCommitSHA: mergeCommitSHA,
IsLocked: pr.IsLocked,
Head: base.PullRequestBranch{
Ref: headRef,
SHA: headSHA,
RepoName: headRepoName,
OwnerName: headUserName,
CloneURL: headCloneURL,
},
Base: base.PullRequestBranch{
Ref: pr.Base.Ref,
SHA: cmp.Or(pr.MergeBase, pr.Base.Sha), // base.sha is the base branch head, which contains the changes once merged
RepoName: g.repoName,
OwnerName: g.repoOwner,
},
ForeignIndex: pr.Index,
Context: pr.Comments,
})
// SECURITY: Ensure that the PR is safe
_ = CheckAndEnsureSafePR(allPRs[len(allPRs)-1], g.baseURL, g)
}
isEnd := len(prs) < perPage
if !g.pagination {
isEnd = len(prs) == 0
}
return allPRs, isEnd, nil
}
// GetReviews returns pull requests review
func (g *GiteaDownloader) GetReviews(ctx context.Context, reviewable base.Reviewable) ([]*base.Review, error) {
if err := g.client.CheckServerVersionConstraint(g.ctx, ">=1.12"); err != nil {
log.Info("GiteaDownloader: instance to old, skip GetReviews")
return nil, nil
}
allReviews := make([]*base.Review, 0, g.maxPerPage)
for i := 1; ; i++ {
// make sure gitea can shutdown gracefully
select {
case <-ctx.Done():
return nil, nil
default:
}
prl, _, err := g.client.PullRequests.ListPullReviews(g.ctx, g.repoOwner, g.repoName, reviewable.GetForeignIndex(), gitea_sdk.ListPullReviewsOptions{ListOptions: gitea_sdk.ListOptions{
Page: i,
PageSize: g.maxPerPage,
}})
if err != nil {
return nil, err
}
for _, pr := range prl {
if pr.Reviewer == nil {
// Presumably this is a team review which we cannot migrate at present but we have to skip this review as otherwise the review will be mapped on to an incorrect user.
// TODO: handle team reviews
continue
}
var rcl []*gitea_sdk.PullReviewComment
if pr.CodeCommentsCount > 0 {
if rcl, _, err = g.client.PullRequests.ListPullReviewComments(g.ctx, g.repoOwner, g.repoName, reviewable.GetForeignIndex(), pr.ID); err != nil {
return nil, err
}
}
var reviewComments []*base.ReviewComment
for i := range rcl {
line := int(rcl[i].LineNum)
if rcl[i].OldLineNum > 0 {
line = int(rcl[i].OldLineNum) * -1
}
reviewComments = append(reviewComments, &base.ReviewComment{
ID: rcl[i].ID,
Content: rcl[i].Body,
TreePath: rcl[i].Path,
DiffHunk: rcl[i].DiffHunk,
Line: line,
CommitID: rcl[i].CommitID,
PosterID: rcl[i].Reviewer.ID,
CreatedAt: rcl[i].Created,
UpdatedAt: rcl[i].Updated,
})
}
review := &base.Review{
ID: pr.ID,
IssueIndex: reviewable.GetLocalIndex(),
ReviewerID: pr.Reviewer.ID,
ReviewerName: pr.Reviewer.UserName,
Official: pr.Official,
CommitID: pr.CommitID,
Content: pr.Body,
CreatedAt: pr.Submitted,
State: string(pr.State),
Dismissed: pr.Dismissed,
Comments: reviewComments,
}
switch pr.State {
case gitea_sdk.ReviewStateComment:
review.State = base.ReviewStateCommented
case gitea_sdk.ReviewStateRequestChanges:
review.State = base.ReviewStateChangesRequested
}
allReviews = append(allReviews, review)
}
if len(prl) < g.maxPerPage {
break
}
}
return allReviews, nil
}