Merge branch 'main' into lunny/issue_dev

services/auth/oauth2.go

@@ -17,7 +17,7 @@ import (
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/timeutil"
 	"code.gitea.io/gitea/modules/web/middleware"
-	"code.gitea.io/gitea/services/auth/source/oauth2"
+	"code.gitea.io/gitea/services/oauth2_provider"
 )
 
 // Ensure the struct implements the interface.
@@ -31,7 +31,7 @@ func CheckOAuthAccessToken(ctx context.Context, accessToken string) int64 {
 	if !strings.Contains(accessToken, ".") {
 		return 0
 	}
-	token, err := oauth2.ParseToken(accessToken, oauth2.DefaultSigningKey)
+	token, err := oauth2_provider.ParseToken(accessToken, oauth2_provider.DefaultSigningKey)
 	if err != nil {
 		log.Trace("oauth2.ParseToken: %v", err)
 		return 0
@@ -40,7 +40,7 @@ func CheckOAuthAccessToken(ctx context.Context, accessToken string) int64 {
 	if grant, err = auth_model.GetOAuth2GrantByID(ctx, token.GrantID); err != nil || grant == nil {
 		return 0
 	}
-	if token.Type != oauth2.TypeAccessToken {
+	if token.Kind != oauth2_provider.KindAccessToken {
 		return 0
 	}
 	if token.ExpiresAt.Before(time.Now()) || token.IssuedAt.After(time.Now()) {

services/auth/source/oauth2/init.go

@@ -30,10 +30,6 @@ const ProviderHeaderKey = "gitea-oauth2-provider"
 
 // Init initializes the oauth source
 func Init(ctx context.Context) error {
-	if err := InitSigningKey(); err != nil {
-		return err
-	}
-
 	// Lock our mutex
 	gothRWMutex.Lock()
 

services/context/repo.go

@@ -404,6 +404,13 @@ func repoAssignment(ctx *Context, repo *repo_model.Repository) {
 	ctx.Data["PushMirrors"] = pushMirrors
 	ctx.Data["RepoName"] = ctx.Repo.Repository.Name
 	ctx.Data["IsEmptyRepo"] = ctx.Repo.Repository.IsEmpty
+
+	repoLicenses, err := repo_model.GetRepoLicenses(ctx, ctx.Repo.Repository)
+	if err != nil {
+		ctx.ServerError("GetRepoLicenses", err)
+		return
+	}
+	ctx.Data["DetectedRepoLicenses"] = repoLicenses.StringList()
 }
 
 // RepoAssignment returns a middleware to handle repository assignment

services/convert/repository.go

@@ -175,6 +175,11 @@ func innerToRepo(ctx context.Context, repo *repo_model.Repository, permissionInR
 		language = repo.PrimaryLanguage.Language
 	}
 
+	repoLicenses, err := repo_model.GetRepoLicenses(ctx, repo)
+	if err != nil {
+		return nil
+	}
+
 	repoAPIURL := repo.APIURL()
 
 	return &api.Repository{
@@ -238,6 +243,7 @@ func innerToRepo(ctx context.Context, repo *repo_model.Repository, permissionInR
 		RepoTransfer:                  transfer,
 		Topics:                        repo.Topics,
 		ObjectFormatName:              repo.ObjectFormatName,
+		Licenses:                      repoLicenses.StringList(),
 	}
 }
 

services/cron/tasks_basic.go

@@ -156,6 +156,16 @@ func registerCleanupPackages() {
 	})
 }
 
+func registerSyncRepoLicenses() {
+	RegisterTaskFatal("sync_repo_licenses", &BaseConfig{
+		Enabled:    false,
+		RunAtStart: false,
+		Schedule:   "@annually",
+	}, func(ctx context.Context, _ *user_model.User, config Config) error {
+		return repo_service.SyncRepoLicenses(ctx)
+	})
+}
+
 func initBasicTasks() {
 	if setting.Mirror.Enabled {
 		registerUpdateMirrorTask()
@@ -172,4 +182,5 @@ func initBasicTasks() {
 	if setting.Packages.Enabled {
 		registerCleanupPackages()
 	}
+	registerSyncRepoLicenses()
 }

services/migrations/gitea_uploader_test.go

@@ -26,6 +26,7 @@ import (
 	"code.gitea.io/gitea/modules/optional"
 	"code.gitea.io/gitea/modules/structs"
 	"code.gitea.io/gitea/modules/test"
+	repo_service "code.gitea.io/gitea/services/repository"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -302,6 +303,8 @@ func TestGiteaUploadUpdateGitForPullRequest(t *testing.T) {
 	toRepoName := "migrated"
 	uploader := NewGiteaLocalUploader(context.Background(), fromRepoOwner, fromRepoOwner.Name, toRepoName)
 	uploader.gitServiceType = structs.GiteaService
+
+	assert.NoError(t, repo_service.Init(context.Background()))
 	assert.NoError(t, uploader.CreateRepo(&base.Repository{
 		Description: "description",
 		OriginalURL: fromRepo.RepoPath(),

services/mirror/mirror_pull.go

@@ -24,6 +24,7 @@ import (
 	"code.gitea.io/gitea/modules/timeutil"
 	"code.gitea.io/gitea/modules/util"
 	notify_service "code.gitea.io/gitea/services/notify"
+	repo_service "code.gitea.io/gitea/services/repository"
 )
 
 // gitShortEmptySha Git short empty SHA
@@ -559,6 +560,14 @@ func SyncPullMirror(ctx context.Context, repoID int64) bool {
 		}
 	}
 
+	// Update License
+	if err = repo_service.AddRepoToLicenseUpdaterQueue(&repo_service.LicenseUpdaterOptions{
+		RepoID: m.Repo.ID,
+	}); err != nil {
+		log.Error("SyncMirrors [repo: %-v]: unable to add repo to license updater queue: %v", m.Repo, err)
+		return false
+	}
+
 	log.Trace("SyncMirrors [repo: %-v]: Successfully updated", m.Repo)
 
 	return true

services/oauth2_provider/access_token.go

@@ -0,0 +1,214 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package oauth2_provider //nolint
+
+import (
+	"context"
+	"fmt"
+
+	auth "code.gitea.io/gitea/models/auth"
+	org_model "code.gitea.io/gitea/models/organization"
+	user_model "code.gitea.io/gitea/models/user"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/timeutil"
+
+	"github.com/golang-jwt/jwt/v5"
+)
+
+// AccessTokenErrorCode represents an error code specified in RFC 6749
+// https://datatracker.ietf.org/doc/html/rfc6749#section-5.2
+type AccessTokenErrorCode string
+
+const (
+	// AccessTokenErrorCodeInvalidRequest represents an error code specified in RFC 6749
+	AccessTokenErrorCodeInvalidRequest AccessTokenErrorCode = "invalid_request"
+	// AccessTokenErrorCodeInvalidClient represents an error code specified in RFC 6749
+	AccessTokenErrorCodeInvalidClient = "invalid_client"
+	// AccessTokenErrorCodeInvalidGrant represents an error code specified in RFC 6749
+	AccessTokenErrorCodeInvalidGrant = "invalid_grant"
+	// AccessTokenErrorCodeUnauthorizedClient represents an error code specified in RFC 6749
+	AccessTokenErrorCodeUnauthorizedClient = "unauthorized_client"
+	// AccessTokenErrorCodeUnsupportedGrantType represents an error code specified in RFC 6749
+	AccessTokenErrorCodeUnsupportedGrantType = "unsupported_grant_type"
+	// AccessTokenErrorCodeInvalidScope represents an error code specified in RFC 6749
+	AccessTokenErrorCodeInvalidScope = "invalid_scope"
+)
+
+// AccessTokenError represents an error response specified in RFC 6749
+// https://datatracker.ietf.org/doc/html/rfc6749#section-5.2
+type AccessTokenError struct {
+	ErrorCode        AccessTokenErrorCode `json:"error" form:"error"`
+	ErrorDescription string               `json:"error_description"`
+}
+
+// Error returns the error message
+func (err AccessTokenError) Error() string {
+	return fmt.Sprintf("%s: %s", err.ErrorCode, err.ErrorDescription)
+}
+
+// TokenType specifies the kind of token
+type TokenType string
+
+const (
+	// TokenTypeBearer represents a token type specified in RFC 6749
+	TokenTypeBearer TokenType = "bearer"
+	// TokenTypeMAC represents a token type specified in RFC 6749
+	TokenTypeMAC = "mac"
+)
+
+// AccessTokenResponse represents a successful access token response
+// https://datatracker.ietf.org/doc/html/rfc6749#section-4.2.2
+type AccessTokenResponse struct {
+	AccessToken  string    `json:"access_token"`
+	TokenType    TokenType `json:"token_type"`
+	ExpiresIn    int64     `json:"expires_in"`
+	RefreshToken string    `json:"refresh_token"`
+	IDToken      string    `json:"id_token,omitempty"`
+}
+
+func NewAccessTokenResponse(ctx context.Context, grant *auth.OAuth2Grant, serverKey, clientKey JWTSigningKey) (*AccessTokenResponse, *AccessTokenError) {
+	if setting.OAuth2.InvalidateRefreshTokens {
+		if err := grant.IncreaseCounter(ctx); err != nil {
+			return nil, &AccessTokenError{
+				ErrorCode:        AccessTokenErrorCodeInvalidGrant,
+				ErrorDescription: "cannot increase the grant counter",
+			}
+		}
+	}
+	// generate access token to access the API
+	expirationDate := timeutil.TimeStampNow().Add(setting.OAuth2.AccessTokenExpirationTime)
+	accessToken := &Token{
+		GrantID: grant.ID,
+		Kind:    KindAccessToken,
+		RegisteredClaims: jwt.RegisteredClaims{
+			ExpiresAt: jwt.NewNumericDate(expirationDate.AsTime()),
+		},
+	}
+	signedAccessToken, err := accessToken.SignToken(serverKey)
+	if err != nil {
+		return nil, &AccessTokenError{
+			ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+			ErrorDescription: "cannot sign token",
+		}
+	}
+
+	// generate refresh token to request an access token after it expired later
+	refreshExpirationDate := timeutil.TimeStampNow().Add(setting.OAuth2.RefreshTokenExpirationTime * 60 * 60).AsTime()
+	refreshToken := &Token{
+		GrantID: grant.ID,
+		Counter: grant.Counter,
+		Kind:    KindRefreshToken,
+		RegisteredClaims: jwt.RegisteredClaims{
+			ExpiresAt: jwt.NewNumericDate(refreshExpirationDate),
+		},
+	}
+	signedRefreshToken, err := refreshToken.SignToken(serverKey)
+	if err != nil {
+		return nil, &AccessTokenError{
+			ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+			ErrorDescription: "cannot sign token",
+		}
+	}
+
+	// generate OpenID Connect id_token
+	signedIDToken := ""
+	if grant.ScopeContains("openid") {
+		app, err := auth.GetOAuth2ApplicationByID(ctx, grant.ApplicationID)
+		if err != nil {
+			return nil, &AccessTokenError{
+				ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+				ErrorDescription: "cannot find application",
+			}
+		}
+		user, err := user_model.GetUserByID(ctx, grant.UserID)
+		if err != nil {
+			if user_model.IsErrUserNotExist(err) {
+				return nil, &AccessTokenError{
+					ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+					ErrorDescription: "cannot find user",
+				}
+			}
+			log.Error("Error loading user: %v", err)
+			return nil, &AccessTokenError{
+				ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+				ErrorDescription: "server error",
+			}
+		}
+
+		idToken := &OIDCToken{
+			RegisteredClaims: jwt.RegisteredClaims{
+				ExpiresAt: jwt.NewNumericDate(expirationDate.AsTime()),
+				Issuer:    setting.AppURL,
+				Audience:  []string{app.ClientID},
+				Subject:   fmt.Sprint(grant.UserID),
+			},
+			Nonce: grant.Nonce,
+		}
+		if grant.ScopeContains("profile") {
+			idToken.Name = user.GetDisplayName()
+			idToken.PreferredUsername = user.Name
+			idToken.Profile = user.HTMLURL()
+			idToken.Picture = user.AvatarLink(ctx)
+			idToken.Website = user.Website
+			idToken.Locale = user.Language
+			idToken.UpdatedAt = user.UpdatedUnix
+		}
+		if grant.ScopeContains("email") {
+			idToken.Email = user.Email
+			idToken.EmailVerified = user.IsActive
+		}
+		if grant.ScopeContains("groups") {
+			groups, err := GetOAuthGroupsForUser(ctx, user)
+			if err != nil {
+				log.Error("Error getting groups: %v", err)
+				return nil, &AccessTokenError{
+					ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+					ErrorDescription: "server error",
+				}
+			}
+			idToken.Groups = groups
+		}
+
+		signedIDToken, err = idToken.SignToken(clientKey)
+		if err != nil {
+			return nil, &AccessTokenError{
+				ErrorCode:        AccessTokenErrorCodeInvalidRequest,
+				ErrorDescription: "cannot sign token",
+			}
+		}
+	}
+
+	return &AccessTokenResponse{
+		AccessToken:  signedAccessToken,
+		TokenType:    TokenTypeBearer,
+		ExpiresIn:    setting.OAuth2.AccessTokenExpirationTime,
+		RefreshToken: signedRefreshToken,
+		IDToken:      signedIDToken,
+	}, nil
+}
+
+// returns a list of "org" and "org:team" strings,
+// that the given user is a part of.
+func GetOAuthGroupsForUser(ctx context.Context, user *user_model.User) ([]string, error) {
+	orgs, err := org_model.GetUserOrgsList(ctx, user)
+	if err != nil {
+		return nil, fmt.Errorf("GetUserOrgList: %w", err)
+	}
+
+	var groups []string
+	for _, org := range orgs {
+		groups = append(groups, org.Name)
+		teams, err := org.LoadTeams(ctx)
+		if err != nil {
+			return nil, fmt.Errorf("LoadTeams: %w", err)
+		}
+		for _, team := range teams {
+			if team.IsMember(ctx, user.ID) {
+				groups = append(groups, org.Name+":"+team.LowerName)
+			}
+		}
+	}
+	return groups, nil
+}

services/oauth2_provider/init.go

@@ -0,0 +1,19 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package oauth2_provider //nolint
+
+import (
+	"context"
+
+	"code.gitea.io/gitea/modules/setting"
+)
+
+// Init initializes the oauth source
+func Init(ctx context.Context) error {
+	if !setting.OAuth2.Enabled {
+		return nil
+	}
+
+	return InitSigningKey()
+}

services/oauth2_provider/jwtsigningkey.go

@@ -1,7 +1,7 @@
 // Copyright 2021 The Gitea Authors. All rights reserved.
 // SPDX-License-Identifier: MIT
 
-package oauth2
+package oauth2_provider //nolint
 
 import (
 	"crypto/ecdsa"

services/oauth2_provider/token.go

@@ -1,7 +1,7 @@
 // Copyright 2021 The Gitea Authors. All rights reserved.
 // SPDX-License-Identifier: MIT
 
-package oauth2
+package oauth2_provider //nolint
 
 import (
 	"fmt"
@@ -12,29 +12,22 @@ import (
 	"github.com/golang-jwt/jwt/v5"
 )
 
-// ___________     __
-// \__    ___/___ |  | __ ____   ____
-//   |    | /  _ \|  |/ // __ \ /    \
-//   |    |(  <_> )    <\  ___/|   |  \
-//   |____| \____/|__|_ \\___  >___|  /
-//                     \/    \/     \/
-
 // Token represents an Oauth grant
 
-// TokenType represents the type of token for an oauth application
-type TokenType int
+// TokenKind represents the type of token for an oauth application
+type TokenKind int
 
 const (
-	// TypeAccessToken is a token with short lifetime to access the api
-	TypeAccessToken TokenType = 0
-	// TypeRefreshToken is token with long lifetime to refresh access tokens obtained by the client
-	TypeRefreshToken = iota
+	// KindAccessToken is a token with short lifetime to access the api
+	KindAccessToken TokenKind = 0
+	// KindRefreshToken is token with long lifetime to refresh access tokens obtained by the client
+	KindRefreshToken = iota
 )
 
 // Token represents a JWT token used to authenticate a client
 type Token struct {
 	GrantID int64     `json:"gnt"`
-	Type    TokenType `json:"tt"`
+	Kind    TokenKind `json:"tt"`
 	Counter int64     `json:"cnt,omitempty"`
 	jwt.RegisteredClaims
 }

services/pull/pull.go

@@ -1013,6 +1013,8 @@ type CommitInfo struct {
 }
 
 // GetPullCommits returns all commits on given pull request and the last review commit sha
+// Attention: The last review commit sha must be from the latest review whose commit id is not empty.
+// So the type of the latest review cannot be "ReviewTypeRequest".
 func GetPullCommits(ctx *gitea_context.Context, issue *issues_model.Issue) ([]CommitInfo, string, error) {
 	pull := issue.PullRequest
 
@@ -1058,7 +1060,11 @@ func GetPullCommits(ctx *gitea_context.Context, issue *issues_model.Issue) ([]Co
 		lastreview, err := issues_model.FindLatestReviews(ctx, issues_model.FindReviewOptions{
 			IssueID:    issue.ID,
 			ReviewerID: ctx.Doer.ID,
-			Type:       issues_model.ReviewTypeUnknown,
+			Types: []issues_model.ReviewType{
+				issues_model.ReviewTypeApprove,
+				issues_model.ReviewTypeComment,
+				issues_model.ReviewTypeReject,
+			},
 		})
 
 		if err != nil && !issues_model.IsErrReviewNotExist(err) {

services/pull/review.go

@@ -348,7 +348,7 @@ func DismissApprovalReviews(ctx context.Context, doer *user_model.User, pull *is
 	reviews, err := issues_model.FindReviews(ctx, issues_model.FindReviewOptions{
 		ListOptions: db.ListOptionsAll,
 		IssueID:     pull.IssueID,
-		Type:        issues_model.ReviewTypeApprove,
+		Types:       []issues_model.ReviewType{issues_model.ReviewTypeApprove},
 		Dismissed:   optional.Some(false),
 	})
 	if err != nil {

services/repository/branch.go

@@ -616,6 +616,14 @@ func SetRepoDefaultBranch(ctx context.Context, repo *repo_model.Repository, gitR
 		return err
 	}
 
+	if !repo.IsEmpty {
+		if err := AddRepoToLicenseUpdaterQueue(&LicenseUpdaterOptions{
+			RepoID: repo.ID,
+		}); err != nil {
+			log.Error("AddRepoToLicenseUpdaterQueue: %v", err)
+		}
+	}
+
 	notify_service.ChangeDefaultBranch(ctx, repo)
 
 	return nil

services/repository/create.go

@@ -303,6 +303,25 @@ func CreateRepositoryDirectly(ctx context.Context, doer, u *user_model.User, opt
 			rollbackRepo.OwnerID = u.ID
 			return fmt.Errorf("CreateRepository(git update-server-info): %w", err)
 		}
+
+		// update licenses
+		var licenses []string
+		if len(opts.License) > 0 {
+			licenses = append(licenses, ConvertLicenseName(opts.License))
+
+			stdout, _, err := git.NewCommand(ctx, "rev-parse", "HEAD").
+				SetDescription(fmt.Sprintf("CreateRepository(git rev-parse HEAD): %s", repoPath)).
+				RunStdString(&git.RunOpts{Dir: repoPath})
+			if err != nil {
+				log.Error("CreateRepository(git rev-parse HEAD) in %v: Stdout: %s\nError: %v", repo, stdout, err)
+				rollbackRepo = repo
+				rollbackRepo.OwnerID = u.ID
+				return fmt.Errorf("CreateRepository(git rev-parse HEAD): %w", err)
+			}
+			if err := repo_model.UpdateRepoLicenses(ctx, repo, stdout, licenses); err != nil {
+				return err
+			}
+		}
 		return nil
 	}); err != nil {
 		if rollbackRepo != nil {

services/repository/delete.go

@@ -140,6 +140,7 @@ func DeleteRepositoryDirectly(ctx context.Context, doer *user_model.User, repoID
 		&git_model.Branch{RepoID: repoID},
 		&git_model.LFSLock{RepoID: repoID},
 		&repo_model.LanguageStat{RepoID: repoID},
+		&repo_model.RepoLicense{RepoID: repoID},
 		&issues_model.Milestone{RepoID: repoID},
 		&repo_model.Mirror{RepoID: repoID},
 		&activities_model.Notification{RepoID: repoID},

services/repository/fork.go

@@ -198,6 +198,9 @@ func ForkRepository(ctx context.Context, doer, owner *user_model.User, opts Fork
 	if err := repo_model.CopyLanguageStat(ctx, opts.BaseRepo, repo); err != nil {
 		log.Error("Copy language stat from oldRepo failed: %v", err)
 	}
+	if err := repo_model.CopyLicense(ctx, opts.BaseRepo, repo); err != nil {
+		return nil, err
+	}
 
 	gitRepo, err := gitrepo.OpenRepository(ctx, repo)
 	if err != nil {

services/repository/license.go

@@ -0,0 +1,205 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package repository
+
+import (
+	"context"
+	"fmt"
+	"io"
+
+	"code.gitea.io/gitea/models/db"
+	repo_model "code.gitea.io/gitea/models/repo"
+	"code.gitea.io/gitea/modules/container"
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/gitrepo"
+	"code.gitea.io/gitea/modules/graceful"
+	"code.gitea.io/gitea/modules/json"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/options"
+	"code.gitea.io/gitea/modules/queue"
+
+	licenseclassifier "github.com/google/licenseclassifier/v2"
+)
+
+var (
+	classifier      *licenseclassifier.Classifier
+	LicenseFileName = "LICENSE"
+	licenseAliases  map[string]string
+
+	// licenseUpdaterQueue represents a queue to handle update repo licenses
+	licenseUpdaterQueue *queue.WorkerPoolQueue[*LicenseUpdaterOptions]
+)
+
+func AddRepoToLicenseUpdaterQueue(opts *LicenseUpdaterOptions) error {
+	if opts == nil {
+		return nil
+	}
+	return licenseUpdaterQueue.Push(opts)
+}
+
+func loadLicenseAliases() error {
+	if licenseAliases != nil {
+		return nil
+	}
+
+	data, err := options.AssetFS().ReadFile("license", "etc", "license-aliases.json")
+	if err != nil {
+		return err
+	}
+	err = json.Unmarshal(data, &licenseAliases)
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func ConvertLicenseName(name string) string {
+	if err := loadLicenseAliases(); err != nil {
+		return name
+	}
+
+	v, ok := licenseAliases[name]
+	if ok {
+		return v
+	}
+	return name
+}
+
+func InitLicenseClassifier() error {
+	// threshold should be 0.84~0.86 or the test will be failed
+	classifier = licenseclassifier.NewClassifier(.85)
+	licenseFiles, err := options.AssetFS().ListFiles("license", true)
+	if err != nil {
+		return err
+	}
+
+	existLicense := make(container.Set[string])
+	if len(licenseFiles) > 0 {
+		for _, licenseFile := range licenseFiles {
+			licenseName := ConvertLicenseName(licenseFile)
+			if existLicense.Contains(licenseName) {
+				continue
+			}
+			existLicense.Add(licenseName)
+			data, err := options.License(licenseFile)
+			if err != nil {
+				return err
+			}
+			classifier.AddContent("License", licenseFile, licenseName, data)
+		}
+	}
+	return nil
+}
+
+type LicenseUpdaterOptions struct {
+	RepoID int64
+}
+
+func repoLicenseUpdater(items ...*LicenseUpdaterOptions) []*LicenseUpdaterOptions {
+	ctx := graceful.GetManager().ShutdownContext()
+
+	for _, opts := range items {
+		repo, err := repo_model.GetRepositoryByID(ctx, opts.RepoID)
+		if err != nil {
+			log.Error("repoLicenseUpdater [%d] failed: GetRepositoryByID: %v", opts.RepoID, err)
+			continue
+		}
+		if repo.IsEmpty {
+			continue
+		}
+
+		gitRepo, err := gitrepo.OpenRepository(ctx, repo)
+		if err != nil {
+			log.Error("repoLicenseUpdater [%d] failed: OpenRepository: %v", opts.RepoID, err)
+			continue
+		}
+		defer gitRepo.Close()
+
+		commit, err := gitRepo.GetBranchCommit(repo.DefaultBranch)
+		if err != nil {
+			log.Error("repoLicenseUpdater [%d] failed: GetBranchCommit: %v", opts.RepoID, err)
+			continue
+		}
+		if err = UpdateRepoLicenses(ctx, repo, commit); err != nil {
+			log.Error("repoLicenseUpdater [%d] failed: updateRepoLicenses: %v", opts.RepoID, err)
+		}
+	}
+	return nil
+}
+
+func SyncRepoLicenses(ctx context.Context) error {
+	log.Trace("Doing: SyncRepoLicenses")
+
+	if err := db.Iterate(
+		ctx,
+		nil,
+		func(ctx context.Context, repo *repo_model.Repository) error {
+			select {
+			case <-ctx.Done():
+				return db.ErrCancelledf("before sync repo licenses for %s", repo.FullName())
+			default:
+			}
+			return AddRepoToLicenseUpdaterQueue(&LicenseUpdaterOptions{RepoID: repo.ID})
+		},
+	); err != nil {
+		log.Trace("Error: SyncRepoLicenses: %v", err)
+		return err
+	}
+
+	log.Trace("Finished: SyncReposLicenses")
+	return nil
+}
+
+// UpdateRepoLicenses will update repository licenses col if license file exists
+func UpdateRepoLicenses(ctx context.Context, repo *repo_model.Repository, commit *git.Commit) error {
+	if commit == nil {
+		return nil
+	}
+
+	b, err := commit.GetBlobByPath(LicenseFileName)
+	if err != nil && !git.IsErrNotExist(err) {
+		return fmt.Errorf("GetBlobByPath: %w", err)
+	}
+
+	if git.IsErrNotExist(err) {
+		return repo_model.CleanRepoLicenses(ctx, repo)
+	}
+
+	licenses := make([]string, 0)
+	if b != nil {
+		r, err := b.DataAsync()
+		if err != nil {
+			return err
+		}
+		defer r.Close()
+
+		licenses, err = detectLicense(r)
+		if err != nil {
+			return fmt.Errorf("detectLicense: %w", err)
+		}
+	}
+	return repo_model.UpdateRepoLicenses(ctx, repo, commit.ID.String(), licenses)
+}
+
+// detectLicense returns the licenses detected by the given content buff
+func detectLicense(r io.Reader) ([]string, error) {
+	if r == nil {
+		return nil, nil
+	}
+
+	matches, err := classifier.MatchFrom(r)
+	if err != nil {
+		return nil, err
+	}
+	if len(matches.Matches) > 0 {
+		results := make(container.Set[string], len(matches.Matches))
+		for _, r := range matches.Matches {
+			if r.MatchType == "License" && !results.Contains(r.Variant) {
+				results.Add(r.Variant)
+			}
+		}
+		return results.Values(), nil
+	}
+	return nil, nil
+}

services/repository/license_test.go

@@ -0,0 +1,73 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package repository
+
+import (
+	"fmt"
+	"strings"
+	"testing"
+
+	repo_module "code.gitea.io/gitea/modules/repository"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func Test_detectLicense(t *testing.T) {
+	type DetectLicenseTest struct {
+		name string
+		arg  string
+		want []string
+	}
+
+	tests := []DetectLicenseTest{
+		{
+			name: "empty",
+			arg:  "",
+			want: nil,
+		},
+		{
+			name: "no detected license",
+			arg:  "Copyright (c) 2023 Gitea",
+			want: nil,
+		},
+	}
+
+	repo_module.LoadRepoConfig()
+	err := loadLicenseAliases()
+	assert.NoError(t, err)
+	for _, licenseName := range repo_module.Licenses {
+		license, err := repo_module.GetLicense(licenseName, &repo_module.LicenseValues{
+			Owner: "Gitea",
+			Email: "teabot@gitea.io",
+			Repo:  "gitea",
+			Year:  "2024",
+		})
+		assert.NoError(t, err)
+
+		tests = append(tests, DetectLicenseTest{
+			name: fmt.Sprintf("single license test: %s", licenseName),
+			arg:  string(license),
+			want: []string{ConvertLicenseName(licenseName)},
+		})
+	}
+
+	err = InitLicenseClassifier()
+	assert.NoError(t, err)
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			license, err := detectLicense(strings.NewReader(tt.arg))
+			assert.NoError(t, err)
+			assert.Equal(t, tt.want, license)
+		})
+	}
+
+	result, err := detectLicense(strings.NewReader(tests[2].arg + tests[3].arg + tests[4].arg))
+	assert.NoError(t, err)
+	t.Run("multiple licenses test", func(t *testing.T) {
+		assert.Equal(t, 3, len(result))
+		assert.Contains(t, result, tests[2].want[0])
+		assert.Contains(t, result, tests[3].want[0])
+		assert.Contains(t, result, tests[4].want[0])
+	})
+}

services/repository/migrate.go

@@ -172,6 +172,11 @@ func MigrateRepositoryGitData(ctx context.Context, u *user_model.User,
 				return repo, fmt.Errorf("StoreMissingLfsObjectsInRepository: %w", err)
 			}
 		}
+
+		// Update repo license
+		if err := AddRepoToLicenseUpdaterQueue(&LicenseUpdaterOptions{RepoID: repo.ID}); err != nil {
+			log.Error("Failed to add repo to license updater queue: %v", err)
+		}
 	}
 
 	ctx, committer, err := db.TxContext(ctx)

services/repository/repository.go

@@ -18,6 +18,7 @@ import (
 	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/graceful"
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/queue"
 	repo_module "code.gitea.io/gitea/modules/repository"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/structs"
@@ -96,6 +97,12 @@ func PushCreateRepo(ctx context.Context, authUser, owner *user_model.User, repoN
 
 // Init start repository service
 func Init(ctx context.Context) error {
+	licenseUpdaterQueue = queue.CreateUniqueQueue(graceful.GetManager().ShutdownContext(), "repo_license_updater", repoLicenseUpdater)
+	if licenseUpdaterQueue == nil {
+		return fmt.Errorf("unable to create repo_license_updater queue")
+	}
+	go graceful.GetManager().RunWithCancel(licenseUpdaterQueue)
+
 	if err := repo_module.LoadRepoConfig(); err != nil {
 		return err
 	}