Merge remote-tracking branch 'upstream/main' into limit-repo-size

models/fixtures/review.yml

@@ -105,3 +105,30 @@
   official: true
   updated_unix: 1603196749
   created_unix: 1603196749
+
+-
+  id: 13
+  type: 1
+  reviewer_id: 5
+  issue_id: 11
+  content: "old review from user5"
+  updated_unix: 946684820
+  created_unix: 946684820
+
+-
+  id: 14
+  type: 1
+  reviewer_id: 5
+  issue_id: 11
+  content: "duplicate review from user5 (latest)"
+  updated_unix: 946684830
+  created_unix: 946684830
+
+-
+  id: 15
+  type: 1
+  reviewer_id: 6
+  issue_id: 11
+  content: "singular review from user6 and final review for this pr"
+  updated_unix: 946684831
+  created_unix: 946684831

models/issues/pull.go

@@ -8,11 +8,13 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"regexp"
 	"strconv"
 	"strings"
 
 	"code.gitea.io/gitea/models/db"
 	git_model "code.gitea.io/gitea/models/git"
+	org_model "code.gitea.io/gitea/models/organization"
 	pull_model "code.gitea.io/gitea/models/pull"
 	repo_model "code.gitea.io/gitea/models/repo"
 	user_model "code.gitea.io/gitea/models/user"
@@ -402,7 +404,7 @@ func (pr *PullRequest) getReviewedByLines(writer io.Writer) error {
 	defer committer.Close()
 
 	// Note: This doesn't page as we only expect a very limited number of reviews
-	reviews, err := FindReviews(ctx, FindReviewOptions{
+	reviews, err := FindLatestReviews(ctx, FindReviewOptions{
 		Type:         ReviewTypeApprove,
 		IssueID:      pr.IssueID,
 		OfficialOnly: setting.Repository.PullRequest.DefaultMergeMessageOfficialApproversOnly,
@@ -887,3 +889,222 @@ func MergeBlockedByOfficialReviewRequests(ctx context.Context, protectBranch *gi
 func MergeBlockedByOutdatedBranch(protectBranch *git_model.ProtectedBranch, pr *PullRequest) bool {
 	return protectBranch.BlockOnOutdatedBranch && pr.CommitsBehind > 0
 }
+
+func PullRequestCodeOwnersReview(ctx context.Context, pull *Issue, pr *PullRequest) error {
+	files := []string{"CODEOWNERS", "docs/CODEOWNERS", ".gitea/CODEOWNERS"}
+
+	if pr.IsWorkInProgress() {
+		return nil
+	}
+
+	if err := pr.LoadBaseRepo(ctx); err != nil {
+		return err
+	}
+
+	repo, err := git.OpenRepository(ctx, pr.BaseRepo.RepoPath())
+	if err != nil {
+		return err
+	}
+	defer repo.Close()
+
+	branch, err := repo.GetDefaultBranch()
+	if err != nil {
+		return err
+	}
+
+	commit, err := repo.GetBranchCommit(branch)
+	if err != nil {
+		return err
+	}
+
+	var data string
+	for _, file := range files {
+		if blob, err := commit.GetBlobByPath(file); err == nil {
+			data, err = blob.GetBlobContent()
+			if err == nil {
+				break
+			}
+		}
+	}
+
+	rules, _ := GetCodeOwnersFromContent(ctx, data)
+	changedFiles, err := repo.GetFilesChangedBetween(git.BranchPrefix+pr.BaseBranch, pr.GetGitRefName())
+	if err != nil {
+		return err
+	}
+
+	uniqUsers := make(map[int64]*user_model.User)
+	uniqTeams := make(map[string]*org_model.Team)
+	for _, rule := range rules {
+		for _, f := range changedFiles {
+			if (rule.Rule.MatchString(f) && !rule.Negative) || (!rule.Rule.MatchString(f) && rule.Negative) {
+				for _, u := range rule.Users {
+					uniqUsers[u.ID] = u
+				}
+				for _, t := range rule.Teams {
+					uniqTeams[fmt.Sprintf("%d/%d", t.OrgID, t.ID)] = t
+				}
+			}
+		}
+	}
+
+	for _, u := range uniqUsers {
+		if u.ID != pull.Poster.ID {
+			if _, err := AddReviewRequest(pull, u, pull.Poster); err != nil {
+				log.Warn("Failed add assignee user: %s to PR review: %s#%d, error: %s", u.Name, pr.BaseRepo.Name, pr.ID, err)
+				return err
+			}
+		}
+	}
+	for _, t := range uniqTeams {
+		if _, err := AddTeamReviewRequest(pull, t, pull.Poster); err != nil {
+			log.Warn("Failed add assignee team: %s to PR review: %s#%d, error: %s", t.Name, pr.BaseRepo.Name, pr.ID, err)
+			return err
+		}
+	}
+
+	return nil
+}
+
+// GetCodeOwnersFromContent returns the code owners configuration
+// Return empty slice if files missing
+// Return warning messages on parsing errors
+// We're trying to do the best we can when parsing a file.
+// Invalid lines are skipped. Non-existent users and teams too.
+func GetCodeOwnersFromContent(ctx context.Context, data string) ([]*CodeOwnerRule, []string) {
+	if len(data) == 0 {
+		return nil, nil
+	}
+
+	rules := make([]*CodeOwnerRule, 0)
+	lines := strings.Split(data, "\n")
+	warnings := make([]string, 0)
+
+	for i, line := range lines {
+		tokens := TokenizeCodeOwnersLine(line)
+		if len(tokens) == 0 {
+			continue
+		} else if len(tokens) < 2 {
+			warnings = append(warnings, fmt.Sprintf("Line: %d: incorrect format", i+1))
+			continue
+		}
+		rule, wr := ParseCodeOwnersLine(ctx, tokens)
+		for _, w := range wr {
+			warnings = append(warnings, fmt.Sprintf("Line: %d: %s", i+1, w))
+		}
+		if rule == nil {
+			continue
+		}
+
+		rules = append(rules, rule)
+	}
+
+	return rules, warnings
+}
+
+type CodeOwnerRule struct {
+	Rule     *regexp.Regexp
+	Negative bool
+	Users    []*user_model.User
+	Teams    []*org_model.Team
+}
+
+func ParseCodeOwnersLine(ctx context.Context, tokens []string) (*CodeOwnerRule, []string) {
+	var err error
+	rule := &CodeOwnerRule{
+		Users:    make([]*user_model.User, 0),
+		Teams:    make([]*org_model.Team, 0),
+		Negative: strings.HasPrefix(tokens[0], "!"),
+	}
+
+	warnings := make([]string, 0)
+
+	rule.Rule, err = regexp.Compile(fmt.Sprintf("^%s$", strings.TrimPrefix(tokens[0], "!")))
+	if err != nil {
+		warnings = append(warnings, fmt.Sprintf("incorrect codeowner regexp: %s", err))
+		return nil, warnings
+	}
+
+	for _, user := range tokens[1:] {
+		user = strings.TrimPrefix(user, "@")
+
+		// Only @org/team can contain slashes
+		if strings.Contains(user, "/") {
+			s := strings.Split(user, "/")
+			if len(s) != 2 {
+				warnings = append(warnings, fmt.Sprintf("incorrect codeowner group: %s", user))
+				continue
+			}
+			orgName := s[0]
+			teamName := s[1]
+
+			org, err := org_model.GetOrgByName(ctx, orgName)
+			if err != nil {
+				warnings = append(warnings, fmt.Sprintf("incorrect codeowner organization: %s", user))
+				continue
+			}
+			teams, err := org.LoadTeams()
+			if err != nil {
+				warnings = append(warnings, fmt.Sprintf("incorrect codeowner team: %s", user))
+				continue
+			}
+
+			for _, team := range teams {
+				if team.Name == teamName {
+					rule.Teams = append(rule.Teams, team)
+				}
+			}
+		} else {
+			u, err := user_model.GetUserByName(ctx, user)
+			if err != nil {
+				warnings = append(warnings, fmt.Sprintf("incorrect codeowner user: %s", user))
+				continue
+			}
+			rule.Users = append(rule.Users, u)
+		}
+	}
+
+	if (len(rule.Users) == 0) && (len(rule.Teams) == 0) {
+		warnings = append(warnings, "no users/groups matched")
+		return nil, warnings
+	}
+
+	return rule, warnings
+}
+
+func TokenizeCodeOwnersLine(line string) []string {
+	if len(line) == 0 {
+		return nil
+	}
+
+	line = strings.TrimSpace(line)
+	line = strings.ReplaceAll(line, "\t", " ")
+
+	tokens := make([]string, 0)
+
+	escape := false
+	token := ""
+	for _, char := range line {
+		if escape {
+			token += string(char)
+			escape = false
+		} else if string(char) == "\\" {
+			escape = true
+		} else if string(char) == "#" {
+			break
+		} else if string(char) == " " {
+			if len(token) > 0 {
+				tokens = append(tokens, token)
+				token = ""
+			}
+		} else {
+			token += string(char)
+		}
+	}
+
+	if len(token) > 0 {
+		tokens = append(tokens, token)
+	}
+
+	return tokens
+}

models/issues/pull_test.go

@@ -10,6 +10,7 @@ import (
 	issues_model "code.gitea.io/gitea/models/issues"
 	"code.gitea.io/gitea/models/unittest"
 	user_model "code.gitea.io/gitea/models/user"
+	"code.gitea.io/gitea/modules/setting"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -303,3 +304,36 @@ func TestDeleteOrphanedObjects(t *testing.T) {
 	assert.NoError(t, err)
 	assert.EqualValues(t, countBefore, countAfter)
 }
+
+func TestParseCodeOwnersLine(t *testing.T) {
+	type CodeOwnerTest struct {
+		Line   string
+		Tokens []string
+	}
+
+	given := []CodeOwnerTest{
+		{Line: "", Tokens: nil},
+		{Line: "# comment", Tokens: []string{}},
+		{Line: "!.* @user1 @org1/team1", Tokens: []string{"!.*", "@user1", "@org1/team1"}},
+		{Line: `.*\\.js @user2 #comment`, Tokens: []string{`.*\.js`, "@user2"}},
+		{Line: `docs/(aws|google|azure)/[^/]*\\.(md|txt) @user3 @org2/team2`, Tokens: []string{`docs/(aws|google|azure)/[^/]*\.(md|txt)`, "@user3", "@org2/team2"}},
+		{Line: `\#path @user3`, Tokens: []string{`#path`, "@user3"}},
+		{Line: `path\ with\ spaces/ @user3`, Tokens: []string{`path with spaces/`, "@user3"}},
+	}
+
+	for _, g := range given {
+		tokens := issues_model.TokenizeCodeOwnersLine(g.Line)
+		assert.Equal(t, g.Tokens, tokens, "Codeowners tokenizer failed")
+	}
+}
+
+func TestGetApprovers(t *testing.T) {
+	assert.NoError(t, unittest.PrepareTestDatabase())
+	pr := unittest.AssertExistsAndLoadBean(t, &issues_model.PullRequest{ID: 5})
+	// Official reviews are already deduplicated. Allow unofficial reviews
+	// to assert that there are no duplicated approvers.
+	setting.Repository.PullRequest.DefaultMergeMessageOfficialApproversOnly = false
+	approvers := pr.GetApprovers()
+	expected := "Reviewed-by: User Five <user5@example.com>\nReviewed-by: User Six <user6@example.com>\n"
+	assert.EqualValues(t, expected, approvers)
+}

models/issues/review.go

@@ -275,6 +275,27 @@ func FindReviews(ctx context.Context, opts FindReviewOptions) ([]*Review, error)
 		Find(&reviews)
 }
 
+// FindLatestReviews returns only latest reviews per user, passing FindReviewOptions
+func FindLatestReviews(ctx context.Context, opts FindReviewOptions) ([]*Review, error) {
+	reviews := make([]*Review, 0, 10)
+	cond := opts.toCond()
+	sess := db.GetEngine(ctx).Where(cond)
+	if opts.Page > 0 {
+		sess = db.SetSessionPagination(sess, &opts)
+	}
+
+	sess.In("id", builder.
+		Select("max ( id ) ").
+		From("review").
+		Where(cond).
+		GroupBy("reviewer_id"))
+
+	return reviews, sess.
+		Asc("created_unix").
+		Asc("id").
+		Find(&reviews)
+}
+
 // CountReviews returns count of reviews passing FindReviewOptions
 func CountReviews(opts FindReviewOptions) (int64, error) {
 	return db.GetEngine(db.DefaultContext).Where(opts.toCond()).Count(&Review{})

models/issues/review_test.go

@@ -71,6 +71,18 @@ func TestFindReviews(t *testing.T) {
 	assert.Equal(t, "Demo Review", reviews[0].Content)
 }
 
+func TestFindLatestReviews(t *testing.T) {
+	assert.NoError(t, unittest.PrepareTestDatabase())
+	reviews, err := issues_model.FindLatestReviews(db.DefaultContext, issues_model.FindReviewOptions{
+		Type:    issues_model.ReviewTypeApprove,
+		IssueID: 11,
+	})
+	assert.NoError(t, err)
+	assert.Len(t, reviews, 2)
+	assert.Equal(t, "duplicate review from user5 (latest)", reviews[0].Content)
+	assert.Equal(t, "singular review from user6 and final review for this pr", reviews[1].Content)
+}
+
 func TestGetCurrentReview(t *testing.T) {
 	assert.NoError(t, unittest.PrepareTestDatabase())
 	issue := unittest.AssertExistsAndLoadBean(t, &issues_model.Issue{ID: 2})

models/packages/container/search.go

@@ -262,6 +262,10 @@ func GetRepositories(ctx context.Context, actor *user_model.User, n int, last st
 		cond = cond.And(builder.Gt{"package_property.value": strings.ToLower(last)})
 	}
 
+	if actor.IsGhost() {
+		actor = nil
+	}
+
 	cond = cond.And(user_model.BuildCanSeeUserCondition(actor))
 
 	sess := db.GetEngine(ctx).