Add archived repository filters to global code search

models/repo/repo_indexer.go

@@ -97,7 +97,7 @@ func UpdateIndexerStatus(ctx context.Context, repo *Repository, indexerType Repo
 		return fmt.Errorf("UpdateIndexerStatus: Unable to getIndexerStatus for repo: %s Error: %w", repo.FullName(), err)
 	}
 
-	if len(status.CommitSha) == 0 {
+	if status.ID == 0 {
 		status.CommitSha = sha
 		if err := db.Insert(ctx, status); err != nil {
 			return fmt.Errorf("UpdateIndexerStatus: Unable to insert repoIndexerStatus for repo: %s Sha: %s Error: %w", repo.FullName(), sha, err)

modules/indexer/code/bleve/bleve.go

@@ -54,6 +54,7 @@ func addUnicodeNormalizeTokenFilter(m *mapping.IndexMappingImpl) error {
 // RepoIndexerData data stored in the repo indexer
 type RepoIndexerData struct {
 	RepoID    int64
+	Archived  bool
 	CommitID  string
 	Content   string
 	Filename  string
@@ -71,7 +72,7 @@ const (
 	filenameIndexerAnalyzer  = "filenameIndexerAnalyzer"
 	filenameIndexerTokenizer = "filenameIndexerTokenizer"
 	repoIndexerDocType       = "repoIndexerDocType"
-	repoIndexerLatestVersion = 9
+	repoIndexerLatestVersion = 10
 )
 
 // generateBleveIndexMapping generates a bleve index mapping for the repo indexer
@@ -81,6 +82,10 @@ func generateBleveIndexMapping() (mapping.IndexMapping, error) {
 	numericFieldMapping.IncludeInAll = false
 	docMapping.AddFieldMappingsAt("RepoID", numericFieldMapping)
 
+	boolFieldMapping := bleve.NewBooleanFieldMapping()
+	boolFieldMapping.IncludeInAll = false
+	docMapping.AddFieldMappingsAt("Archived", boolFieldMapping)
+
 	textFieldMapping := bleve.NewTextFieldMapping()
 	textFieldMapping.IncludeInAll = false
 	docMapping.AddFieldMappingsAt("Content", textFieldMapping)
@@ -195,6 +200,7 @@ func (b *Indexer) addUpdate(ctx context.Context, catFileBatch git.CatFileBatch,
 	id := internal.FilenameIndexerID(repo.ID, update.Filename)
 	return batch.Index(id, &RepoIndexerData{
 		RepoID:    repo.ID,
+		Archived:  repo.IsArchived,
 		CommitID:  commitSha,
 		Filename:  update.Filename,
 		Content:   string(charset.ToUTF8DropErrors(fileContents)),
@@ -298,6 +304,13 @@ func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int
 		indexerQuery = keywordQuery
 	}
 
+	if opts.Archived.Has() {
+		indexerQuery = bleve.NewConjunctionQuery(
+			indexerQuery,
+			inner_bleve.BoolFieldQuery(opts.Archived.Value(), "Archived"),
+		)
+	}
+
 	// Save for reuse without language filter
 	facetQuery := indexerQuery
 	if len(opts.Language) > 0 {

modules/indexer/code/elasticsearch/elasticsearch.go

@@ -32,7 +32,7 @@ import (
 )
 
 const (
-	esRepoIndexerLatestVersion = 3
+	esRepoIndexerLatestVersion = 4
 	// multi-match-types, currently only 2 types are used
 	// Reference: https://www.elastic.co/guide/en/elasticsearch/reference/7.0/query-dsl-multi-match-query.html#multi-match-types
 	esMultiMatchTypeBestFields   = "best_fields"
@@ -100,6 +100,10 @@ const (
 					"type": "long",
 					"index": true
 				},
+				"archived": {
+					"type": "boolean",
+					"index": true
+				},
 				"filename": {
 					"type": "text",
 					"term_vector": "with_positions_offsets",
@@ -185,6 +189,7 @@ func (b *Indexer) addUpdate(ctx context.Context, catFileBatch git.CatFileBatch,
 			Id(id).
 			Doc(map[string]any{
 				"repo_id":    repo.ID,
+				"archived":   repo.IsArchived,
 				"filename":   update.Filename,
 				"content":    string(charset.ToUTF8DropErrors(fileContents)),
 				"commit_id":  sha,
@@ -377,6 +382,9 @@ func (b *Indexer) Search(ctx context.Context, opts *internal.SearchOptions) (int
 		repoQuery := elastic.NewTermsQuery("repo_id", repoStrs...)
 		query = query.Must(repoQuery)
 	}
+	if opts.Archived.Has() {
+		query = query.Must(elastic.NewTermQuery("archived", opts.Archived.Value()))
+	}
 
 	var (
 		start, pageSize = opts.GetSkipTake()

modules/indexer/code/indexer_test.go

@@ -10,11 +10,13 @@ import (
 	"testing"
 
 	"code.gitea.io/gitea/models/db"
+	repo_model "code.gitea.io/gitea/models/repo"
 	"code.gitea.io/gitea/models/unittest"
 	indexer_module "code.gitea.io/gitea/modules/indexer"
 	"code.gitea.io/gitea/modules/indexer/code/bleve"
 	"code.gitea.io/gitea/modules/indexer/code/elasticsearch"
 	"code.gitea.io/gitea/modules/indexer/code/internal"
+	"code.gitea.io/gitea/modules/optional"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/test"
 	"code.gitea.io/gitea/modules/util"
@@ -280,6 +282,110 @@ func testIndexer(name string, t *testing.T, indexer internal.Indexer) {
 
 		assert.NoError(t, tearDownRepositoryIndexes(t.Context(), indexer))
 	})
+
+	t.Run(name+"_archived_filter", func(t *testing.T) {
+		assert.NoError(t, setupRepositoryIndexes(t.Context(), indexer))
+		t.Cleanup(func() {
+			assert.NoError(t, tearDownRepositoryIndexes(context.Background(), indexer))
+		})
+
+		repo1, err := repo_model.GetRepositoryByID(t.Context(), 1)
+		require.NoError(t, err)
+		repo62, err := repo_model.GetRepositoryByID(t.Context(), 62)
+		require.NoError(t, err)
+
+		originalRepo1Archived := repo1.IsArchived
+		originalRepo62Archived := repo62.IsArchived
+		t.Cleanup(func() {
+			ctx := context.Background()
+			require.NoError(t, repo_model.SetArchiveRepoState(ctx, repo1, originalRepo1Archived))
+			require.NoError(t, repo_model.SetArchiveRepoState(ctx, repo62, originalRepo62Archived))
+			require.NoError(t, repo_model.UpdateIndexerStatus(ctx, repo1, repo_model.RepoIndexerTypeCode, ""))
+			require.NoError(t, repo_model.UpdateIndexerStatus(ctx, repo62, repo_model.RepoIndexerTypeCode, ""))
+			require.NoError(t, index(ctx, indexer, repo1.ID))
+			require.NoError(t, index(ctx, indexer, repo62.ID))
+		})
+
+		require.NoError(t, repo_model.SetArchiveRepoState(t.Context(), repo1, false))
+		require.NoError(t, repo_model.SetArchiveRepoState(t.Context(), repo62, true))
+		require.NoError(t, repo_model.UpdateIndexerStatus(t.Context(), repo1, repo_model.RepoIndexerTypeCode, ""))
+		require.NoError(t, repo_model.UpdateIndexerStatus(t.Context(), repo62, repo_model.RepoIndexerTypeCode, ""))
+		require.NoError(t, index(t.Context(), indexer, repo1.ID))
+		require.NoError(t, index(t.Context(), indexer, repo62.ID))
+
+		testCases := []struct {
+			name      string
+			keyword   string
+			archived  optional.Option[bool]
+			total     int64
+			filenames []string
+		}{
+			{
+				name:      "exclude_archived_repo_results",
+				keyword:   "cucumber",
+				archived:  optional.Some(false),
+				total:     0,
+				filenames: []string{},
+			},
+			{
+				name:      "include_archived_repo_results",
+				keyword:   "cucumber",
+				archived:  optional.None[bool](),
+				total:     2,
+				filenames: []string{"cucumber.md", "avocado.md"},
+			},
+			{
+				name:      "only_archived_repo_results",
+				keyword:   "cucumber",
+				archived:  optional.Some(true),
+				total:     2,
+				filenames: []string{"cucumber.md", "avocado.md"},
+			},
+			{
+				name:      "exclude_keeps_non_archived_repo_results",
+				keyword:   "Description",
+				archived:  optional.Some(false),
+				total:     1,
+				filenames: []string{"README.md"},
+			},
+			{
+				name:      "include_keeps_non_archived_repo_results",
+				keyword:   "Description",
+				archived:  optional.None[bool](),
+				total:     1,
+				filenames: []string{"README.md"},
+			},
+			{
+				name:      "only_excludes_non_archived_repo_results",
+				keyword:   "Description",
+				archived:  optional.Some(true),
+				total:     0,
+				filenames: []string{},
+			},
+		}
+
+		for _, tc := range testCases {
+			t.Run(tc.name, func(t *testing.T) {
+				total, res, _, err := indexer.Search(t.Context(), &internal.SearchOptions{
+					Keyword:    tc.keyword,
+					SearchMode: indexer_module.SearchModeWords,
+					Archived:   tc.archived,
+					Paginator: &db.ListOptions{
+						Page:     1,
+						PageSize: 10,
+					},
+				})
+				require.NoError(t, err)
+				assert.Equal(t, tc.total, total)
+
+				filenames := make([]string, 0, len(res))
+				for _, hit := range res {
+					filenames = append(filenames, hit.Filename)
+				}
+				assert.Equal(t, tc.filenames, filenames)
+			})
+		}
+	})
 }
 
 func TestBleveIndexAndSearch(t *testing.T) {
@@ -332,6 +438,13 @@ func tearDownRepositoryIndexes(ctx context.Context, indexer internal.Indexer) er
 		if err := indexer.Delete(ctx, repoID); err != nil {
 			return err
 		}
+		repo, err := repo_model.GetRepositoryByID(ctx, repoID)
+		if err != nil {
+			return err
+		}
+		if err := repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, ""); err != nil {
+			return err
+		}
 	}
 	return nil
 }

modules/indexer/code/internal/indexer.go

@@ -11,6 +11,7 @@ import (
 	repo_model "code.gitea.io/gitea/models/repo"
 	"code.gitea.io/gitea/modules/indexer"
 	"code.gitea.io/gitea/modules/indexer/internal"
+	"code.gitea.io/gitea/modules/optional"
 )
 
 // Indexer defines an interface to index and search code contents
@@ -28,6 +29,7 @@ type SearchOptions struct {
 	Language string
 
 	SearchMode indexer.SearchModeType
+	Archived   optional.Option[bool]
 
 	db.Paginator
 }

options/locale/locale_en-US.json

@@ -161,6 +161,8 @@
   "search.type_tooltip": "Search type",
   "search.fuzzy": "Fuzzy",
   "search.fuzzy_tooltip": "Include results that closely match the search term",
+  "search.code_archived_filter.all": "Archived and not archived",
+  "search.code_archived_filter.not_archived": "Not Archived",
   "search.words": "Words",
   "search.words_tooltip": "Include only results that match the search term words",
   "search.regexp": "Regexp",

routers/api/v1/repo/repo.go

@@ -22,6 +22,7 @@ import (
 	unit_model "code.gitea.io/gitea/models/unit"
 	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/gitrepo"
+	code_indexer "code.gitea.io/gitea/modules/indexer/code"
 	"code.gitea.io/gitea/modules/label"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/optional"
@@ -1014,6 +1015,12 @@ func updateRepoArchivedState(ctx *context.APIContext, opts api.EditRepoOption) e
 			if err := actions_service.CleanRepoScheduleTasks(ctx, repo); err != nil {
 				log.Error("CleanRepoScheduleTasks for archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
 			}
+			if setting.Indexer.RepoIndexerEnabled {
+				if err := repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, ""); err != nil {
+					log.Error("Reset code indexer status for archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
+				}
+				code_indexer.UpdateRepoIndexer(repo)
+			}
 			log.Trace("Repository was archived: %s/%s", ctx.Repo.Owner.Name, repo.Name)
 		} else {
 			if err := repo_model.SetArchiveRepoState(ctx, repo, *opts.Archived); err != nil {
@@ -1026,6 +1033,12 @@ func updateRepoArchivedState(ctx *context.APIContext, opts api.EditRepoOption) e
 					log.Error("DetectAndHandleSchedules for un-archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
 				}
 			}
+			if setting.Indexer.RepoIndexerEnabled {
+				if err := repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, ""); err != nil {
+					log.Error("Reset code indexer status for un-archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
+				}
+				code_indexer.UpdateRepoIndexer(repo)
+			}
 			log.Trace("Repository was un-archived: %s/%s", ctx.Repo.Owner.Name, repo.Name)
 		}
 	}

routers/web/explore/code.go

@@ -10,6 +10,7 @@ import (
 	"code.gitea.io/gitea/models/db"
 	repo_model "code.gitea.io/gitea/models/repo"
 	code_indexer "code.gitea.io/gitea/modules/indexer/code"
+	"code.gitea.io/gitea/modules/optional"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/templates"
 	"code.gitea.io/gitea/routers/common"
@@ -19,8 +20,27 @@ import (
 const (
 	// tplExploreCode explore code page template
 	tplExploreCode templates.TplName = "explore/code"
+
+	archivedModeExclude = "exclude"
+	archivedModeInclude = "include"
+	archivedModeOnly    = "only"
 )
 
+func parseArchivedMode(value string) (string, optional.Option[bool]) {
+	switch value {
+	case archivedModeInclude:
+		return archivedModeInclude, optional.None[bool]()
+	case "":
+		return archivedModeInclude, optional.None[bool]()
+	case archivedModeOnly:
+		return archivedModeOnly, optional.Some(true)
+	case archivedModeExclude:
+		return archivedModeExclude, optional.Some(false)
+	default:
+		return archivedModeInclude, optional.None[bool]()
+	}
+}
+
 // Code render explore code page
 func Code(ctx *context.Context) {
 	if !setting.Indexer.RepoIndexerEnabled || setting.Service.Explore.DisableCodePage {
@@ -36,6 +56,10 @@ func Code(ctx *context.Context) {
 	ctx.Data["PageIsExploreCode"] = true
 	ctx.Data["PageIsViewCode"] = true
 
+	archivedMode, archivedFilter := parseArchivedMode(ctx.FormTrim("archived"))
+	ctx.Data["ArchivedMode"] = archivedMode
+	ctx.Data["IsArchived"] = archivedFilter
+
 	prepareSearch := common.PrepareCodeSearch(ctx)
 	if prepareSearch.Keyword == "" {
 		ctx.HTML(http.StatusOK, tplExploreCode)
@@ -77,6 +101,7 @@ func Code(ctx *context.Context) {
 			Keyword:    prepareSearch.Keyword,
 			SearchMode: prepareSearch.SearchMode,
 			Language:   prepareSearch.Language,
+			Archived:   archivedFilter,
 			Paginator: &db.ListOptions{
 				Page:     page,
 				PageSize: setting.UI.RepoSearchPagingNum,

routers/web/repo/setting/setting.go

@@ -962,6 +962,12 @@ func handleSettingsPostArchive(ctx *context.Context) {
 
 	// update issue indexer
 	issue_indexer.UpdateRepoIndexer(ctx, repo.ID)
+	if setting.Indexer.RepoIndexerEnabled {
+		if err := repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, ""); err != nil {
+			log.Error("Reset code indexer status for archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
+		}
+		code.UpdateRepoIndexer(repo)
+	}
 
 	ctx.Flash.Success(ctx.Tr("repo.settings.archive.success"))
 
@@ -991,6 +997,12 @@ func handleSettingsPostUnarchive(ctx *context.Context) {
 
 	// update issue indexer
 	issue_indexer.UpdateRepoIndexer(ctx, repo.ID)
+	if setting.Indexer.RepoIndexerEnabled {
+		if err := repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, ""); err != nil {
+			log.Error("Reset code indexer status for un-archived repo %s/%s: %v", ctx.Repo.Owner.Name, repo.Name, err)
+		}
+		code.UpdateRepoIndexer(repo)
+	}
 
 	ctx.Flash.Success(ctx.Tr("repo.settings.unarchive.success"))
 

templates/shared/search/code/results.tmpl

@@ -1,7 +1,7 @@
 <div class="flex-text-block tw-flex-wrap">
 	{{range $term := .SearchResultLanguages}}
 	<a class="ui {{if eq $.Language $term.Language}}primary{{end}} basic label tw-m-0"
-		href="?q={{$.Keyword}}{{if ne $.Language $term.Language}}&l={{$term.Language}}{{end}}&search_mode={{$.SelectedSearchMode}}">
+		href="?q={{$.Keyword}}{{if ne $.Language $term.Language}}&l={{$term.Language}}{{end}}{{if $.SelectedSearchMode}}&search_mode={{$.SelectedSearchMode}}{{end}}{{if $.ArchivedMode}}&archived={{$.ArchivedMode}}{{end}}">
 		<i class="color-icon tw-mr-2" style="background-color: {{$term.Color}}"></i>
 		{{$term.Language}}
 		<div class="detail">{{$term.Count}}</div>

templates/shared/search/code/search.tmpl

@@ -1,12 +1,26 @@
-<form class="ui form ignore-dirty">
+<div class="ui small secondary filter menu">
-	{{template "shared/search/combo" (dict
+	<form id="repo-search-form" class="ui form ignore-dirty tw-flex-1 tw-flex tw-items-center tw-gap-x-2">
-	"Disabled" .CodeIndexerUnavailable
+		{{if .Language}}<input type="hidden" name="l" value="{{.Language}}">{{end}}
-	"Value" .Keyword
+		<div class="tw-flex-1">
-	"Placeholder" (ctx.Locale.Tr "search.code_kind")
+			{{template "shared/search/combo" (dict
-	"SearchModes" .SearchModes
+			"Disabled" .CodeIndexerUnavailable
-	"SelectedSearchMode" .SelectedSearchMode
+			"Value" .Keyword
-	)}}
+			"Placeholder" (ctx.Locale.Tr "search.code_kind")
-</form>
+			"SearchModes" .SearchModes
+			"SelectedSearchMode" .SelectedSearchMode
+			)}}
+		</div>
+		<div class="item ui small dropdown jump">
+			<span class="text">{{ctx.Locale.Tr "filter_title"}}</span>
+			{{svg "octicon-triangle-down" 14 "dropdown icon"}}
+			<div class="menu flex-items-menu">
+				<label class="item"><input type="radio" name="archived" {{if eq .ArchivedMode "include"}}checked{{end}} value="include"> {{ctx.Locale.Tr "search.code_archived_filter.all"}}</label>
+				<label class="item"><input type="radio" name="archived" {{if eq .ArchivedMode "exclude"}}checked{{end}} value="exclude"> {{ctx.Locale.Tr "search.code_archived_filter.not_archived"}}</label>
+				<label class="item"><input type="radio" name="archived" {{if .IsArchived.Value}}checked{{end}} value="only"> {{ctx.Locale.Tr "filter.is_archived"}}</label>
+			</div>
+		</div>
+	</form>
+</div>
 <div class="divider"></div>
 <div class="ui list">
 	{{template "base/alert" .}}