first-commit

modules/indexer/issues/bleve/bleve.go

@@ -0,0 +1,322 @@
+// Copyright 2018 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package bleve
+
+import (
+	"context"
+	"strconv"
+
+	"code.gitea.io/gitea/modules/indexer"
+	indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
+	inner_bleve "code.gitea.io/gitea/modules/indexer/internal/bleve"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/util"
+
+	"github.com/blevesearch/bleve/v2"
+	"github.com/blevesearch/bleve/v2/analysis/analyzer/custom"
+	"github.com/blevesearch/bleve/v2/analysis/token/camelcase"
+	"github.com/blevesearch/bleve/v2/analysis/token/lowercase"
+	"github.com/blevesearch/bleve/v2/analysis/token/unicodenorm"
+	"github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"
+	"github.com/blevesearch/bleve/v2/mapping"
+	"github.com/blevesearch/bleve/v2/search/query"
+)
+
+const (
+	issueIndexerAnalyzer      = "issueIndexer"
+	issueIndexerDocType       = "issueIndexerDocType"
+	issueIndexerLatestVersion = 5
+)
+
+const unicodeNormalizeName = "unicodeNormalize"
+
+func addUnicodeNormalizeTokenFilter(m *mapping.IndexMappingImpl) error {
+	return m.AddCustomTokenFilter(unicodeNormalizeName, map[string]any{
+		"type": unicodenorm.Name,
+		"form": unicodenorm.NFC,
+	})
+}
+
+const maxBatchSize = 16
+
+// IndexerData an update to the issue indexer
+type IndexerData internal.IndexerData
+
+// Type returns the document type, for bleve's mapping.Classifier interface.
+func (i *IndexerData) Type() string {
+	return issueIndexerDocType
+}
+
+// generateIssueIndexMapping generates the bleve index mapping for issues
+func generateIssueIndexMapping() (mapping.IndexMapping, error) {
+	mapping := bleve.NewIndexMapping()
+	docMapping := bleve.NewDocumentMapping()
+
+	numericFieldMapping := bleve.NewNumericFieldMapping()
+	numericFieldMapping.Store = false
+	numericFieldMapping.IncludeInAll = false
+	docMapping.AddFieldMappingsAt("repo_id", numericFieldMapping)
+
+	textFieldMapping := bleve.NewTextFieldMapping()
+	textFieldMapping.Store = false
+	textFieldMapping.IncludeInAll = false
+
+	boolFieldMapping := bleve.NewBooleanFieldMapping()
+	boolFieldMapping.Store = false
+	boolFieldMapping.IncludeInAll = false
+
+	numberFieldMapping := bleve.NewNumericFieldMapping()
+	numberFieldMapping.Store = false
+	numberFieldMapping.IncludeInAll = false
+
+	docMapping.AddFieldMappingsAt("is_public", boolFieldMapping)
+
+	docMapping.AddFieldMappingsAt("title", textFieldMapping)
+	docMapping.AddFieldMappingsAt("content", textFieldMapping)
+	docMapping.AddFieldMappingsAt("comments", textFieldMapping)
+
+	docMapping.AddFieldMappingsAt("is_pull", boolFieldMapping)
+	docMapping.AddFieldMappingsAt("is_closed", boolFieldMapping)
+	docMapping.AddFieldMappingsAt("is_archived", boolFieldMapping)
+	docMapping.AddFieldMappingsAt("label_ids", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("no_label", boolFieldMapping)
+	docMapping.AddFieldMappingsAt("milestone_id", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("project_id", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("project_board_id", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("poster_id", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("assignee_id", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("mention_ids", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("reviewed_ids", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("review_requested_ids", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("subscriber_ids", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("updated_unix", numberFieldMapping)
+
+	docMapping.AddFieldMappingsAt("created_unix", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("deadline_unix", numberFieldMapping)
+	docMapping.AddFieldMappingsAt("comment_count", numberFieldMapping)
+
+	if err := addUnicodeNormalizeTokenFilter(mapping); err != nil {
+		return nil, err
+	} else if err = mapping.AddCustomAnalyzer(issueIndexerAnalyzer, map[string]any{
+		"type":          custom.Name,
+		"char_filters":  []string{},
+		"tokenizer":     unicode.Name,
+		"token_filters": []string{unicodeNormalizeName, camelcase.Name, lowercase.Name},
+	}); err != nil {
+		return nil, err
+	}
+
+	mapping.DefaultAnalyzer = issueIndexerAnalyzer
+	mapping.AddDocumentMapping(issueIndexerDocType, docMapping)
+	mapping.AddDocumentMapping("_all", bleve.NewDocumentDisabledMapping())
+	mapping.DefaultMapping = bleve.NewDocumentDisabledMapping() // disable default mapping, avoid indexing unexpected structs
+
+	return mapping, nil
+}
+
+var _ internal.Indexer = &Indexer{}
+
+// Indexer implements Indexer interface
+type Indexer struct {
+	inner                    *inner_bleve.Indexer
+	indexer_internal.Indexer // do not composite inner_bleve.Indexer directly to avoid exposing too much
+}
+
+func (b *Indexer) SupportedSearchModes() []indexer.SearchMode {
+	return indexer.SearchModesExactWordsFuzzy()
+}
+
+// NewIndexer creates a new bleve local indexer
+func NewIndexer(indexDir string) *Indexer {
+	inner := inner_bleve.NewIndexer(indexDir, issueIndexerLatestVersion, generateIssueIndexMapping)
+	return &Indexer{
+		Indexer: inner,
+		inner:   inner,
+	}
+}
+
+// Index will save the index data
+func (b *Indexer) Index(_ context.Context, issues ...*internal.IndexerData) error {
+	batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
+	for _, issue := range issues {
+		if err := batch.Index(indexer_internal.Base36(issue.ID), (*IndexerData)(issue)); err != nil {
+			return err
+		}
+	}
+	return batch.Flush()
+}
+
+// Delete deletes indexes by ids
+func (b *Indexer) Delete(_ context.Context, ids ...int64) error {
+	batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
+	for _, id := range ids {
+		if err := batch.Delete(indexer_internal.Base36(id)); err != nil {
+			return err
+		}
+	}
+	return batch.Flush()
+}
+
+// Search searches for issues by given conditions.
+// Returns the matching issue IDs
+func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
+	var queries []query.Query
+
+	if options.Keyword != "" {
+		searchMode := util.IfZero(options.SearchMode, b.SupportedSearchModes()[0].ModeValue)
+		if searchMode == indexer.SearchModeWords || searchMode == indexer.SearchModeFuzzy {
+			fuzziness := 0
+			if searchMode == indexer.SearchModeFuzzy {
+				fuzziness = inner_bleve.GuessFuzzinessByKeyword(options.Keyword)
+			}
+			queries = append(queries, bleve.NewDisjunctionQuery([]query.Query{
+				inner_bleve.MatchAndQuery(options.Keyword, "title", issueIndexerAnalyzer, fuzziness),
+				inner_bleve.MatchAndQuery(options.Keyword, "content", issueIndexerAnalyzer, fuzziness),
+				inner_bleve.MatchAndQuery(options.Keyword, "comments", issueIndexerAnalyzer, fuzziness),
+			}...))
+		} else /* exact */ {
+			queries = append(queries, bleve.NewDisjunctionQuery([]query.Query{
+				inner_bleve.MatchPhraseQuery(options.Keyword, "title", issueIndexerAnalyzer, 0),
+				inner_bleve.MatchPhraseQuery(options.Keyword, "content", issueIndexerAnalyzer, 0),
+				inner_bleve.MatchPhraseQuery(options.Keyword, "comments", issueIndexerAnalyzer, 0),
+			}...))
+		}
+	}
+
+	if len(options.RepoIDs) > 0 || options.AllPublic {
+		var repoQueries []query.Query
+		for _, repoID := range options.RepoIDs {
+			repoQueries = append(repoQueries, inner_bleve.NumericEqualityQuery(repoID, "repo_id"))
+		}
+		if options.AllPublic {
+			repoQueries = append(repoQueries, inner_bleve.BoolFieldQuery(true, "is_public"))
+		}
+		queries = append(queries, bleve.NewDisjunctionQuery(repoQueries...))
+	}
+
+	if options.IsPull.Has() {
+		queries = append(queries, inner_bleve.BoolFieldQuery(options.IsPull.Value(), "is_pull"))
+	}
+	if options.IsClosed.Has() {
+		queries = append(queries, inner_bleve.BoolFieldQuery(options.IsClosed.Value(), "is_closed"))
+	}
+	if options.IsArchived.Has() {
+		queries = append(queries, inner_bleve.BoolFieldQuery(options.IsArchived.Value(), "is_archived"))
+	}
+
+	if options.NoLabelOnly {
+		queries = append(queries, inner_bleve.BoolFieldQuery(true, "no_label"))
+	} else {
+		if len(options.IncludedLabelIDs) > 0 {
+			var includeQueries []query.Query
+			for _, labelID := range options.IncludedLabelIDs {
+				includeQueries = append(includeQueries, inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
+			}
+			queries = append(queries, bleve.NewConjunctionQuery(includeQueries...))
+		} else if len(options.IncludedAnyLabelIDs) > 0 {
+			var includeQueries []query.Query
+			for _, labelID := range options.IncludedAnyLabelIDs {
+				includeQueries = append(includeQueries, inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
+			}
+			queries = append(queries, bleve.NewDisjunctionQuery(includeQueries...))
+		}
+		if len(options.ExcludedLabelIDs) > 0 {
+			var excludeQueries []query.Query
+			for _, labelID := range options.ExcludedLabelIDs {
+				q := bleve.NewBooleanQuery()
+				q.AddMustNot(inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
+				excludeQueries = append(excludeQueries, q)
+			}
+			queries = append(queries, bleve.NewConjunctionQuery(excludeQueries...))
+		}
+	}
+
+	if len(options.MilestoneIDs) > 0 {
+		var milestoneQueries []query.Query
+		for _, milestoneID := range options.MilestoneIDs {
+			milestoneQueries = append(milestoneQueries, inner_bleve.NumericEqualityQuery(milestoneID, "milestone_id"))
+		}
+		queries = append(queries, bleve.NewDisjunctionQuery(milestoneQueries...))
+	}
+
+	if options.ProjectID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ProjectID.Value(), "project_id"))
+	}
+	if options.ProjectColumnID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ProjectColumnID.Value(), "project_board_id"))
+	}
+
+	if options.PosterID != "" {
+		// "(none)" becomes 0, it means no poster
+		posterIDInt64, _ := strconv.ParseInt(options.PosterID, 10, 64)
+		queries = append(queries, inner_bleve.NumericEqualityQuery(posterIDInt64, "poster_id"))
+	}
+
+	if options.AssigneeID != "" {
+		if options.AssigneeID == "(any)" {
+			queries = append(queries, inner_bleve.NumericRangeInclusiveQuery(optional.Some[int64](1), optional.None[int64](), "assignee_id"))
+		} else {
+			// "(none)" becomes 0, it means no assignee
+			assigneeIDInt64, _ := strconv.ParseInt(options.AssigneeID, 10, 64)
+			queries = append(queries, inner_bleve.NumericEqualityQuery(assigneeIDInt64, "assignee_id"))
+		}
+	}
+
+	if options.MentionID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.MentionID.Value(), "mention_ids"))
+	}
+
+	if options.ReviewedID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ReviewedID.Value(), "reviewed_ids"))
+	}
+	if options.ReviewRequestedID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ReviewRequestedID.Value(), "review_requested_ids"))
+	}
+
+	if options.SubscriberID.Has() {
+		queries = append(queries, inner_bleve.NumericEqualityQuery(options.SubscriberID.Value(), "subscriber_ids"))
+	}
+
+	if options.UpdatedAfterUnix.Has() || options.UpdatedBeforeUnix.Has() {
+		queries = append(queries, inner_bleve.NumericRangeInclusiveQuery(
+			options.UpdatedAfterUnix,
+			options.UpdatedBeforeUnix,
+			"updated_unix"))
+	}
+
+	var indexerQuery query.Query = bleve.NewConjunctionQuery(queries...)
+	if len(queries) == 0 {
+		indexerQuery = bleve.NewMatchAllQuery()
+	}
+
+	skip, limit := indexer_internal.ParsePaginator(options.Paginator)
+	search := bleve.NewSearchRequestOptions(indexerQuery, limit, skip, false)
+
+	if options.SortBy == "" {
+		options.SortBy = internal.SortByCreatedAsc
+	}
+
+	search.SortBy([]string{string(options.SortBy), "-_id"})
+
+	result, err := b.inner.Indexer.SearchInContext(ctx, search)
+	if err != nil {
+		return nil, err
+	}
+
+	ret := &internal.SearchResult{
+		Total: int64(result.Total),
+		Hits:  make([]internal.Match, 0, len(result.Hits)),
+	}
+	for _, hit := range result.Hits {
+		id, err := indexer_internal.ParseBase36(hit.ID)
+		if err != nil {
+			return nil, err
+		}
+		ret.Hits = append(ret.Hits, internal.Match{
+			ID: id,
+		})
+	}
+	return ret, nil
+}

modules/indexer/issues/bleve/bleve_test.go

@@ -0,0 +1,18 @@
+// Copyright 2018 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package bleve
+
+import (
+	"testing"
+
+	"code.gitea.io/gitea/modules/indexer/issues/internal/tests"
+)
+
+func TestBleveIndexer(t *testing.T) {
+	dir := t.TempDir()
+	indexer := NewIndexer(dir)
+	defer indexer.Close()
+
+	tests.TestIndexer(t, indexer)
+}

modules/indexer/issues/db/db.go

@@ -0,0 +1,143 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package db
+
+import (
+	"context"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/models/db"
+	issue_model "code.gitea.io/gitea/models/issues"
+	"code.gitea.io/gitea/modules/indexer"
+	indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
+	inner_db "code.gitea.io/gitea/modules/indexer/internal/db"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/util"
+
+	"xorm.io/builder"
+)
+
+var _ internal.Indexer = (*Indexer)(nil)
+
+// Indexer implements Indexer interface to use database's like search
+type Indexer struct {
+	indexer_internal.Indexer
+}
+
+func (i *Indexer) SupportedSearchModes() []indexer.SearchMode {
+	return indexer.SearchModesExactWords()
+}
+
+var GetIndexer = sync.OnceValue(func() *Indexer {
+	return &Indexer{Indexer: &inner_db.Indexer{}}
+})
+
+// Index dummy function
+func (i *Indexer) Index(_ context.Context, _ ...*internal.IndexerData) error {
+	return nil
+}
+
+// Delete dummy function
+func (i *Indexer) Delete(_ context.Context, _ ...int64) error {
+	return nil
+}
+
+func buildMatchQuery(mode indexer.SearchModeType, colName, keyword string) builder.Cond {
+	if mode == indexer.SearchModeExact {
+		return db.BuildCaseInsensitiveLike(colName, keyword)
+	}
+
+	// match words
+	cond := builder.NewCond()
+	fields := strings.Fields(keyword)
+	if len(fields) == 0 {
+		return builder.Expr("1=1")
+	}
+	for _, field := range fields {
+		if field == "" {
+			continue
+		}
+		cond = cond.And(db.BuildCaseInsensitiveLike(colName, field))
+	}
+	return cond
+}
+
+// Search searches for issues
+func (i *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
+	// FIXME: I tried to avoid importing models here, but it seems to be impossible.
+	//        We can provide a function to register the search function, so models/issues can register it.
+	//        So models/issues will import modules/indexer/issues, it's OK because it's by design.
+	//        But modules/indexer/issues has already imported models/issues to do UpdateRepoIndexer and UpdateIssueIndexer.
+	//        And to avoid circular import, we have to move the functions to another package.
+	//        I believe it should be services/indexer, sounds great!
+	//        But the two functions are used in modules/notification/indexer, that means we will import services/indexer in modules/notification/indexer.
+	//        So that's the root problem:
+	//        The notification is defined in modules, but it's using lots of things should be in services.
+
+	cond := builder.NewCond()
+
+	if options.Keyword != "" {
+		repoCond := builder.In("repo_id", options.RepoIDs)
+		if len(options.RepoIDs) == 1 {
+			repoCond = builder.Eq{"repo_id": options.RepoIDs[0]}
+		}
+		subQuery := builder.Select("id").From("issue").Where(repoCond)
+		searchMode := util.IfZero(options.SearchMode, i.SupportedSearchModes()[0].ModeValue)
+		cond = builder.Or(
+			buildMatchQuery(searchMode, "issue.name", options.Keyword),
+			buildMatchQuery(searchMode, "issue.content", options.Keyword),
+			builder.In("issue.id", builder.Select("issue_id").
+				From("comment").
+				Where(builder.And(
+					builder.Eq{"type": issue_model.CommentTypeComment},
+					builder.In("issue_id", subQuery),
+					buildMatchQuery(searchMode, "content", options.Keyword),
+				)),
+			),
+		)
+
+		if options.IsKeywordNumeric() {
+			cond = cond.Or(
+				builder.Eq{"`index`": options.Keyword},
+			)
+		}
+	}
+
+	opt, err := ToDBOptions(ctx, options)
+	if err != nil {
+		return nil, err
+	}
+
+	// If pagesize == 0, return total count only. It's a special case for search count.
+	if options.Paginator != nil && options.Paginator.PageSize == 0 {
+		total, err := issue_model.CountIssues(ctx, opt, cond)
+		if err != nil {
+			return nil, err
+		}
+		return &internal.SearchResult{
+			Total: total,
+		}, nil
+	}
+
+	return i.FindWithIssueOptions(ctx, opt, cond)
+}
+
+func (i *Indexer) FindWithIssueOptions(ctx context.Context, opt *issue_model.IssuesOptions, otherConds ...builder.Cond) (*internal.SearchResult, error) {
+	ids, total, err := issue_model.IssueIDs(ctx, opt, otherConds...)
+	if err != nil {
+		return nil, err
+	}
+
+	hits := make([]internal.Match, 0, len(ids))
+	for _, id := range ids {
+		hits = append(hits, internal.Match{
+			ID: id,
+		})
+	}
+	return &internal.SearchResult{
+		Total: total,
+		Hits:  hits,
+	}, nil
+}

modules/indexer/issues/db/options.go

@@ -0,0 +1,116 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package db
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	"code.gitea.io/gitea/models/db"
+	issue_model "code.gitea.io/gitea/models/issues"
+	"code.gitea.io/gitea/modules/container"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/optional"
+)
+
+func ToDBOptions(ctx context.Context, options *internal.SearchOptions) (*issue_model.IssuesOptions, error) {
+	var sortType string
+	switch options.SortBy {
+	case internal.SortByCreatedAsc:
+		sortType = "oldest"
+	case internal.SortByUpdatedAsc:
+		sortType = "leastupdate"
+	case internal.SortByCommentsAsc:
+		sortType = "leastcomment"
+	case internal.SortByDeadlineDesc:
+		sortType = "farduedate"
+	case internal.SortByCreatedDesc:
+		sortType = "newest"
+	case internal.SortByUpdatedDesc:
+		sortType = "recentupdate"
+	case internal.SortByCommentsDesc:
+		sortType = "mostcomment"
+	case internal.SortByDeadlineAsc:
+		sortType = "nearduedate"
+	default:
+		if strings.HasPrefix(string(options.SortBy), issue_model.ScopeSortPrefix) {
+			sortType = string(options.SortBy)
+		} else {
+			sortType = "newest"
+		}
+	}
+
+	// See the comment of issues_model.SearchOptions for the reason why we need to convert
+	convertID := func(id optional.Option[int64]) int64 {
+		if !id.Has() {
+			return 0
+		}
+		value := id.Value()
+		if value == 0 {
+			return db.NoConditionID
+		}
+		return value
+	}
+
+	opts := &issue_model.IssuesOptions{
+		Paginator:          options.Paginator,
+		RepoIDs:            options.RepoIDs,
+		AllPublic:          options.AllPublic,
+		RepoCond:           nil,
+		AssigneeID:         options.AssigneeID,
+		PosterID:           options.PosterID,
+		MentionedID:        convertID(options.MentionID),
+		ReviewRequestedID:  convertID(options.ReviewRequestedID),
+		ReviewedID:         convertID(options.ReviewedID),
+		SubscriberID:       convertID(options.SubscriberID),
+		ProjectID:          convertID(options.ProjectID),
+		ProjectColumnID:    convertID(options.ProjectColumnID),
+		IsClosed:           options.IsClosed,
+		IsPull:             options.IsPull,
+		IncludedLabelNames: nil,
+		ExcludedLabelNames: nil,
+		IncludeMilestones:  nil,
+		SortType:           sortType,
+		UpdatedAfterUnix:   options.UpdatedAfterUnix.Value(),
+		UpdatedBeforeUnix:  options.UpdatedBeforeUnix.Value(),
+		PriorityRepoID:     0,
+		IsArchived:         options.IsArchived,
+		Owner:              nil,
+		Team:               nil,
+		Doer:               nil,
+	}
+
+	if len(options.MilestoneIDs) == 1 && options.MilestoneIDs[0] == 0 {
+		opts.MilestoneIDs = []int64{db.NoConditionID}
+	} else {
+		opts.MilestoneIDs = options.MilestoneIDs
+	}
+
+	if options.NoLabelOnly {
+		opts.LabelIDs = []int64{0} // Be careful, it's zero, not db.NoConditionID
+	} else {
+		opts.LabelIDs = make([]int64, 0, len(options.IncludedLabelIDs)+len(options.ExcludedLabelIDs))
+		opts.LabelIDs = append(opts.LabelIDs, options.IncludedLabelIDs...)
+		for _, id := range options.ExcludedLabelIDs {
+			opts.LabelIDs = append(opts.LabelIDs, -id)
+		}
+
+		if len(options.IncludedLabelIDs) == 0 && len(options.IncludedAnyLabelIDs) > 0 {
+			labels, err := issue_model.GetLabelsByIDs(ctx, options.IncludedAnyLabelIDs, "name")
+			if err != nil {
+				return nil, fmt.Errorf("GetLabelsByIDs: %v", err)
+			}
+			set := container.Set[string]{}
+			for _, label := range labels {
+				if !set.Contains(label.Name) {
+					set.Add(label.Name)
+					opts.IncludedLabelNames = append(opts.IncludedLabelNames, label.Name)
+				}
+			}
+		}
+	}
+
+	return opts, nil
+}

modules/indexer/issues/dboptions.go

@@ -0,0 +1,113 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package issues
+
+import (
+	"strings"
+
+	"code.gitea.io/gitea/models/db"
+	issues_model "code.gitea.io/gitea/models/issues"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/setting"
+)
+
+func ToSearchOptions(keyword string, opts *issues_model.IssuesOptions) *SearchOptions {
+	if opts.IssueIDs != nil {
+		setting.PanicInDevOrTesting("Indexer SearchOptions doesn't support IssueIDs")
+	}
+	searchOpt := &SearchOptions{
+		Keyword:    keyword,
+		RepoIDs:    opts.RepoIDs,
+		AllPublic:  opts.AllPublic,
+		IsPull:     opts.IsPull,
+		IsClosed:   opts.IsClosed,
+		IsArchived: opts.IsArchived,
+	}
+
+	if len(opts.LabelIDs) == 1 && opts.LabelIDs[0] == 0 {
+		searchOpt.NoLabelOnly = true
+	} else {
+		for _, labelID := range opts.LabelIDs {
+			if labelID > 0 {
+				searchOpt.IncludedLabelIDs = append(searchOpt.IncludedLabelIDs, labelID)
+			} else {
+				searchOpt.ExcludedLabelIDs = append(searchOpt.ExcludedLabelIDs, -labelID)
+			}
+		}
+		// opts.IncludedLabelNames and opts.ExcludedLabelNames are not supported here.
+		// It's not a TO DO, it's just unnecessary.
+	}
+
+	if len(opts.MilestoneIDs) == 1 && opts.MilestoneIDs[0] == db.NoConditionID {
+		searchOpt.MilestoneIDs = []int64{0}
+	} else {
+		searchOpt.MilestoneIDs = opts.MilestoneIDs
+	}
+
+	if opts.ProjectID > 0 {
+		searchOpt.ProjectID = optional.Some(opts.ProjectID)
+	} else if opts.ProjectID == db.NoConditionID { // FIXME: this is inconsistent from other places
+		searchOpt.ProjectID = optional.Some[int64](0) // Those issues with no project(projectid==0)
+	}
+
+	searchOpt.AssigneeID = opts.AssigneeID
+
+	// See the comment of issues_model.SearchOptions for the reason why we need to convert
+	convertID := func(id int64) optional.Option[int64] {
+		if id > 0 {
+			return optional.Some(id)
+		}
+		if id == db.NoConditionID {
+			return optional.None[int64]()
+		}
+		return nil
+	}
+
+	searchOpt.ProjectColumnID = convertID(opts.ProjectColumnID)
+	searchOpt.PosterID = opts.PosterID
+	searchOpt.MentionID = convertID(opts.MentionedID)
+	searchOpt.ReviewedID = convertID(opts.ReviewedID)
+	searchOpt.ReviewRequestedID = convertID(opts.ReviewRequestedID)
+	searchOpt.SubscriberID = convertID(opts.SubscriberID)
+
+	if opts.UpdatedAfterUnix > 0 {
+		searchOpt.UpdatedAfterUnix = optional.Some(opts.UpdatedAfterUnix)
+	}
+	if opts.UpdatedBeforeUnix > 0 {
+		searchOpt.UpdatedBeforeUnix = optional.Some(opts.UpdatedBeforeUnix)
+	}
+
+	searchOpt.Paginator = opts.Paginator
+
+	switch opts.SortType {
+	case "", "latest":
+		searchOpt.SortBy = SortByCreatedDesc
+	case "oldest":
+		searchOpt.SortBy = SortByCreatedAsc
+	case "recentupdate":
+		searchOpt.SortBy = SortByUpdatedDesc
+	case "leastupdate":
+		searchOpt.SortBy = SortByUpdatedAsc
+	case "mostcomment":
+		searchOpt.SortBy = SortByCommentsDesc
+	case "leastcomment":
+		searchOpt.SortBy = SortByCommentsAsc
+	case "nearduedate":
+		searchOpt.SortBy = SortByDeadlineAsc
+	case "farduedate":
+		searchOpt.SortBy = SortByDeadlineDesc
+	case "priority", "priorityrepo", "project-column-sorting":
+		// Unsupported sort type for search
+		fallthrough
+	default:
+		if strings.HasPrefix(opts.SortType, issues_model.ScopeSortPrefix) {
+			searchOpt.SortBy = internal.SortBy(opts.SortType)
+		} else {
+			searchOpt.SortBy = SortByUpdatedDesc
+		}
+	}
+
+	return searchOpt
+}

modules/indexer/issues/elasticsearch/elasticsearch.go

@@ -0,0 +1,310 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package elasticsearch
+
+import (
+	"context"
+	"strconv"
+	"strings"
+
+	"code.gitea.io/gitea/modules/graceful"
+	"code.gitea.io/gitea/modules/indexer"
+	indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
+	inner_elasticsearch "code.gitea.io/gitea/modules/indexer/internal/elasticsearch"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/util"
+
+	"github.com/olivere/elastic/v7"
+)
+
+const (
+	issueIndexerLatestVersion = 2
+	// 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"
+	esMultiMatchTypePhrasePrefix = "phrase_prefix"
+)
+
+var _ internal.Indexer = &Indexer{}
+
+// Indexer implements Indexer interface
+type Indexer struct {
+	inner                    *inner_elasticsearch.Indexer
+	indexer_internal.Indexer // do not composite inner_elasticsearch.Indexer directly to avoid exposing too much
+}
+
+func (b *Indexer) SupportedSearchModes() []indexer.SearchMode {
+	// TODO: es supports fuzzy search, but our code doesn't at the moment, and actually the default fuzziness is already "AUTO"
+	return indexer.SearchModesExactWords()
+}
+
+// NewIndexer creates a new elasticsearch indexer
+func NewIndexer(url, indexerName string) *Indexer {
+	inner := inner_elasticsearch.NewIndexer(url, indexerName, issueIndexerLatestVersion, defaultMapping)
+	indexer := &Indexer{
+		inner:   inner,
+		Indexer: inner,
+	}
+	return indexer
+}
+
+const (
+	defaultMapping = `
+{
+	"mappings": {
+		"properties": {
+			"id": { "type": "integer", "index": true },
+			"repo_id": { "type": "integer", "index": true },
+			"is_public": { "type": "boolean", "index": true },
+
+			"title": {  "type": "text", "index": true },
+			"content": { "type": "text", "index": true },
+			"comments": { "type" : "text", "index": true },
+
+			"is_pull": { "type": "boolean", "index": true },
+			"is_closed": { "type": "boolean", "index": true },
+			"is_archived": { "type": "boolean", "index": true },
+			"label_ids": { "type": "integer", "index": true },
+			"no_label": { "type": "boolean", "index": true },
+			"milestone_id": { "type": "integer", "index": true },
+			"project_id": { "type": "integer", "index": true },
+			"project_board_id": { "type": "integer", "index": true },
+			"poster_id": { "type": "integer", "index": true },
+			"assignee_id": { "type": "integer", "index": true },
+			"mention_ids": { "type": "integer", "index": true },
+			"reviewed_ids": { "type": "integer", "index": true },
+			"review_requested_ids": { "type": "integer", "index": true },
+			"subscriber_ids": { "type": "integer", "index": true },
+			"updated_unix": { "type": "integer", "index": true },
+
+			"created_unix": { "type": "integer", "index": true },
+			"deadline_unix": { "type": "integer", "index": true },
+			"comment_count": { "type": "integer", "index": true }
+		}
+	}
+}
+`
+)
+
+// Index will save the index data
+func (b *Indexer) Index(ctx context.Context, issues ...*internal.IndexerData) error {
+	if len(issues) == 0 {
+		return nil
+	} else if len(issues) == 1 {
+		issue := issues[0]
+		_, err := b.inner.Client.Index().
+			Index(b.inner.VersionedIndexName()).
+			Id(strconv.FormatInt(issue.ID, 10)).
+			BodyJson(issue).
+			Do(ctx)
+		return err
+	}
+
+	reqs := make([]elastic.BulkableRequest, 0)
+	for _, issue := range issues {
+		reqs = append(reqs,
+			elastic.NewBulkIndexRequest().
+				Index(b.inner.VersionedIndexName()).
+				Id(strconv.FormatInt(issue.ID, 10)).
+				Doc(issue),
+		)
+	}
+
+	_, err := b.inner.Client.Bulk().
+		Index(b.inner.VersionedIndexName()).
+		Add(reqs...).
+		Do(graceful.GetManager().HammerContext())
+	return err
+}
+
+// Delete deletes indexes by ids
+func (b *Indexer) Delete(ctx context.Context, ids ...int64) error {
+	if len(ids) == 0 {
+		return nil
+	} else if len(ids) == 1 {
+		_, err := b.inner.Client.Delete().
+			Index(b.inner.VersionedIndexName()).
+			Id(strconv.FormatInt(ids[0], 10)).
+			Do(ctx)
+		return err
+	}
+
+	reqs := make([]elastic.BulkableRequest, 0)
+	for _, id := range ids {
+		reqs = append(reqs,
+			elastic.NewBulkDeleteRequest().
+				Index(b.inner.VersionedIndexName()).
+				Id(strconv.FormatInt(id, 10)),
+		)
+	}
+
+	_, err := b.inner.Client.Bulk().
+		Index(b.inner.VersionedIndexName()).
+		Add(reqs...).
+		Do(graceful.GetManager().HammerContext())
+	return err
+}
+
+// Search searches for issues by given conditions.
+// Returns the matching issue IDs
+func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
+	query := elastic.NewBoolQuery()
+
+	if options.Keyword != "" {
+		searchMode := util.IfZero(options.SearchMode, b.SupportedSearchModes()[0].ModeValue)
+		if searchMode == indexer.SearchModeExact {
+			query.Must(elastic.NewMultiMatchQuery(options.Keyword, "title", "content", "comments").Type(esMultiMatchTypePhrasePrefix))
+		} else /* words */ {
+			query.Must(elastic.NewMultiMatchQuery(options.Keyword, "title", "content", "comments").Type(esMultiMatchTypeBestFields).Operator("and"))
+		}
+	}
+
+	if len(options.RepoIDs) > 0 {
+		q := elastic.NewBoolQuery()
+		q.Should(elastic.NewTermsQuery("repo_id", toAnySlice(options.RepoIDs)...))
+		if options.AllPublic {
+			q.Should(elastic.NewTermQuery("is_public", true))
+		}
+		query.Must(q)
+	}
+
+	if options.IsPull.Has() {
+		query.Must(elastic.NewTermQuery("is_pull", options.IsPull.Value()))
+	}
+	if options.IsClosed.Has() {
+		query.Must(elastic.NewTermQuery("is_closed", options.IsClosed.Value()))
+	}
+	if options.IsArchived.Has() {
+		query.Must(elastic.NewTermQuery("is_archived", options.IsArchived.Value()))
+	}
+
+	if options.NoLabelOnly {
+		query.Must(elastic.NewTermQuery("no_label", true))
+	} else {
+		if len(options.IncludedLabelIDs) > 0 {
+			q := elastic.NewBoolQuery()
+			for _, labelID := range options.IncludedLabelIDs {
+				q.Must(elastic.NewTermQuery("label_ids", labelID))
+			}
+			query.Must(q)
+		} else if len(options.IncludedAnyLabelIDs) > 0 {
+			query.Must(elastic.NewTermsQuery("label_ids", toAnySlice(options.IncludedAnyLabelIDs)...))
+		}
+		if len(options.ExcludedLabelIDs) > 0 {
+			q := elastic.NewBoolQuery()
+			for _, labelID := range options.ExcludedLabelIDs {
+				q.MustNot(elastic.NewTermQuery("label_ids", labelID))
+			}
+			query.Must(q)
+		}
+	}
+
+	if len(options.MilestoneIDs) > 0 {
+		query.Must(elastic.NewTermsQuery("milestone_id", toAnySlice(options.MilestoneIDs)...))
+	}
+
+	if options.ProjectID.Has() {
+		query.Must(elastic.NewTermQuery("project_id", options.ProjectID.Value()))
+	}
+	if options.ProjectColumnID.Has() {
+		query.Must(elastic.NewTermQuery("project_board_id", options.ProjectColumnID.Value()))
+	}
+
+	if options.PosterID != "" {
+		// "(none)" becomes 0, it means no poster
+		posterIDInt64, _ := strconv.ParseInt(options.PosterID, 10, 64)
+		query.Must(elastic.NewTermQuery("poster_id", posterIDInt64))
+	}
+
+	if options.AssigneeID != "" {
+		if options.AssigneeID == "(any)" {
+			q := elastic.NewRangeQuery("assignee_id")
+			q.Gte(1)
+			query.Must(q)
+		} else {
+			// "(none)" becomes 0, it means no assignee
+			assigneeIDInt64, _ := strconv.ParseInt(options.AssigneeID, 10, 64)
+			query.Must(elastic.NewTermQuery("assignee_id", assigneeIDInt64))
+		}
+	}
+
+	if options.MentionID.Has() {
+		query.Must(elastic.NewTermQuery("mention_ids", options.MentionID.Value()))
+	}
+
+	if options.ReviewedID.Has() {
+		query.Must(elastic.NewTermQuery("reviewed_ids", options.ReviewedID.Value()))
+	}
+	if options.ReviewRequestedID.Has() {
+		query.Must(elastic.NewTermQuery("review_requested_ids", options.ReviewRequestedID.Value()))
+	}
+
+	if options.SubscriberID.Has() {
+		query.Must(elastic.NewTermQuery("subscriber_ids", options.SubscriberID.Value()))
+	}
+
+	if options.UpdatedAfterUnix.Has() || options.UpdatedBeforeUnix.Has() {
+		q := elastic.NewRangeQuery("updated_unix")
+		if options.UpdatedAfterUnix.Has() {
+			q.Gte(options.UpdatedAfterUnix.Value())
+		}
+		if options.UpdatedBeforeUnix.Has() {
+			q.Lte(options.UpdatedBeforeUnix.Value())
+		}
+		query.Must(q)
+	}
+
+	if options.SortBy == "" {
+		options.SortBy = internal.SortByCreatedAsc
+	}
+	sortBy := []elastic.Sorter{
+		parseSortBy(options.SortBy),
+		elastic.NewFieldSort("id").Desc(),
+	}
+
+	// See https://stackoverflow.com/questions/35206409/elasticsearch-2-1-result-window-is-too-large-index-max-result-window/35221900
+	// TODO: make it configurable since it's configurable in elasticsearch
+	const maxPageSize = 10000
+
+	skip, limit := indexer_internal.ParsePaginator(options.Paginator, maxPageSize)
+	searchResult, err := b.inner.Client.Search().
+		Index(b.inner.VersionedIndexName()).
+		Query(query).
+		SortBy(sortBy...).
+		From(skip).Size(limit).
+		Do(ctx)
+	if err != nil {
+		return nil, err
+	}
+
+	hits := make([]internal.Match, 0, limit)
+	for _, hit := range searchResult.Hits.Hits {
+		id, _ := strconv.ParseInt(hit.Id, 10, 64)
+		hits = append(hits, internal.Match{
+			ID: id,
+		})
+	}
+
+	return &internal.SearchResult{
+		Total: searchResult.TotalHits(),
+		Hits:  hits,
+	}, nil
+}
+
+func toAnySlice[T any](s []T) []any {
+	ret := make([]any, 0, len(s))
+	for _, item := range s {
+		ret = append(ret, item)
+	}
+	return ret
+}
+
+func parseSortBy(sortBy internal.SortBy) elastic.Sorter {
+	field := strings.TrimPrefix(string(sortBy), "-")
+	ret := elastic.NewFieldSort(field)
+	if strings.HasPrefix(string(sortBy), "-") {
+		ret.Desc()
+	}
+	return ret
+}

modules/indexer/issues/elasticsearch/elasticsearch_test.go

@@ -0,0 +1,40 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package elasticsearch
+
+import (
+	"fmt"
+	"net/http"
+	"os"
+	"testing"
+	"time"
+
+	"code.gitea.io/gitea/modules/indexer/issues/internal/tests"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestElasticsearchIndexer(t *testing.T) {
+	// The elasticsearch instance started by pull-db-tests.yml > test-unit > services > elasticsearch
+	url := "http://elastic:changeme@elasticsearch:9200"
+
+	if os.Getenv("CI") == "" {
+		// Make it possible to run tests against a local elasticsearch instance
+		url = os.Getenv("TEST_ELASTICSEARCH_URL")
+		if url == "" {
+			t.Skip("TEST_ELASTICSEARCH_URL not set and not running in CI")
+			return
+		}
+	}
+
+	require.Eventually(t, func() bool {
+		resp, err := http.Get(url)
+		return err == nil && resp.StatusCode == http.StatusOK
+	}, time.Minute, time.Second, "Expected elasticsearch to be up")
+
+	indexer := NewIndexer(url, fmt.Sprintf("test_elasticsearch_indexer_%d", time.Now().Unix()))
+	defer indexer.Close()
+
+	tests.TestIndexer(t, indexer)
+}

modules/indexer/issues/indexer.go

@@ -0,0 +1,326 @@
+// Copyright 2018 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package issues
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"runtime/pprof"
+	"sync/atomic"
+	"time"
+
+	db_model "code.gitea.io/gitea/models/db"
+	repo_model "code.gitea.io/gitea/models/repo"
+	"code.gitea.io/gitea/modules/graceful"
+	"code.gitea.io/gitea/modules/indexer"
+	"code.gitea.io/gitea/modules/indexer/issues/bleve"
+	"code.gitea.io/gitea/modules/indexer/issues/db"
+	"code.gitea.io/gitea/modules/indexer/issues/elasticsearch"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/indexer/issues/meilisearch"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/process"
+	"code.gitea.io/gitea/modules/queue"
+	"code.gitea.io/gitea/modules/setting"
+)
+
+// IndexerMetadata is used to send data to the queue, so it contains only the ids.
+// It may look weired, because it has to be compatible with the old queue data format.
+// If the IsDelete flag is true, the IDs specify the issues to delete from the index without querying the database.
+// If the IsDelete flag is false, the ID specify the issue to index, so Indexer will query the database to get the issue data.
+// It should be noted that if the id is not existing in the database, it's index will be deleted too even if IsDelete is false.
+// Valid values:
+//   - IsDelete = true, IDs = [1, 2, 3], and ID will be ignored
+//   - IsDelete = false, ID = 1, and IDs will be ignored
+type IndexerMetadata struct {
+	ID int64 `json:"id"`
+
+	IsDelete bool    `json:"is_delete"`
+	IDs      []int64 `json:"ids"`
+}
+
+var (
+	// issueIndexerQueue queue of issue ids to be updated
+	issueIndexerQueue *queue.WorkerPoolQueue[*IndexerMetadata]
+	// globalIndexer is the global indexer, it cannot be nil.
+	// When the real indexer is not ready, it will be a dummy indexer which will return error to explain it's not ready.
+	// So it's always safe use it as *globalIndexer.Load() and call its methods.
+	globalIndexer atomic.Pointer[internal.Indexer]
+	dummyIndexer  *internal.Indexer
+)
+
+func init() {
+	i := internal.NewDummyIndexer()
+	dummyIndexer = &i
+	globalIndexer.Store(dummyIndexer)
+}
+
+// InitIssueIndexer initialize issue indexer, syncReindex is true then reindex until
+// all issue index done.
+func InitIssueIndexer(syncReindex bool) {
+	ctx, _, finished := process.GetManager().AddTypedContext(context.Background(), "Service: IssueIndexer", process.SystemProcessType, false)
+
+	indexerInitWaitChannel := make(chan time.Duration, 1)
+
+	// Create the Queue
+	issueIndexerQueue = queue.CreateUniqueQueue(ctx, "issue_indexer", getIssueIndexerQueueHandler(ctx))
+
+	graceful.GetManager().RunAtTerminate(finished)
+
+	// Create the Indexer
+	go func() {
+		pprof.SetGoroutineLabels(ctx)
+		start := time.Now()
+		log.Info("PID %d: Initializing Issue Indexer: %s", os.Getpid(), setting.Indexer.IssueType)
+		var (
+			issueIndexer internal.Indexer
+			existed      bool
+			err          error
+		)
+		switch setting.Indexer.IssueType {
+		case "bleve":
+			defer func() {
+				if err := recover(); err != nil {
+					log.Error("PANIC whilst initializing issue indexer: %v\nStacktrace: %s", err, log.Stack(2))
+					log.Error("The indexer files are likely corrupted and may need to be deleted")
+					log.Error("You can completely remove the %q directory to make Gitea recreate the indexes", setting.Indexer.IssuePath)
+					globalIndexer.Store(dummyIndexer)
+					log.Fatal("PID: %d Unable to initialize the Bleve Issue Indexer at path: %s Error: %v", os.Getpid(), setting.Indexer.IssuePath, err)
+				}
+			}()
+			issueIndexer = bleve.NewIndexer(setting.Indexer.IssuePath)
+			existed, err = issueIndexer.Init(ctx)
+			if err != nil {
+				log.Fatal("Unable to initialize Bleve Issue Indexer at path: %s Error: %v", setting.Indexer.IssuePath, err)
+			}
+		case "elasticsearch":
+			issueIndexer = elasticsearch.NewIndexer(setting.Indexer.IssueConnStr, setting.Indexer.IssueIndexerName)
+			existed, err = issueIndexer.Init(ctx)
+			if err != nil {
+				log.Fatal("Unable to issueIndexer.Init with connection %s Error: %v", setting.Indexer.IssueConnStr, err)
+			}
+		case "db":
+			issueIndexer = db.GetIndexer()
+		case "meilisearch":
+			issueIndexer = meilisearch.NewIndexer(setting.Indexer.IssueConnStr, setting.Indexer.IssueConnAuth, setting.Indexer.IssueIndexerName)
+			existed, err = issueIndexer.Init(ctx)
+			if err != nil {
+				log.Fatal("Unable to issueIndexer.Init with connection %s Error: %v", setting.Indexer.IssueConnStr, err)
+			}
+		default:
+			log.Fatal("Unknown issue indexer type: %s", setting.Indexer.IssueType)
+		}
+		globalIndexer.Store(&issueIndexer)
+
+		graceful.GetManager().RunAtTerminate(func() {
+			log.Debug("Closing issue indexer")
+			(*globalIndexer.Load()).Close()
+			log.Info("PID: %d Issue Indexer closed", os.Getpid())
+		})
+
+		// Start processing the queue
+		go graceful.GetManager().RunWithCancel(issueIndexerQueue)
+
+		// Populate the index
+		if !existed {
+			if syncReindex {
+				graceful.GetManager().RunWithShutdownContext(populateIssueIndexer)
+			} else {
+				go graceful.GetManager().RunWithShutdownContext(populateIssueIndexer)
+			}
+		}
+
+		indexerInitWaitChannel <- time.Since(start)
+		close(indexerInitWaitChannel)
+	}()
+
+	if syncReindex {
+		select {
+		case <-indexerInitWaitChannel:
+		case <-graceful.GetManager().IsShutdown():
+		}
+	} else if setting.Indexer.StartupTimeout > 0 {
+		go func() {
+			pprof.SetGoroutineLabels(ctx)
+			timeout := setting.Indexer.StartupTimeout
+			if graceful.GetManager().IsChild() && setting.GracefulHammerTime > 0 {
+				timeout += setting.GracefulHammerTime
+			}
+			select {
+			case duration := <-indexerInitWaitChannel:
+				log.Info("Issue Indexer Initialization took %v", duration)
+			case <-graceful.GetManager().IsShutdown():
+				log.Warn("Shutdown occurred before issue index initialisation was complete")
+			case <-time.After(timeout):
+				issueIndexerQueue.ShutdownWait(5 * time.Second)
+				log.Fatal("Issue Indexer Initialization timed-out after: %v", timeout)
+			}
+		}()
+	}
+}
+
+func getIssueIndexerQueueHandler(ctx context.Context) func(items ...*IndexerMetadata) []*IndexerMetadata {
+	return func(items ...*IndexerMetadata) []*IndexerMetadata {
+		var unhandled []*IndexerMetadata
+
+		indexer := *globalIndexer.Load()
+		for _, item := range items {
+			log.Trace("IndexerMetadata Process: %d %v %t", item.ID, item.IDs, item.IsDelete)
+			if item.IsDelete {
+				if err := indexer.Delete(ctx, item.IDs...); err != nil {
+					log.Error("Issue indexer handler: failed to from index: %v Error: %v", item.IDs, err)
+					unhandled = append(unhandled, item)
+				}
+				continue
+			}
+			data, existed, err := getIssueIndexerData(ctx, item.ID)
+			if err != nil {
+				log.Error("Issue indexer handler: failed to get issue data of %d: %v", item.ID, err)
+				unhandled = append(unhandled, item)
+				continue
+			}
+			if !existed {
+				if err := indexer.Delete(ctx, item.ID); err != nil {
+					log.Error("Issue indexer handler: failed to delete issue %d from index: %v", item.ID, err)
+					unhandled = append(unhandled, item)
+				}
+				continue
+			}
+			if err := indexer.Index(ctx, data); err != nil {
+				log.Error("Issue indexer handler: failed to index issue %d: %v", item.ID, err)
+				unhandled = append(unhandled, item)
+				continue
+			}
+		}
+
+		return unhandled
+	}
+}
+
+// populateIssueIndexer populate the issue indexer with issue data
+func populateIssueIndexer(ctx context.Context) {
+	ctx, _, finished := process.GetManager().AddTypedContext(ctx, "Service: PopulateIssueIndexer", process.SystemProcessType, true)
+	defer finished()
+	ctx = contextWithKeepRetry(ctx) // keep retrying since it's a background task
+	if err := PopulateIssueIndexer(ctx); err != nil {
+		log.Error("Issue indexer population failed: %v", err)
+	}
+}
+
+func PopulateIssueIndexer(ctx context.Context) error {
+	for page := 1; ; page++ {
+		select {
+		case <-ctx.Done():
+			return fmt.Errorf("shutdown before completion: %w", ctx.Err())
+		default:
+		}
+		repos, _, err := repo_model.SearchRepositoryByName(ctx, repo_model.SearchRepoOptions{
+			ListOptions: db_model.ListOptions{Page: page, PageSize: repo_model.RepositoryListDefaultPageSize},
+			OrderBy:     db_model.SearchOrderByID,
+			Private:     true,
+			Collaborate: optional.Some(false),
+		})
+		if err != nil {
+			log.Error("SearchRepositoryByName: %v", err)
+			continue
+		}
+		if len(repos) == 0 {
+			log.Debug("Issue Indexer population complete")
+			return nil
+		}
+
+		for _, repo := range repos {
+			if err := updateRepoIndexer(ctx, repo.ID); err != nil {
+				return fmt.Errorf("populate issue indexer for repo %d: %v", repo.ID, err)
+			}
+		}
+	}
+}
+
+// UpdateRepoIndexer add/update all issues of the repositories
+func UpdateRepoIndexer(ctx context.Context, repoID int64) {
+	if err := updateRepoIndexer(ctx, repoID); err != nil {
+		log.Error("Unable to push repo %d to issue indexer: %v", repoID, err)
+	}
+}
+
+// UpdateIssueIndexer add/update an issue to the issue indexer
+func UpdateIssueIndexer(ctx context.Context, issueID int64) {
+	if err := updateIssueIndexer(ctx, issueID); err != nil {
+		log.Error("Unable to push issue %d to issue indexer: %v", issueID, err)
+	}
+}
+
+// DeleteRepoIssueIndexer deletes repo's all issues indexes
+func DeleteRepoIssueIndexer(ctx context.Context, repoID int64) {
+	if err := deleteRepoIssueIndexer(ctx, repoID); err != nil {
+		log.Error("Unable to push deleted repo %d to issue indexer: %v", repoID, err)
+	}
+}
+
+// IsAvailable checks if issue indexer is available
+func IsAvailable(ctx context.Context) bool {
+	return (*globalIndexer.Load()).Ping(ctx) == nil
+}
+
+// SearchOptions indicates the options for searching issues
+type SearchOptions = internal.SearchOptions
+
+const (
+	SortByCreatedDesc  = internal.SortByCreatedDesc
+	SortByUpdatedDesc  = internal.SortByUpdatedDesc
+	SortByCommentsDesc = internal.SortByCommentsDesc
+	SortByDeadlineDesc = internal.SortByDeadlineDesc
+	SortByCreatedAsc   = internal.SortByCreatedAsc
+	SortByUpdatedAsc   = internal.SortByUpdatedAsc
+	SortByCommentsAsc  = internal.SortByCommentsAsc
+	SortByDeadlineAsc  = internal.SortByDeadlineAsc
+)
+
+// SearchIssues search issues by options.
+func SearchIssues(ctx context.Context, opts *SearchOptions) ([]int64, int64, error) {
+	ix := *globalIndexer.Load()
+
+	if opts.Keyword == "" || opts.IsKeywordNumeric() {
+		// This is a conservative shortcut.
+		// If the keyword is empty or an integer, db has better (at least not worse) performance to filter issues.
+		// When the keyword is empty, it tends to listing rather than searching issues.
+		// So if the user creates an issue and list issues immediately, the issue may not be listed because the indexer needs time to index the issue.
+		// Even worse, the external indexer like elastic search may not be available for a while,
+		// and the user may not be able to list issues completely until it is available again.
+		ix = db.GetIndexer()
+	}
+
+	result, err := ix.Search(ctx, opts)
+	if err != nil {
+		return nil, 0, err
+	}
+	return SearchResultToIDSlice(result), result.Total, nil
+}
+
+func SearchResultToIDSlice(result *internal.SearchResult) []int64 {
+	ret := make([]int64, 0, len(result.Hits))
+	for _, hit := range result.Hits {
+		ret = append(ret, hit.ID)
+	}
+	return ret
+}
+
+// CountIssues counts issues by options. It is a shortcut of SearchIssues(ctx, opts) but only returns the total count.
+func CountIssues(ctx context.Context, opts *SearchOptions) (int64, error) {
+	opts = opts.Copy(func(options *SearchOptions) { options.Paginator = &db_model.ListOptions{PageSize: 0} })
+
+	_, total, err := SearchIssues(ctx, opts)
+	return total, err
+}
+
+func SupportedSearchModes() []indexer.SearchMode {
+	gi := globalIndexer.Load()
+	if gi == nil {
+		return nil
+	}
+	return (*gi).SupportedSearchModes()
+}

modules/indexer/issues/indexer_test.go

@@ -0,0 +1,487 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package issues
+
+import (
+	"testing"
+
+	"code.gitea.io/gitea/models/db"
+	"code.gitea.io/gitea/models/issues"
+	"code.gitea.io/gitea/models/unittest"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/setting"
+
+	_ "code.gitea.io/gitea/models"
+	_ "code.gitea.io/gitea/models/actions"
+	_ "code.gitea.io/gitea/models/activities"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestMain(m *testing.M) {
+	unittest.MainTest(m)
+}
+
+func TestDBSearchIssues(t *testing.T) {
+	require.NoError(t, unittest.PrepareTestDatabase())
+
+	setting.Indexer.IssueType = "db"
+	InitIssueIndexer(true)
+
+	t.Run("search issues with keyword", searchIssueWithKeyword)
+	t.Run("search issues by index", searchIssueByIndex)
+	t.Run("search issues in repo", searchIssueInRepo)
+	t.Run("search issues by ID", searchIssueByID)
+	t.Run("search issues is pr", searchIssueIsPull)
+	t.Run("search issues is closed", searchIssueIsClosed)
+	t.Run("search issues is archived", searchIssueIsArchived)
+	t.Run("search issues by milestone", searchIssueByMilestoneID)
+	t.Run("search issues by label", searchIssueByLabelID)
+	t.Run("search issues by time", searchIssueByTime)
+	t.Run("search issues with order", searchIssueWithOrder)
+	t.Run("search issues in project", searchIssueInProject)
+	t.Run("search issues with paginator", searchIssueWithPaginator)
+	t.Run("search issues with any assignee", searchIssueWithAnyAssignee)
+}
+
+func searchIssueWithKeyword(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				Keyword: "issue2",
+				RepoIDs: []int64{1},
+			},
+			[]int64{2},
+		},
+		{
+			SearchOptions{
+				Keyword: "first",
+				RepoIDs: []int64{1},
+			},
+			[]int64{1},
+		},
+		{
+			SearchOptions{
+				Keyword: "for",
+				RepoIDs: []int64{1},
+			},
+			[]int64{11, 5, 3, 2, 1},
+		},
+		{
+			SearchOptions{
+				Keyword: "good",
+				RepoIDs: []int64{1},
+			},
+			[]int64{1},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.opts.Keyword, func(t *testing.T) {
+			issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+			require.NoError(t, err)
+			assert.Equal(t, test.expectedIDs, issueIDs)
+		})
+	}
+}
+
+func searchIssueByIndex(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				Keyword: "1000",
+				RepoIDs: []int64{1},
+			},
+			[]int64{},
+		},
+		{
+			SearchOptions{
+				Keyword: "2",
+				RepoIDs: []int64{1, 2, 3, 32},
+			},
+			[]int64{17, 12, 7, 2},
+		},
+		{
+			SearchOptions{
+				Keyword: "1",
+				RepoIDs: []int64{58},
+			},
+			[]int64{19},
+		},
+	}
+
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueInRepo(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				RepoIDs: []int64{1},
+			},
+			[]int64{11, 5, 3, 2, 1},
+		},
+		{
+			SearchOptions{
+				RepoIDs: []int64{2},
+			},
+			[]int64{7, 4},
+		},
+		{
+			SearchOptions{
+				RepoIDs: []int64{3},
+			},
+			[]int64{12, 6},
+		},
+		{
+			SearchOptions{
+				RepoIDs: []int64{4},
+			},
+			[]int64{},
+		},
+		{
+			SearchOptions{
+				RepoIDs: []int64{5},
+			},
+			[]int64{15},
+		},
+	}
+
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueByID(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			opts: SearchOptions{
+				PosterID: "1",
+			},
+			expectedIDs: []int64{11, 6, 3, 2, 1},
+		},
+		{
+			opts: SearchOptions{
+				AssigneeID: "1",
+			},
+			expectedIDs: []int64{6, 1},
+		},
+		{
+			// NOTE: This tests no assignees filtering and also ToSearchOptions() to ensure it handles the filter correctly
+			opts:        *ToSearchOptions("", &issues.IssuesOptions{AssigneeID: "(none)"}),
+			expectedIDs: []int64{22, 21, 16, 15, 14, 13, 12, 11, 20, 5, 19, 18, 10, 7, 4, 9, 8, 3, 2},
+		},
+		{
+			opts: SearchOptions{
+				MentionID: optional.Some(int64(4)),
+			},
+			expectedIDs: []int64{1},
+		},
+		{
+			opts: SearchOptions{
+				ReviewedID: optional.Some(int64(1)),
+			},
+			expectedIDs: []int64{},
+		},
+		{
+			opts: SearchOptions{
+				ReviewRequestedID: optional.Some(int64(1)),
+			},
+			expectedIDs: []int64{12},
+		},
+		{
+			opts: SearchOptions{
+				SubscriberID: optional.Some(int64(1)),
+			},
+			expectedIDs: []int64{11, 6, 5, 3, 2, 1},
+		},
+		{
+			// issue 20 request user 15 and team 5 which user 15 belongs to
+			// the review request number of issue 20 should be 1
+			opts: SearchOptions{
+				ReviewRequestedID: optional.Some(int64(15)),
+			},
+			expectedIDs: []int64{12, 20},
+		},
+		{
+			// user 20 approved the issue 20, so return nothing
+			opts: SearchOptions{
+				ReviewRequestedID: optional.Some(int64(20)),
+			},
+			expectedIDs: []int64{},
+		},
+	}
+
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueIsPull(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				IsPull: optional.Some(false),
+			},
+			[]int64{17, 16, 15, 14, 13, 6, 5, 18, 10, 7, 4, 1},
+		},
+		{
+			SearchOptions{
+				IsPull: optional.Some(true),
+			},
+			[]int64{22, 21, 12, 11, 20, 19, 9, 8, 3, 2},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueIsClosed(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				IsClosed: optional.Some(false),
+			},
+			[]int64{22, 21, 17, 16, 15, 14, 13, 12, 11, 20, 6, 19, 18, 10, 7, 9, 8, 3, 2, 1},
+		},
+		{
+			SearchOptions{
+				IsClosed: optional.Some(true),
+			},
+			[]int64{5, 4},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueIsArchived(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				IsArchived: optional.Some(false),
+			},
+			[]int64{22, 21, 17, 16, 15, 13, 12, 11, 20, 6, 5, 19, 18, 10, 7, 4, 9, 8, 3, 2, 1},
+		},
+		{
+			SearchOptions{
+				IsArchived: optional.Some(true),
+			},
+			[]int64{14},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueByMilestoneID(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				MilestoneIDs: []int64{1},
+			},
+			[]int64{2},
+		},
+		{
+			SearchOptions{
+				MilestoneIDs: []int64{3},
+			},
+			[]int64{3},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueByLabelID(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				IncludedLabelIDs: []int64{1},
+			},
+			[]int64{2, 1},
+		},
+		{
+			SearchOptions{
+				IncludedLabelIDs: []int64{4},
+			},
+			[]int64{2},
+		},
+		{
+			SearchOptions{
+				ExcludedLabelIDs: []int64{1},
+			},
+			[]int64{22, 21, 17, 16, 15, 14, 13, 12, 11, 20, 6, 5, 19, 18, 10, 7, 4, 9, 8, 3},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueByTime(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				UpdatedAfterUnix: optional.Some(int64(0)),
+			},
+			[]int64{22, 21, 17, 16, 15, 14, 13, 12, 11, 20, 6, 5, 19, 18, 10, 7, 4, 9, 8, 3, 2, 1},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueWithOrder(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				SortBy: internal.SortByCreatedAsc,
+			},
+			[]int64{1, 2, 3, 8, 9, 4, 7, 10, 18, 19, 5, 6, 20, 11, 12, 13, 14, 15, 16, 17, 21, 22},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueInProject(t *testing.T) {
+	tests := []struct {
+		opts        SearchOptions
+		expectedIDs []int64
+	}{
+		{
+			SearchOptions{
+				ProjectID: optional.Some(int64(1)),
+			},
+			[]int64{5, 3, 2, 1},
+		},
+		{
+			SearchOptions{
+				ProjectColumnID: optional.Some(int64(1)),
+			},
+			[]int64{1},
+		},
+		{
+			SearchOptions{
+				ProjectColumnID: optional.Some(int64(0)), // issue with in default column
+			},
+			[]int64{2},
+		},
+	}
+	for _, test := range tests {
+		issueIDs, _, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+	}
+}
+
+func searchIssueWithPaginator(t *testing.T) {
+	tests := []struct {
+		opts          SearchOptions
+		expectedIDs   []int64
+		expectedTotal int64
+	}{
+		{
+			SearchOptions{
+				Paginator: &db.ListOptions{
+					PageSize: 5,
+				},
+			},
+			[]int64{22, 21, 17, 16, 15},
+			22,
+		},
+	}
+	for _, test := range tests {
+		issueIDs, total, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+		assert.Equal(t, test.expectedTotal, total)
+	}
+}
+
+func searchIssueWithAnyAssignee(t *testing.T) {
+	tests := []struct {
+		opts          SearchOptions
+		expectedIDs   []int64
+		expectedTotal int64
+	}{
+		{
+			SearchOptions{
+				AssigneeID: "(any)",
+			},
+			[]int64{17, 6, 1},
+			3,
+		},
+	}
+	for _, test := range tests {
+		issueIDs, total, err := SearchIssues(t.Context(), &test.opts)
+		require.NoError(t, err)
+		assert.Equal(t, test.expectedIDs, issueIDs)
+		assert.Equal(t, test.expectedTotal, total)
+	}
+}

modules/indexer/issues/internal/indexer.go

@@ -0,0 +1,48 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package internal
+
+import (
+	"context"
+	"errors"
+
+	"code.gitea.io/gitea/modules/indexer"
+	"code.gitea.io/gitea/modules/indexer/internal"
+)
+
+// Indexer defines an interface to indexer issues contents
+type Indexer interface {
+	internal.Indexer
+	Index(ctx context.Context, issue ...*IndexerData) error
+	Delete(ctx context.Context, ids ...int64) error
+	Search(ctx context.Context, options *SearchOptions) (*SearchResult, error)
+	SupportedSearchModes() []indexer.SearchMode
+}
+
+// NewDummyIndexer returns a dummy indexer
+func NewDummyIndexer() Indexer {
+	return &dummyIndexer{
+		Indexer: internal.NewDummyIndexer(),
+	}
+}
+
+type dummyIndexer struct {
+	internal.Indexer
+}
+
+func (d *dummyIndexer) SupportedSearchModes() []indexer.SearchMode {
+	return nil
+}
+
+func (d *dummyIndexer) Index(_ context.Context, _ ...*IndexerData) error {
+	return errors.New("indexer is not ready")
+}
+
+func (d *dummyIndexer) Delete(_ context.Context, _ ...int64) error {
+	return errors.New("indexer is not ready")
+}
+
+func (d *dummyIndexer) Search(_ context.Context, _ *SearchOptions) (*SearchResult, error) {
+	return nil, errors.New("indexer is not ready")
+}

modules/indexer/issues/internal/model.go

@@ -0,0 +1,160 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package internal
+
+import (
+	"strconv"
+
+	"code.gitea.io/gitea/models/db"
+	"code.gitea.io/gitea/modules/indexer"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/timeutil"
+)
+
+// IndexerData data stored in the issue indexer
+type IndexerData struct {
+	ID       int64 `json:"id"`
+	RepoID   int64 `json:"repo_id"`
+	IsPublic bool  `json:"is_public"` // If the repo is public
+
+	// Fields used for keyword searching
+	Title    string   `json:"title"`
+	Content  string   `json:"content"`
+	Comments []string `json:"comments"`
+
+	// Fields used for filtering
+	IsPull             bool               `json:"is_pull"`
+	IsClosed           bool               `json:"is_closed"`
+	IsArchived         bool               `json:"is_archived"`
+	LabelIDs           []int64            `json:"label_ids"`
+	NoLabel            bool               `json:"no_label"` // True if LabelIDs is empty
+	MilestoneID        int64              `json:"milestone_id"`
+	ProjectID          int64              `json:"project_id"`
+	ProjectColumnID    int64              `json:"project_board_id"` // the key should be kept as project_board_id to keep compatible
+	PosterID           int64              `json:"poster_id"`
+	AssigneeID         int64              `json:"assignee_id"`
+	MentionIDs         []int64            `json:"mention_ids"`
+	ReviewedIDs        []int64            `json:"reviewed_ids"`
+	ReviewRequestedIDs []int64            `json:"review_requested_ids"`
+	SubscriberIDs      []int64            `json:"subscriber_ids"`
+	UpdatedUnix        timeutil.TimeStamp `json:"updated_unix"`
+
+	// Fields used for sorting
+	// UpdatedUnix is both used for filtering and sorting.
+	// ID is used for sorting too, to make the sorting stable.
+	CreatedUnix  timeutil.TimeStamp `json:"created_unix"`
+	DeadlineUnix timeutil.TimeStamp `json:"deadline_unix"`
+	CommentCount int64              `json:"comment_count"`
+}
+
+// Match represents on search result
+type Match struct {
+	ID    int64   `json:"id"`
+	Score float64 `json:"score"`
+}
+
+// SearchResult represents search results
+type SearchResult struct {
+	Total int64
+	Hits  []Match
+}
+
+// SearchOptions represents search options.
+//
+// It has a slightly different design from database query options.
+// In database query options, a field is never a pointer, so it could be confusing when it's zero value:
+// Do you want to find data with a field value of 0, or do you not specify the field in the options?
+// To avoid this confusion, db introduced db.NoConditionID(-1).
+// So zero value means the field is not specified in the search options, and db.NoConditionID means "== 0" or "id NOT IN (SELECT id FROM ...)"
+// It's still not ideal, it trapped developers many times.
+// And sometimes -1 could be a valid value, like issue ID, negative numbers indicate exclusion.
+// Since db.NoConditionID is for "db" (the package name is db), it makes sense not to use it in the indexer:
+// Why do bleve/elasticsearch/meilisearch indexers need to know about db.NoConditionID?
+// So in SearchOptions, we use pointer for fields which could be not specified,
+// and always use the value to filter if it's not nil, even if it's zero or negative.
+// It can handle almost all cases, if there is an exception, we can add a new field, like NoLabelOnly.
+// Unfortunately, we still use db for the indexer and have to convert between db.NoConditionID and nil for legacy reasons.
+type SearchOptions struct {
+	Keyword string // keyword to search
+
+	SearchMode indexer.SearchModeType
+
+	RepoIDs   []int64 // repository IDs which the issues belong to
+	AllPublic bool    // if include all public repositories
+
+	IsPull     optional.Option[bool] // if the issues is a pull request
+	IsClosed   optional.Option[bool] // if the issues is closed
+	IsArchived optional.Option[bool] // if the repo is archived
+
+	IncludedLabelIDs    []int64 // labels the issues have
+	ExcludedLabelIDs    []int64 // labels the issues don't have
+	IncludedAnyLabelIDs []int64 // labels the issues have at least one. It will be ignored if IncludedLabelIDs is not empty. It's an uncommon filter, but it has been supported accidentally by issues.IssuesOptions.IncludedLabelNames.
+	NoLabelOnly         bool    // if the issues have no label, if true, IncludedLabelIDs and ExcludedLabelIDs, IncludedAnyLabelIDs will be ignored
+
+	MilestoneIDs []int64 // milestones the issues have
+
+	ProjectID       optional.Option[int64] // project the issues belong to
+	ProjectColumnID optional.Option[int64] // project column the issues belong to
+
+	PosterID   string // poster of the issues, "(none)" or "(any)" or a user ID
+	AssigneeID string // assignee of the issues, "(none)" or "(any)" or a user ID
+
+	MentionID optional.Option[int64] // mentioned user of the issues
+
+	ReviewedID        optional.Option[int64] // reviewer of the issues
+	ReviewRequestedID optional.Option[int64] // requested reviewer of the issues
+
+	SubscriberID optional.Option[int64] // subscriber of the issues
+
+	UpdatedAfterUnix  optional.Option[int64]
+	UpdatedBeforeUnix optional.Option[int64]
+
+	Paginator *db.ListOptions
+
+	SortBy SortBy // sort by field
+}
+
+// Copy returns a copy of the options.
+// Be careful, it's not a deep copy, so `SearchOptions.RepoIDs = {...}` is OK while `SearchOptions.RepoIDs[0] = ...` is not.
+func (o *SearchOptions) Copy(edit ...func(options *SearchOptions)) *SearchOptions {
+	if o == nil {
+		return nil
+	}
+	v := *o
+	for _, e := range edit {
+		e(&v)
+	}
+	return &v
+}
+
+// used for optimized issue index based search
+func (o *SearchOptions) IsKeywordNumeric() bool {
+	_, err := strconv.Atoi(o.Keyword)
+	return err == nil
+}
+
+type SortBy string
+
+const (
+	SortByCreatedDesc  SortBy = "-created_unix"
+	SortByUpdatedDesc  SortBy = "-updated_unix"
+	SortByCommentsDesc SortBy = "-comment_count"
+	SortByDeadlineDesc SortBy = "-deadline_unix"
+	SortByCreatedAsc   SortBy = "created_unix"
+	SortByUpdatedAsc   SortBy = "updated_unix"
+	SortByCommentsAsc  SortBy = "comment_count"
+	SortByDeadlineAsc  SortBy = "deadline_unix"
+	// Unsupported sort types which are supported by issues.IssuesOptions.SortType:
+	//
+	//  - "priorityrepo":
+	//                    It's impossible to support it in the indexer.
+	//                    It is based on the specified repository in the request, so we cannot add static field to the indexer.
+	//                    If we do something like that query the issues in the specified repository first then append other issues,
+	//                    it will break the pagination.
+	//
+	// - "project-column-sorting":
+	//                    Although it's possible to support it by adding project.ProjectIssue.Sorting to the indexer,
+	//                    but what if the issue belongs to multiple projects?
+	//                    Since it's unsupported to search issues with keyword in project page, we don't need to support it.
+)

modules/indexer/issues/internal/tests/tests.go

@@ -0,0 +1,759 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+// This package contains tests for issues indexer modules.
+// All the code in this package is only used for testing.
+// Do not put any production code in this package to avoid it being included in the final binary.
+
+package tests
+
+import (
+	"fmt"
+	"slices"
+	"testing"
+	"time"
+
+	"code.gitea.io/gitea/models/db"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/optional"
+	"code.gitea.io/gitea/modules/timeutil"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestIndexer(t *testing.T, indexer internal.Indexer) {
+	_, err := indexer.Init(t.Context())
+	require.NoError(t, err)
+
+	require.NoError(t, indexer.Ping(t.Context()))
+
+	var (
+		ids  []int64
+		data = map[int64]*internal.IndexerData{}
+	)
+	{
+		d := generateDefaultIndexerData()
+		for _, v := range d {
+			ids = append(ids, v.ID)
+			data[v.ID] = v
+		}
+		require.NoError(t, indexer.Index(t.Context(), d...))
+		waitData(t, indexer, int64(len(data)))
+	}
+
+	defer func() {
+		require.NoError(t, indexer.Delete(t.Context(), ids...))
+	}()
+
+	for _, c := range cases {
+		t.Run(c.Name, func(t *testing.T) {
+			if len(c.ExtraData) > 0 {
+				require.NoError(t, indexer.Index(t.Context(), c.ExtraData...))
+				for _, v := range c.ExtraData {
+					data[v.ID] = v
+				}
+				waitData(t, indexer, int64(len(data)))
+				defer func() {
+					for _, v := range c.ExtraData {
+						require.NoError(t, indexer.Delete(t.Context(), v.ID))
+						delete(data, v.ID)
+					}
+					waitData(t, indexer, int64(len(data)))
+				}()
+			}
+
+			result, err := indexer.Search(t.Context(), c.SearchOptions)
+			require.NoError(t, err)
+
+			if c.Expected != nil {
+				c.Expected(t, data, result)
+			} else {
+				ids := make([]int64, 0, len(result.Hits))
+				for _, hit := range result.Hits {
+					ids = append(ids, hit.ID)
+				}
+				assert.Equal(t, c.ExpectedIDs, ids)
+				assert.Equal(t, c.ExpectedTotal, result.Total)
+			}
+
+			// test counting
+			c.SearchOptions.Paginator = &db.ListOptions{PageSize: 0}
+			countResult, err := indexer.Search(t.Context(), c.SearchOptions)
+			require.NoError(t, err)
+			assert.Empty(t, countResult.Hits)
+			assert.Equal(t, result.Total, countResult.Total)
+		})
+	}
+}
+
+var cases = []*testIndexerCase{
+	{
+		Name:          "default",
+		SearchOptions: &internal.SearchOptions{},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+		},
+	},
+	{
+		Name: "empty",
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "f1dfac73-fda6-4a6b-b8a4-2408fcb8ef69",
+		},
+		ExpectedIDs:   []int64{},
+		ExpectedTotal: 0,
+	},
+	{
+		Name: "with limit",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			assert.Equal(t, len(data), int(result.Total))
+		},
+	},
+	{
+		Name: "Keyword",
+		ExtraData: []*internal.IndexerData{
+			{ID: 1000, Title: "hi hello world"},
+			{ID: 1001, Content: "hi hello world"},
+			{ID: 1002, Comments: []string{"hi", "hello world"}},
+		},
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "hello",
+		},
+		ExpectedIDs:   []int64{1002, 1001, 1000},
+		ExpectedTotal: 3,
+	},
+	{
+		Name: "RepoIDs",
+		ExtraData: []*internal.IndexerData{
+			{ID: 1001, Title: "hello world", RepoID: 1, IsPublic: false},
+			{ID: 1002, Title: "hello world", RepoID: 1, IsPublic: false},
+			{ID: 1003, Title: "hello world", RepoID: 2, IsPublic: true},
+			{ID: 1004, Title: "hello world", RepoID: 2, IsPublic: true},
+			{ID: 1005, Title: "hello world", RepoID: 3, IsPublic: true},
+			{ID: 1006, Title: "hello world", RepoID: 4, IsPublic: false},
+			{ID: 1007, Title: "hello world", RepoID: 5, IsPublic: false},
+		},
+		SearchOptions: &internal.SearchOptions{
+			Keyword: "hello",
+			RepoIDs: []int64{1, 4},
+		},
+		ExpectedIDs:   []int64{1006, 1002, 1001},
+		ExpectedTotal: 3,
+	},
+	{
+		Name: "RepoIDs and AllPublic",
+		ExtraData: []*internal.IndexerData{
+			{ID: 1001, Title: "hello world", RepoID: 1, IsPublic: false},
+			{ID: 1002, Title: "hello world", RepoID: 1, IsPublic: false},
+			{ID: 1003, Title: "hello world", RepoID: 2, IsPublic: true},
+			{ID: 1004, Title: "hello world", RepoID: 2, IsPublic: true},
+			{ID: 1005, Title: "hello world", RepoID: 3, IsPublic: true},
+			{ID: 1006, Title: "hello world", RepoID: 4, IsPublic: false},
+			{ID: 1007, Title: "hello world", RepoID: 5, IsPublic: false},
+		},
+		SearchOptions: &internal.SearchOptions{
+			Keyword:   "hello",
+			RepoIDs:   []int64{1, 4},
+			AllPublic: true,
+		},
+		ExpectedIDs:   []int64{1006, 1005, 1004, 1003, 1002, 1001},
+		ExpectedTotal: 6,
+	},
+	{
+		Name: "issue only",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			IsPull: optional.Some(false),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.False(t, data[v.ID].IsPull)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool { return !v.IsPull }), result.Total)
+		},
+	},
+	{
+		Name: "pull only",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			IsPull: optional.Some(true),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.True(t, data[v.ID].IsPull)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool { return v.IsPull }), result.Total)
+		},
+	},
+	{
+		Name: "opened only",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			IsClosed: optional.Some(false),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.False(t, data[v.ID].IsClosed)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool { return !v.IsClosed }), result.Total)
+		},
+	},
+	{
+		Name: "closed only",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			IsClosed: optional.Some(true),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.True(t, data[v.ID].IsClosed)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool { return v.IsClosed }), result.Total)
+		},
+	},
+	{
+		Name: "labels",
+		ExtraData: []*internal.IndexerData{
+			{ID: 1000, Title: "hello a", LabelIDs: []int64{2000, 2001, 2002}},
+			{ID: 1001, Title: "hello b", LabelIDs: []int64{2000, 2001}},
+			{ID: 1002, Title: "hello c", LabelIDs: []int64{2000, 2001, 2003}},
+			{ID: 1003, Title: "hello d", LabelIDs: []int64{2000}},
+			{ID: 1004, Title: "hello e", LabelIDs: []int64{}},
+		},
+		SearchOptions: &internal.SearchOptions{
+			Keyword:          "hello",
+			IncludedLabelIDs: []int64{2000, 2001},
+			ExcludedLabelIDs: []int64{2003},
+		},
+		ExpectedIDs:   []int64{1001, 1000},
+		ExpectedTotal: 2,
+	},
+	{
+		Name: "include any labels",
+		ExtraData: []*internal.IndexerData{
+			{ID: 1000, Title: "hello a", LabelIDs: []int64{2000, 2001, 2002}},
+			{ID: 1001, Title: "hello b", LabelIDs: []int64{2001}},
+			{ID: 1002, Title: "hello c", LabelIDs: []int64{2000, 2001, 2003}},
+			{ID: 1003, Title: "hello d", LabelIDs: []int64{2002}},
+			{ID: 1004, Title: "hello e", LabelIDs: []int64{}},
+		},
+		SearchOptions: &internal.SearchOptions{
+			Keyword:             "hello",
+			IncludedAnyLabelIDs: []int64{2001, 2002},
+			ExcludedLabelIDs:    []int64{2003},
+		},
+		ExpectedIDs:   []int64{1003, 1001, 1000},
+		ExpectedTotal: 3,
+	},
+	{
+		Name: "MilestoneIDs",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			MilestoneIDs: []int64{1, 2, 6},
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Contains(t, []int64{1, 2, 6}, data[v.ID].MilestoneID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.MilestoneID == 1 || v.MilestoneID == 2 || v.MilestoneID == 6
+			}), result.Total)
+		},
+	},
+	{
+		Name: "no MilestoneIDs",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			MilestoneIDs: []int64{0},
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(0), data[v.ID].MilestoneID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.MilestoneID == 0
+			}), result.Total)
+		},
+	},
+	{
+		Name: "ProjectID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ProjectID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(1), data[v.ID].ProjectID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.ProjectID == 1
+			}), result.Total)
+		},
+	},
+	{
+		Name: "no ProjectID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ProjectID: optional.Some(int64(0)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(0), data[v.ID].ProjectID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.ProjectID == 0
+			}), result.Total)
+		},
+	},
+	{
+		Name: "ProjectColumnID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ProjectColumnID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(1), data[v.ID].ProjectColumnID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.ProjectColumnID == 1
+			}), result.Total)
+		},
+	},
+	{
+		Name: "no ProjectColumnID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ProjectColumnID: optional.Some(int64(0)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(0), data[v.ID].ProjectColumnID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.ProjectColumnID == 0
+			}), result.Total)
+		},
+	},
+	{
+		Name: "PosterID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			PosterID: "1",
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(1), data[v.ID].PosterID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.PosterID == 1
+			}), result.Total)
+		},
+	},
+	{
+		Name: "AssigneeID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			AssigneeID: "1",
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(1), data[v.ID].AssigneeID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.AssigneeID == 1
+			}), result.Total)
+		},
+	},
+	{
+		Name: "no AssigneeID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			AssigneeID: "(none)",
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Equal(t, int64(0), data[v.ID].AssigneeID)
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.AssigneeID == 0
+			}), result.Total)
+		},
+	},
+	{
+		Name: "MentionID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			MentionID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Contains(t, data[v.ID].MentionIDs, int64(1))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return slices.Contains(v.MentionIDs, 1)
+			}), result.Total)
+		},
+	},
+	{
+		Name: "ReviewedID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ReviewedID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Contains(t, data[v.ID].ReviewedIDs, int64(1))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return slices.Contains(v.ReviewedIDs, 1)
+			}), result.Total)
+		},
+	},
+	{
+		Name: "ReviewRequestedID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			ReviewRequestedID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Contains(t, data[v.ID].ReviewRequestedIDs, int64(1))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return slices.Contains(v.ReviewRequestedIDs, 1)
+			}), result.Total)
+		},
+	},
+	{
+		Name: "SubscriberID",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			SubscriberID: optional.Some(int64(1)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.Contains(t, data[v.ID].SubscriberIDs, int64(1))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return slices.Contains(v.SubscriberIDs, 1)
+			}), result.Total)
+		},
+	},
+	{
+		Name: "updated",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptions{
+				PageSize: 5,
+			},
+			UpdatedAfterUnix:  optional.Some(int64(20)),
+			UpdatedBeforeUnix: optional.Some(int64(30)),
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 5)
+			for _, v := range result.Hits {
+				assert.GreaterOrEqual(t, data[v.ID].UpdatedUnix, int64(20))
+				assert.LessOrEqual(t, data[v.ID].UpdatedUnix, int64(30))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return data[v.ID].UpdatedUnix >= 20 && data[v.ID].UpdatedUnix <= 30
+			}), result.Total)
+		},
+	},
+	{
+		Name: "SortByCreatedDesc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByCreatedDesc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.GreaterOrEqual(t, data[v.ID].CreatedUnix, data[result.Hits[i+1].ID].CreatedUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByUpdatedDesc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByUpdatedDesc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.GreaterOrEqual(t, data[v.ID].UpdatedUnix, data[result.Hits[i+1].ID].UpdatedUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByCommentsDesc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByCommentsDesc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.GreaterOrEqual(t, data[v.ID].CommentCount, data[result.Hits[i+1].ID].CommentCount)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByDeadlineDesc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByDeadlineDesc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.GreaterOrEqual(t, data[v.ID].DeadlineUnix, data[result.Hits[i+1].ID].DeadlineUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByCreatedAsc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByCreatedAsc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.LessOrEqual(t, data[v.ID].CreatedUnix, data[result.Hits[i+1].ID].CreatedUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByUpdatedAsc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByUpdatedAsc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.LessOrEqual(t, data[v.ID].UpdatedUnix, data[result.Hits[i+1].ID].UpdatedUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByCommentsAsc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByCommentsAsc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.LessOrEqual(t, data[v.ID].CommentCount, data[result.Hits[i+1].ID].CommentCount)
+				}
+			}
+		},
+	},
+	{
+		Name: "SortByDeadlineAsc",
+		SearchOptions: &internal.SearchOptions{
+			Paginator: &db.ListOptionsAll,
+			SortBy:    internal.SortByDeadlineAsc,
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, len(data))
+			assert.Equal(t, len(data), int(result.Total))
+			for i, v := range result.Hits {
+				if i < len(result.Hits)-1 {
+					assert.LessOrEqual(t, data[v.ID].DeadlineUnix, data[result.Hits[i+1].ID].DeadlineUnix)
+				}
+			}
+		},
+	},
+	{
+		Name: "SearchAnyAssignee",
+		SearchOptions: &internal.SearchOptions{
+			AssigneeID: "(any)",
+		},
+		Expected: func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) {
+			assert.Len(t, result.Hits, 180)
+			for _, v := range result.Hits {
+				assert.GreaterOrEqual(t, data[v.ID].AssigneeID, int64(1))
+			}
+			assert.Equal(t, countIndexerData(data, func(v *internal.IndexerData) bool {
+				return v.AssigneeID >= 1
+			}), result.Total)
+		},
+	},
+}
+
+type testIndexerCase struct {
+	Name      string
+	ExtraData []*internal.IndexerData
+
+	SearchOptions *internal.SearchOptions
+
+	Expected      func(t *testing.T, data map[int64]*internal.IndexerData, result *internal.SearchResult) // if nil, use ExpectedIDs, ExpectedTotal
+	ExpectedIDs   []int64
+	ExpectedTotal int64
+}
+
+func generateDefaultIndexerData() []*internal.IndexerData {
+	var id int64
+	var data []*internal.IndexerData
+	for repoID := int64(1); repoID <= 10; repoID++ {
+		for issueIndex := int64(1); issueIndex <= 20; issueIndex++ {
+			id++
+
+			comments := make([]string, id%4)
+			for i := range comments {
+				comments[i] = fmt.Sprintf("comment%d", i)
+			}
+
+			labelIDs := make([]int64, id%5)
+			for i := range labelIDs {
+				labelIDs[i] = int64(i) + 1 // LabelID should not be 0
+			}
+			mentionIDs := make([]int64, id%6)
+			for i := range mentionIDs {
+				mentionIDs[i] = int64(i) + 1 // MentionID should not be 0
+			}
+			reviewedIDs := make([]int64, id%7)
+			for i := range reviewedIDs {
+				reviewedIDs[i] = int64(i) + 1 // ReviewID should not be 0
+			}
+			reviewRequestedIDs := make([]int64, id%8)
+			for i := range reviewRequestedIDs {
+				reviewRequestedIDs[i] = int64(i) + 1 // ReviewRequestedID should not be 0
+			}
+			subscriberIDs := make([]int64, id%9)
+			for i := range subscriberIDs {
+				subscriberIDs[i] = int64(i) + 1 // SubscriberID should not be 0
+			}
+
+			data = append(data, &internal.IndexerData{
+				ID:                 id,
+				RepoID:             repoID,
+				IsPublic:           repoID%2 == 0,
+				Title:              fmt.Sprintf("issue%d of repo%d", issueIndex, repoID),
+				Content:            fmt.Sprintf("content%d", issueIndex),
+				Comments:           comments,
+				IsPull:             issueIndex%2 == 0,
+				IsClosed:           issueIndex%3 == 0,
+				LabelIDs:           labelIDs,
+				NoLabel:            len(labelIDs) == 0,
+				MilestoneID:        issueIndex % 4,
+				ProjectID:          issueIndex % 5,
+				ProjectColumnID:    issueIndex % 6,
+				PosterID:           id%10 + 1, // PosterID should not be 0
+				AssigneeID:         issueIndex % 10,
+				MentionIDs:         mentionIDs,
+				ReviewedIDs:        reviewedIDs,
+				ReviewRequestedIDs: reviewRequestedIDs,
+				SubscriberIDs:      subscriberIDs,
+				UpdatedUnix:        timeutil.TimeStamp(id + issueIndex),
+				CreatedUnix:        timeutil.TimeStamp(id),
+				DeadlineUnix:       timeutil.TimeStamp(id + issueIndex + repoID),
+				CommentCount:       int64(len(comments)),
+			})
+		}
+	}
+
+	return data
+}
+
+func countIndexerData(data map[int64]*internal.IndexerData, f func(v *internal.IndexerData) bool) int64 {
+	var count int64
+	for _, v := range data {
+		if f(v) {
+			count++
+		}
+	}
+	return count
+}
+
+// waitData waits for the indexer to index all data.
+// Some engines like Elasticsearch index data asynchronously, so we need to wait for a while.
+func waitData(t *testing.T, indexer internal.Indexer, total int64) {
+	assert.Eventually(t, func() bool {
+		result, err := indexer.Search(t.Context(), &internal.SearchOptions{Paginator: &db.ListOptions{}})
+		require.NoError(t, err)
+		return result.Total == total
+	}, 10*time.Second, 100*time.Millisecond, "expected total=%d", total)
+}

modules/indexer/issues/meilisearch/meilisearch.go

@@ -0,0 +1,317 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package meilisearch
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"strconv"
+	"strings"
+
+	"code.gitea.io/gitea/modules/indexer"
+	indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
+	inner_meilisearch "code.gitea.io/gitea/modules/indexer/internal/meilisearch"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+
+	"github.com/meilisearch/meilisearch-go"
+)
+
+const (
+	issueIndexerLatestVersion = 4
+
+	// TODO: make this configurable if necessary
+	maxTotalHits = 10000
+)
+
+// ErrMalformedResponse is never expected as we initialize the indexer ourself and so define the types.
+var ErrMalformedResponse = errors.New("meilisearch returned unexpected malformed content")
+
+var _ internal.Indexer = &Indexer{}
+
+// Indexer implements Indexer interface
+type Indexer struct {
+	inner                    *inner_meilisearch.Indexer
+	indexer_internal.Indexer // do not composite inner_meilisearch.Indexer directly to avoid exposing too much
+}
+
+func (b *Indexer) SupportedSearchModes() []indexer.SearchMode {
+	return indexer.SearchModesExactWords()
+}
+
+// NewIndexer creates a new meilisearch indexer
+func NewIndexer(url, apiKey, indexerName string) *Indexer {
+	settings := &meilisearch.Settings{
+		// The default ranking rules of meilisearch are: ["words", "typo", "proximity", "attribute", "sort", "exactness"]
+		// So even if we specify the sort order, it could not be respected because the priority of "sort" is so low.
+		// So we need to specify the ranking rules to make sure the sort order is respected.
+		// See https://www.meilisearch.com/docs/learn/core_concepts/relevancy
+		RankingRules: []string{"sort", // make sure "sort" has the highest priority
+			"words", "typo", "proximity", "attribute", "exactness"},
+
+		SearchableAttributes: []string{
+			"title",
+			"content",
+			"comments",
+		},
+		DisplayedAttributes: []string{
+			"id",
+			"title",
+			"content",
+			"comments",
+		},
+		FilterableAttributes: []string{
+			"repo_id",
+			"is_public",
+			"is_pull",
+			"is_closed",
+			"is_archived",
+			"label_ids",
+			"no_label",
+			"milestone_id",
+			"project_id",
+			"project_board_id",
+			"poster_id",
+			"assignee_id",
+			"mention_ids",
+			"reviewed_ids",
+			"review_requested_ids",
+			"subscriber_ids",
+			"updated_unix",
+		},
+		SortableAttributes: []string{
+			"updated_unix",
+			"created_unix",
+			"deadline_unix",
+			"comment_count",
+			"id",
+		},
+		Pagination: &meilisearch.Pagination{
+			MaxTotalHits: maxTotalHits,
+		},
+	}
+
+	inner := inner_meilisearch.NewIndexer(url, apiKey, indexerName, issueIndexerLatestVersion, settings)
+	indexer := &Indexer{
+		inner:   inner,
+		Indexer: inner,
+	}
+	return indexer
+}
+
+// Index will save the index data
+func (b *Indexer) Index(_ context.Context, issues ...*internal.IndexerData) error {
+	if len(issues) == 0 {
+		return nil
+	}
+	for _, issue := range issues {
+		_, err := b.inner.Client.Index(b.inner.VersionedIndexName()).AddDocuments(issue)
+		if err != nil {
+			return err
+		}
+	}
+	// TODO: bulk send index data
+	return nil
+}
+
+// Delete deletes indexes by ids
+func (b *Indexer) Delete(_ context.Context, ids ...int64) error {
+	if len(ids) == 0 {
+		return nil
+	}
+
+	for _, id := range ids {
+		_, err := b.inner.Client.Index(b.inner.VersionedIndexName()).DeleteDocument(strconv.FormatInt(id, 10))
+		if err != nil {
+			return err
+		}
+	}
+	// TODO: bulk send deletes
+	return nil
+}
+
+// Search searches for issues by given conditions.
+// Returns the matching issue IDs
+func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
+	query := inner_meilisearch.FilterAnd{}
+
+	if len(options.RepoIDs) > 0 {
+		q := &inner_meilisearch.FilterOr{}
+		q.Or(inner_meilisearch.NewFilterIn("repo_id", options.RepoIDs...))
+		if options.AllPublic {
+			q.Or(inner_meilisearch.NewFilterEq("is_public", true))
+		}
+		query.And(q)
+	}
+
+	if options.IsPull.Has() {
+		query.And(inner_meilisearch.NewFilterEq("is_pull", options.IsPull.Value()))
+	}
+	if options.IsClosed.Has() {
+		query.And(inner_meilisearch.NewFilterEq("is_closed", options.IsClosed.Value()))
+	}
+	if options.IsArchived.Has() {
+		query.And(inner_meilisearch.NewFilterEq("is_archived", options.IsArchived.Value()))
+	}
+
+	if options.NoLabelOnly {
+		query.And(inner_meilisearch.NewFilterEq("no_label", true))
+	} else {
+		if len(options.IncludedLabelIDs) > 0 {
+			q := &inner_meilisearch.FilterAnd{}
+			for _, labelID := range options.IncludedLabelIDs {
+				q.And(inner_meilisearch.NewFilterEq("label_ids", labelID))
+			}
+			query.And(q)
+		} else if len(options.IncludedAnyLabelIDs) > 0 {
+			query.And(inner_meilisearch.NewFilterIn("label_ids", options.IncludedAnyLabelIDs...))
+		}
+		if len(options.ExcludedLabelIDs) > 0 {
+			q := &inner_meilisearch.FilterAnd{}
+			for _, labelID := range options.ExcludedLabelIDs {
+				q.And(inner_meilisearch.NewFilterNot(inner_meilisearch.NewFilterEq("label_ids", labelID)))
+			}
+			query.And(q)
+		}
+	}
+
+	if len(options.MilestoneIDs) > 0 {
+		query.And(inner_meilisearch.NewFilterIn("milestone_id", options.MilestoneIDs...))
+	}
+
+	if options.ProjectID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("project_id", options.ProjectID.Value()))
+	}
+	if options.ProjectColumnID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("project_board_id", options.ProjectColumnID.Value()))
+	}
+
+	if options.PosterID != "" {
+		// "(none)" becomes 0, it means no poster
+		posterIDInt64, _ := strconv.ParseInt(options.PosterID, 10, 64)
+		query.And(inner_meilisearch.NewFilterEq("poster_id", posterIDInt64))
+	}
+
+	if options.AssigneeID != "" {
+		if options.AssigneeID == "(any)" {
+			query.And(inner_meilisearch.NewFilterGte("assignee_id", 1))
+		} else {
+			// "(none)" becomes 0, it means no assignee
+			assigneeIDInt64, _ := strconv.ParseInt(options.AssigneeID, 10, 64)
+			query.And(inner_meilisearch.NewFilterEq("assignee_id", assigneeIDInt64))
+		}
+	}
+
+	if options.MentionID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("mention_ids", options.MentionID.Value()))
+	}
+
+	if options.ReviewedID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("reviewed_ids", options.ReviewedID.Value()))
+	}
+	if options.ReviewRequestedID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("review_requested_ids", options.ReviewRequestedID.Value()))
+	}
+
+	if options.SubscriberID.Has() {
+		query.And(inner_meilisearch.NewFilterEq("subscriber_ids", options.SubscriberID.Value()))
+	}
+
+	if options.UpdatedAfterUnix.Has() {
+		query.And(inner_meilisearch.NewFilterGte("updated_unix", options.UpdatedAfterUnix.Value()))
+	}
+	if options.UpdatedBeforeUnix.Has() {
+		query.And(inner_meilisearch.NewFilterLte("updated_unix", options.UpdatedBeforeUnix.Value()))
+	}
+
+	if options.SortBy == "" {
+		options.SortBy = internal.SortByCreatedAsc
+	}
+	sortBy := []string{
+		parseSortBy(options.SortBy),
+		"id:desc",
+	}
+
+	skip, limit := indexer_internal.ParsePaginator(options.Paginator, maxTotalHits)
+
+	counting := limit == 0
+	if counting {
+		// If set limit to 0, it will be 20 by default, and -1 is not allowed.
+		// See https://www.meilisearch.com/docs/reference/api/search#limit
+		// So set limit to 1 to make the cost as low as possible, then clear the result before returning.
+		limit = 1
+	}
+
+	keyword := options.Keyword // default to match "words"
+	if options.SearchMode == indexer.SearchModeExact {
+		// https://www.meilisearch.com/docs/reference/api/search#phrase-search
+		keyword = doubleQuoteKeyword(keyword)
+	}
+
+	searchRes, err := b.inner.Client.Index(b.inner.VersionedIndexName()).Search(keyword, &meilisearch.SearchRequest{
+		Filter:           query.Statement(),
+		Limit:            int64(limit),
+		Offset:           int64(skip),
+		Sort:             sortBy,
+		MatchingStrategy: "all",
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	if counting {
+		searchRes.Hits = nil
+	}
+
+	hits, err := convertHits(searchRes)
+	if err != nil {
+		return nil, err
+	}
+
+	return &internal.SearchResult{
+		Total: searchRes.EstimatedTotalHits,
+		Hits:  hits,
+	}, nil
+}
+
+func parseSortBy(sortBy internal.SortBy) string {
+	field := strings.TrimPrefix(string(sortBy), "-")
+	if strings.HasPrefix(string(sortBy), "-") {
+		return field + ":desc"
+	}
+	return field + ":asc"
+}
+
+func doubleQuoteKeyword(k string) string {
+	kp := strings.Split(k, " ")
+	parts := 0
+	for i := range kp {
+		part := strings.Trim(kp[i], "\"")
+		if part != "" {
+			kp[parts] = fmt.Sprintf(`"%s"`, part)
+			parts++
+		}
+	}
+	return strings.Join(kp[:parts], " ")
+}
+
+func convertHits(searchRes *meilisearch.SearchResponse) ([]internal.Match, error) {
+	hits := make([]internal.Match, 0, len(searchRes.Hits))
+	for _, hit := range searchRes.Hits {
+		hit, ok := hit.(map[string]any)
+		if !ok {
+			return nil, ErrMalformedResponse
+		}
+
+		issueID, ok := hit["id"].(float64)
+		if !ok {
+			return nil, ErrMalformedResponse
+		}
+
+		hits = append(hits, internal.Match{
+			ID: int64(issueID),
+		})
+	}
+	return hits, nil
+}

modules/indexer/issues/meilisearch/meilisearch_test.go

@@ -0,0 +1,86 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package meilisearch
+
+import (
+	"fmt"
+	"net/http"
+	"os"
+	"testing"
+	"time"
+
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/indexer/issues/internal/tests"
+
+	"github.com/meilisearch/meilisearch-go"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestMeilisearchIndexer(t *testing.T) {
+	// The meilisearch instance started by pull-db-tests.yml > test-unit > services > meilisearch
+	url := "http://meilisearch:7700"
+	key := "" // auth has been disabled in test environment
+
+	if os.Getenv("CI") == "" {
+		// Make it possible to run tests against a local meilisearch instance
+		url = os.Getenv("TEST_MEILISEARCH_URL")
+		if url == "" {
+			t.Skip("TEST_MEILISEARCH_URL not set and not running in CI")
+			return
+		}
+		key = os.Getenv("TEST_MEILISEARCH_KEY")
+	}
+
+	require.Eventually(t, func() bool {
+		resp, err := http.Get(url)
+		return err == nil && resp.StatusCode == http.StatusOK
+	}, time.Minute, time.Second, "Expected meilisearch to be up")
+
+	indexer := NewIndexer(url, key, fmt.Sprintf("test_meilisearch_indexer_%d", time.Now().Unix()))
+	defer indexer.Close()
+
+	tests.TestIndexer(t, indexer)
+}
+
+func TestConvertHits(t *testing.T) {
+	_, err := convertHits(&meilisearch.SearchResponse{
+		Hits: []any{"aa", "bb", "cc", "dd"},
+	})
+	assert.ErrorIs(t, err, ErrMalformedResponse)
+
+	validResponse := &meilisearch.SearchResponse{
+		Hits: []any{
+			map[string]any{
+				"id":       float64(11),
+				"title":    "a title",
+				"content":  "issue body with no match",
+				"comments": []any{"hey whats up?", "I'm currently bowling", "nice"},
+			},
+			map[string]any{
+				"id":       float64(22),
+				"title":    "Bowling as title",
+				"content":  "",
+				"comments": []any{},
+			},
+			map[string]any{
+				"id":       float64(33),
+				"title":    "Bowl-ing as fuzzy match",
+				"content":  "",
+				"comments": []any{},
+			},
+		},
+	}
+	hits, err := convertHits(validResponse)
+	assert.NoError(t, err)
+	assert.Equal(t, []internal.Match{{ID: 11}, {ID: 22}, {ID: 33}}, hits)
+}
+
+func TestDoubleQuoteKeyword(t *testing.T) {
+	assert.Empty(t, doubleQuoteKeyword(""))
+	assert.Equal(t, `"a" "b" "c"`, doubleQuoteKeyword("a b c"))
+	assert.Equal(t, `"a" "d" "g"`, doubleQuoteKeyword("a  d g"))
+	assert.Equal(t, `"a" "d" "g"`, doubleQuoteKeyword("a  d g"))
+	assert.Equal(t, `"a" "d" "g"`, doubleQuoteKeyword(`a  "" "d" """g`))
+}

modules/indexer/issues/util.go

@@ -0,0 +1,199 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package issues
+
+import (
+	"context"
+	"errors"
+	"fmt"
+
+	"code.gitea.io/gitea/models/db"
+	issue_model "code.gitea.io/gitea/models/issues"
+	"code.gitea.io/gitea/modules/container"
+	"code.gitea.io/gitea/modules/indexer/issues/internal"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/queue"
+)
+
+// getIssueIndexerData returns the indexer data of an issue and a bool value indicating whether the issue exists.
+func getIssueIndexerData(ctx context.Context, issueID int64) (*internal.IndexerData, bool, error) {
+	issue, err := issue_model.GetIssueByID(ctx, issueID)
+	if err != nil {
+		if issue_model.IsErrIssueNotExist(err) {
+			return nil, false, nil
+		}
+		return nil, false, err
+	}
+
+	// FIXME: what if users want to search for a review comment of a pull request?
+	//        The comment type is CommentTypeCode or CommentTypeReview.
+	//        But LoadDiscussComments only loads CommentTypeComment.
+	if err := issue.LoadDiscussComments(ctx); err != nil {
+		return nil, false, err
+	}
+
+	comments := make([]string, 0, len(issue.Comments))
+	for _, comment := range issue.Comments {
+		if comment.Content != "" {
+			// what ever the comment type is, index the content if it is not empty.
+			comments = append(comments, comment.Content)
+		}
+	}
+
+	if err := issue.LoadAttributes(ctx); err != nil {
+		return nil, false, err
+	}
+
+	labels := make([]int64, 0, len(issue.Labels))
+	for _, label := range issue.Labels {
+		labels = append(labels, label.ID)
+	}
+
+	mentionIDs, err := issue_model.GetIssueMentionIDs(ctx, issueID)
+	if err != nil {
+		return nil, false, err
+	}
+
+	var (
+		reviewedIDs        []int64
+		reviewRequestedIDs []int64
+	)
+	{
+		reviews, err := issue_model.FindReviews(ctx, issue_model.FindReviewOptions{
+			ListOptions:  db.ListOptionsAll,
+			IssueID:      issueID,
+			OfficialOnly: false,
+		})
+		if err != nil {
+			return nil, false, err
+		}
+
+		reviewedIDsSet := make(container.Set[int64], len(reviews))
+		reviewRequestedIDsSet := make(container.Set[int64], len(reviews))
+		for _, review := range reviews {
+			if review.Type == issue_model.ReviewTypeRequest {
+				reviewRequestedIDsSet.Add(review.ReviewerID)
+			} else {
+				reviewedIDsSet.Add(review.ReviewerID)
+			}
+		}
+		reviewedIDs = reviewedIDsSet.Values()
+		reviewRequestedIDs = reviewRequestedIDsSet.Values()
+	}
+
+	subscriberIDs, err := issue_model.GetIssueWatchersIDs(ctx, issue.ID, true)
+	if err != nil {
+		return nil, false, err
+	}
+
+	var projectID int64
+	if issue.Project != nil {
+		projectID = issue.Project.ID
+	}
+
+	projectColumnID, err := issue.ProjectColumnID(ctx)
+	if err != nil {
+		return nil, false, err
+	}
+
+	return &internal.IndexerData{
+		ID:                 issue.ID,
+		RepoID:             issue.RepoID,
+		IsPublic:           !issue.Repo.IsPrivate,
+		Title:              issue.Title,
+		Content:            issue.Content,
+		Comments:           comments,
+		IsPull:             issue.IsPull,
+		IsClosed:           issue.IsClosed,
+		IsArchived:         issue.Repo.IsArchived,
+		LabelIDs:           labels,
+		NoLabel:            len(labels) == 0,
+		MilestoneID:        issue.MilestoneID,
+		ProjectID:          projectID,
+		ProjectColumnID:    projectColumnID,
+		PosterID:           issue.PosterID,
+		AssigneeID:         issue.AssigneeID,
+		MentionIDs:         mentionIDs,
+		ReviewedIDs:        reviewedIDs,
+		ReviewRequestedIDs: reviewRequestedIDs,
+		SubscriberIDs:      subscriberIDs,
+		UpdatedUnix:        issue.UpdatedUnix,
+		CreatedUnix:        issue.CreatedUnix,
+		DeadlineUnix:       issue.DeadlineUnix,
+		CommentCount:       int64(len(issue.Comments)),
+	}, true, nil
+}
+
+func updateRepoIndexer(ctx context.Context, repoID int64) error {
+	ids, err := issue_model.GetIssueIDsByRepoID(ctx, repoID)
+	if err != nil {
+		return fmt.Errorf("issue_model.GetIssueIDsByRepoID: %w", err)
+	}
+	for _, id := range ids {
+		if err := updateIssueIndexer(ctx, id); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+func updateIssueIndexer(ctx context.Context, issueID int64) error {
+	return pushIssueIndexerQueue(ctx, &IndexerMetadata{ID: issueID})
+}
+
+func deleteRepoIssueIndexer(ctx context.Context, repoID int64) error {
+	var ids []int64
+	ids, err := issue_model.GetIssueIDsByRepoID(ctx, repoID)
+	if err != nil {
+		return fmt.Errorf("issue_model.GetIssueIDsByRepoID: %w", err)
+	}
+
+	if len(ids) == 0 {
+		return nil
+	}
+	return pushIssueIndexerQueue(ctx, &IndexerMetadata{
+		IDs:      ids,
+		IsDelete: true,
+	})
+}
+
+type keepRetryKey struct{}
+
+// contextWithKeepRetry returns a context with a key indicating that the indexer should keep retrying.
+// Please note that it's for background tasks only, and it should not be used for user requests, or it may cause blocking.
+func contextWithKeepRetry(ctx context.Context) context.Context {
+	return context.WithValue(ctx, keepRetryKey{}, true)
+}
+
+func pushIssueIndexerQueue(ctx context.Context, data *IndexerMetadata) error {
+	if issueIndexerQueue == nil {
+		// Some unit tests will trigger indexing, but the queue is not initialized.
+		// It's OK to ignore it, but log a warning message in case it's not a unit test.
+		log.Warn("Trying to push %+v to issue indexer queue, but the queue is not initialized, it's OK if it's a unit test", data)
+		return nil
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		default:
+		}
+		err := issueIndexerQueue.Push(data)
+		if errors.Is(err, queue.ErrAlreadyInQueue) {
+			return nil
+		}
+		if errors.Is(err, context.DeadlineExceeded) { // the queue is full
+			log.Warn("It seems that issue indexer is slow and the queue is full. Please check the issue indexer or increase the queue size.")
+			if ctx.Value(keepRetryKey{}) == nil {
+				return err
+			}
+			// It will be better to increase the queue size instead of retrying, but users may ignore the previous warning message.
+			// However, even it retries, it may still cause index loss when there's a deadline in the context.
+			log.Debug("Retry to push %+v to issue indexer queue", data)
+			continue
+		}
+		return err
+	}
+}