Add API route for explore/code search

modules/indexer/code/internal/indexer.go

@@ -26,6 +26,7 @@ type SearchOptions struct {
 	Language string
 
 	IsKeywordFuzzy bool
+	IsHTMLSafe     bool
 
 	db.Paginator
 }

modules/indexer/code/search.go

@@ -28,18 +28,19 @@ type Result struct {
 type ResultLine struct {
 	Num              int
 	FormattedContent template.HTML
+	RawContent       string
 }
 
 type SearchResultLanguages = internal.SearchResultLanguages
 
 type SearchOptions = internal.SearchOptions
 
-func indices(content string, selectionStartIndex, selectionEndIndex int) (int, int) {
+func indices(content string, selectionStartIndex, selectionEndIndex, numLinesBuffer int) (int, int) {
 	startIndex := selectionStartIndex
 	numLinesBefore := 0
 	for ; startIndex > 0; startIndex-- {
 		if content[startIndex-1] == '\n' {
-			if numLinesBefore == 1 {
+			if numLinesBefore == numLinesBuffer {
 				break
 			}
 			numLinesBefore++
@@ -50,7 +51,7 @@ func indices(content string, selectionStartIndex, selectionEndIndex int) (int, i
 	numLinesAfter := 0
 	for ; endIndex < len(content); endIndex++ {
 		if content[endIndex] == '\n' {
-			if numLinesAfter == 1 {
+			if numLinesAfter == numLinesBuffer {
 				break
 			}
 			numLinesAfter++
@@ -86,7 +87,20 @@ func HighlightSearchResultCode(filename, language string, lineNums []int, code s
 	return lines
 }
 
-func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Result, error) {
+func rawSearchResultCode(lineNums []int, code string) []*ResultLine {
+	rawLines := strings.Split(code, "\n")
+
+	lines := make([]*ResultLine, min(len(rawLines), len(lineNums)))
+	for i := 0; i < len(lines); i++ {
+		lines[i] = &ResultLine{
+			Num:        lineNums[i],
+			RawContent: rawLines[i],
+		}
+	}
+	return lines
+}
+
+func searchResult(result *internal.SearchResult, startIndex, endIndex int, escapeHTML bool) (*Result, error) {
 	startLineNum := 1 + strings.Count(result.Content[:startIndex], "\n")
 
 	var formattedLinesBuffer bytes.Buffer
@@ -117,14 +131,21 @@ func searchResult(result *internal.SearchResult, startIndex, endIndex int) (*Res
 		index += len(line)
 	}
 
+	var lines []*ResultLine
+	if escapeHTML {
+		lines = HighlightSearchResultCode(result.Filename, result.Language, lineNums, formattedLinesBuffer.String())
+	} else {
+		lines = rawSearchResultCode(lineNums, formattedLinesBuffer.String())
+	}
+
 	return &Result{
 		RepoID:      result.RepoID,
 		Filename:    result.Filename,
 		CommitID:    result.CommitID,
 		UpdatedUnix: result.UpdatedUnix,
 		Language:    result.Language,
 		Color:       result.Color,
-		Lines:       HighlightSearchResultCode(result.Filename, result.Language, lineNums, formattedLinesBuffer.String()),
+		Lines:       lines,
 	}, nil
 }
 
@@ -142,9 +163,14 @@ func PerformSearch(ctx context.Context, opts *SearchOptions) (int, []*Result, []
 
 	displayResults := make([]*Result, len(results))
 
+	nLinesBuffer := 0
+	if opts.IsHTMLSafe {
+		nLinesBuffer = 1
+	}
+
 	for i, result := range results {
-		startIndex, endIndex := indices(result.Content, result.StartIndex, result.EndIndex)
-		displayResults[i], err = searchResult(result, startIndex, endIndex)
+		startIndex, endIndex := indices(result.Content, result.StartIndex, result.EndIndex, nLinesBuffer)
+		displayResults[i], err = searchResult(result, startIndex, endIndex, opts.IsHTMLSafe)
 		if err != nil {
 			return 0, nil, nil, err
 		}

modules/structs/explore.go

@@ -0,0 +1,20 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package structs
+
+// ExploreCodeSearchItem A code search match
+// swagger:model
+type ExploreCodeSearchItem struct {
+	RepoName   string `json:"repoName"`
+	FilePath   string `json:"path"`
+	LineNumber int    `json:"lineNumber"`
+	LineText   string `json:"lineText"`
+}
+
+// ExploreCodeResult all returned code search results
+// swagger:model
+type ExploreCodeResult struct {
+	Total   int                     `json:"total"`
+	Results []ExploreCodeSearchItem `json:"results"`
+}

routers/api/v1/api.go

@@ -85,6 +85,7 @@ import (
 	"code.gitea.io/gitea/modules/web"
 	"code.gitea.io/gitea/routers/api/v1/activitypub"
 	"code.gitea.io/gitea/routers/api/v1/admin"
+	"code.gitea.io/gitea/routers/api/v1/explore"
 	"code.gitea.io/gitea/routers/api/v1/misc"
 	"code.gitea.io/gitea/routers/api/v1/notify"
 	"code.gitea.io/gitea/routers/api/v1/org"
@@ -890,6 +891,14 @@ func Routes() *web.Router {
 		// Misc (public accessible)
 		m.Group("", func() {
 			m.Get("/version", misc.Version)
+			m.Group("/explore", func() {
+				m.Get("/code", func(ctx *context.APIContext) {
+					if unit.TypeCode.UnitGlobalDisabled() {
+						ctx.NotFound("Repo unit code is disabled", nil)
+						return
+					}
+				}, explore.Code)
+			})
 			m.Get("/signing-key.gpg", misc.SigningKey)
 			m.Post("/markup", reqToken(), bind(api.MarkupOption{}), misc.Markup)
 			m.Post("/markdown", reqToken(), bind(api.MarkdownOption{}), misc.Markdown)

routers/api/v1/explore/code.go

@@ -0,0 +1,143 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package explore
+
+import (
+	"net/http"
+	"slices"
+
+	"code.gitea.io/gitea/models/db"
+	repo_model "code.gitea.io/gitea/models/repo"
+	code_indexer "code.gitea.io/gitea/modules/indexer/code"
+	"code.gitea.io/gitea/modules/setting"
+	api "code.gitea.io/gitea/modules/structs"
+	"code.gitea.io/gitea/services/context"
+	"code.gitea.io/gitea/services/convert"
+)
+
+// Code explore code
+func Code(ctx *context.APIContext) {
+	// swagger:operation GET /explore/code explore codeSearch
+	// ---
+	// summary: Search for code
+	// produces:
+	// - application/json
+	// parameters:
+	// - name: q
+	//   in: query
+	//   description: keyword
+	//   type: string
+	// - name: page
+	//   in: query
+	//   description: page number of results to return (1-based)
+	//   type: integer
+	// - name: fuzzy
+	//   in: query
+	//   description: whether to search fuzzy or strict (defaults to true)
+	//   type: boolean
+	// responses:
+	//   "200":
+	//     description: "SearchResults of a successful search"
+	//     schema:
+	//			 "$ref": "#/definitions/ExploreCodeResult"
+	if !setting.Indexer.RepoIndexerEnabled {
+		ctx.NotFound("Indexer not enabled")
+		return
+	}
+
+	keyword := ctx.FormTrim("q")
+
+	isFuzzy := ctx.FormOptionalBool("fuzzy").ValueOrDefault(true)
+
+	if keyword == "" {
+		ctx.JSON(http.StatusOK, api.ExploreCodeResult{
+			Total:   0,
+			Results: make([]api.ExploreCodeSearchItem, 0),
+		})
+		return
+	}
+
+	page := ctx.FormInt("page")
+	if page <= 0 {
+		page = 1
+	}
+
+	var (
+		repoIDs []int64
+		err     error
+		isAdmin bool
+	)
+	if ctx.Doer != nil {
+		isAdmin = ctx.Doer.IsAdmin
+	}
+
+	if ctx.Doer == nil || !isAdmin {
+		repoIDs, err = repo_model.FindUserCodeAccessibleRepoIDs(ctx, ctx.Doer)
+		if err != nil {
+			ctx.JSON(http.StatusInternalServerError, api.SearchError{
+				OK:    false,
+				Error: err.Error(),
+			})
+			return
+		}
+	}
+
+	var (
+		total         int
+		searchResults []*code_indexer.Result
+		repoMaps      map[int64]*repo_model.Repository
+	)
+
+	if (len(repoIDs) > 0) || isAdmin {
+		total, searchResults, _, err = code_indexer.PerformSearch(ctx, &code_indexer.SearchOptions{
+			RepoIDs:        repoIDs,
+			Keyword:        keyword,
+			IsKeywordFuzzy: isFuzzy,
+			IsHTMLSafe:     false,
+			Paginator: &db.ListOptions{
+				Page:     page,
+				PageSize: setting.API.DefaultPagingNum,
+			},
+		})
+		if err != nil {
+			if code_indexer.IsAvailable(ctx) {
+				ctx.JSON(http.StatusInternalServerError, api.SearchError{
+					OK:    false,
+					Error: err.Error(),
+				})
+				return
+			}
+		}
+
+		loadRepoIDs := make([]int64, 0, len(searchResults))
+		for _, result := range searchResults {
+			if !slices.Contains(loadRepoIDs, result.RepoID) {
+				loadRepoIDs = append(loadRepoIDs, result.RepoID)
+			}
+		}
+
+		repoMaps, err = repo_model.GetRepositoriesMapByIDs(ctx, loadRepoIDs)
+		if err != nil {
+			ctx.JSON(http.StatusInternalServerError, api.SearchError{
+				OK:    false,
+				Error: err.Error(),
+			})
+			return
+		}
+
+		if len(loadRepoIDs) != len(repoMaps) {
+			// Remove deleted repos from search results
+			cleanedSearchResults := make([]*code_indexer.Result, 0, len(repoMaps))
+			for _, sr := range searchResults {
+				if _, found := repoMaps[sr.RepoID]; found {
+					cleanedSearchResults = append(cleanedSearchResults, sr)
+				}
+			}
+
+			searchResults = cleanedSearchResults
+		}
+	}
+
+	ctx.JSON(http.StatusOK, convert.ToExploreCodeSearchResults(total, searchResults, repoMaps))
+}

routers/api/v1/swagger/explore.go

@@ -0,0 +1,15 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package swagger
+
+import (
+	api "code.gitea.io/gitea/modules/structs"
+)
+
+// ExploreCode
+// swagger:response ExploreCode
+type swaggerResponseExploreCode struct {
+	// in:body
+	Body api.ExploreCodeResult `json:"body"`
+}

routers/web/explore/code.go

@@ -81,6 +81,7 @@ func Code(ctx *context.Context) {
 			RepoIDs:        repoIDs,
 			Keyword:        keyword,
 			IsKeywordFuzzy: isFuzzy,
+			IsHTMLSafe:     true,
 			Language:       language,
 			Paginator: &db.ListOptions{
 				Page:     page,

routers/web/repo/search.go

@@ -59,6 +59,7 @@ func Search(ctx *context.Context) {
 			RepoIDs:        []int64{ctx.Repo.Repository.ID},
 			Keyword:        keyword,
 			IsKeywordFuzzy: isFuzzy,
+			IsHTMLSafe:     true,
 			Language:       language,
 			Paginator: &db.ListOptions{
 				Page:     page,

routers/web/user/code.go

@@ -80,6 +80,7 @@ func CodeSearch(ctx *context.Context) {
 			Keyword:        keyword,
 			IsKeywordFuzzy: isFuzzy,
 			Language:       language,
+			IsHTMLSafe:     true,
 			Paginator: &db.ListOptions{
 				Page:     page,
 				PageSize: setting.UI.RepoSearchPagingNum,

services/convert/explore.go

@@ -0,0 +1,30 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package convert
+
+import (
+	repo_model "code.gitea.io/gitea/models/repo"
+	code_indexer "code.gitea.io/gitea/modules/indexer/code"
+	api "code.gitea.io/gitea/modules/structs"
+)
+
+func ToExploreCodeSearchResults(total int, results []*code_indexer.Result, repoMaps map[int64]*repo_model.Repository) api.ExploreCodeResult {
+	out := api.ExploreCodeResult{
+		Total:   total,
+		Results: make([]api.ExploreCodeSearchItem, 0, len(results)),
+	}
+	for _, res := range results {
+		if repo := repoMaps[res.RepoID]; repo != nil {
+			for _, r := range res.Lines {
+				out.Results = append(out.Results, api.ExploreCodeSearchItem{
+					RepoName:   repo.FullName(),
+					FilePath:   res.Filename,
+					LineNumber: r.Num,
+					LineText:   r.RawContent,
+				})
+			}
+		}
+	}
+	return out
+}