[FEAT] support searching non default branches/tags when using git-grep (#3654)

resolves https://codeberg.org/forgejo/forgejo/pulls/3639#issuecomment-1806676 and https://codeberg.org/forgejo/forgejo/pulls/3513#issuecomment-1794990

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/3654
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: Shiny Nematoda <snematoda.751k2@aleeas.com>
Co-committed-by: Shiny Nematoda <snematoda.751k2@aleeas.com>

modules/git/grep.go

@@ -29,6 +29,7 @@ type GrepOptions struct {
 	MaxResultLimit    int
 	ContextLineNumber int
 	IsFuzzy           bool
+	PathSpec          []setting.Glob
 }
 
 func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepOptions) ([]*GrepResult, error) {
@@ -61,15 +62,20 @@ func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepO
 	} else {
 		cmd.AddOptionValues("-e", strings.TrimLeft(search, "-"))
 	}
+
 	// pathspec
-	files := make([]string, 0, len(setting.Indexer.IncludePatterns)+len(setting.Indexer.ExcludePatterns))
-	for _, expr := range setting.Indexer.IncludePatterns {
-		files = append(files, expr.Pattern())
+	files := make([]string, 0,
+		len(setting.Indexer.IncludePatterns)+
+			len(setting.Indexer.ExcludePatterns)+
+			len(opts.PathSpec))
+	for _, expr := range append(setting.Indexer.IncludePatterns, opts.PathSpec...) {
+		files = append(files, ":"+expr.Pattern())
 	}
 	for _, expr := range setting.Indexer.ExcludePatterns {
 		files = append(files, ":^"+expr.Pattern())
 	}
 	cmd.AddDynamicArguments(cmp.Or(opts.RefName, "HEAD")).AddDashesAndList(files...)
+
 	opts.MaxResultLimit = cmp.Or(opts.MaxResultLimit, 50)
 	stderr := bytes.Buffer{}
 	err = cmd.Run(&RunOpts{

modules/git/grep_test.go

@@ -76,3 +76,33 @@ func TestGrepLongFiles(t *testing.T) {
 	assert.Len(t, res, 1)
 	assert.Len(t, res[0].LineCodes[0], 65*1024)
 }
+
+func TestGrepRefs(t *testing.T) {
+	tmpDir := t.TempDir()
+
+	err := InitRepository(DefaultContext, tmpDir, false, Sha1ObjectFormat.Name())
+	assert.NoError(t, err)
+
+	gitRepo, err := openRepositoryWithDefaultContext(tmpDir)
+	assert.NoError(t, err)
+	defer gitRepo.Close()
+
+	assert.NoError(t, os.WriteFile(path.Join(tmpDir, "README.md"), []byte{'A'}, 0o666))
+	assert.NoError(t, AddChanges(tmpDir, true))
+
+	err = CommitChanges(tmpDir, CommitChangesOptions{Message: "add A"})
+	assert.NoError(t, err)
+
+	assert.NoError(t, gitRepo.CreateTag("v1", "HEAD"))
+
+	assert.NoError(t, os.WriteFile(path.Join(tmpDir, "README.md"), []byte{'A', 'B', 'C', 'D'}, 0o666))
+	assert.NoError(t, AddChanges(tmpDir, true))
+
+	err = CommitChanges(tmpDir, CommitChangesOptions{Message: "add BCD"})
+	assert.NoError(t, err)
+
+	res, err := GrepSearch(context.Background(), gitRepo, "a", GrepOptions{RefName: "v1"})
+	assert.NoError(t, err)
+	assert.Len(t, res, 1)
+	assert.Equal(t, res[0].LineCodes[0], "A")
+}

release-notes/8.0.0/feat/3654.md

@@ -0,0 +1 @@
+Code Search for non-default branches and tags when repository indexer is disabled

routers/web/repo/search.go

@@ -64,7 +64,11 @@ func Search(ctx *context.Context) {
 			ctx.Data["CodeIndexerUnavailable"] = !code_indexer.IsAvailable(ctx)
 		}
 	} else {
-		res, err := git.GrepSearch(ctx, ctx.Repo.GitRepo, keyword, git.GrepOptions{ContextLineNumber: 3, IsFuzzy: isFuzzy})
+		res, err := git.GrepSearch(ctx, ctx.Repo.GitRepo, keyword, git.GrepOptions{
+			ContextLineNumber: 1,
+			IsFuzzy:           isFuzzy,
+			RefName:           ctx.Repo.RefName,
+		})
 		if err != nil {
 			ctx.ServerError("GrepSearch", err)
 			return

routers/web/repo/view.go

@@ -1140,6 +1140,7 @@ PostRecentBranchCheck:
 	ctx.Data["TreeLink"] = treeLink
 	ctx.Data["TreeNames"] = treeNames
 	ctx.Data["BranchLink"] = branchLink
+	ctx.Data["CodeIndexerDisabled"] = !setting.Indexer.RepoIndexerEnabled
 	ctx.HTML(http.StatusOK, tplRepoHome)
 }
 

routers/web/web.go

@@ -1570,11 +1570,17 @@ func registerRoutes(m *web.Route) {
 
 	m.Group("/{username}/{reponame}", func() {
 		if !setting.Repository.DisableStars {
-			m.Get("/stars", repo.Stars)
+			m.Get("/stars", context.RepoRef(), repo.Stars)
 		}
-		m.Get("/watchers", repo.Watchers)
-		m.Get("/search", reqRepoCodeReader, repo.Search)
-	}, ignSignIn, context.RepoAssignment, context.RepoRef(), context.UnitTypes())
+		m.Get("/watchers", context.RepoRef(), repo.Watchers)
+		m.Group("/search", func() {
+			m.Get("", context.RepoRef(), repo.Search)
+			if !setting.Indexer.RepoIndexerEnabled {
+				m.Get("/branch/*", context.RepoRefByType(context.RepoRefBranch), repo.Search)
+				m.Get("/tag/*", context.RepoRefByType(context.RepoRefTag), repo.Search)
+			}
+		}, reqRepoCodeReader)
+	}, ignSignIn, context.RepoAssignment, context.UnitTypes())
 
 	m.Group("/{username}", func() {
 		m.Group("/{reponame}", func() {

templates/repo/home.tmpl

@@ -8,10 +8,10 @@
 		<div class="repo-description">
 			<div id="repo-desc" class="gt-word-break tw-text-16">
 				{{$description := .Repository.DescriptionHTML $.Context}}
-				{{if $description}}<span class="description">{{$description | RenderCodeBlock}}</span>{{else if .IsRepositoryAdmin}}<span class="no-description text-italic">{{ctx.Locale.Tr "repo.no_desc"}}</span>{{end}}
-				<a class="link" href="{{.Repository.Website}}">{{.Repository.Website}}</a>
+				{{if $description}}<span class="description">{{$description | RenderCodeBlock}}</span>{{else}}<span class="no-description text-italic">{{ctx.Locale.Tr "repo.no_desc"}}</span>{{end}}
+				{{if .Repository.Website}}<a class="link" href="{{.Repository.Website}}">{{.Repository.Website}}</a>{{end}}
 			</div>
-			<form class="ignore-dirty" action="{{.RepoLink}}/search" method="get">
+			<form class="ignore-dirty" action="{{.RepoLink}}/search/{{if .CodeIndexerDisabled}}{{.BranchNameSubURL}}{{end}}" method="get" data-test-tag="codesearch">
 				<div class="ui small action input">
 					<input name="q" value="{{.Keyword}}" placeholder="{{ctx.Locale.Tr "search.code_kind"}}">
 					{{template "shared/search/button"}}
@@ -106,16 +106,7 @@
 						{{ctx.Locale.Tr "repo.use_template"}}
 					</a>
 				{{end}}
-				{{if $isHomepage}}
-					{{/* only show the "code search" on the repo home page, it only does global search,
-						so do not show it when viewing file or directory to avoid misleading users (it doesn't search in a directory) */}}
-					<form class="ignore-dirty" action="{{.RepoLink}}/search" method="get">
-						<div class="ui small action input">
-							<input name="q" placeholder="{{ctx.Locale.Tr "search.code_kind"}}">
-							{{template "shared/search/button"}}
-						</div>
-					</form>
-				{{else}}
+				{{if (not $isHomepage)}}
 					<span class="breadcrumb repo-path tw-ml-1">
 						<a class="section" href="{{.RepoLink}}/src/{{.BranchNameSubURL}}" title="{{.Repository.Name}}">{{StringUtils.EllipsisString .Repository.Name 30}}</a>
 						{{- range $i, $v := .TreeNames -}}

tests/integration/repo_search_test.go

@@ -12,6 +12,7 @@ import (
 	code_indexer "code.gitea.io/gitea/modules/indexer/code"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/test"
+	"code.gitea.io/gitea/routers"
 	"code.gitea.io/gitea/tests"
 
 	"github.com/PuerkitoBio/goquery"
@@ -38,6 +39,7 @@ func TestSearchRepoNoIndexer(t *testing.T) {
 func testSearchRepo(t *testing.T, indexer bool) {
 	defer tests.PrepareTestEnv(t)()
 	defer test.MockVariableValue(&setting.Indexer.RepoIndexerEnabled, indexer)()
+	defer test.MockVariableValue(&testWebRoutes, routers.NormalRoutes())()
 
 	repo, err := repo_model.GetRepositoryByOwnerAndName(db.DefaultContext, "user2", "repo1")
 	assert.NoError(t, err)
@@ -48,6 +50,13 @@ func testSearchRepo(t *testing.T, indexer bool) {
 
 	testSearch(t, "/user2/repo1/search?q=Description&page=1", []string{"README.md"}, indexer)
 
+	req := NewRequest(t, "HEAD", "/user2/repo1/search/branch/"+repo.DefaultBranch)
+	if indexer {
+		MakeRequest(t, req, http.StatusNotFound)
+	} else {
+		MakeRequest(t, req, http.StatusOK)
+	}
+
 	defer test.MockVariableValue(&setting.Indexer.IncludePatterns, setting.IndexerGlobFromString("**.txt"))()
 	defer test.MockVariableValue(&setting.Indexer.ExcludePatterns, setting.IndexerGlobFromString("**/y/**"))()
 

tests/integration/repo_test.go

@@ -190,10 +190,7 @@ func TestViewRepoWithSymlinks(t *testing.T) {
 
 // TestViewAsRepoAdmin tests PR #2167
 func TestViewAsRepoAdmin(t *testing.T) {
-	for user, expectedNoDescription := range map[string]bool{
-		"user2": true,
-		"user4": false,
-	} {
+	for _, user := range []string{"user2", "user4"} {
 		defer tests.PrepareTestEnv(t)()
 
 		session := loginUser(t, user)
@@ -206,7 +203,7 @@ func TestViewAsRepoAdmin(t *testing.T) {
 		repoTopics := htmlDoc.doc.Find("#repo-topics").Children()
 		repoSummary := htmlDoc.doc.Find(".repository-summary").Children()
 
-		assert.Equal(t, expectedNoDescription, noDescription.HasClass("no-description"))
+		assert.True(t, noDescription.HasClass("no-description"))
 		assert.True(t, repoTopics.HasClass("repo-topic"))
 		assert.True(t, repoSummary.HasClass("repository-menu"))
 	}
@@ -995,3 +992,33 @@ func TestViewRepoOpenWith(t *testing.T) {
 		testOpenWith([]string{"test://"})
 	})
 }
+
+func TestRepoCodeSearchForm(t *testing.T) {
+	defer tests.PrepareTestEnv(t)()
+
+	testSearchForm := func(t *testing.T, indexer bool) {
+		defer test.MockVariableValue(&setting.Indexer.RepoIndexerEnabled, indexer)()
+		req := NewRequest(t, "GET", "/user2/repo1/src/branch/master")
+		resp := MakeRequest(t, req, http.StatusOK)
+
+		htmlDoc := NewHTMLParser(t, resp.Body)
+		action, exists := htmlDoc.doc.Find("form[data-test-tag=codesearch]").Attr("action")
+		assert.True(t, exists)
+
+		branchSubURL := "/branch/master"
+
+		if indexer {
+			assert.NotContains(t, action, branchSubURL)
+		} else {
+			assert.Contains(t, action, branchSubURL)
+		}
+	}
+
+	t.Run("indexer disabled", func(t *testing.T) {
+		testSearchForm(t, false)
+	})
+
+	t.Run("indexer enabled", func(t *testing.T) {
+		testSearchForm(t, true)
+	})
+}