Merge pull request 'feat(search): подстрока/AND, scope по поддереву, лимит 1000 (#49)' (#50) from fix/es-substring#49 into main

cmd/oo/search.go

@@ -18,14 +18,15 @@ func init() {
 // content search needs Elasticsearch (see docs/elasticsearch.md).
 func searchCmd() *cobra.Command {
 	var (
-		content bool
+		content   bool
-		folder  string
+		folder    string
-		limit   int
+		limit     int
-		backend string
+		backend   string
-		asJSON  bool
+		asJSON    bool
+		substring bool
 	)
 	cmd := &cobra.Command{
-		Use:   "search QUERY",
+		Use:   "search QUERY...",
 		Short: "Full-text search over documents by name, optionally by content (Elasticsearch)",
 		Long: "Search the OnlyOffice Documents index.\n\n" +
 			"By default only file names are matched. With --content the query also\n" +
@@ -36,7 +37,7 @@ func searchCmd() *cobra.Command {
 			"PDFs and scans (see docs/elasticsearch.md).\n\n" +
 			"Requires ONLYOFFICE_ES_URL (and optionally ONLYOFFICE_ES_INDEX,\n" +
 			"ONLYOFFICE_TENANT). See docs/elasticsearch.md for the tunnel setup.",
-		Args: cobra.ExactArgs(1),
+		Args: cobra.MinimumNArgs(1),
 		RunE: func(cmd *cobra.Command, args []string) error {
 			if asJSON {
 				outputFormat = "json"
@@ -59,10 +60,11 @@ func searchCmd() *cobra.Command {
 				return err
 			}
 			hits, err := searcher.Search(cmd.Context(), onlyoffice.SearchQuery{
-				Text:      args[0],
+				Text:      strings.Join(args, " "),
 				InContent: content,
 				FolderID:  folder,
 				Limit:     limit,
+				Substring: substring,
 			})
 			if err != nil {
 				return err
@@ -86,7 +88,8 @@ func searchCmd() *cobra.Command {
 		},
 	}
 	cmd.Flags().BoolVar(&content, "content", false, "also match extracted document content")
-	cmd.Flags().StringVar(&folder, "folder", "", "limit to a Documents folder id")
+	cmd.Flags().BoolVar(&substring, "substring", false, "case-insensitive *term* title match; multiple QUERY args are ANDed")
+	cmd.Flags().StringVar(&folder, "folder", "", "limit to a Documents folder id (matches the folder subtree)")
 	cmd.Flags().IntVar(&limit, "limit", 20, "maximum number of results")
 	cmd.Flags().StringVar(&backend, "backend", "oo", "index to query: oo (OnlyOffice) | own (oo index)")
 	cmd.Flags().BoolVar(&asJSON, "json", false, "shorthand for --output json")

file_core.go

@@ -80,13 +80,16 @@ type FileStore interface {
 }
 
 // SearchQuery narrows a Searcher request. InContent asks the backend to match
-// document bodies, not just titles.
+// document bodies, not just titles. Substring switches title matching from the
+// analyzer's whole-token match to a case-insensitive "*term*" wildcard and ANDs
+// every whitespace-separated term (e.g. "rechnung 2025").
 type SearchQuery struct {
 	Text       string
 	InContent  bool
 	FolderID   string
 	Extensions []string
 	Limit      int
+	Substring  bool
 }
 
 // SearchHit is one Searcher result: the matching entry plus backend-specific

file_es.go

@@ -28,7 +28,7 @@ import (
 const (
 	defaultESIndex    = "files_file"
 	defaultESLimit    = 20
-	maxESLimit        = 200
+	maxESLimit        = 1000
 	maxESResponseSize = 8 << 20
 )
 
@@ -119,14 +119,30 @@ func esSearchRequest(q SearchQuery, tenant string) esRequest {
 	if q.InContent {
 		fields = append(fields, "document.attachment.content")
 	}
-	must := []esClause{{MultiMatch: &esMultiMatch{Query: q.Text, Fields: fields}}}
+	var must []esClause
+	if q.Substring {
+		for _, term := range strings.Fields(strings.ToLower(q.Text)) {
+			if term = escapeWildcard(term); term != "" {
+				must = append(must, esClause{Wildcard: map[string]any{"title": "*" + term + "*"}})
+			}
+		}
+	}
+	if len(must) == 0 {
+		must = []esClause{{MultiMatch: &esMultiMatch{Query: q.Text, Fields: fields}}}
+	}
 
 	var filter []esClause
 	if t := strings.TrimSpace(tenant); t != "" {
 		filter = append(filter, esClause{Term: map[string]any{"tenantId": numericOrString(t)}})
 	}
 	if f := strings.TrimSpace(q.FolderID); f != "" {
-		filter = append(filter, esClause{Term: map[string]any{"folders.folderId": f}})
+		// folders is an ES nested field; a plain term on folders.folderId would
+		// not match. The stored Folders list holds every ancestor id, so
+		// filtering by a project root id scopes to its whole subtree.
+		filter = append(filter, esClause{Nested: &esNested{
+			Path:  "folders",
+			Query: esNestedTerm{Term: map[string]any{"folders.folderId": f}},
+		}})
 	}
 	for _, ext := range normalizeExtensions(q.Extensions) {
 		filter = append(filter, esClause{Wildcard: map[string]any{"title": "*." + ext}})
@@ -190,6 +206,22 @@ type esClause struct {
 	Term       map[string]any `json:"term,omitempty"`
 	Terms      map[string]any `json:"terms,omitempty"`
 	Wildcard   map[string]any `json:"wildcard,omitempty"`
+	Nested     *esNested      `json:"nested,omitempty"`
+}
+
+type esNested struct {
+	Path  string       `json:"path"`
+	Query esNestedTerm `json:"query"`
+}
+
+type esNestedTerm struct {
+	Term map[string]any `json:"term,omitempty"`
+}
+
+// escapeWildcard strips ES wildcard metacharacters from a user term so a query
+// cannot inject wildcard syntax. Pure, so it is unit-tested.
+func escapeWildcard(s string) string {
+	return strings.NewReplacer("*", "", "?", "", `\`, "").Replace(s)
 }
 
 type esMultiMatch struct {

file_es_test.go

@@ -55,7 +55,7 @@ func TestESSearchRequestFiltersAndLimit(t *testing.T) {
 		Text:       "Storchen",
 		FolderID:   "649",
 		Extensions: []string{".PDF", "pdf", "docx"},
-		Limit:      999,
+		Limit:      5000,
 	}, "42")
 	if got.Size != maxESLimit {
 		t.Errorf("size = %d, want cap %d", got.Size, maxESLimit)
@@ -65,10 +65,10 @@ func TestESSearchRequestFiltersAndLimit(t *testing.T) {
 		switch {
 		case f.Term != nil && f.Term["tenantId"] != nil:
 			tenant++
-		case f.Term != nil && f.Term["folders.folderId"] != nil:
+		case f.Nested != nil:
 			folder++
-			if f.Term["folders.folderId"] != "649" {
+			if f.Nested.Path != "folders" || f.Nested.Query.Term["folders.folderId"] != "649" {
-				t.Errorf("folder filter = %+v", f.Term)
+				t.Errorf("folder filter = %+v, want nested folders term 649", f.Nested)
 			}
 		case f.Wildcard != nil:
 			wildcards++
@@ -82,6 +82,34 @@ func TestESSearchRequestFiltersAndLimit(t *testing.T) {
 	}
 }
 
+func TestESSearchRequestSubstringAndsTerms(t *testing.T) {
+	got := esSearchRequest(SearchQuery{Text: "Rechnung 2025", Substring: true, FolderID: "522"}, "")
+	if got.Query.Bool.Must[0].MultiMatch != nil {
+		t.Fatalf("substring must not use multi_match: %+v", got.Query.Bool.Must)
+	}
+	if len(got.Query.Bool.Must) != 2 {
+		t.Fatalf("must = %+v, want two ANDed wildcard terms", got.Query.Bool.Must)
+	}
+	want := []string{"*rechnung*", "*2025*"}
+	for i, m := range got.Query.Bool.Must {
+		if m.Wildcard == nil || m.Wildcard["title"] != want[i] {
+			t.Errorf("must[%d] = %+v, want title wildcard %q", i, m, want[i])
+		}
+	}
+	if len(got.Query.Bool.Filter) != 1 || got.Query.Bool.Filter[0].Nested == nil {
+		t.Errorf("folder filter = %+v, want nested", got.Query.Bool.Filter)
+	}
+}
+
+func TestEscapeWildcard(t *testing.T) {
+	cases := map[string]string{"*rechnung*": "rechnung", "a?b\\c": "abc", "plain": "plain"}
+	for in, want := range cases {
+		if got := escapeWildcard(in); got != want {
+			t.Errorf("escapeWildcard(%q) = %q, want %q", in, got, want)
+		}
+	}
+}
+
 func TestESSearchRequestRejectsEmptyTextAtSearch(t *testing.T) {
 	s, err := NewESSearcher(ESConfig{URL: "http://localhost:9200"})
 	if err != nil {