feat(search): Elasticsearch searcher (name+content) and oo search (#37)

cmd/oo/main.go

@@ -18,6 +18,7 @@
 //	oo docs          tools | convert | optimize | ocr | hocr | as-md | put-md | put-txt | put-xlsx
 //	oo catalog       match | merge | apply | scan-contacts | scan-projects | scan-thunderbird
 //	oo dav           ls | move | copy | mkdir | rename-file | rename-folder | download | fileops
+//	oo search        QUERY [--content] [--folder ID] [--limit N] [--json]
 //
 // CRM association rules: docs/crm-associations.md
 //

cmd/oo/search.go

@@ -0,0 +1,72 @@
+package main
+
+import (
+	onlyoffice "github.com/eslider/go-onlyoffice"
+	"github.com/spf13/cobra"
+)
+
+func init() {
+	rootCmd.AddCommand(searchCmd())
+}
+
+// searchCmd queries the OnlyOffice Elasticsearch index directly. The REST
+// /api/2.0/files/@search endpoint only searches file names in the database;
+// content search needs ES (see docs/elasticsearch.md).
+func searchCmd() *cobra.Command {
+	var (
+		content bool
+		folder  string
+		limit   int
+		asJSON  bool
+	)
+	cmd := &cobra.Command{
+		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" +
+			"matches extracted document text (document.attachment.content); this covers\n" +
+			"Office formats (docx/xlsx/pptx) and is slower.\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),
+		RunE: func(cmd *cobra.Command, args []string) error {
+			if asJSON {
+				outputFormat = "json"
+			}
+			es, err := onlyoffice.NewESSearcher(onlyoffice.ESConfigFromEnv())
+			if err != nil {
+				return err
+			}
+			hits, err := es.Search(cmd.Context(), onlyoffice.SearchQuery{
+				Text:      args[0],
+				InContent: content,
+				FolderID:  folder,
+				Limit:     limit,
+			})
+			if err != nil {
+				return err
+			}
+			rows := make([]map[string]any, 0, len(hits))
+			for _, h := range hits {
+				rows = append(rows, map[string]any{
+					"id":        h.ID,
+					"title":     h.Title,
+					"folder":    h.ParentID,
+					"score":     h.Score,
+					"highlight": h.Highlight,
+				})
+			}
+			if outputFormat == "json" {
+				printJSON(rows)
+				return nil
+			}
+			printTable([]string{"id", "title", "folder", "score", "highlight"}, rows)
+			return nil
+		},
+	}
+	cmd.Flags().BoolVar(&content, "content", false, "also match extracted document content")
+	cmd.Flags().StringVar(&folder, "folder", "", "limit to a Documents folder id")
+	cmd.Flags().IntVar(&limit, "limit", 20, "maximum number of results")
+	cmd.Flags().BoolVar(&asJSON, "json", false, "shorthand for --output json")
+	return cmd
+}

cmd/oo/search_test.go

@@ -0,0 +1,46 @@
+package main
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+)
+
+func TestSearchCommandRegisteredWithFlags(t *testing.T) {
+	cmd, _, err := rootCmd.Find([]string{"search"})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if cmd.Name() != "search" {
+		t.Fatalf("search resolved to %q", cmd.Name())
+	}
+	for _, name := range []string{"content", "folder", "limit", "json"} {
+		if cmd.Flags().Lookup(name) == nil {
+			t.Errorf("search: missing --%s flag", name)
+		}
+	}
+	if got := cmd.Flags().Lookup("limit").DefValue; got != "20" {
+		t.Errorf("--limit default = %q, want 20", got)
+	}
+}
+
+func TestSearchWithoutESURLIsClearError(t *testing.T) {
+	clearEnv(t, "ONLYOFFICE_ES_URL", "ONLYOFFICE_ES_INDEX", "ONLYOFFICE_TENANT")
+	errBuf := &bytes.Buffer{}
+	rootCmd.SetErr(errBuf)
+	rootCmd.SetOut(&bytes.Buffer{})
+	rootCmd.SetArgs([]string{"search", "Rechnung"})
+	t.Cleanup(func() {
+		rootCmd.SetArgs(nil)
+		rootCmd.SetOut(nil)
+		rootCmd.SetErr(nil)
+	})
+
+	err := rootCmd.Execute()
+	if err == nil {
+		t.Fatal("expected error without ONLYOFFICE_ES_URL")
+	}
+	if !strings.Contains(err.Error(), "ONLYOFFICE_ES_URL") {
+		t.Fatalf("error %q missing ONLYOFFICE_ES_URL", err.Error())
+	}
+}