feat(search): Elasticsearch-клиент (имя + контент) + oo search (#37) #41

Merged
eSlider merged 13 commits from feat/es-search#37 into main 2026-09-16 17:43:21 +01:00
20 changed files with 2330 additions and 25 deletions
+7
View File
@@ -34,3 +34,10 @@ ONLYOFFICE_PROJECT_ID=33
# MINIO_BUCKET=office # MINIO_BUCKET=office
# MINIO_ACCESS_KEY= # MINIO_ACCESS_KEY=
# MINIO_SECRET_KEY= # MINIO_SECRET_KEY=
# oo search — direct Elasticsearch access for name + content search. ES lives
# inside the OnlyOffice VM on localhost:9200; expose it with an SSH tunnel
# (see docs/elasticsearch.md). ONLYOFFICE_ES_INDEX defaults to files_file.
# ONLYOFFICE_ES_URL=http://127.0.0.1:9200
# ONLYOFFICE_ES_INDEX=files_file
# ONLYOFFICE_TENANT=
+1 -1
View File
@@ -1,3 +1,3 @@
{ {
".": "0.17.0" ".": "0.18.0"
} }
+17
View File
@@ -6,6 +6,23 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## Unreleased ## Unreleased
## [0.18.0](https://github.com/eSlider/go-onlyoffice/compare/v0.17.0...v0.18.0) (2026-09-04)
### Features
* **crm:** add UpdateContactName and CloseCRMTask helpers ([4d8af7a](https://github.com/eSlider/go-onlyoffice/commit/4d8af7a2fef1fcfd96ad913cefc6497e3389963b))
### Bug Fixes
* **crm:** deterministic sortBy=id in contact paged lists ([4d91726](https://github.com/eSlider/go-onlyoffice/commit/4d917261792203732ac739644c9c82124b2166eb))
### Documentation
* **funding:** eSlider support links (reverse-import GitHub e9c969a) ([d650a16](https://github.com/eSlider/go-onlyoffice/commit/d650a16a36037949505eb017c92bd3283e6e2f0f))
## [0.17.0](https://github.com/eSlider/go-onlyoffice/compare/v0.16.0...v0.17.0) (2026-08-31) ## [0.17.0](https://github.com/eSlider/go-onlyoffice/compare/v0.16.0...v0.17.0) (2026-08-31)
+20
View File
@@ -678,6 +678,25 @@ oo dav download 22881 --to ./copy.pdf # default path: ./<server title>
oo dav fileops # active move/copy operations (status polling) oo dav fileops # active move/copy operations (status polling)
``` ```
### Search (`oo search`)
Full-text search over the Documents index. The REST endpoint
`/api/2.0/files/@search/{query}` only searches file names in the database, so
`oo search` talks to the OnlyOffice **Elasticsearch** directly (index
`files_file`). Name search is default; `--content` also matches extracted
document text (`document.attachment.content`, Office formats only).
See [`docs/elasticsearch.md`](docs/elasticsearch.md) for the tunnel setup.
```bash
oo search "Rechnung" # names only
oo search "Mahngebühr" --content # names + document text
oo search "Rechnung" --folder 649 --limit 50
oo search "Rechnung" --json # shorthand for -o json
```
Requires `ONLYOFFICE_ES_URL` (plus optional `ONLYOFFICE_ES_INDEX`,
`ONLYOFFICE_TENANT`).
### Bulk tools (`cmd/`) ### Bulk tools (`cmd/`)
Small single-purpose binaries for bulk Documents work. All of them pace Small single-purpose binaries for bulk Documents work. All of them pace
@@ -714,6 +733,7 @@ kontolink IN.xlsx oo-index.tsv OUT.xlsx [FILE_ID] [AMOUNTS_TSV]
| `docs` | `tools`, `convert`, `optimize`, `ocr`, `hocr`, `as-md`, `put-md`, `put-txt`, `put-xlsx` | | `docs` | `tools`, `convert`, `optimize`, `ocr`, `hocr`, `as-md`, `put-md`, `put-txt`, `put-xlsx` |
| `catalog` | `match`, `merge`, `apply`, `scan-contacts`, `scan-projects`, `scan-thunderbird` | | `catalog` | `match`, `merge`, `apply`, `scan-contacts`, `scan-projects`, `scan-thunderbird` |
| `dav` | `ls`, `move`, `copy`, `mkdir`, `rename-file`, `rename-folder`, `download`, `fileops` | | `dav` | `ls`, `move`, `copy`, `mkdir`, `rename-file`, `rename-folder`, `download`, `fileops` |
| `search` | `QUERY` (`--content`, `--folder ID`, `--limit N`, `--json`) |
The CLI reads only `.env` from the current working directory (godotenv is a The CLI reads only `.env` from the current working directory (godotenv is a
CLI-only concern — the library itself never loads dotfiles). CLI-only concern — the library itself never loads dotfiles).
+27
View File
@@ -25,6 +25,7 @@ func davCmd() *cobra.Command {
cmd.AddCommand(davMoveCmd()) cmd.AddCommand(davMoveCmd())
cmd.AddCommand(davCopyCmd()) cmd.AddCommand(davCopyCmd())
cmd.AddCommand(davMkdirCmd()) cmd.AddCommand(davMkdirCmd())
cmd.AddCommand(davRemoveCmd())
cmd.AddCommand(davRenameFileCmd()) cmd.AddCommand(davRenameFileCmd())
cmd.AddCommand(davRenameFolderCmd()) cmd.AddCommand(davRenameFolderCmd())
cmd.AddCommand(davDownloadCmd()) cmd.AddCommand(davDownloadCmd())
@@ -184,6 +185,32 @@ func davMkdirCmd() *cobra.Command {
} }
} }
func davRemoveCmd() *cobra.Command {
var folderIDs []string
cmd := &cobra.Command{
Use: "rm [FILE_ID...]",
Aliases: []string{"delete"},
Short: "Permanently delete file(s) and/or folder(s) from Documents",
Args: cobra.ArbitraryArgs,
RunE: func(cmd *cobra.Command, args []string) error {
if len(args) == 0 && len(folderIDs) == 0 {
return fmt.Errorf("dav rm: give at least one FILE_ID or --folders")
}
c, err := newOO(cmd)
if err != nil {
return err
}
if err := c.DeleteDavItems(cmd.Context(), folderIDs, args); err != nil {
return err
}
printObject(map[string]any{"deleted_files": args, "deleted_folders": folderIDs})
return nil
},
}
cmd.Flags().StringSliceVar(&folderIDs, "folders", nil, "folder ids to delete")
return cmd
}
func davRenameFileCmd() *cobra.Command { func davRenameFileCmd() *cobra.Command {
return &cobra.Command{ return &cobra.Command{
Use: "rename-file FILE_ID NEW_TITLE", Use: "rename-file FILE_ID NEW_TITLE",
+1
View File
@@ -18,6 +18,7 @@
// oo docs tools | convert | optimize | ocr | hocr | as-md | put-md | put-txt | put-xlsx // 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 catalog match | merge | apply | scan-contacts | scan-projects | scan-thunderbird
// oo dav ls | move | copy | mkdir | rename-file | rename-folder | download | fileops // 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 // CRM association rules: docs/crm-associations.md
// //
+72
View File
@@ -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
}
+46
View File
@@ -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())
}
}
+197 -13
View File
@@ -18,7 +18,52 @@ import (
onlyoffice "github.com/eslider/go-onlyoffice" onlyoffice "github.com/eslider/go-onlyoffice"
) )
var amountRe = regexp.MustCompile(`(?i)(zu zahlender betrag|rechnungsbetrag)\s*[:\s]*([0-9][0-9.]*,[0-9]{2})`) // amountPat is the amount capture shared by every amount regex.
const amountPat = `([0-9]+(?:[.,][0-9]+)*)`
// amountRE builds "<label> [optional (comment)] [: -] <number>".
func amountRE(label string) *regexp.Regexp {
return regexp.MustCompile(
`(?i)\b` + regexp.QuoteMeta(label) + `\b\s*(?:\([^)]*\))?\s*[:\-]?\s*` + amountPat)
}
// amountRes lists the payable-amount patterns in strict priority order: the
// first pattern with a usable amount wins, and a lower-priority label can never
// override a higher-priority one ("zu zahlender betrag" > "rechnungsbetrag" >
// "rechnungsendbetrag" > "gesamtbetrag" > "gesamtsumme (inkl. steuern)").
//
// "gesamtbetrag" and "gesamtsumme" are not in the original set but are the real
// labels on Diashop invoices ("Gesamtsumme (inkl. Steuern)"). The inclusive
// variant is matched before a plain "gesamtsumme". Everything after those
// primary labels is the broader fallback set, consulted only when no primary
// label yields an amount. Within one pattern the last usable amount is taken,
// because totals usually come last.
var amountRes = []*regexp.Regexp{
amountRE("zu zahlender betrag"),
amountRE("rechnungsbetrag"),
amountRE("rechnungsendbetrag"),
amountRE("gesamtbetrag"),
regexp.MustCompile(`(?i)\bgesamtsumme\b\s*\(\s*inkl\.?\s*steuern\s*\)\s*[:\-]?\s*` + amountPat),
amountRE("gesamtsumme"),
amountRE("endbetrag"),
amountRE("zahlbetrag"),
amountRE("bruttobetrag"),
amountRE("betrag"),
amountRE("total"),
amountRE("summe"),
}
// taxLineRe marks a line whose number is a tax rate/percentage: an explicit
// percent sign or a VAT/tax keyword. "Steuern" (plural, as in "inkl. Steuern")
// is handled separately so the inclusive total stays usable.
var taxLineRe = regexp.MustCompile(`(?i)%|\bMwSt\b|\bUSt\b|\bProzent\b`)
// steuerRe finds "Steuer"/"Umsatzsteuer" etc. RE2 has no lookahead, so the
// plural "Steuern" is excluded in isTaxLine.
var steuerRe = regexp.MustCompile(`(?i)steuer`)
// percentAfterRe detects a percent sign directly after a number (spaces ok).
var percentAfterRe = regexp.MustCompile(`^\s*%`)
func main() { func main() {
if len(os.Args) < 2 { if len(os.Args) < 2 {
@@ -105,20 +150,159 @@ func pdfAmount(ctx context.Context, c *onlyoffice.Client, id string) (string, er
if err := cmd.Run(); err != nil { if err := cmd.Run(); err != nil {
return "", err return "", err
} }
m := amountRe.FindStringSubmatch(buf.String()) return extractAmount(buf.String()), nil
if m == nil {
return "", nil
}
return parseDe(m[2]), nil
} }
// parseDe turns "1.234,56" into 1234.56. // extractAmount returns the normalised ("1234.56") payable amount found in
func parseDe(s string) string { // text, or "" if no usable amount matches.
s = strings.ReplaceAll(s, ".", "") //
s = strings.ReplaceAll(s, ",", ".") // DKV invoices are special-cased first: they repeat a per-vehicle "TOTAL:" line
v, err := strconv.ParseFloat(s, 64) // and carry the real total only in the "Gesamtsummenaufstellung" section.
func extractAmount(text string) string {
if v, ok := dkvGrandTotal(text); ok {
return v
}
for _, re := range amountRes {
if v, ok := lastUsableAmount(text, re); ok {
return v
}
}
return ""
}
// dkvGrandTotal extracts the total of a DKV "Gesamtsummenaufstellung" section.
//
// Rule: DKV invoices repeat a per-vehicle "TOTAL:" line, so the last TOTAL is
// not the invoice total. When a "Gesamtsummenaufstellung" section exists, its
// total wins over every "TOTAL:" line: the first amount after the "»" marker,
// or, if there is none, the last amount in the section. The section ends at the
// page break (form feed) or end of text.
func dkvGrandTotal(text string) (string, bool) {
idx := strings.Index(strings.ToLower(text), "gesamtsummenaufstellung")
if idx < 0 {
return "", false
}
section := text[idx:]
if ff := strings.IndexByte(section, '\f'); ff >= 0 {
section = section[:ff]
}
if m := strings.Index(section, "»"); m >= 0 {
if v, ok := firstAmount(section[m:]); ok {
return v, true
}
}
return lastAmount(section)
}
// lastUsableAmount returns the last amount matched by re that is not a tax rate
// or percentage. Within one label the last usable amount wins.
func lastUsableAmount(text string, re *regexp.Regexp) (string, bool) {
ms := re.FindAllStringSubmatchIndex(text, -1)
for i := len(ms) - 1; i >= 0; i-- {
m := ms[i]
if isTaxRate(text, m[2], m[3]) {
continue
}
if v, ok := normalizeAmount(text[m[2]:m[3]]); ok {
return v, true
}
}
return "", false
}
// isTaxRate reports whether the number at text[start:end] is a tax rate or a
// percentage instead of a payable amount. A candidate is rejected when the
// token right after the number is "%" or the number's line carries a percent
// sign or a tax keyword. Rejecting is deliberate: office matching treats a
// known-but-different amount as a hard disqualifier, so an empty result is
// safer than the VAT rate.
func isTaxRate(text string, start, end int) bool {
if percentAfterRe.MatchString(text[end:]) {
return true
}
lineStart := strings.LastIndexByte(text[:start], '\n') + 1
line := text[lineStart:]
if n := strings.IndexByte(text[end:], '\n'); n >= 0 {
line = text[lineStart : end+n]
}
return isTaxLine(line)
}
// isTaxLine reports whether a line looks like a tax rate rather than a payable
// amount. "Steuern" is treated as a qualifier ("inkl. Steuern"), not a rate.
func isTaxLine(line string) bool {
if taxLineRe.MatchString(line) {
return true
}
for _, loc := range steuerRe.FindAllStringIndex(line, -1) {
if loc[1] >= len(line) || (line[loc[1]] != 'n' && line[loc[1]] != 'N') {
return true
}
}
return false
}
// numberRe finds bare numbers (with optional thousands/decimal separators).
var numberRe = regexp.MustCompile(`[0-9]+(?:[.,][0-9]+)*`)
func firstAmount(s string) (string, bool) {
for _, m := range numberRe.FindAllString(s, -1) {
if v, ok := normalizeAmount(m); ok {
return v, true
}
}
return "", false
}
func lastAmount(s string) (string, bool) {
ms := numberRe.FindAllString(s, -1)
for i := len(ms) - 1; i >= 0; i-- {
if v, ok := normalizeAmount(ms[i]); ok {
return v, true
}
}
return "", false
}
// normalizeAmount turns "1.234,56" (DE), "1,234.56" (EN) or "1234.56" into
// "1234.56". The rightmost separator is decimal only when followed by one or
// two digits; otherwise every separator is a thousands separator.
func normalizeAmount(s string) (string, bool) {
last := -1
for i := 0; i < len(s); i++ {
if s[i] == '.' || s[i] == ',' {
last = i
}
}
var dec byte
if last >= 0 {
digits := 0
for i := last + 1; i < len(s); i++ {
if s[i] < '0' || s[i] > '9' {
return "", false
}
digits++
}
if digits == 1 || digits == 2 {
dec = s[last]
}
}
var b strings.Builder
for i := 0; i < len(s); i++ {
switch c := s[i]; {
case c >= '0' && c <= '9':
b.WriteByte(c)
case (c == '.' || c == ',') && c == dec:
b.WriteByte('.')
case c == '.' || c == ',':
// thousands separator
default:
return "", false
}
}
v, err := strconv.ParseFloat(b.String(), 64)
if err != nil { if err != nil {
return s return "", false
} }
return strconv.FormatFloat(v, 'f', 2, 64) return strconv.FormatFloat(v, 'f', 2, 64), true
} }
+175
View File
@@ -0,0 +1,175 @@
package main
import "testing"
func TestExtractAmount(t *testing.T) {
tests := []struct {
name, text, want string
}{
{
name: "rechnungsbetrag de format",
text: "Rechnungsbetrag: 1.234,56 €",
want: "1234.56",
},
{
name: "rechnungsbetrag en thousands and dot",
text: "Rechnungsbetrag: 1,234.56",
want: "1234.56",
},
{
name: "rechnungsbetrag plain dot",
text: "Rechnungsbetrag: 1234.56",
want: "1234.56",
},
{
name: "rechnungsbetrag de comma only",
text: "Rechnungsbetrag: 1234,56",
want: "1234.56",
},
{
name: "currency suffix eur",
text: "Rechnungsbetrag: 1.234,56 EUR",
want: "1234.56",
},
{
name: "zu zahlender betrag wins over rechnungsbetrag",
text: "Zu zahlender Betrag: 10,00\nRechnungsbetrag: 99,00",
want: "10.00",
},
{
name: "rechnungsbetrag wins over endbetrag",
text: "Endbetrag: 20,00\nRechnungsbetrag: 30,00",
want: "30.00",
},
{
name: "bruttobetrag wins over bare betrag",
text: "Bruttobetrag: 50,00\nBetrag: 10,00",
want: "50.00",
},
{
name: "gesamtbetrag wins over bare betrag",
text: "Gesamtbetrag: 80,00\nBetrag: 10,00",
want: "80.00",
},
{
name: "last occurrence of same label wins",
text: "Rechnungsbetrag: 10,00\nRechnungsbetrag: 20,00",
want: "20.00",
},
{
name: "endbetrag fallback",
text: "Endbetrag: 42,00",
want: "42.00",
},
{
name: "zahlbetrag fallback without colon",
text: "Zahlbetrag 7,50 €",
want: "7.50",
},
{
name: "rechnungsendbetrag beats endbetrag",
text: "Rechnungsendbetrag: 12,00\nEndbetrag: 13,00",
want: "12.00",
},
{
name: "dkv style total line",
text: "Kundenbezogene Daten\n» TOTAL: 123,45 100,00 23,45 123,45\n",
want: "123.45",
},
{
name: "dkv gesamtsummenaufstellung grand total after marker",
text: "» TOTAL: 111,11 100,00 11,11 111,11\n" +
"» TOTAL: 222,22 200,00 22,22 222,22\n" +
"Gesamtsummenaufstellung\n" +
"Netto 240,00\n" +
"MwSt 47,25\n" +
"» 287,25\n",
want: "287.25",
},
{
name: "dkv gesamtsummenaufstellung total on next line",
text: "» TOTAL: 111,11\nGesamtsummenaufstellung\n»\n287,25\n",
want: "287.25",
},
{
name: "tax rate with percent sign is not an amount",
text: "Betrag: 19,00 % MwSt",
want: "",
},
{
name: "mehrwertsteuer rate is not an amount",
text: "Gesamtsumme: 19,00% MwSt",
want: "",
},
{
name: "steuer word on the number line rejects it",
text: "Betrag: 2,83 Steuer",
want: "",
},
{
name: "rejected primary falls back to a usable label",
text: "Gesamtsumme: 19,00 % MwSt\nEndbetrag: 42,00",
want: "42.00",
},
{
name: "labeled zu zahlender betrag beats unlabeled larger number",
text: "unlabeled 999,99\nZu zahlender Betrag: 10,00",
want: "10.00",
},
{
name: "labeled zu zahlender betrag beats lower label larger number",
text: "Endbetrag: 999,99\nZu zahlender Betrag: 10,00",
want: "10.00",
},
{
name: "diashop style gesamtsumme with comment",
text: "Zwischensumme\n12,34 €\nZwischensumme\n12,34 €\nVersand & Bearbeitung\n4,95 €\nGesamtsumme (inkl. Steuern)\n17,29 €\n",
want: "17.29",
},
{
name: "diashop picks inclusive total last",
text: "Gesamtsumme (exkl. Steuern)\n12,34 €\nGesamtsumme (inkl. Steuern)\n17,29 €",
want: "17.29",
},
{
name: "no label",
text: "some text without any amount label 12,34",
want: "",
},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
if got := extractAmount(tc.text); got != tc.want {
t.Fatalf("extractAmount()=%q want %q", got, tc.want)
}
})
}
}
func TestNormalizeAmount(t *testing.T) {
tests := []struct {
in string
want string
ok bool
}{
{"1.234,56", "1234.56", true},
{"1,234.56", "1234.56", true},
{"1234.56", "1234.56", true},
{"1234,56", "1234.56", true},
{"1.234.567,89", "1234567.89", true},
{"1,234,567.89", "1234567.89", true},
{"1.234", "1234.00", true},
{"12,5", "12.50", true},
{"12", "12.00", true},
{"", "0.00", false},
}
for _, tc := range tests {
got, ok := normalizeAmount(tc.in)
if ok != tc.ok {
t.Fatalf("normalizeAmount(%q) ok=%v want %v", tc.in, ok, tc.ok)
}
if ok && got != tc.want {
t.Fatalf("normalizeAmount(%q)=%q want %q", tc.in, got, tc.want)
}
}
}
+125
View File
@@ -0,0 +1,125 @@
---
type: reference
status: current
related:
- README.md
- file_es.go
---
# Elasticsearch — полнотекстовый поиск OnlyOffice
## Что это
Полнотекстовый поиск OnlyOffice Workspace работает на **Elasticsearch**.
Клиент на сервере — NEST. Индекс — имя таблицы.
Для файлов индекс `files_file`:
| поле | тип | смысл |
|------|-----|-------|
| `id` | integer | id файла (тот же, что в REST/Documents) |
| `title` | text (`whitespacecustom`) | имя файла |
| `tenantId` | integer | тенант (портал) |
| `folders` | nested | список папок: `folderId` (строка), `id`, `tenantId` |
| `document.attachment.content` | text (`document`) | извлеченный текст (ingest-attachment) |
| `document.attachment.content_type` | text | MIME |
Важно:
- Живой сервер — **Elasticsearch 7.16.3**, кластер `elasticsearch`.
- REST `GET /api/2.0/files/@search/{query}` ищет **только по имени в БД**
(`fileDao.Search`), ES не задействует. Для поиска по содержимому нужен
прямой ES — это и делает `oo search`.
- `title` analyzer `whitespacecustom` режет по пробелам и lower-case. Полное
имя файла — один токен (`Rechnung-4711.pdf`), поэтому поиск по имени ищет
слово целиком, а не подстроку.
- `document.attachment.content` заполняется **только для Office-форматов**
(docx / xlsx / pptx). У PDF/txt, залитых через API, контент не извлекается.
- Индексация асинхронная (TeamLabSvc) — файл появляется в ES не мгновенно.
## Доступ
ES слушает `127.0.0.1:9200` **внутри** VM OnlyOffice. Снаружи порт закрыт,
SSH в VM открыт на хосте как `127.0.0.1:32` (контейнер `onlyoffice-v2`,
QEMU). Схема — SSH-туннель.
```bash
# из корня go-onlyoffice (ключ и хост — как в infra-доках)
ssh -f -N -o ControlMaster=no -o ControlPath=none \
-p 32 -i ~/.ssh/id_ed25519 \
-L 9200:127.0.0.1:9200 root@127.0.0.1
curl -s http://127.0.0.1:9200/ | head # tagline + version
curl -s 'http://127.0.0.1:9200/_cat/indices?h=index,docs.count'
```
`-o ControlMaster=no -o ControlPath=none` обязательны: иначе forward уходит
в persistent master-соединение из `~/.ssh/config` и порт остаётся занят.
Проверить, что туннель жив:
```bash
curl -s http://127.0.0.1:9200/files_file/_count
```
## Переменные
| env | default | смысл |
|-----|---------|-------|
| `ONLYOFFICE_ES_URL` | — (обязателен) | `scheme://host:port` ES |
| `ONLYOFFICE_ES_INDEX` | `files_file` | индекс |
| `ONLYOFFICE_TENANT` | пусто (все) | фильтр `tenantId` |
Имена — в [`.env.example`](../.env.example). Секретов нет: ES без пароля.
## CLI
```bash
ONLYOFFICE_ES_URL=http://127.0.0.1:9200 oo search "Rechnung"
ONLYOFFICE_ES_URL=http://127.0.0.1:9200 oo search "Mahngebühr" --content
oo search "Rechnung" --folder 649 --limit 50 --json
```
Флаги: `--content` (искать и по тексту), `--folder ID` (папка
`folders.folderId`), `--limit N` (по умолчанию 20, максимум 200),
`--json` = `-o json`.
## Библиотека
`file_es.go` — `ESSearcher` (`Name() = "elasticsearch"`), прямой ES REST на
stdlib `net/http`:
```go
es, _ := onlyoffice.NewESSearcher(onlyoffice.ESConfigFromEnv())
hits, _ := es.Search(ctx, onlyoffice.SearchQuery{
Text: "Rechnung", InContent: true, Limit: 20,
})
```
Запрос: `multi_match` по `title^2` (+ `document.attachment.content` при
`InContent`), фильтры `tenantId` и `folders.folderId`, `_source`
id/title/folders, `highlight` для фрагмента. Ответ → `[]SearchHit` (модель из
эпика #34; пока объявлена в `file_es.go`, переедет в `file_core.go` с F1 #35).
## Тесты
```bash
# unit — чистые builders/парсеры, без сети
go test ./ -run ES
# integration — нужен ONLYOFFICE_ES_URL (+ креды REST для залива)
set -a; . .env; set +a
ONLYOFFICE_ES_URL=http://127.0.0.1:9200 ONLYOFFICE_TENANT=1 \
go test -tags=integration -run TestIntegrationESSearch -v .
```
Интеграционный тест заливает временный xlsx (в имени и в ячейке — уникальные
токены), ждёт индексации, проверяет поиск по имени и по содержимому, затем
удаляет проект.
## Грабли
- `locale`/версия ES: 7.16.3, `_search` совместим с REST 7.x.
- ES без auth и слушает только localhost — туннель обязателен.
- Фильтр `tenantId` сузит выдачу; без него видны документы всех тенантов.
- Поиск по содержимому PDF, залитых через API, не работает (нет
`attachment.content`) — только Office-форматы.
+199
View File
@@ -0,0 +1,199 @@
package onlyoffice
// Canonical file model and the backend-agnostic store interface. REST
// (files.go), WebDAV (files_webdav.go) and future backends (PostgreSQL,
// Elasticsearch) implement FileStore/Searcher so callers stop depending on a
// concrete transport. This file holds only types and pure conversions — no IO.
import (
"context"
"io"
"mime"
"path/filepath"
"strconv"
"strings"
"time"
)
// Kind distinguishes files from folders in the canonical model.
type Kind int
const (
File Kind = iota
Folder
)
// String renders the kind for logs and table output.
func (k Kind) String() string {
switch k {
case File:
return "file"
case Folder:
return "folder"
default:
return "unknown"
}
}
// Provider names for the FileStore adapters.
const (
ProviderREST = "rest"
ProviderDAV = "dav"
)
// Entry is the backend-independent representation of a document or folder.
// Fields that a backend cannot supply stay at their zero value.
type Entry struct {
ID string
ParentID string
Title string
Kind Kind
Size int64
MIME string
Created time.Time
Modified time.Time
Version int
Provider string
}
// FileStore is the operation surface every file backend implements.
type FileStore interface {
Name() string
List(ctx context.Context, parentID string) ([]Entry, error)
Stat(ctx context.Context, id string) (Entry, error)
CreateFolder(ctx context.Context, parentID, title string) (Entry, error)
Upload(ctx context.Context, parentID, title string, r io.Reader) (Entry, error)
Download(ctx context.Context, id string, w io.Writer) (int64, error)
Move(ctx context.Context, ids []string, parentID string) error
Copy(ctx context.Context, ids []string, parentID string) error
Rename(ctx context.Context, id, title string) error
Delete(ctx context.Context, ids []string) error
}
// SearchQuery narrows a Searcher request. InContent asks the backend to match
// document bodies, not just titles.
type SearchQuery struct {
Text string
InContent bool
FolderID string
Extensions []string
Limit int
}
// SearchHit is one Searcher result: the matching entry plus backend-specific
// ranking metadata.
type SearchHit struct {
Entry
Score float64
Highlight string
Path []string
}
// Searcher is the optional content/name search surface. Only some backends
// (for example Elasticsearch) provide it.
type Searcher interface {
Search(ctx context.Context, q SearchQuery) ([]SearchHit, error)
Name() string
}
// FileStore returns the adapter for a backend name: ProviderREST (default) or
// ProviderDAV. Unknown or empty names select the REST backend. The full facade
// (backend composition) is deliberately left to a later change.
func (c *Client) FileStore(backend string) FileStore {
switch strings.ToLower(strings.TrimSpace(backend)) {
case ProviderDAV, "webdav":
return &davStore{c: c}
default:
return &restStore{c: c}
}
}
// Files returns the default (REST) file store.
func (c *Client) Files() FileStore { return c.FileStore(ProviderREST) }
// retryStoreOp runs one store operation under the shared deterministic
// transient-error policy (429/502/503/504).
func retryStoreOp(ctx context.Context, fn func() error) error {
return DoRetry(ctx, DefaultRetryPolicy(), fn)
}
// FileEntryToEntry converts a Files-module file row to the canonical model.
func FileEntryToEntry(f *FileEntry, provider string) Entry {
e := Entry{Kind: File, Provider: provider}
if f == nil {
return e
}
if f.ID != nil {
e.ID = f.ID.String()
}
e.ParentID = FileFolderID(f)
if f.Title != nil {
e.Title = *f.Title
}
if f.ContentLength != nil {
e.Size = parseContentLength(*f.ContentLength)
}
exst := ""
if f.FileExst != nil {
exst = *f.FileExst
}
e.MIME = mimeForTitle(e.Title, exst)
if f.Updated != nil {
e.Modified = *f.Updated
}
return e
}
// DavFileToEntry converts a WebDAV file row to the canonical model.
func DavFileToEntry(f DavFile, provider string) Entry {
return Entry{
ID: f.ID,
Title: f.Title,
Kind: File,
Size: f.Size,
MIME: mimeForTitle(f.Title, ""),
Modified: f.ModTime(),
Provider: provider,
}
}
// DavFolderToEntry converts a WebDAV folder row to the canonical model.
func DavFolderToEntry(f DavFolder, provider string) Entry {
return Entry{
ID: f.ID,
ParentID: f.ParentID,
Title: f.Title,
Kind: Folder,
Modified: f.ModTime(),
Provider: provider,
}
}
// parseContentLength reads the leading integer of an OnlyOffice contentLength
// string (the API sometimes appends a unit, e.g. "12345 b").
func parseContentLength(s string) int64 {
fields := strings.Fields(s)
if len(fields) == 0 {
return 0
}
n, err := strconv.ParseInt(fields[0], 10, 64)
if err != nil {
return 0
}
return n
}
// mimeForTitle derives a MIME type from an explicit extension or the title.
func mimeForTitle(title, exst string) string {
ext := strings.TrimSpace(exst)
if ext == "" {
ext = filepath.Ext(title)
}
if ext == "" {
return ""
}
if !strings.HasPrefix(ext, ".") {
ext = "." + ext
}
return mime.TypeByExtension(strings.ToLower(ext))
}
+193
View File
@@ -0,0 +1,193 @@
package onlyoffice
import (
"encoding/json"
"strings"
"testing"
"time"
)
func TestFileEntryToEntry(t *testing.T) {
id := json.Number("42")
title := "invoice.pdf"
exst := ".pdf"
size := "12345"
parent := json.Number("7")
updated := time.Date(2026, 1, 2, 3, 4, 5, 0, time.UTC)
f := &FileEntry{
ID: &id,
Title: &title,
FileExst: &exst,
ContentLength: &size,
FolderID: &parent,
Updated: &updated,
}
e := FileEntryToEntry(f, ProviderREST)
if e.ID != "42" {
t.Errorf("ID = %q, want 42", e.ID)
}
if e.ParentID != "7" {
t.Errorf("ParentID = %q, want 7", e.ParentID)
}
if e.Title != title {
t.Errorf("Title = %q, want %q", e.Title, title)
}
if e.Kind != File {
t.Errorf("Kind = %v, want file", e.Kind)
}
if e.Size != 12345 {
t.Errorf("Size = %d, want 12345", e.Size)
}
if e.MIME != "application/pdf" {
t.Errorf("MIME = %q, want application/pdf", e.MIME)
}
if !e.Modified.Equal(updated) {
t.Errorf("Modified = %v, want %v", e.Modified, updated)
}
if e.Provider != ProviderREST {
t.Errorf("Provider = %q, want %q", e.Provider, ProviderREST)
}
}
func TestFileEntryToEntryNil(t *testing.T) {
e := FileEntryToEntry(nil, ProviderDAV)
if e.Kind != File {
t.Errorf("Kind = %v, want file", e.Kind)
}
if e.ID != "" || e.Title != "" {
t.Errorf("nil entry should be empty: %+v", e)
}
if e.Provider != ProviderDAV {
t.Errorf("Provider = %q, want %q", e.Provider, ProviderDAV)
}
}
func TestFileEntryToEntrySizeFormats(t *testing.T) {
cases := map[string]int64{
"12345": 12345,
"12345 b": 12345,
"0": 0,
"": 0,
"notanum": 0,
}
for in, want := range cases {
got := parseContentLength(in)
if got != want {
t.Errorf("parseContentLength(%q) = %d, want %d", in, got, want)
}
}
}
func TestDavFileToEntry(t *testing.T) {
f := DavFile{
ID: "9",
Title: "note.txt",
Size: 10,
Updated: "2026-01-02T03:04:05.0000000+01:00",
}
e := DavFileToEntry(f, ProviderDAV)
if e.ID != "9" || e.Title != "note.txt" {
t.Errorf("identity mismatch: %+v", e)
}
if e.Kind != File {
t.Errorf("Kind = %v, want file", e.Kind)
}
if e.Size != 10 {
t.Errorf("Size = %d, want 10", e.Size)
}
if !strings.HasPrefix(e.MIME, "text/plain") {
t.Errorf("MIME = %q, want text/plain*", e.MIME)
}
if e.Modified.IsZero() {
t.Error("Modified not parsed")
}
if e.Provider != ProviderDAV {
t.Errorf("Provider = %q, want %q", e.Provider, ProviderDAV)
}
}
func TestDavFolderToEntry(t *testing.T) {
f := DavFolder{
ID: "5",
Title: "inbox",
ParentID: "1",
Updated: "2026-01-02T03:04:05.0000000+01:00",
}
e := DavFolderToEntry(f, ProviderDAV)
if e.ID != "5" || e.Title != "inbox" || e.ParentID != "1" {
t.Errorf("identity mismatch: %+v", e)
}
if e.Kind != Folder {
t.Errorf("Kind = %v, want folder", e.Kind)
}
if e.MIME != "" {
t.Errorf("folder MIME = %q, want empty", e.MIME)
}
if e.Modified.IsZero() {
t.Error("Modified not parsed")
}
}
func TestEntriesFromFolderMap(t *testing.T) {
m := map[string]any{
"files": []any{
map[string]any{"id": float64(42), "title": "a.pdf", "pureContentLength": float64(7)},
},
"folders": []any{
map[string]any{"id": float64(7), "title": "sub", "parentId": float64(1)},
},
}
entries, err := entriesFromFolderMap(m, ProviderREST)
if err != nil {
t.Fatalf("entriesFromFolderMap: %v", err)
}
if len(entries) != 2 {
t.Fatalf("got %d entries, want 2: %+v", len(entries), entries)
}
byID := map[string]Entry{}
for _, e := range entries {
byID[e.ID] = e
}
if got := byID["42"]; got.Kind != File || got.Size != 7 || got.Title != "a.pdf" {
t.Errorf("file entry = %+v", got)
}
if got := byID["7"]; got.Kind != Folder || got.ParentID != "1" || got.Title != "sub" {
t.Errorf("folder entry = %+v", got)
}
}
func TestEntriesFromFolderMapNil(t *testing.T) {
entries, err := entriesFromFolderMap(nil, ProviderREST)
if err != nil || entries != nil {
t.Fatalf("got %v, %v; want nil, nil", entries, err)
}
}
func TestKindString(t *testing.T) {
if File.String() != "file" || Folder.String() != "folder" {
t.Errorf("kind strings: %q %q", File.String(), Folder.String())
}
if Kind(9).String() != "unknown" {
t.Errorf("unknown kind = %q", Kind(9).String())
}
}
func TestClientFileStoreSelection(t *testing.T) {
c := NewClient(Credentials{})
if got := c.FileStore(ProviderDAV).Name(); got != ProviderDAV {
t.Errorf("FileStore(dav).Name() = %q", got)
}
if got := c.FileStore("webdav").Name(); got != ProviderDAV {
t.Errorf("FileStore(webdav).Name() = %q", got)
}
if got := c.FileStore(ProviderREST).Name(); got != ProviderREST {
t.Errorf("FileStore(rest).Name() = %q", got)
}
if got := c.FileStore("").Name(); got != ProviderREST {
t.Errorf("FileStore(\"\").Name() = %q", got)
}
if got := c.Files().Name(); got != ProviderREST {
t.Errorf("Files().Name() = %q", got)
}
}
+189
View File
@@ -0,0 +1,189 @@
package onlyoffice
// davStore implements FileStore on top of the Documents/WebDAV methods in
// files_webdav.go. The Documents fileops calls need folder and file ids
// separated, so ids are classified through Stat before move/copy/rename/delete.
import (
"bytes"
"context"
"fmt"
"io"
)
// davStore is a FileStore over the WebDAV-oriented Documents API.
type davStore struct{ c *Client }
// Name reports the backend name.
func (s *davStore) Name() string { return ProviderDAV }
// List returns the files and folders directly below parentID.
func (s *davStore) List(ctx context.Context, parentID string) ([]Entry, error) {
var out []Entry
err := retryStoreOp(ctx, func() error {
l, err := s.c.ListDavFolder(ctx, parentID)
if err != nil {
return err
}
entries := make([]Entry, 0, len(l.Folders)+len(l.Files))
for _, f := range l.Folders {
entries = append(entries, DavFolderToEntry(f, ProviderDAV))
}
for _, f := range l.Files {
entries = append(entries, DavFileToEntry(f, ProviderDAV))
}
out = entries
return nil
})
return out, err
}
// Stat resolves a folder or file entry by id. A folder answers ListDavFolder
// with its own metadata in Current; otherwise the file metadata API is used.
func (s *davStore) Stat(ctx context.Context, id string) (Entry, error) {
return s.stat(ctx, id)
}
// CreateFolder creates a subfolder under parentID.
func (s *davStore) CreateFolder(ctx context.Context, parentID, title string) (Entry, error) {
var out Entry
err := retryStoreOp(ctx, func() error {
f, err := s.c.CreateDavFolder(ctx, parentID, title)
if err != nil {
return err
}
if f == nil {
return fmt.Errorf("onlyoffice: dav store: empty create-folder response")
}
out = DavFolderToEntry(*f, ProviderDAV)
return nil
})
return out, err
}
// Upload streams r into parentID as title. The reader is buffered once so a
// retry re-sends the same bytes instead of an exhausted stream.
func (s *davStore) Upload(ctx context.Context, parentID, title string, r io.Reader) (Entry, error) {
data, err := io.ReadAll(r)
if err != nil {
return Entry{}, err
}
var out Entry
err = retryStoreOp(ctx, func() error {
f, err := s.c.UploadDavFile(ctx, parentID, title, bytes.NewReader(data))
if err != nil {
return err
}
if f == nil {
return fmt.Errorf("onlyoffice: dav store: empty upload response")
}
out = DavFileToEntry(*f, ProviderDAV)
return nil
})
return out, err
}
// Download streams the file bytes into w.
func (s *davStore) Download(ctx context.Context, id string, w io.Writer) (int64, error) {
var n int64
err := retryStoreOp(ctx, func() error {
var e error
n, e = s.c.DownloadDavFile(ctx, id, w)
return e
})
return n, err
}
// Move moves ids into parentID, splitting folders from files.
func (s *davStore) Move(ctx context.Context, ids []string, parentID string) error {
folders, files, err := s.split(ctx, ids)
if err != nil {
return err
}
if len(folders) == 0 && len(files) == 0 {
return nil
}
return retryStoreOp(ctx, func() error {
return s.c.MoveDavItems(ctx, folders, files, parentID)
})
}
// Copy copies ids into parentID, splitting folders from files.
func (s *davStore) Copy(ctx context.Context, ids []string, parentID string) error {
folders, files, err := s.split(ctx, ids)
if err != nil {
return err
}
if len(folders) == 0 && len(files) == 0 {
return nil
}
return retryStoreOp(ctx, func() error {
return s.c.CopyDavItems(ctx, folders, files, parentID)
})
}
// Rename renames a folder or file.
func (s *davStore) Rename(ctx context.Context, id, title string) error {
e, err := s.stat(ctx, id)
if err != nil {
return err
}
return retryStoreOp(ctx, func() error {
if e.Kind == Folder {
return s.c.RenameDavFolder(ctx, id, title)
}
return s.c.RenameDavFile(ctx, id, title)
})
}
// Delete removes ids, splitting folders from files.
func (s *davStore) Delete(ctx context.Context, ids []string) error {
folders, files, err := s.split(ctx, ids)
if err != nil {
return err
}
if len(folders) == 0 && len(files) == 0 {
return nil
}
return retryStoreOp(ctx, func() error {
return s.c.DeleteDavItems(ctx, folders, files)
})
}
// stat resolves a single id to a folder or file Entry.
func (s *davStore) stat(ctx context.Context, id string) (Entry, error) {
var out Entry
err := retryStoreOp(ctx, func() error {
if l, err := s.c.ListDavFolder(ctx, id); err == nil {
if l != nil && l.Current.ID != "" && l.Current.ID == id {
out = DavFolderToEntry(l.Current, ProviderDAV)
return nil
}
} else if Transient(err) {
return err
}
f, err := s.c.GetFile(ctx, id)
if err != nil {
return err
}
out = FileEntryToEntry(f, ProviderDAV)
return nil
})
return out, err
}
// split classifies ids into folder and file id lists.
func (s *davStore) split(ctx context.Context, ids []string) (folders, files []string, err error) {
for _, id := range ids {
e, err := s.stat(ctx, id)
if err != nil {
return nil, nil, err
}
if e.Kind == Folder {
folders = append(folders, id)
} else {
files = append(files, id)
}
}
return folders, files, nil
}
+285
View File
@@ -0,0 +1,285 @@
package onlyoffice
// Elasticsearch backend of the unified file client (epic #34, F3 #37).
//
// OnlyOffice full-text search runs on Elasticsearch (index `files_file`, NEST
// client on the server). The REST endpoint GET /api/2.0/files/@search/{query}
// only searches file names in the database, so content search needs a direct
// ES query. The live server is Elasticsearch 7.16.3; the request shape below
// is plain REST and stays stdlib-only, matching the repo's no-extra-deps rule.
import (
"bytes"
"context"
"encoding/json"
"fmt"
"io"
"net/http"
"os"
"regexp"
"strconv"
"strings"
"time"
)
// The canonical model (Kind, Entry, SearchQuery, SearchHit, Searcher) lives in
// file_core.go (F1 #35).
const (
defaultESIndex = "files_file"
defaultESLimit = 20
maxESLimit = 200
maxESResponseSize = 8 << 20
)
// ESConfig configures the direct Elasticsearch searcher.
type ESConfig struct {
URL string // scheme://host:port of the ES HTTP endpoint
Index string // index name, default files_file
Tenant string // tenantId filter, empty means all tenants
}
// ESConfigFromEnv reads ONLYOFFICE_ES_URL, ONLYOFFICE_ES_INDEX (default
// files_file) and ONLYOFFICE_TENANT. The library never loads dotfiles — the
// CLI does that.
func ESConfigFromEnv() ESConfig {
return ESConfig{
URL: strings.TrimRight(strings.TrimSpace(os.Getenv("ONLYOFFICE_ES_URL")), "/"),
Index: firstNonEmpty(os.Getenv("ONLYOFFICE_ES_INDEX"), defaultESIndex),
Tenant: strings.TrimSpace(os.Getenv("ONLYOFFICE_TENANT")),
}
}
// ESSearcher queries OnlyOffice's Elasticsearch index directly for file name
// and document content.
type ESSearcher struct {
cfg ESConfig
http *http.Client
}
// NewESSearcher returns a searcher for the OnlyOffice Elasticsearch index.
// The URL is required; an empty index falls back to files_file.
func NewESSearcher(cfg ESConfig) (*ESSearcher, error) {
if strings.TrimSpace(cfg.URL) == "" {
return nil, fmt.Errorf("onlyoffice: elasticsearch URL is empty (set ONLYOFFICE_ES_URL)")
}
cfg.URL = strings.TrimRight(cfg.URL, "/")
if cfg.Index == "" {
cfg.Index = defaultESIndex
}
return &ESSearcher{cfg: cfg, http: &http.Client{Timeout: 30 * time.Second}}, nil
}
// Name implements Searcher.
func (s *ESSearcher) Name() string { return "elasticsearch" }
// Search runs a multi_match over title (and, when q.InContent is set,
// document.attachment.content), filtered by tenant and optional folder.
func (s *ESSearcher) Search(ctx context.Context, q SearchQuery) ([]SearchHit, error) {
q.Text = strings.TrimSpace(q.Text)
if q.Text == "" {
return nil, fmt.Errorf("onlyoffice: empty search query")
}
body, err := json.Marshal(esSearchRequest(q, s.cfg.Tenant))
if err != nil {
return nil, fmt.Errorf("onlyoffice: build elasticsearch query: %w", err)
}
endpoint := s.cfg.URL + "/" + s.cfg.Index + "/_search"
req, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewReader(body))
if err != nil {
return nil, err
}
req.Header.Set("Content-Type", "application/json")
req.Header.Set("Accept", "application/json")
resp, err := s.http.Do(req)
if err != nil {
return nil, fmt.Errorf("onlyoffice: elasticsearch search: %w", err)
}
defer resp.Body.Close()
raw, err := io.ReadAll(io.LimitReader(resp.Body, maxESResponseSize))
if err != nil {
return nil, err
}
if resp.StatusCode >= 400 {
return nil, fmt.Errorf("onlyoffice: elasticsearch search: %d %s", resp.StatusCode, truncate(string(raw), 400))
}
return parseESSearchResponse(raw)
}
// esSearchRequest builds the ES query body. Pure, so it is unit-tested.
func esSearchRequest(q SearchQuery, tenant string) esRequest {
limit := q.Limit
if limit <= 0 {
limit = defaultESLimit
}
if limit > maxESLimit {
limit = maxESLimit
}
fields := []string{"title^2"}
if q.InContent {
fields = append(fields, "document.attachment.content")
}
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}})
}
for _, ext := range normalizeExtensions(q.Extensions) {
filter = append(filter, esClause{Wildcard: map[string]any{"title": "*." + ext}})
}
highlightFields := map[string]struct{}{"title": {}}
if q.InContent {
highlightFields["document.attachment.content"] = struct{}{}
}
return esRequest{
Size: limit,
Source: []string{"id", "title", "folders"},
Query: esQuery{Bool: esBool{Must: must, Filter: filter}},
Highlight: esHighlight{PreTags: []string{"<em>"}, PostTags: []string{"</em>"}, Fields: highlightFields},
}
}
// normalizeExtensions lowercases, trims leading dots and drops empties.
func normalizeExtensions(exts []string) []string {
out := make([]string, 0, len(exts))
seen := map[string]bool{}
for _, e := range exts {
e = strings.ToLower(strings.TrimSpace(strings.TrimPrefix(strings.TrimSpace(e), ".")))
if e == "" || seen[e] {
continue
}
seen[e] = true
out = append(out, e)
}
return out
}
// numericOrString keeps an integer-looking filter value numeric (tenantId is
// a long) and leaves anything else as a string (folderId is a text token).
func numericOrString(s string) any {
if n, err := strconv.ParseInt(s, 10, 64); err == nil {
return n
}
return s
}
// esRequest is the subset of the ES query DSL this client emits.
type esRequest struct {
Size int `json:"size"`
Source []string `json:"_source"`
Query esQuery `json:"query"`
Highlight esHighlight `json:"highlight"`
}
type esQuery struct {
Bool esBool `json:"bool"`
}
type esBool struct {
Must []esClause `json:"must,omitempty"`
Filter []esClause `json:"filter,omitempty"`
}
type esClause struct {
MultiMatch *esMultiMatch `json:"multi_match,omitempty"`
Term map[string]any `json:"term,omitempty"`
Wildcard map[string]any `json:"wildcard,omitempty"`
}
type esMultiMatch struct {
Query string `json:"query"`
Fields []string `json:"fields"`
}
type esHighlight struct {
PreTags []string `json:"pre_tags,omitempty"`
PostTags []string `json:"post_tags,omitempty"`
Fields map[string]struct{} `json:"fields"`
}
// esResponse is the subset of an ES search response we consume.
type esResponse struct {
Took int `json:"took"`
Hits struct {
Total struct {
Value int `json:"value"`
Relation string `json:"relation"`
} `json:"total"`
Hits []esResponseHit `json:"hits"`
} `json:"hits"`
}
type esResponseHit struct {
ID string `json:"_id"`
Score float64 `json:"_score"`
Source struct {
ID int `json:"id"`
Title string `json:"title"`
Folders []struct {
FolderID string `json:"folderId"`
ID int `json:"id"`
} `json:"folders"`
} `json:"_source"`
Highlight map[string][]string `json:"highlight"`
}
// parseESSearchResponse converts an ES search response into SearchHit values.
// Pure, so it is unit-tested.
func parseESSearchResponse(raw []byte) ([]SearchHit, error) {
var r esResponse
if err := json.Unmarshal(raw, &r); err != nil {
return nil, fmt.Errorf("onlyoffice: decode elasticsearch response: %w", err)
}
hits := make([]SearchHit, 0, len(r.Hits.Hits))
for _, h := range r.Hits.Hits {
id := strconv.Itoa(h.Source.ID)
if h.Source.ID == 0 {
id = h.ID
}
var parent string
path := make([]string, 0, len(h.Source.Folders))
for i, f := range h.Source.Folders {
path = append(path, f.FolderID)
if i == 0 {
parent = f.FolderID
}
}
hits = append(hits, SearchHit{
Entry: Entry{
ID: id,
ParentID: parent,
Title: h.Source.Title,
Kind: File,
Provider: "elasticsearch",
},
Score: h.Score,
Highlight: esHighlightText(h.Highlight),
Path: path,
})
}
return hits, nil
}
var esHighlightTag = regexp.MustCompile(`</?em[^>]*>`)
// esHighlightText flattens a highlight map into one plain-text snippet,
// preferring the content fragment over the title.
func esHighlightText(hl map[string][]string) string {
for _, key := range []string{"document.attachment.content", "title"} {
frags := hl[key]
if len(frags) == 0 {
continue
}
clean := make([]string, 0, len(frags))
for _, f := range frags {
clean = append(clean, esHighlightTag.ReplaceAllString(f, ""))
}
return strings.Join(clean, " … ")
}
return ""
}
+135
View File
@@ -0,0 +1,135 @@
//go:build integration
package onlyoffice
import (
"context"
"os"
"path/filepath"
"strconv"
"strings"
"testing"
"time"
"github.com/xuri/excelize/v2"
)
// TestIntegrationESSearch uploads a throwaway workbook and verifies that the
// direct Elasticsearch search finds it by file name and by content.
//
// The content index (document.attachment.content) is only populated for Office
// formats (docx/xlsx/pptx), so the fixture is an xlsx whose cell carries a
// unique token. Requires ONLYOFFICE_ES_URL (a reachable ES endpoint — in the
// current setup a tunnel to the ES inside the OnlyOffice VM, see
// docs/elasticsearch.md) plus the regular REST credentials for the upload.
// Skips when either is missing.
func TestIntegrationESSearch(t *testing.T) {
esURL := strings.TrimSpace(os.Getenv("ONLYOFFICE_ES_URL"))
if esURL == "" {
t.Skip("ONLYOFFICE_ES_URL not set — skipping Elasticsearch integration test")
}
c := liveClient(t)
t.Cleanup(func() { cleanupTestProjects(t, c) })
stamp := time.Now().UTC().Format("20060102-150405")
nameToken := "goesname" + stamp
contentToken := "goescontent" + stamp
ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
defer cancel()
project, err := c.CreateProject(NewProjectRequest{
Title: testProjectPrefix + "es-" + stamp,
Description: "go-onlyoffice elasticsearch integration",
})
if err != nil {
t.Fatalf("CreateProject: %v", err)
}
if project.ID == nil {
t.Fatal("created project without id")
}
pid := strconv.Itoa(*project.ID)
title := nameToken + ".xlsx"
localPath := filepath.Join(t.TempDir(), title)
book := excelize.NewFile()
if err := book.SetCellValue("Sheet1", "A1", "OnlyOffice Elasticsearch content fixture "+contentToken); err != nil {
t.Fatalf("SetCellValue: %v", err)
}
if err := book.SaveAs(localPath); err != nil {
t.Fatalf("SaveAs: %v", err)
}
entry, err := c.UploadProjectFile(ctx, pid, localPath)
if err != nil {
t.Fatalf("UploadProjectFile: %v", err)
}
fileID := strconv.Itoa(int(FileEntryNumericID(entry)))
if fileID == "0" {
t.Fatalf("upload returned no file id: %+v", entry)
}
es, err := NewESSearcher(ESConfig{
URL: esURL,
Index: os.Getenv("ONLYOFFICE_ES_INDEX"),
Tenant: os.Getenv("ONLYOFFICE_TENANT"),
})
if err != nil {
t.Fatalf("NewESSearcher: %v", err)
}
// Indexing is asynchronous on the server; poll until the file shows up.
// The server's title analyzer splits on whitespace, so the name query is
// the full file name token (including extension), as a user would type it.
nameHit := waitForHit(t, ctx, es, SearchQuery{Text: title}, fileID)
if nameHit.Title != title {
t.Errorf("name hit title = %q, want %q", nameHit.Title, title)
}
contentHit := waitForHit(t, ctx, es, SearchQuery{Text: contentToken, InContent: true}, fileID)
if contentHit.Highlight == "" {
t.Error("content hit has no highlight fragment")
}
if !strings.Contains(contentHit.Title, nameToken) {
t.Errorf("content hit title = %q, want the uploaded workbook", contentHit.Title)
}
// The content token is absent from the title, so a name-only search must
// not return the file — this proves the content field is really queried.
if hits := searchQuiet(t, es, SearchQuery{Text: contentToken}); len(hits) != 0 {
t.Errorf("name-only search for content token returned %d hits, want 0", len(hits))
}
}
// waitForHit polls ES until the file with fileID appears and returns that hit.
func waitForHit(t *testing.T, ctx context.Context, s *ESSearcher, q SearchQuery, fileID string) SearchHit {
t.Helper()
var lastErr error
for {
hits, err := s.Search(ctx, q)
if err != nil {
lastErr = err
} else {
for _, h := range hits {
if h.ID == fileID {
return h
}
}
}
select {
case <-ctx.Done():
t.Fatalf("search %q: file %s not indexed in time (last err: %v)", q.Text, fileID, lastErr)
case <-time.After(3 * time.Second):
}
}
}
func searchQuiet(t *testing.T, s *ESSearcher, q SearchQuery) []SearchHit {
t.Helper()
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()
hits, err := s.Search(ctx, q)
if err != nil {
t.Fatalf("Search(%q): %v", q.Text, err)
}
return hits
}
+188
View File
@@ -0,0 +1,188 @@
package onlyoffice
import (
"encoding/json"
"reflect"
"testing"
)
func TestESSearchRequestNameOnly(t *testing.T) {
got := esSearchRequest(SearchQuery{Text: "Rechnung"}, "1")
if got.Size != defaultESLimit {
t.Errorf("size = %d, want %d", got.Size, defaultESLimit)
}
if !reflect.DeepEqual(got.Source, []string{"id", "title", "folders"}) {
t.Errorf("_source = %v", got.Source)
}
if len(got.Query.Bool.Must) != 1 || got.Query.Bool.Must[0].MultiMatch == nil {
t.Fatalf("must = %+v, want one multi_match", got.Query.Bool.Must)
}
mm := got.Query.Bool.Must[0].MultiMatch
if mm.Query != "Rechnung" {
t.Errorf("query = %q", mm.Query)
}
if !reflect.DeepEqual(mm.Fields, []string{"title^2"}) {
t.Errorf("fields = %v, want title only", mm.Fields)
}
if _, ok := got.Highlight.Fields["document.attachment.content"]; ok {
t.Error("content highlight present without InContent")
}
if _, ok := got.Highlight.Fields["title"]; !ok {
t.Error("title highlight missing")
}
if len(got.Query.Bool.Filter) != 1 || got.Query.Bool.Filter[0].Term["tenantId"] != int64(1) {
t.Errorf("tenant filter = %+v, want numeric tenantId=1", got.Query.Bool.Filter)
}
}
func TestESSearchRequestContentFields(t *testing.T) {
got := esSearchRequest(SearchQuery{Text: "Mahnung", InContent: true}, "")
mm := got.Query.Bool.Must[0].MultiMatch
want := []string{"title^2", "document.attachment.content"}
if !reflect.DeepEqual(mm.Fields, want) {
t.Errorf("fields = %v, want %v", mm.Fields, want)
}
if _, ok := got.Highlight.Fields["document.attachment.content"]; !ok {
t.Error("content highlight missing with InContent")
}
if len(got.Query.Bool.Filter) != 0 {
t.Errorf("filter = %+v, want none without tenant/folder", got.Query.Bool.Filter)
}
}
func TestESSearchRequestFiltersAndLimit(t *testing.T) {
got := esSearchRequest(SearchQuery{
Text: "Storchen",
FolderID: "649",
Extensions: []string{".PDF", "pdf", "docx"},
Limit: 999,
}, "42")
if got.Size != maxESLimit {
t.Errorf("size = %d, want cap %d", got.Size, maxESLimit)
}
var tenant, folder, wildcards int
for _, f := range got.Query.Bool.Filter {
switch {
case f.Term != nil && f.Term["tenantId"] != nil:
tenant++
case f.Term != nil && f.Term["folders.folderId"] != nil:
folder++
if f.Term["folders.folderId"] != "649" {
t.Errorf("folder filter = %+v", f.Term)
}
case f.Wildcard != nil:
wildcards++
}
}
if tenant != 1 || folder != 1 {
t.Errorf("term filters tenant=%d folder=%d, want 1 each", tenant, folder)
}
if wildcards != 2 {
t.Errorf("wildcard filters = %d, want deduped PDF+docx", wildcards)
}
}
func TestESSearchRequestRejectsEmptyTextAtSearch(t *testing.T) {
s, err := NewESSearcher(ESConfig{URL: "http://localhost:9200"})
if err != nil {
t.Fatalf("NewESSearcher: %v", err)
}
if _, err := s.Search(t.Context(), SearchQuery{Text: " "}); err == nil {
t.Error("empty query: want error")
}
}
func TestNewESSearcherRequiresURL(t *testing.T) {
if _, err := NewESSearcher(ESConfig{}); err == nil {
t.Error("empty URL: want error")
}
s, err := NewESSearcher(ESConfig{URL: "http://es:9200/"})
if err != nil {
t.Fatalf("NewESSearcher: %v", err)
}
if s.cfg.Index != defaultESIndex {
t.Errorf("index = %q, want %q", s.cfg.Index, defaultESIndex)
}
if s.cfg.URL != "http://es:9200" {
t.Errorf("url = %q, want trimmed", s.cfg.URL)
}
if s.Name() != "elasticsearch" {
t.Errorf("Name() = %q", s.Name())
}
}
func TestNormalizeExtensions(t *testing.T) {
got := normalizeExtensions([]string{" .PDF ", "pdf", "", "xlsx"})
want := []string{"pdf", "xlsx"}
if !reflect.DeepEqual(got, want) {
t.Errorf("normalizeExtensions = %v, want %v", got, want)
}
}
func TestParseESSearchResponse(t *testing.T) {
raw := []byte(`{
"took": 12,
"hits": {
"total": {"value": 2, "relation": "eq"},
"hits": [
{
"_id": "2395",
"_score": 7.31,
"_source": {"id": 2395, "title": "Rechnung-4711.pdf",
"folders": [{"folderId": "438", "id": 0}, {"folderId": "11", "id": 0}]},
"highlight": {
"title": ["<em>Rechnung</em>-4711.pdf"],
"document.attachment.content": ["… Zahlung der <em>Rechnung</em> …"]
}
},
{
"_id": "2318",
"_score": 6.02,
"_source": {"id": 2318, "title": "Mahnung.pdf", "folders": []},
"highlight": {"title": ["<em>Mahnung</em>.pdf"]}
}
]
}
}`)
hits, err := parseESSearchResponse(raw)
if err != nil {
t.Fatalf("parseESSearchResponse: %v", err)
}
if len(hits) != 2 {
t.Fatalf("hits = %d, want 2", len(hits))
}
h0 := hits[0]
if h0.ID != "2395" || h0.Title != "Rechnung-4711.pdf" || h0.Kind != File {
t.Errorf("hit0 entry = %+v", h0.Entry)
}
if h0.ParentID != "438" || !reflect.DeepEqual(h0.Path, []string{"438", "11"}) {
t.Errorf("hit0 path = %v parent = %q", h0.Path, h0.ParentID)
}
if h0.Score != 7.31 {
t.Errorf("hit0 score = %v", h0.Score)
}
if h0.Highlight != "… Zahlung der Rechnung …" {
t.Errorf("hit0 highlight = %q, want content fragment", h0.Highlight)
}
if hits[1].Highlight != "Mahnung.pdf" {
t.Errorf("hit1 highlight = %q, want title without tags", hits[1].Highlight)
}
if hits[1].ParentID != "" || len(hits[1].Path) != 0 {
t.Errorf("hit1 path = %v", hits[1].Path)
}
}
func TestESSearchRequestJSONShape(t *testing.T) {
got := esSearchRequest(SearchQuery{Text: "Rechnung", InContent: true}, "1")
b, err := json.Marshal(got)
if err != nil {
t.Fatalf("marshal: %v", err)
}
var back map[string]any
if err := json.Unmarshal(b, &back); err != nil {
t.Fatalf("unmarshal: %v", err)
}
if _, ok := back["query"].(map[string]any)["bool"]; !ok {
t.Errorf("query.bool missing: %s", b)
}
}
+230
View File
@@ -0,0 +1,230 @@
package onlyoffice
// restStore implements FileStore on top of the REST Documents methods in
// files.go. It is a thin adapter: no endpoint logic lives here, and every call
// is wrapped in DoRetry.
import (
"context"
"encoding/json"
"fmt"
"io"
"os"
"path/filepath"
"strconv"
"strings"
)
// restStore is a FileStore over the REST Documents API.
type restStore struct{ c *Client }
// Name reports the backend name.
func (s *restStore) Name() string { return ProviderREST }
// List returns the files and folders directly below parentID.
func (s *restStore) List(ctx context.Context, parentID string) ([]Entry, error) {
var out []Entry
err := retryStoreOp(ctx, func() error {
raw, err := s.c.ListFolder(ctx, parentID)
if err != nil {
return err
}
entries, err := entriesFromFolderMap(raw, ProviderREST)
if err != nil {
return err
}
out = entries
return nil
})
return out, err
}
// Stat returns file metadata. The REST adapter resolves files only; folders
// are listed by their parent (use List).
func (s *restStore) Stat(ctx context.Context, id string) (Entry, error) {
var out Entry
err := retryStoreOp(ctx, func() error {
f, err := s.c.GetFile(ctx, id)
if err != nil {
return err
}
out = FileEntryToEntry(f, ProviderREST)
return nil
})
return out, err
}
// CreateFolder creates a subfolder under parentID.
func (s *restStore) CreateFolder(ctx context.Context, parentID, title string) (Entry, error) {
var out Entry
err := retryStoreOp(ctx, func() error {
m, err := s.c.CreateFolder(ctx, parentID, title)
if err != nil {
return err
}
e, err := folderEntryFromMap(m, parentID, ProviderREST)
if err != nil {
return err
}
if e.ParentID == "" {
e.ParentID = parentID
}
if e.Title == "" {
e.Title = title
}
out = e
return nil
})
return out, err
}
// Upload streams r into parentID as title. UploadToFolder is path based, so
// the reader is spooled to a temporary file first (ponytail: OnlyOffice
// multipart upload buffers the whole body anyway).
func (s *restStore) Upload(ctx context.Context, parentID, title string, r io.Reader) (Entry, error) {
dir, err := os.MkdirTemp("", "oo-rest-upload-")
if err != nil {
return Entry{}, err
}
defer os.RemoveAll(dir)
local := filepath.Join(dir, SafeLocalFileName(title))
f, err := os.Create(local)
if err != nil {
return Entry{}, err
}
if _, err := io.Copy(f, r); err != nil {
f.Close()
return Entry{}, err
}
if err := f.Close(); err != nil {
return Entry{}, err
}
var out Entry
err = retryStoreOp(ctx, func() error {
fe, err := s.c.UploadToFolder(ctx, parentID, local)
if err != nil {
return err
}
out = FileEntryToEntry(fe, ProviderREST)
return nil
})
return out, err
}
// Download streams the file bytes into w.
func (s *restStore) Download(ctx context.Context, id string, w io.Writer) (int64, error) {
var n int64
err := retryStoreOp(ctx, func() error {
var e error
n, e = s.c.DownloadFile(ctx, id, w)
return e
})
return n, err
}
// Move moves file ids into parentID. The REST MoveFiles endpoint handles files
// only; folder moves are not exposed by this adapter.
func (s *restStore) Move(ctx context.Context, ids []string, parentID string) error {
dest, err := strconv.Atoi(strings.TrimSpace(parentID))
if err != nil {
return fmt.Errorf("onlyoffice: rest store: move: non-numeric destination folder id %q", parentID)
}
fileIDs, err := numericIDs(ids)
if err != nil {
return err
}
return retryStoreOp(ctx, func() error {
_, err := s.c.MoveFiles(ctx, dest, fileIDs)
return err
})
}
// Copy copies file ids into parentID. files.go has no copy method, so the
// shared REST fileops copy endpoint (CopyDavItems) is used.
func (s *restStore) Copy(ctx context.Context, ids []string, parentID string) error {
if len(ids) == 0 {
return nil
}
return retryStoreOp(ctx, func() error {
return s.c.CopyDavItems(ctx, nil, ids, parentID)
})
}
// Rename sets a new title (including extension) for a file.
func (s *restStore) Rename(ctx context.Context, id, title string) error {
return retryStoreOp(ctx, func() error {
_, err := s.c.RenameFile(ctx, id, title)
return err
})
}
// Delete permanently deletes file ids.
func (s *restStore) Delete(ctx context.Context, ids []string) error {
fileIDs, err := numericIDs(ids)
if err != nil {
return err
}
if len(fileIDs) == 0 {
return nil
}
return retryStoreOp(ctx, func() error {
return s.c.DeleteFiles(ctx, fileIDs)
})
}
// entriesFromFolderMap converts a ListFolder response map into canonical
// entries, reusing the DavFile/DavFolder decoders for robust size handling.
func entriesFromFolderMap(m map[string]any, provider string) ([]Entry, error) {
if m == nil {
return nil, nil
}
b, err := json.Marshal(m)
if err != nil {
return nil, err
}
var listing DavListing
if err := json.Unmarshal(b, &listing); err != nil {
return nil, err
}
out := make([]Entry, 0, len(listing.Folders)+len(listing.Files))
for _, f := range listing.Folders {
out = append(out, DavFolderToEntry(f, provider))
}
for _, f := range listing.Files {
out = append(out, DavFileToEntry(f, provider))
}
return out, nil
}
// folderEntryFromMap converts a CreateFolder response map into a folder Entry.
func folderEntryFromMap(m map[string]any, parentID, provider string) (Entry, error) {
e := Entry{Kind: Folder, Provider: provider, ParentID: parentID}
if m == nil {
return e, nil
}
b, err := json.Marshal(m)
if err != nil {
return e, err
}
var f DavFolder
if err := json.Unmarshal(b, &f); err != nil {
return e, err
}
e = DavFolderToEntry(f, provider)
return e, nil
}
// numericIDs parses Documents numeric ids from strings.
func numericIDs(ids []string) ([]int, error) {
out := make([]int, 0, len(ids))
for _, id := range ids {
n, err := strconv.Atoi(strings.TrimSpace(id))
if err != nil {
return nil, fmt.Errorf("onlyoffice: rest store: non-numeric id %q", id)
}
out = append(out, n)
}
return out, nil
}
+209
View File
@@ -0,0 +1,209 @@
//go:build integration
package onlyoffice
import (
"bytes"
"context"
"strconv"
"testing"
"time"
)
// TestIntegrationFileStores runs the same operation set (create folder, upload,
// list, stat, download, move, copy, rename, delete) through the REST and DAV
// FileStore adapters against a throwaway project Documents folder. Destructive
// — only run against instances you own.
//
// The Documents fileops API is asynchronous: a move/copy/delete is accepted
// immediately and becomes visible a moment later, so effects are polled.
func TestIntegrationFileStores(t *testing.T) {
c := liveClient(t)
t.Cleanup(func() { cleanupTestProjects(t, c) })
ctx := context.Background()
suffix := time.Now().UTC().Format("20060102-150405")
project, err := c.CreateProject(NewProjectRequest{
Title: testProjectPrefix + "store-" + suffix,
Description: "go-onlyoffice file store integration",
})
if err != nil {
t.Fatalf("CreateProject: %v", err)
}
if project.ID == nil {
t.Fatal("created project without id")
}
root, err := c.projectFolderID(ctx, strconv.Itoa(*project.ID))
if err != nil {
t.Fatalf("projectFolderID: %v", err)
}
for _, backend := range []string{ProviderREST, ProviderDAV} {
t.Run(backend, func(t *testing.T) {
testFileStoreOps(t, ctx, c, c.FileStore(backend), root, suffix)
})
}
}
func testFileStoreOps(t *testing.T, ctx context.Context, c *Client, store FileStore, root, suffix string) {
t.Helper()
content := []byte("file store " + store.Name() + " " + suffix + "\n")
src, err := store.CreateFolder(ctx, root, "fs-src-"+suffix)
if err != nil {
t.Fatalf("CreateFolder src: %v", err)
}
if src.Kind != Folder || src.ID == "" {
t.Fatalf("created src folder: %+v", src)
}
dst, err := store.CreateFolder(ctx, root, "fs-dst-"+suffix)
if err != nil {
t.Fatalf("CreateFolder dst: %v", err)
}
if dst.Kind != Folder || dst.ID == "" {
t.Fatalf("created dst folder: %+v", dst)
}
t.Cleanup(func() {
if err := c.DeleteDavItems(ctx, []string{src.ID, dst.ID}, nil); err != nil {
t.Logf("cleanup folders: %v", err)
}
})
up, err := store.Upload(ctx, src.ID, "doc-"+suffix+".txt", bytes.NewReader(content))
if err != nil {
t.Fatalf("Upload: %v", err)
}
if up.Kind != File || up.ID == "" {
t.Fatalf("uploaded entry: %+v", up)
}
if !waitEntry(ctx, store, src.ID, up.ID, 15*time.Second) {
t.Fatalf("uploaded %s not listed in src", up.ID)
}
st, err := store.Stat(ctx, up.ID)
if err != nil {
t.Fatalf("Stat: %v", err)
}
if st.ID != up.ID || st.Kind != File {
t.Fatalf("stat = %+v", st)
}
var buf bytes.Buffer
n, err := store.Download(ctx, up.ID, &buf)
if err != nil {
t.Fatalf("Download: %v", err)
}
if n != int64(len(content)) || !bytes.Equal(buf.Bytes(), content) {
t.Fatalf("download mismatch: got %d bytes %q want %d", n, buf.String(), len(content))
}
moveEventually(t, ctx, store, up.ID, dst.ID)
if !waitEntry(ctx, store, dst.ID, up.ID, 20*time.Second) {
t.Fatalf("moved file %s not in dst", up.ID)
}
if err := store.Copy(ctx, []string{up.ID}, src.ID); err != nil {
t.Fatalf("Copy: %v", err)
}
copied := waitOtherFile(ctx, store, src.ID, up.ID, 20*time.Second)
if copied == nil {
t.Fatalf("no copy found in src after Copy")
}
newTitle := "renamed-" + suffix + ".txt"
renameEventually(t, ctx, store, up.ID, newTitle)
if err := store.Delete(ctx, []string{up.ID, copied.ID}); err != nil {
t.Fatalf("Delete: %v", err)
}
if !waitNoEntry(ctx, store, dst.ID, up.ID, 20*time.Second) {
t.Fatalf("file %s still present in dst after delete", up.ID)
}
if !waitNoEntry(ctx, store, src.ID, copied.ID, 20*time.Second) {
t.Fatalf("copy %s still present in src after delete", copied.ID)
}
}
// moveEventually issues Move and retries while the operation is not visible yet
// (the fileops API accepts asynchronously and occasionally rejects a move that
// raced the just-finished upload).
func moveEventually(t *testing.T, ctx context.Context, store FileStore, id, dstID string) {
t.Helper()
var lastErr error
for attempt := 0; attempt < 5; attempt++ {
if lastErr = store.Move(ctx, []string{id}, dstID); lastErr == nil {
if waitEntry(ctx, store, dstID, id, 6*time.Second) {
return
}
}
time.Sleep(time.Second)
}
t.Fatalf("Move %s -> %s: %v", id, dstID, lastErr)
}
func renameEventually(t *testing.T, ctx context.Context, store FileStore, id, title string) {
t.Helper()
var lastErr error
for attempt := 0; attempt < 5; attempt++ {
if lastErr = store.Rename(ctx, id, title); lastErr == nil {
if e, err := store.Stat(ctx, id); err == nil && e.Title == title {
return
}
}
time.Sleep(time.Second)
}
t.Fatalf("Rename %s -> %q: %v", id, title, lastErr)
}
func waitEntry(ctx context.Context, store FileStore, parentID, id string, d time.Duration) bool {
deadline := time.Now().Add(d)
for time.Now().Before(deadline) {
if list, err := store.List(ctx, parentID); err == nil && entryByID(list, id) != nil {
return true
}
time.Sleep(500 * time.Millisecond)
}
return false
}
func waitNoEntry(ctx context.Context, store FileStore, parentID, id string, d time.Duration) bool {
deadline := time.Now().Add(d)
for time.Now().Before(deadline) {
if list, err := store.List(ctx, parentID); err == nil && entryByID(list, id) == nil {
return true
}
time.Sleep(500 * time.Millisecond)
}
return false
}
func waitOtherFile(ctx context.Context, store FileStore, parentID, id string, d time.Duration) *Entry {
deadline := time.Now().Add(d)
for time.Now().Before(deadline) {
if list, err := store.List(ctx, parentID); err == nil {
if e := firstFileOtherThan(list, id); e != nil {
return e
}
}
time.Sleep(500 * time.Millisecond)
}
return nil
}
func entryByID(entries []Entry, id string) *Entry {
for i := range entries {
if entries[i].ID == id {
return &entries[i]
}
}
return nil
}
func firstFileOtherThan(entries []Entry, id string) *Entry {
for i := range entries {
if entries[i].Kind == File && entries[i].ID != id {
return &entries[i]
}
}
return nil
}
+9 -6
View File
@@ -324,19 +324,22 @@ func (c *Client) MoveFiles(ctx context.Context, destFolderID int, fileIDs []int)
"resolveType": "Skip", "resolveType": "Skip",
"holdResult": true, "holdResult": true,
} }
out, err := c.putJSONObject(ctx, "/api/2.0/files/fileops/move.json", body) // fileops/move answers an operations envelope (like MoveDavItems), not a
// single object, so parse the raw body before unwrapping and surface any
// per-operation error. Unwrapping first (putJSONObject) made fileopsError
// look for a "response" key that was already stripped.
raw, err := c.putJSON(ctx, "/api/2.0/files/fileops/move", body)
if err != nil { if err != nil {
out, err = c.putJSONObject(ctx, "/api/2.0/files/fileops/move", body) raw, err = c.putJSON(ctx, "/api/2.0/files/fileops/move.json", body)
}
if err != nil { if err != nil {
return nil, err return nil, err
} }
if raw, merr := json.Marshal(out); merr == nil { }
if ferr := fileopsError(raw); ferr != nil { if ferr := fileopsError(raw); ferr != nil {
return nil, ferr return nil, ferr
} }
} out, _ := unmarshalResponseObject(raw)
return out, err return out, nil
} }
// UploadToFolder uploads a local file into an arbitrary Documents folder id. // UploadToFolder uploads a local file into an arbitrary Documents folder id.