feat(oo): kontoblatt bulk tools, oo dav, update, retry (#22) #23

Merged
eSlider merged 1 commits from feat/kontoblatt-tools#22 into main 2026-09-14 22:31:23 +01:00
9 changed files with 1320 additions and 10 deletions
+220
View File
@@ -0,0 +1,220 @@
// Command kontoblatt builds a summary ("сводная таблица") of a Kontoblatt XLSX
// (Datum, Gegenkonto, Buchungstext, Beleg, Soll, Haben, Bemerkung) and uploads
// it back to the same OnlyOffice folder as the source file.
//
// Usage: kontoblatt <FILE_ID> <LOCAL_XLSX>
package main
import (
"context"
"fmt"
"os"
"regexp"
"sort"
"strconv"
"strings"
onlyoffice "github.com/eslider/go-onlyoffice"
"github.com/xuri/excelize/v2"
)
type agg struct {
count int
soll float64
haben float64
reFehlt int
}
type rec struct {
date, month, konto, text string
soll, haben float64
reFehlt bool
}
var dateRe = regexp.MustCompile(`^\d{2}\.\d{2}\.\d{4}$`)
func parseAmount(s string) float64 {
s = strings.TrimSpace(s)
s = strings.ReplaceAll(s, "€", "")
s = strings.ReplaceAll(s, " ", "")
s = strings.ReplaceAll(s, ",", "") // German thousands separator
s = strings.TrimSpace(s)
if s == "" {
return 0
}
v, err := strconv.ParseFloat(s, 64)
if err != nil {
return 0
}
return v
}
func cell(row []string, i int) string {
if i < len(row) {
return strings.TrimSpace(row[i])
}
return ""
}
func main() {
if len(os.Args) < 3 {
fmt.Fprintln(os.Stderr, "usage: kontoblatt <FILE_ID> <LOCAL_XLSX>")
os.Exit(2)
}
fileID, path := os.Args[1], os.Args[2]
ctx := context.Background()
f, err := excelize.OpenFile(path)
if err != nil {
panic(err)
}
defer f.Close()
var recs []rec
for _, sh := range f.GetSheetList() {
rows, err := f.GetRows(sh)
if err != nil {
continue
}
for _, r := range rows {
d := cell(r, 0)
if !dateRe.MatchString(d) {
continue
}
text := cell(r, 2)
recs = append(recs, rec{
date: d,
month: d[3:10],
konto: cell(r, 1),
text: text,
soll: parseAmount(cell(r, 4)),
haben: parseAmount(cell(r, 5)),
reFehlt: strings.Contains(strings.ToUpper(text), "FEHLT"),
})
}
}
byKonto := map[string]*agg{}
byMonth := map[string]*agg{}
getK := func(k string) *agg {
if byKonto[k] == nil {
byKonto[k] = &agg{}
}
return byKonto[k]
}
getM := func(k string) *agg {
if byMonth[k] == nil {
byMonth[k] = &agg{}
}
return byMonth[k]
}
var tot agg
for _, r := range recs {
k := getK(r.konto)
k.count++
k.soll += r.soll
k.haben += r.haben
if r.reFehlt {
k.reFehlt++
}
m := getM(r.month)
m.count++
m.soll += r.soll
m.haben += r.haben
if r.reFehlt {
m.reFehlt++
}
tot.count++
tot.soll += r.soll
tot.haben += r.haben
if r.reFehlt {
tot.reFehlt++
}
}
out := excelize.NewFile()
defer out.Close()
writeSheet(out, "Nach Gegenkonto", "Gegenkonto", byKonto, tot)
writeSheet(out, "Nach Monat", "Monat", byMonth, tot)
outPath := "/tmp/opencode/kontoblatt-zusammenfassung.xlsx"
if err := out.SaveAs(outPath); err != nil {
panic(err)
}
// upload next to the source file
creds := onlyoffice.GetEnvironmentCredentials()
c := onlyoffice.NewClient(creds)
var src *onlyoffice.FileEntry
if derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
var err error
src, err = c.GetFile(ctx, fileID)
return err
}); derr != nil {
panic(derr)
}
folder := ""
if src.FolderID != nil {
folder = src.FolderID.String()
}
title := ""
if src.Title != nil {
title = *src.Title
}
fmt.Printf("source: id=%s title=%q folder=%s\n", fileID, title, folder)
name := "Kontoblatt-1591-2025-Zusammenfassung.xlsx"
tmp := "/tmp/opencode/" + name
data, _ := os.ReadFile(outPath)
if err := os.WriteFile(tmp, data, 0o600); err != nil {
panic(err)
}
var entry *onlyoffice.FileEntry
if derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
var err error
entry, _, err = c.UploadToFolderReplacing(ctx, folder, tmp)
return err
}); derr != nil {
panic(derr)
}
fmt.Printf("uploaded: %s -> folder %s (id %v)\n", name, folder, entry.ID)
// print the summary
printAgg("Nach Gegenkonto", byKonto, tot)
printAgg("Nach Monat", byMonth, tot)
}
func writeSheet(f *excelize.File, sheet, key string, m map[string]*agg, tot agg) {
f.NewSheet(sheet)
rows := [][]any{{key, "Anzahl", "Soll", "Haben", "Saldo", `davon "fehlt"`}}
keys := make([]string, 0, len(m))
for k := range m {
keys = append(keys, k)
}
sort.Strings(keys)
for _, k := range keys {
a := m[k]
rows = append(rows, []any{k, a.count, a.soll, a.haben, a.soll - a.haben, a.reFehlt})
}
rows = append(rows, []any{"GESAMT", tot.count, tot.soll, tot.haben, tot.soll - tot.haben, tot.reFehlt})
for i, row := range rows {
for j, v := range row {
cellRef, _ := excelize.CoordinatesToCellName(j+1, i+1)
_ = f.SetCellValue(sheet, cellRef, v)
}
}
}
func printAgg(title string, m map[string]*agg, tot agg) {
fmt.Printf("\n== %s ==\n", title)
keys := make([]string, 0, len(m))
for k := range m {
keys = append(keys, k)
}
sort.Strings(keys)
fmt.Printf("%-12s %6s %12s %12s %12s %7s\n", "key", "count", "soll", "haben", "saldo", "fehlt")
for _, k := range keys {
a := m[k]
fmt.Printf("%-12s %6d %12.2f %12.2f %12.2f %7d\n", k, a.count, a.soll, a.haben, a.soll-a.haben, a.reFehlt)
}
fmt.Printf("%-12s %6d %12.2f %12.2f %12.2f %7d\n", "GESAMT", tot.count, tot.soll, tot.haben, tot.soll-tot.haben, tot.reFehlt)
}
+478
View File
@@ -0,0 +1,478 @@
// Command kontolink fills the "Link" column of a Kontoblatt ("ungeklärte
// Posten") XLSX by matching each row to an OnlyOffice document.
//
// Strategy (deterministic, conservative — no LLM):
// 1. Beleg token (letters/digits from the "Beleg" column) appears in the file
// title; among candidates prefer (a) the row's month, (b) real invoices over
// copies/dupes, and require the result to be unique;
// 2. else supplier + row month + "rechnung", again unique.
//
// A file is linked at most once (rows already carrying a link are kept and their
// file counts as used). Ambiguous rows are left UNLINKED for manual review.
//
// Usage: kontolink <IN_XLSX> <INDEX_TSV> <OUT_XLSX>
package main
import (
"context"
"fmt"
"os"
"path/filepath"
"regexp"
"strconv"
"strings"
"time"
onlyoffice "github.com/eslider/go-onlyoffice"
"github.com/xuri/excelize/v2"
)
var (
dateRe = regexp.MustCompile(`^\d{2}\.\d{2}\.\d{4}$`)
nonAln = regexp.MustCompile(`[^0-9a-z]+`)
fileID = regexp.MustCompile(`fileid=(\d+)`)
)
func parseDay(s string) (time.Time, bool) {
t, err := time.Parse("02.01.2006", strings.TrimSpace(s))
return t, err == nil
}
func titleDay(title string) (time.Time, bool) {
if len(title) >= 10 {
if t, err := time.Parse("2006-01-02", title[:10]); err == nil {
return t, true
}
}
return time.Time{}, false
}
// nearest picks the candidate whose title date is closest to rd. Ties and
// undated candidates (when >1) are rejected.
func nearest(cands []entry, rd time.Time) (entry, bool) {
if len(cands) == 1 {
return cands[0], true
}
best, bestD, tie := -1, 0.0, false
for i, e := range cands {
td, ok := titleDay(e.title)
if !ok {
continue
}
d := td.Sub(rd).Hours() / 24
if d < 0 {
d = -d
}
if best < 0 || d < bestD {
best, bestD, tie = i, d, false
} else if d == bestD {
tie = true
}
}
if best < 0 || tie {
return entry{}, false
}
return cands[best], true
}
const linkPrefix = "https://office.pro-dukt.de/Products/Files/DocEditor.aspx?fileid="
type entry struct {
id, path, title, norm string
}
func norm(s string) string { return nonAln.ReplaceAllString(strings.ToLower(s), "") }
func main() {
if len(os.Args) < 4 {
fmt.Fprintln(os.Stderr, "usage: kontolink <IN_XLSX> <INDEX_TSV> <OUT_XLSX>")
os.Exit(2)
}
in, idxPath, out := os.Args[1], os.Args[2], os.Args[3]
idxRaw, err := os.ReadFile(idxPath)
if err != nil {
panic(err)
}
var entries []entry
for _, line := range strings.Split(string(idxRaw), "\n") {
parts := strings.Split(line, "\t")
if len(parts) < 4 || parts[0] == "" {
continue
}
entries = append(entries, entry{id: parts[0], path: parts[2], title: parts[3], norm: norm(parts[3])})
}
f, err := excelize.OpenFile(in)
if err != nil {
panic(err)
}
defer f.Close()
sheet := f.GetSheetList()[0]
rows, err := f.GetRows(sheet)
if err != nil {
panic(err)
}
// optional 5th arg: amounts TSV "file_id\ttitle\tamount" (see cmd/pdfamount)
var amts []amtEntry
if len(os.Args) >= 6 && os.Args[5] != "" {
amts = loadAmounts(os.Args[5])
}
used := map[string]bool{}
for _, r := range rows {
if m := fileID.FindStringSubmatch(cell(r, 7)); m != nil {
used[m[1]] = true
}
}
var linked, byBeleg, bySupplier, byAmount, unmatched, ambiguous int
for i, r := range rows {
if i == 0 || !dateRe.MatchString(cell(r, 0)) || strings.TrimSpace(cell(r, 7)) != "" {
continue
}
beleg := norm(cell(r, 3))
supplier := supplierNorm(cell(r, 2))
month := monthYear(cell(r, 0))
rd, _ := parseDay(cell(r, 0))
e, kind, ok := pick(entries, used, beleg, supplier, month, rd)
if !ok {
if ae, aok := amountPick(amts, used, supplier, rowAmount(r), rd); aok {
e, kind, ok = entry{id: ae.id, title: ae.title}, "amount", true
}
}
if !ok {
if beleg != "" {
ambiguous++
} else {
unmatched++
}
continue
}
ref, _ := excelize.CoordinatesToCellName(8, i+1)
if err := f.SetCellValue(sheet, ref, linkPrefix+e.id); err != nil {
panic(err)
}
used[e.id] = true
linked++
switch kind {
case "beleg":
byBeleg++
case "supplier":
bySupplier++
case "amount":
byAmount++
}
fmt.Printf("row %3d %-30s -> %s [%s]\n", i+1, cell(r, 2), e.title, kind)
}
if err := f.SaveAs(out); err != nil {
panic(err)
}
fmt.Printf("\nlinked=%d (beleg=%d, supplier=%d, amount=%d), ambiguous=%d, no-candidate=%d\n",
linked, byBeleg, bySupplier, byAmount, ambiguous, unmatched)
// Optional 4th arg: source OnlyOffice file id. Try to update it in place;
// if it is locked (OnlyOffice 500), upload a "(links)" copy next to it.
if len(os.Args) >= 5 && os.Args[4] != "" {
c := onlyoffice.NewClient(onlyoffice.GetEnvironmentCredentials())
ctx, cancel := context.WithTimeout(context.Background(), 120*time.Second)
defer cancel()
var src *onlyoffice.FileEntry
if derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
var err error
src, err = c.GetFile(ctx, os.Args[4])
return err
}); derr != nil {
panic(derr)
}
folder, title := "", ""
if src.FolderID != nil {
folder = src.FolderID.String()
}
if src.Title != nil {
title = *src.Title
}
uderr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
_, err := c.UpdateFile(ctx, os.Args[4], out)
return err
})
if uderr == nil {
fmt.Printf("updated file %s in place\n", os.Args[4])
return
}
fmt.Printf("in-place update failed (locked?); uploading a copy to folder %s\n", folder)
ext := filepath.Ext(title)
name := strings.TrimSuffix(title, ext) + " (links)" + ext
tmp := filepath.Join(os.TempDir(), name)
data, _ := os.ReadFile(out)
if err := os.WriteFile(tmp, data, 0o600); err != nil {
panic(err)
}
if derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
_, _, err := c.UploadToFolderReplacing(ctx, folder, tmp)
return err
}); derr != nil {
panic(derr)
}
fmt.Printf("uploaded copy: %s -> folder %s\n", name, folder)
}
}
func cell(r []string, i int) string {
if i < len(r) {
return strings.TrimSpace(r[i])
}
return ""
}
func supplierNorm(s string) string {
s = strings.ToUpper(s)
if i := strings.Index(s, ","); i >= 0 {
s = s[:i]
}
for _, w := range []string{"RE FEHLT", "GS FEHLT", "WOFR", "WOFÜR"} {
s = strings.ReplaceAll(s, w, "")
}
return norm(s)
}
func monthYear(date string) string {
if len(date) == 10 {
return date[6:10] + "-" + date[3:5]
}
return ""
}
// pick returns an unused candidate. Beleg match wins; supplier+month is a
// fallback. When several candidates qualify, the one closest in time to the row
// date wins; a tie is rejected (ambiguous) rather than guessed.
func pick(entries []entry, used map[string]bool, beleg, supplier, month string, rd time.Time) (entry, string, bool) {
free := func(e entry) bool { return !used[e.id] }
if len(beleg) >= 5 {
var inMonth []entry
for _, e := range entries {
if free(e) && belegMatches(e.norm, beleg) &&
(month == "" || strings.Contains(e.title, month)) {
inMonth = append(inMonth, e)
}
}
if supplier != "" {
var s []entry
for _, e := range inMonth {
if strings.Contains(e.norm, supplier) {
s = append(s, e)
}
}
if len(s) > 0 {
inMonth = s
}
}
inMonth = topRank(inMonth)
if e, ok := nearest(inMonth, rd); ok {
return e, "beleg", true
}
// A Beleg is present but no file carries it: do NOT fall back to a
// supplier guess (that links the wrong invoice).
return entry{}, "", false
}
if supplier != "" && month != "" {
var c []entry
for _, e := range entries {
if free(e) && strings.Contains(e.norm, supplier) &&
strings.Contains(e.title, month) && strings.Contains(e.norm, "rechnung") {
c = append(c, e)
}
}
c = topRank(c)
if e, ok := nearest(c, rd); ok {
return e, "supplier", true
}
}
return entry{}, "", false
}
// topRank keeps only the highest-ranked candidates (real invoice over copy /
// dupe / op), so a tie with a duplicate does not mask the real file.
func topRank(cands []entry) []entry {
if len(cands) < 2 {
return cands
}
best := 0
for _, e := range cands {
if rank(e) > best {
best = rank(e)
}
}
out := cands[:0]
for _, e := range cands {
if rank(e) == best {
out = append(out, e)
}
}
return out
}
func rank(e entry) int {
s := 0
if strings.Contains(e.path, "/2025") || strings.Contains(e.path, "/2024") {
s += 4
}
if strings.Contains(e.norm, "rechnung") {
s += 2
}
if strings.Contains(e.norm, "dupe") || strings.Contains(e.norm, "copy") ||
strings.Contains(e.norm, "op") {
s--
}
return s
}
// belegMatches reports whether a Beleg identifies the file: the whole normalized
// Beleg appears, or (for long numeric Belege, e.g. "24/641393110") an 8-digit
// window of its longest digit run appears.
func belegMatches(titleNorm, beleg string) bool {
if strings.Contains(titleNorm, beleg) {
return true
}
run := longestDigitRun(beleg)
for i := 0; i+8 <= len(run); i++ {
if strings.Contains(titleNorm, run[i:i+8]) {
return true
}
}
return false
}
func longestDigitRun(s string) string {
var best, cur strings.Builder
for _, r := range s {
if r >= '0' && r <= '9' {
cur.WriteRune(r)
if cur.Len() > best.Len() {
best.Reset()
best.WriteString(cur.String())
}
} else {
cur.Reset()
}
}
return best.String()
}
type amtEntry struct {
id string
title string
norm string
amount float64
date time.Time
hasDate bool
}
func loadAmounts(path string) []amtEntry {
raw, err := os.ReadFile(path)
if err != nil {
return nil
}
var out []amtEntry
for _, line := range strings.Split(string(raw), "\n") {
p := strings.Split(line, "\t")
if len(p) < 3 {
continue
}
v, err := strconv.ParseFloat(strings.TrimSpace(p[2]), 64)
if err != nil {
continue
}
e := amtEntry{id: p[0], title: p[1], norm: norm(p[1]), amount: v}
if len(p[1]) >= 10 {
if t, err := time.Parse("2006-01-02", p[1][:10]); err == nil {
e.date, e.hasDate = t, true
}
}
out = append(out, e)
}
return out
}
func rowAmount(r []string) float64 {
if v := parseAmount(cell(r, 4)); v != 0 {
return v
}
return parseAmount(cell(r, 5))
}
func parseAmount(s string) float64 {
s = strings.ReplaceAll(s, "€", "")
s = strings.ReplaceAll(s, " ", "")
s = strings.ReplaceAll(s, ",", ".")
if s == "" {
return 0
}
v, err := strconv.ParseFloat(s, 64)
if err != nil {
return 0
}
return v
}
// amountPick matches a row to an O2 invoice by amount + nearest date. Scoped to
// Telefonica/O2 rows and O2 files, so it cannot cross-link other suppliers.
func amountPick(amts []amtEntry, used map[string]bool, supplier string, amt float64, rd time.Time) (amtEntry, bool) {
if amt <= 0 || len(amts) == 0 {
return amtEntry{}, false
}
if !strings.Contains(supplier, "telefonica") && !strings.Contains(supplier, "o2") {
return amtEntry{}, false
}
var cands []amtEntry
for _, a := range amts {
if used[a.id] || !strings.Contains(a.norm, "o2") {
continue
}
d := a.amount - amt
if d < 0 {
d = -d
}
if d > 0.005 {
continue
}
if a.hasDate && !rd.IsZero() {
days := a.date.Sub(rd).Hours() / 24
if days < 0 {
days = -days
}
if days > 75 {
continue
}
}
cands = append(cands, a)
}
if len(cands) == 1 {
return cands[0], true
}
best, bestD, tie := -1, 0.0, false
for i, a := range cands {
if !a.hasDate {
continue
}
d := a.date.Sub(rd).Hours() / 24
if d < 0 {
d = -d
}
if best < 0 || d < bestD {
best, bestD, tie = i, d, false
} else if d == bestD {
tie = true
}
}
if best < 0 || tie {
return amtEntry{}, false
}
return cands[best], true
}
+291
View File
@@ -0,0 +1,291 @@
package main
import (
"fmt"
"os"
onlyoffice "github.com/eslider/go-onlyoffice"
"github.com/spf13/cobra"
)
func init() {
rootCmd.AddCommand(davCmd())
}
// davCmd exposes the Documents module through the same Dav calls that back
// oo-webdav (ListDavFolder / MoveDavItems / CopyDavItems / DownloadDavFile).
// MoveDavItems sends resolveType=Skip + holdResult=true, which the legacy
// fileops/move call without those params silently ignores (200 without move).
func davCmd() *cobra.Command {
cmd := &cobra.Command{
Use: "dav",
Short: "Documents module by folder/file id (oo-webdav proven path)",
}
cmd.AddCommand(davLsCmd())
cmd.AddCommand(davMoveCmd())
cmd.AddCommand(davCopyCmd())
cmd.AddCommand(davMkdirCmd())
cmd.AddCommand(davRenameFileCmd())
cmd.AddCommand(davRenameFolderCmd())
cmd.AddCommand(davDownloadCmd())
cmd.AddCommand(davFileOpsCmd())
return cmd
}
func davLsCmd() *cobra.Command {
cmd := &cobra.Command{
Use: "ls FOLDER_ID",
Short: "List a Documents folder (@root for virtual sections)",
Args: cobra.ExactArgs(1),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
ctx := cmd.Context()
if args[0] == "@root" {
sections, err := c.ListDavSections(ctx)
if err != nil {
return err
}
rows := make([]map[string]any, 0, len(sections))
for _, s := range sections {
rows = append(rows, map[string]any{
"id": s.ID,
"title": s.Title,
"filesCount": s.FilesCount,
"foldersCount": s.FoldersCount,
})
}
printTable([]string{"id", "title", "filesCount", "foldersCount"}, rows)
return nil
}
l, err := c.ListDavFolder(ctx, args[0])
if err != nil {
return err
}
if outputFormat == "json" {
folders := make([]map[string]any, 0, len(l.Folders))
for _, f := range l.Folders {
folders = append(folders, map[string]any{
"id": f.ID,
"title": f.Title,
"filesCount": f.FilesCount,
"foldersCount": f.FoldersCount,
})
}
files := make([]map[string]any, 0, len(l.Files))
for _, f := range l.Files {
files = append(files, map[string]any{
"id": f.ID,
"title": f.Title,
"size": f.Size,
"updated": f.Updated,
})
}
printObject(map[string]any{"folders": folders, "files": files})
return nil
}
if len(l.Folders) > 0 {
frows := make([]map[string]any, 0, len(l.Folders))
for _, f := range l.Folders {
frows = append(frows, map[string]any{
"id": f.ID,
"title": f.Title,
"filesCount": f.FilesCount,
"foldersCount": f.FoldersCount,
})
}
if outputFormat == "table" {
fmt.Println("folders:")
}
printTable([]string{"id", "title", "filesCount", "foldersCount"}, frows)
}
rows := make([]map[string]any, 0, len(l.Files))
for _, f := range l.Files {
rows = append(rows, map[string]any{
"id": f.ID,
"title": f.Title,
"size": f.Size,
"updated": f.Updated,
})
}
if outputFormat == "table" {
fmt.Println("files:")
}
printTable([]string{"id", "title", "size", "updated"}, rows)
return nil
},
}
return cmd
}
func davMoveCmd() *cobra.Command {
var folderIDs []string
cmd := &cobra.Command{
Use: "move DEST_FOLDER_ID FILE_ID [FILE_ID...]",
Short: "Move file(s) into a Documents folder (resolveType=Skip, holdResult)",
Args: cobra.MinimumNArgs(2),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
if err := c.MoveDavItems(cmd.Context(), folderIDs, args[1:], args[0]); err != nil {
return err
}
printObject(map[string]any{"moved_files": args[1:], "moved_folders": folderIDs, "dest": args[0]})
return nil
},
}
cmd.Flags().StringSliceVar(&folderIDs, "folders", nil, "folder ids to move along with the files")
return cmd
}
func davCopyCmd() *cobra.Command {
var folderIDs []string
cmd := &cobra.Command{
Use: "copy DEST_FOLDER_ID FILE_ID [FILE_ID...]",
Short: "Copy file(s) into a Documents folder (conflictResolveType=Skip)",
Args: cobra.MinimumNArgs(2),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
if err := c.CopyDavItems(cmd.Context(), folderIDs, args[1:], args[0]); err != nil {
return err
}
printObject(map[string]any{"copied_files": args[1:], "copied_folders": folderIDs, "dest": args[0]})
return nil
},
}
cmd.Flags().StringSliceVar(&folderIDs, "folders", nil, "folder ids to copy along with the files")
return cmd
}
func davMkdirCmd() *cobra.Command {
return &cobra.Command{
Use: "mkdir PARENT_FOLDER_ID TITLE",
Short: "Create a subfolder in Documents",
Args: cobra.ExactArgs(2),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
f, err := c.CreateDavFolder(cmd.Context(), args[0], args[1])
if err != nil {
return err
}
printObject(map[string]any{"id": f.ID, "title": f.Title, "parent": args[0]})
return nil
},
}
}
func davRenameFileCmd() *cobra.Command {
return &cobra.Command{
Use: "rename-file FILE_ID NEW_TITLE",
Short: "Rename a Documents file (include extension in NEW_TITLE)",
Args: cobra.ExactArgs(2),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
if err := c.RenameDavFile(cmd.Context(), args[0], args[1]); err != nil {
return err
}
printObject(map[string]any{"id": args[0], "title": args[1]})
return nil
},
}
}
func davRenameFolderCmd() *cobra.Command {
return &cobra.Command{
Use: "rename-folder FOLDER_ID NEW_TITLE",
Short: "Rename a Documents folder",
Args: cobra.ExactArgs(2),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
if err := c.RenameDavFolder(cmd.Context(), args[0], args[1]); err != nil {
return err
}
printObject(map[string]any{"id": args[0], "title": args[1]})
return nil
},
}
}
func davDownloadCmd() *cobra.Command {
var to string
cmd := &cobra.Command{
Use: "download FILE_ID",
Short: "Download Documents file bytes (default path: ./<title>)",
Args: cobra.ExactArgs(1),
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
ctx := cmd.Context()
f, err := c.GetFile(ctx, args[0])
if err != nil {
return err
}
path := to
if path == "" {
path = onlyoffice.SafeLocalFileName(onlyoffice.FileEntryTitle(f))
}
out, err := os.Create(path)
if err != nil {
return err
}
defer out.Close()
n, err := c.DownloadDavFile(ctx, args[0], out)
if err != nil {
_ = os.Remove(path)
return err
}
printObject(map[string]any{"path": path, "bytes": n})
return nil
},
}
cmd.Flags().StringVar(&to, "to", "", "output path (default: ./<server title>)")
return cmd
}
func davFileOpsCmd() *cobra.Command {
return &cobra.Command{
Use: "fileops",
Short: "List active file operations (move/copy status polling)",
Args: cobra.NoArgs,
RunE: func(cmd *cobra.Command, args []string) error {
c, err := newOO(cmd)
if err != nil {
return err
}
ops, err := c.ListFileOps(cmd.Context())
if err != nil {
return err
}
rows := make([]map[string]any, 0, len(ops))
for _, op := range ops {
rows = append(rows, map[string]any{
"id": fmt.Sprint(op["id"]),
"operation": fmt.Sprint(op["operation"]),
"progress": fmt.Sprint(op["progress"]),
"finished": fmt.Sprint(op["finished"]),
"error": fmt.Sprint(op["error"]),
})
}
printTable([]string{"id", "operation", "progress", "finished", "error"}, rows)
return nil
},
}
}
+1
View File
@@ -16,6 +16,7 @@
// oo mails accounts | folders | list | get | download-attachment | draft | attach | draft-invoice | delete // oo mails accounts | folders | list | get | download-attachment | draft | attach | draft-invoice | delete
// oo invoices list | get | create | update | pdf | pdf-cleanup | status | delete | items … // oo invoices list | get | create | update | pdf | pdf-cleanup | status | delete | items …
// oo docs tools | convert | ocr | as-md | put-md // oo docs tools | convert | ocr | as-md | put-md
// oo dav ls | move | copy | mkdir | rename-file | rename-folder | download | fileops
// //
// CRM association rules: docs/crm-associations.md // CRM association rules: docs/crm-associations.md
// //
+50
View File
@@ -0,0 +1,50 @@
// Command ooscan recursively lists OnlyOffice Documents folders into a TSV
// index: file_id, folder_id, path, title.
//
// Usage: ooscan <FOLDER_ID> [<FOLDER_ID>...]
package main
import (
"context"
"fmt"
"os"
"time"
onlyoffice "github.com/eslider/go-onlyoffice"
)
func main() {
ctx := context.Background()
c := onlyoffice.NewClient(onlyoffice.GetEnvironmentCredentials())
seen := map[string]bool{}
for _, root := range os.Args[1:] {
walk(ctx, c, root, "", 0, seen)
}
}
func walk(ctx context.Context, c *onlyoffice.Client, folderID, path string, depth int, seen map[string]bool) {
if depth > 8 || seen[folderID] {
return
}
seen[folderID] = true
// Throttle: OnlyOffice rate-limits (429) and the host must not be flooded.
time.Sleep(350 * time.Millisecond)
ctx, cancel := context.WithTimeout(ctx, 60*time.Second)
defer cancel()
var l *onlyoffice.DavListing
derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
var err error
l, err = c.ListDavFolder(ctx, folderID)
return err
})
if derr != nil {
fmt.Fprintf(os.Stderr, "list %s (%s): %v\n", path, folderID, derr)
return
}
for _, f := range l.Files {
fmt.Printf("%s\t%s\t%s\t%s\n", f.ID, folderID, path, f.Title)
}
for _, sub := range l.Folders {
walk(ctx, c, sub.ID, path+"/"+sub.Title, depth+1, seen)
}
}
+124
View File
@@ -0,0 +1,124 @@
// Command pdfamount walks a Documents folder, downloads matching PDFs and
// extracts the payable amount, printing "file_id\ttitle\tamount".
//
// Usage: pdfamount <FOLDER_ID> [TITLE_FILTER_REGEX]
package main
import (
"bytes"
"context"
"fmt"
"os"
"os/exec"
"regexp"
"strconv"
"strings"
"time"
onlyoffice "github.com/eslider/go-onlyoffice"
)
var amountRe = regexp.MustCompile(`(?i)(zu zahlender betrag|rechnungsbetrag)\s*[:\s]*([0-9][0-9.]*,[0-9]{2})`)
func main() {
if len(os.Args) < 2 {
fmt.Fprintln(os.Stderr, "usage: pdfamount <FOLDER_ID> [TITLE_FILTER_REGEX]")
os.Exit(2)
}
folder := os.Args[1]
filter := regexp.MustCompile(`(?i)rechnung`)
if len(os.Args) >= 3 {
filter = regexp.MustCompile(os.Args[2])
}
ctx := context.Background()
c := onlyoffice.NewClient(onlyoffice.GetEnvironmentCredentials())
files := listAll(ctx, c, folder)
for _, f := range files {
if !filter.MatchString(f.title) {
continue
}
if !strings.HasSuffix(strings.ToLower(f.title), ".pdf") {
continue
}
amount, err := pdfAmount(ctx, c, f.id)
if err != nil {
fmt.Fprintf(os.Stderr, "%s: %v\n", f.title, err)
continue
}
if amount == "" {
continue
}
fmt.Printf("%s\t%s\t%s\n", f.id, f.title, amount)
}
}
type file struct{ id, title string }
func listAll(ctx context.Context, c *onlyoffice.Client, folder string) []file {
seen := map[string]bool{}
var out []file
var walk func(string)
walk = func(id string) {
if seen[id] {
return
}
seen[id] = true
time.Sleep(300 * time.Millisecond)
l, err := c.ListDavFolder(ctx, id)
if err != nil {
fmt.Fprintf(os.Stderr, "list %s: %v\n", id, err)
return
}
for _, f := range l.Files {
out = append(out, file{f.ID, f.Title})
}
for _, sub := range l.Folders {
walk(sub.ID)
}
}
walk(folder)
return out
}
func pdfAmount(ctx context.Context, c *onlyoffice.Client, id string) (string, error) {
time.Sleep(time.Second)
tmp, err := os.CreateTemp("", "pdf-*.pdf")
if err != nil {
return "", err
}
defer os.Remove(tmp.Name())
derr := onlyoffice.DoRetry(ctx, onlyoffice.DefaultRetryPolicy(), func() error {
_ = tmp.Truncate(0)
_, _ = tmp.Seek(0, 0)
_, err := c.DownloadFile(ctx, id, tmp)
return err
})
if derr != nil {
tmp.Close()
return "", derr
}
tmp.Close()
var buf bytes.Buffer
cmd := exec.CommandContext(ctx, "pdftotext", "-layout", tmp.Name(), "-")
cmd.Stdout = &buf
if err := cmd.Run(); err != nil {
return "", err
}
m := amountRe.FindStringSubmatch(buf.String())
if m == nil {
return "", nil
}
return parseDe(m[2]), nil
}
// parseDe turns "1.234,56" into 1234.56.
func parseDe(s string) string {
s = strings.ReplaceAll(s, ".", "")
s = strings.ReplaceAll(s, ",", ".")
v, err := strconv.ParseFloat(s, 64)
if err != nil {
return s
}
return strconv.FormatFloat(v, 'f', 2, 64)
}
+28
View File
@@ -321,11 +321,21 @@ func (c *Client) MoveFiles(ctx context.Context, destFolderID int, fileIDs []int)
"folderIds": []int{}, "folderIds": []int{},
"fileIds": fileIDs, "fileIds": fileIDs,
"destFolderId": destFolderID, "destFolderId": destFolderID,
"resolveType": "Skip",
"holdResult": true,
} }
out, err := c.putJSONObject(ctx, "/api/2.0/files/fileops/move.json", body) out, err := c.putJSONObject(ctx, "/api/2.0/files/fileops/move.json", body)
if err != nil { if err != nil {
out, err = c.putJSONObject(ctx, "/api/2.0/files/fileops/move", body) out, err = c.putJSONObject(ctx, "/api/2.0/files/fileops/move", body)
} }
if err != nil {
return nil, err
}
if raw, merr := json.Marshal(out); merr == nil {
if ferr := fileopsError(raw); ferr != nil {
return nil, ferr
}
}
return out, err return out, err
} }
@@ -346,6 +356,24 @@ func (c *Client) UploadToFolder(ctx context.Context, folderID, localPath string)
return decodeResponseFileEntry(raw) return decodeResponseFileEntry(raw)
} }
// UpdateFile uploads a new version of an existing file (same id, name and
// folder). It does not delete and does not create a second file.
func (c *Client) UpdateFile(ctx context.Context, fileID, localPath string) (*FileEntry, error) {
if fileID == "" || localPath == "" {
return nil, fmt.Errorf("file id and local path are required")
}
uploadPath := fmt.Sprintf("/api/2.0/files/%s/update", url.PathEscape(fileID))
raw, err := c.uploadMultipart(ctx, uploadPath, "file", localPath)
if err != nil {
uploadPath = fmt.Sprintf("/api/2.0/files/%s/update.json", url.PathEscape(fileID))
raw, err = c.uploadMultipart(ctx, uploadPath, "file", localPath)
if err != nil {
return nil, err
}
}
return decodeResponseFileEntry(raw)
}
// FileFolderID returns the parent folder id string for a file entry, if known. // FileFolderID returns the parent folder id string for a file entry, if known.
func FileFolderID(f *FileEntry) string { func FileFolderID(f *FileEntry) string {
if f == nil || f.FolderID == nil { if f == nil || f.FolderID == nil {
+60 -2
View File
@@ -144,27 +144,85 @@ func (c *Client) RenameDavFile(ctx context.Context, id, title string) error {
} }
// MoveDavItems moves the given folders and/or files into destFolderID. // MoveDavItems moves the given folders and/or files into destFolderID.
// The fileops API answers 200 with per-operation error strings even when
// nothing moves (e.g. missing permission), so the response is parsed and the
// first operation error is returned instead of a silent nil.
func (c *Client) MoveDavItems(ctx context.Context, folderIDs, fileIDs []string, destFolderID string) error { func (c *Client) MoveDavItems(ctx context.Context, folderIDs, fileIDs []string, destFolderID string) error {
_, err := c.putJSON(ctx, "/api/2.0/files/fileops/move", map[string]any{ raw, err := c.putJSON(ctx, "/api/2.0/files/fileops/move", map[string]any{
"folderIds": nums(folderIDs), "folderIds": nums(folderIDs),
"fileIds": nums(fileIDs), "fileIds": nums(fileIDs),
"destFolderId": num(destFolderID), "destFolderId": num(destFolderID),
"resolveType": "Skip", "resolveType": "Skip",
"holdResult": true, "holdResult": true,
}) })
if err != nil {
return err return err
}
return fileopsError(raw)
} }
// CopyDavItems copies the given folders and/or files into destFolderID. // CopyDavItems copies the given folders and/or files into destFolderID.
// Per-operation errors are surfaced like in MoveDavItems.
func (c *Client) CopyDavItems(ctx context.Context, folderIDs, fileIDs []string, destFolderID string) error { func (c *Client) CopyDavItems(ctx context.Context, folderIDs, fileIDs []string, destFolderID string) error {
_, err := c.putJSON(ctx, "/api/2.0/files/fileops/copy", map[string]any{ raw, err := c.putJSON(ctx, "/api/2.0/files/fileops/copy", map[string]any{
"folderIds": nums(folderIDs), "folderIds": nums(folderIDs),
"fileIds": nums(fileIDs), "fileIds": nums(fileIDs),
"destFolderId": num(destFolderID), "destFolderId": num(destFolderID),
"conflictResolveType": "Skip", "conflictResolveType": "Skip",
"deleteAfter": true, "deleteAfter": true,
}) })
if err != nil {
return err return err
}
return fileopsError(raw)
}
// ListFileOps returns the currently active file operations
// (GET /api/2.0/files/fileops) for status polling.
func (c *Client) ListFileOps(ctx context.Context) ([]map[string]any, error) {
raw, err := c.getJSON(ctx, "/api/2.0/files/fileops")
if err != nil {
return nil, err
}
resp, err := responseField(raw, "response")
if err != nil {
return nil, err
}
if len(resp) == 0 || string(resp) == "null" {
return nil, nil
}
var ops []map[string]any
if err := json.Unmarshal(resp, &ops); err != nil {
return nil, err
}
return ops, nil
}
// fileopsError extracts per-operation "error" strings from a fileops/move or
// fileops/copy envelope. A 200 with error entries means nothing moved.
func fileopsError(raw json.RawMessage) error {
resp, err := responseField(raw, "response")
if err != nil {
return err
}
var ops []struct {
Error *string `json:"error"`
Finished *bool `json:"finished"`
Progress *int `json:"progress"`
}
if err := json.Unmarshal(resp, &ops); err != nil {
return nil // not an operations envelope — nothing to report
}
var errs []string
for _, op := range ops {
if op.Error != nil && *op.Error != "" {
errs = append(errs, *op.Error)
}
}
if len(errs) > 0 {
return fmt.Errorf("onlyoffice: fileops: %s", strings.Join(errs, "; "))
}
return nil
} }
// DeleteDavItems deletes the given folders and/or files. // DeleteDavItems deletes the given folders and/or files.
+60
View File
@@ -0,0 +1,60 @@
package onlyoffice
import (
"context"
"regexp"
"time"
)
// RetryPolicy controls deterministic retries against OnlyOffice: fixed linear
// backoff without jitter, so repeated runs wait exactly the same schedule.
// OnlyOffice throttles bulk reads/writes with 429 (and occasional 502/503/504
// from openresty), so every bulk tool routes API calls through DoRetry.
type RetryPolicy struct {
Attempts int // total attempts, including the first try
Base time.Duration // wait before retry N is N*Base
Max time.Duration // per-wait cap
}
// DefaultRetryPolicy retries up to 5 times with 1s, 2s, 3s, 4s waits.
func DefaultRetryPolicy() RetryPolicy {
return RetryPolicy{Attempts: 5, Base: time.Second, Max: 30 * time.Second}
}
var transientRe = regexp.MustCompile(`:\s*(429|502|503|504)\b`)
// Transient reports whether err looks like a transient OnlyOffice answer
// (an HTTP 429/502/503/504 surfaced as "...: <code> ...").
func Transient(err error) bool {
if err == nil {
return false
}
return transientRe.MatchString(err.Error())
}
// DoRetry runs fn until it succeeds, fails non-transiently, or attempts run
// out. Waits are deterministic: N*Base capped at Max, no jitter.
func DoRetry(ctx context.Context, p RetryPolicy, fn func() error) error {
if p.Attempts < 1 {
p.Attempts = 1
}
var err error
for attempt := 1; attempt <= p.Attempts; attempt++ {
if ctx.Err() != nil {
return ctx.Err()
}
if err = fn(); err == nil || !Transient(err) || attempt == p.Attempts {
return err
}
wait := time.Duration(attempt) * p.Base
if wait > p.Max {
wait = p.Max
}
select {
case <-ctx.Done():
return ctx.Err()
case <-time.After(wait):
}
}
return err
}