Drei while ((m = re.exec(t)) !== null)-Schleifen (dkv-parser.service.ts,
dkv-parser.validate.ts, icon-discovery.service.ts) sind die korrekte
Standardform fuer globale Regexe - kein verrutschtes "=". Umgeschrieben auf
eine verhaltensgleiche for-Schleife, die ohne noAssignInExpressions-
Unterdrueckung auskommt: Zuweisung wandert in Initialisierung und
Fortschaltung der for-Schleife, Bedingung prueft weiterhin auf null.
Abfolge der exec-Aufrufe, lastIndex-Fortschritt und Rumpfinhalte
unveraendert. Neue Spezifikation dkv-parser.service.spec.ts deckt
parseDkvText erstmals eigenstaendig ab (zwei Fahrzeugbloecke, Rechnungsnummer
und -datum aus einer gemockten pdf-parse-Attrappe) - das Rueckfall-Tor fuer
diesen Umbau. dkv-parser.validate.ts bleibt bei 27 Fahrzeugbloecken/66
Transaktionen gegen die reale invoice.pdf identisch.
LdapService.escapeLdapFilterValue bleibt zeichengleich: der NUL-Treffer in
der Regel ist die von RFC 4515 vorgeschriebene \00-Maskierung, kein Fehler.
Ein biome-ignore-Kommentar dokumentiert das, statt die Funktion zu aendern.
locale-switcher.tsx setzt jetzt SameSite=Lax auf dem NEXT_LOCALE-Cookie -
path=/ und max-age waren bereits korrekt, es lag also kein Persistenzdefekt
vor. Ohne SameSite haengt die Uebertragung am Browservorgabewert statt an
einer Festlegung. Neue Spezifikation locale-switcher.test.tsx haelt die
vollstaendige geschriebene Cookie-Zeichenkette fest.
Biome-Warnungen 446 -> 434 (noControlCharactersInRegex/useIterableCallbackReturn/
noGlobalIsNan auf 0, noAssignInExpressions auf 4 und noDocumentCookie auf 17 -
beide Reste ausschliesslich in Testdateien, suppressions/unused auf 0).
Quick-Vorgang 260921-i8x, Task 3/3.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TPPB4ApQxzSU1rwV2Ffj9J
- Aufgabe 2: vier sichere Biome-Regeln (useImportType pfadgebunden auf
apps/web+packages, noUselessEscapeInRegex, useConst,
useExponentiationOperator) sowie fuenf ungesicherte Regeln
(useNodejsImportProtocol, useLiteralKeys, useOptionalChain, useTemplate,
useParseIntRadix) angewendet und den gesamten Diff von Hand gelesen
(ldap.service.ts zeichenweise gegen Gross-/Kleinschreibung der
AD-Merkmale, auth.service.ts/jwt.strategy.ts gegen Durchwinken bei
fehlender Sitzung geprueft)
- noUselessSwitchCase bleibt bewusst stehen (tender-normalizer.service.ts:60,
die Fallmarke dokumentiert Absicht)
- Toter Code (D-03): fuenf folgenlose Auffangvariablen entfernt, eine
nicht benutzte Funktion (forSystemQuery, Pruefskript) entfernt, ein
positionsgebundener Dekoratorparameter umbenannt (current-user.decorator.ts),
fuenf Symptomfunde entfernt und als Folgeaufgaben zu melden (siehe unten)
- Sechs weitere, im Plan nicht namentlich gelistete aber
gleich-kategorische Dead-Code-Fundstellen in Testdateien zusaetzlich
bereinigt (groups.service.spec.ts, cert-manager.test.tsx,
ldap.service.spec.ts, prisma-tenant.extension.spec.ts x3) — noetig, um
die vom Plan selbst verlangten Nullstaende bei noUnusedVariables/
noUnusedImports/noUnusedFunctionParameters zu erreichen
Dekoratordaten aus apps/api unveraendert (593 Zeilen, sha256 6e1583f1...).
Endstand 620 Befunde (541 echt, 79 Test) statt der im Plan geschaetzten
621/542 — eine Differenz von 1, weil das Streichen des Namens aus
`catch (e: any)` in calendar.service.ts (Symptom-Fix) den dort ebenfalls
gemeldeten noExplicitAny-Befund miteliminiert; das ist eine erwuenschte
Nebenwirkung, keine Regression. Fehlerstufe 0, beide Testlaeufe
punktgleich gruen (69/1124, 66/459), pnpm type-check 4/4, pnpm lint
--force 5/5.
Folgeaufgaben aus D-03 (nicht in diesem Vorgang behoben):
- force-password-change.interceptor.ts: Freigabeliste prueft nur den Pfad,
nicht die HTTP-Methode
- change-password/page.tsx: nach erzwungenem Wechsel bleibt die Person auf
der Seite stehen (keine Weiterleitung, keine Aktualisierung der
Benutzerablage)
- VehicleTable.tsx: Loeschschaltflaeche hat keinen Besetztzustand, laesst
sich doppelt ausloesen
- SplitTab.tsx: downloadAllAsZip erhielt eine ungenutzte
Uebersetzungsfunktion, Hinweis auf fest verdrahtete Texte im Zip-Pfad
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TPPB4ApQxzSU1rwV2Ffj9J
Die vorige Aufgabe-2-Teilcommit (11f5731) hatte nur die Loeschung von
tenant.middleware.ts und die neue tenant.guard.spec.ts erfasst — ein
`git add` mit mehreren Pfaden schlug wegen eines bereits entfernten
Pfads fataler fehl und liess die restlichen fuenf Dateien unstaged,
ohne dass das beim Commit auffiel (Rule 1 — Prozessfehler, hier
korrigiert). Dieser Commit traegt den eigentlichen Umbau nach:
tenant.guard.ts ohne Prisma-Abhaengigkeit, die geleerte
FORTENANT_ASSIGNMENT_EXCEPTIONS samt Wachhund-Test in
rls-access-inventory.spec.ts, und die drei berichtigten
Kommentarzeilen (app.module.ts, module.guard.ts, dkv.controller.ts).
Inhaltlich identisch mit dem, was bereits verifiziert wurde (891 Tests
gruen, Typpruefung sauber) — nur die Staging-Reihenfolge war fehlerhaft.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AMASaSxv5QMY7RncqZriRR
- apps/api/src/dkv/dkv.service.spec.ts (neu): Zwei-Klienten-Nachweis nach
dem Muster aus groups.service.spec.ts/tender-triage.service.spec.ts —
dieser Bereich hatte vorher KEINE Testdatei (Befund J). 7 Testfaelle
decken getConfigForApi, saveConfig (Zugangsdaten-Erhaltung), testConnection,
die Verarbeitungsstrecke und den bewusst ungebundenen Planer-Startpfad ab
- dkv.service.ts: loadConfig(tenantId?) in zwei Methoden geteilt —
loadConfig(tenantId) [Pflicht-Mandant, gebunden] und die neue, eigene
Methode loadAnyActiveConfigForScheduler() [bewusst UNGEBUNDEN, eigener
Kopfkommentar mit beiden Zustaenden]. getConfigForApi/saveConfig/
testConnection/_runPipeline binden je EINEN Klienten pro Methode
vollstaendig ueber forTenant()
- dkv-scheduler.service.ts: Kopfkommentar fortgeschrieben (beide Zustaende,
Praezedenzfall, Unsymmetrie), Aufruf auf loadAnyActiveConfigForScheduler()
umgestellt — an der Ablauflogik des Planers nichts geaendert
- .planning/WINDOWS.md: Eintrag #21 (deviation) fuer die benannte Altlast
des Planer-Startpfads angelegt
- docs/mandantentrennung-zugriffsklassifikation.md: dkvModuleConfig-Zeile
auf den jetzt gemessenen Stand "gemischt" nachgezogen (Rule 3 — noetig,
damit rls-access-inventory.spec.ts nach der Aufteilung von loadConfig()
gruen bleibt; die uebrigen zwei dkv-Zeilen und die Uebersichtstabelle
bleiben Aufgabe 3 vorbehalten)
- Falsifizierungsnachweis erbracht: getConfigForApi's erster gebundener
Client probeweise durch this.prisma ersetzt, genau Test 1 wurde rot
(6 andere blieben gruen), Rueckbau zurueckgenommen, Dateien identisch
zum Ausgangsstand bestaetigt
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AMASaSxv5QMY7RncqZriRR
CALENDAR_ENCRYPTION_KEY was named after the calendar module because that
module needed encryption first, in Phase 5. Every feature since has shared the
same key -- SMTP, the DKV and tender mailboxes, and as of today the LDAP bind
password -- so the name has been describing one of five users rather than the
thing itself, and each new feature inherited the confusion.
TESSERA_ENCRYPTION_KEY is the name now. The old one is still read, because
renaming outright would stop every existing installation at the next start:
their .env carries the old name, and compose was just made to fail hard on a
missing key. When only the old name is present the API logs a deprecation
warning naming both, and when both are set the new one wins -- otherwise a
half-migrated .env would encrypt with one key and decrypt with the other.
CalendarCryptoService becomes CryptoService in its own global CryptoModule.
Four modules used to import CalendarModule purely to reach the provider, which
read as a dependency on calendars where there was none; that import is gone.
Compose keeps the hard failure: without either name the stack refuses to
start. Verified in both files for all three cases -- neither name set (abort),
only the old name (starts), only the new name (starts).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- Move ImapProvider, ExchangeInboxProvider, InboxProvider into apps/api/src/inbox/
- Move InboxConfig/InboxAttachment/InboxEmail into new inbox.types.ts
- dkv.types.ts re-exports the moved types so existing DKV imports keep compiling
- DKV switches import paths to ../inbox/... and imports InboxModule
- Pure move + import-path swap: fetchPdfAttachments and all DKV logic unchanged (D-01/D-02)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
FindFolder was searching only under inbox DistinguishedFolderId, missing
folders at mailbox root level. Now searches msgfolderroot (full mailbox)
so custom folders like DKV are found regardless of placement.
Also adds HTTP status check and debug logging for FindFolder responses.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Custom folder names (e.g. "DKV" or "INBOX/DKV") now resolved by calling
EWS FindFolder deep-search under inbox. Well-known names still map to
DistinguishedFolderId directly. Falls back to inbox with a warning log
when the subfolder cannot be found.
IMAP already supported subfolder paths natively via ImapFlow.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
loadConfig used CONFIG_SAFE_SELECT which excludes encryptedInboxCreds entirely,
so username was never returned to the frontend — form always showed empty username.
Added getConfigForApi() which loads the safe config + decrypts encryptedInboxCreds
to extract username (never password) and adds hasPassword boolean. Controller
getConfig now calls getConfigForApi instead of loadConfig.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Date fix: previous regex matched payment-due date ("10 Tage nach Rechnungsdatum...
10.04.2026") instead of actual Rechnungsdatum. New approach anchors on the
invoice number line (DD/DDDDDDDDD/DDD) and takes the date on the next line,
which is always the actual Rechnungsdatum in DKV PDFs.
Exchange dedup: FindItem now filters IsRead=false (combined with sender filter
via <t:And>), so already-processed emails are skipped automatically.
After downloading attachments, UpdateItem marks the message as read
(using ItemId + ChangeKey from GetItem response), mirroring IMAP \Seen behavior.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Parser now extracts Rechnungsnummer (DD/DDDDDDDDD/DDD) and Rechnungsdatum
from PDF text, so filename doesn't rely on email subject
- Export filename changed from DKV_YYYY-MM_... to RG-DKV-{nr}-{YYMMDD}.xlsx
e.g. RG-DKV-26-650869002-002-260331.xlsx
- Subject fallback now also matches slash-separated invoice numbers (26/NNN/NNN)
- writeAndPrune simplified to accept baseName instead of separate fields
- Validation regex and prune prefix updated to match new RG-DKV- pattern
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Analyzed Invoice-4302486921-26_650869002_000.pdf text structure. Three cases:
1. FUEL (fields[4] = numeric tx-nr): km+product merged in fields[5], unit in
fields[6]. Already working; no change.
2. EV CHARGING (fields contains "DDDD KWH" or "DDDD MIN" unit): column layout
shifts — no km field, station+ort sometimes merged in fields[1]. Detected by
regex on unit field; kwhIdx drives relative offset for menge/netto/brutto.
Ort extracted from fields[2] (kwhIdx>=5) or fields[1] (kwhIdx=4, compact).
Kilometerstand = 0 (EV chargers don't record odometer).
3. SERVICE ROWS (e.g. "DKV Analytics Premiu"): appear inside a VEHICLE: block but
fields[4] is non-numeric (product description, not a transaction number). These
were being parsed as fake vehicle transactions producing wrong ort/km values.
Now filtered out (return null).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Three issues fixed:
1. Kennzeichen normalization: DKV PDF extracts plates without hyphens
("GP JL 740E" vs CSV-imported "GP-JL 740E"). Added _normalizeKennzeichen()
which strips hyphens, spaces, and dots before lookup — resolves vehicle
master match failure that caused Marke/Modell/Fahrer to appear empty.
2. Empty-string NaN: parser used ?? '0' which doesn't catch empty strings,
causing parseDE('') = NaN. Changed to || '0' for km, menge, and totals.
3. Invalid km values: EV charging rows from DKV have misaligned columns —
km position contains a decimal price (e.g. 18.64 EUR or kWh). Added
sanity check: non-integer km values are written as null (empty cell)
instead of a misleading decimal. ExportRow.kilometerstand is now number|null.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
extractAttrs(block, 't:FileAttachment', 'Id') always returned empty array
because the attachment Id lives in a child <t:AttachmentId Id="..."/>, not
on the <t:FileAttachment> tag itself. This caused all Exchange inbox checks
to silently find zero PDF attachments and report "no matching emails".
Fixed by iterating FileAttachment blocks individually and extracting
t:AttachmentId/@Id from within each block. Also added filename (.pdf)
as fallback when ContentType is application/octet-stream.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Fix orphaned DkvModuleConfig: tenantId pointed to deleted tenant, updated to Default tenant
- DKV controller: return 404 instead of HTTP 200 null when no config exists
- IMAP provider: also detect PDFs sent as application/octet-stream (check filename extension)
- IMAP provider: add seen:false filter so already-processed emails are skipped on re-poll
- IMAP provider: mark email as \Seen after successful PDF download to prevent reprocessing
- Frontend dkv-api: handle 404 from fetchConfig as "not yet configured" (returns null)
- InboxConfigForm: show warning banner when config not yet saved in DB
- InboxConfigForm: add "Jetzt prüfen" button to manually trigger POST /dkv/check-now
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- ExchangeInboxProvider rewritten to use httpntlm + raw EWS SOAP:
FindItem / GetItem / GetAttachment via NTLM challenge-response.
No longer requires Basic Auth on Exchange EWS virtual directory.
Folder name mapped to EWS DistinguishedFolderId (Inbox/SentItems/etc).
- CalendarCryptoService: move key init from onModuleInit to constructor
so MailModule.forRootAsync() factory can call decrypt() before NestJS
lifecycle hooks execute (startup crash when SmtpConfig row has password).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- testConnection now returns { success, message? } instead of boolean so
admins see the actual EWS/IMAP error in the UI rather than "Unbekannter Fehler"
- Exchange provider: resolveFolder() maps folder string to WellKnownFolderName
(Inbox, SentItems, DeletedItems, Drafts, JunkEmail + German aliases)
- InboxConfigForm: folder field now shown for both IMAP and Exchange protocols
with Exchange-specific help text listing valid well-known names
- Controller returns testConnection result directly (no more redundant wrapping)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add `cron@4.4.0` as direct dep (pnpm strict isolation blocks transitive access)
- Import SettingsModule in DkvModule so DkvMailService can inject SettingsService
- Fix dkv.service.ts return key: `count` → `imported` to match declared return type
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
FileInterceptor used multer's default memory storage with no size limit.
An oversized file could exhaust Node.js heap before parsing begins.
Add fileSize: 5*1024*1024 (5 MB) — sufficient for any realistic vehicle list.
Exchange EWS UniqueIds are base64-encoded and can contain +, /, = characters.
When used as the fallback rechnungsnummer (email-{uid}), a slash would cause
path.join() to resolve into a subdirectory, making writeFileSync fail silently.
Sanitise uid to [a-zA-Z0-9-] before it reaches the filesystem write path.
HTTP query params arrive as strings. Without @Type(() => Number),
class-transformer never coerces page/limit before @IsInt() runs,
causing HTTP 400 for any request that explicitly passes ?page or ?limit.
Also adds @Max(100) on limit to bound result-set size.
Each sendExportEmail call created a new nodemailer transport which was never
closed, leaving the internal SMTP connection pool alive. With 3-retry backoff,
up to 3 leaked transports per invoice accumulate over time and can exhaust OS
socket limits. Add transport.close() in a finally block.
If getMailboxLock() failed (e.g. folder not found) the try/finally cleanup
block was never entered, leaving the ImapFlow connection open and leaking.
Move the lock acquisition inside the try block and use lock?.release() in
finally so client.logout() is always called regardless of lock success.
Values >=60 in the cron minute field silently misbehave (*/60 fires once per
hour, */90 fires once per hour, etc.). Use the hours field for intervals >=60:
<60 min → */N * * * *
>=60 min → 0 */H * * * (H = floor(N/60))
Also adds @Max(1440) to DkvConfigDto to bound the field at 24 h.
Backend dkv.service.ts returned { imported } but frontend read result.count,
causing the success toast to always display "undefined Fahrzeuge importiert".
Align backend field name to count and update the dkv-api.ts return type to
include mode for completeness.
- DkvSchedulerService: SchedulerRegistry.addCronJob (dynamic interval, not static @Cron)
- onModuleInit loads first active config (v1 single-tenant, documented in SUMMARY)
- setInterval() replaces existing job and registers new one with */ cron expression
- stopJob() removes job when config.isActive=false
- cron package resolved via require() workaround (pnpm strict isolation: transitive dep)
- DkvController: 12 handlers all carrying @Roles(Role.ADMIN, Role.SUPER_ADMIN)
- Routes: GET/PUT config, POST check-now, POST test-connection, GET history,
GET exports/:filename, GET/POST/PUT/DELETE vehicles, POST vehicles/import
- vehicles/import uses FileInterceptor('file') for CSV multipart upload
- exports/:filename streams file as attachment; traversal guard in DkvService
- Controller coordinates scheduler after PUT /dkv/config (no circular dep)
- createTransport() called per-send from DB SmtpConfig (Pitfall 3 mitigation — not at startup)
- Injects SettingsService to load decrypted SMTP config per tenant
- secure/requireTLS mapped from encryption field (ssl-tls / starttls / none)
- Auth omitted when username absent (anonymous relay support)
- Attachment contentType: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet
- Error path rethrows after generic log (T-07-10) so DkvService can run 3-retry backoff (D-16)
- Does not import @nestjs-modules/mailer abstractions
- buildExcelBuffer: 5-column xlsx per D-13 (Lieferdatum as string, never Date), SheetJS aoa_to_sheet
- resolveFahrzeug: replaces {Marke}/{Modell}/{Kennzeichen}/{Fahrer} tokens in format string (D-19)
- writeAndPrune: server-side filename DKV_YYYY-MM_<nr>.xlsx (T-07-09 path-traversal prevention),
writes to user-files/ (resolved from monorepo root, not request input), prunes to last 10 DKV_*.xlsx
files sorted by mtime ascending (D-15, Pitfall 7 atomicity)
- dkv-parser.validate.ts: empirical validation script against user-files/invoice.pdf
- 27 vehicle blocks, 66 transactions extracted (matches expected count)
- Handles two PDF extraction formats: single-tx (tab-separated) + multi-tx (columnar)
- German number parsing: replace(/\./g,'').replace(',','.') applied to km and menge
- Exits 1 with full raw text dump if zero vehicle blocks parsed (assertion guard)
- dkv-parser.service.ts: @Injectable() NestJS service wrapping validated logic
- parsePdf(buffer: Buffer): Promise<DkvVehicleBlock[]>
- Uses pdf-parse v2 class API: new PDFParse({data:buffer}) — NOT v1 pdfParse()
- Calls destroy() after extraction (T-07-01 memory safety)
- Generic error messages only on parse failure (T-07-02 info disclosure)
Regex adjustment vs Research Pattern 4: space-based regex replaced with
tab-split (single-tx) + columnar transpose (multi-tx) after empirical analysis
of actual invoice.pdf text extraction output.
schalli