91e2c66afc
* fix(parties): readable suggestions from assistant-written vouchers, auto-build queue, SCB fetch after promotion Live feedback on a real company (2026-09-03): the queue showed 35 one-off suggestions with sentence-long names, wide empty rows, a "Hämta förslag" step nobody could predict, no SCB fetch after promotion, and an empty supplier created from a Finansinspektionen fee line. - ledger_key v2 (migration 20260904002000): keep the counterpart head of "<counterpart> · <note>" descriptions, drop bank method tokens and long references before normalising; JS mirror in lib/parties/ledger-key.ts with shared LEDGER_KEY_CASES. Suggested parties nobody has touched are rebuilt under the new keys (repair in the same migration). - apply_party_suggestions attaches by VAT number too, so ledger keys with a VAT number but no org number reach existing roles. - Queue: fixed name/reason column widths, inline "Hitta i företagsregistret" for rows without an org number. - Page: builds the queue automatically on first visit when nothing has been suggested yet; after promotion, fetches SCB facts for every promoted legal person (spaced under the 10 calls/10 s limit) and fills the role's VAT number; confirm dialog says how many rows lack an org number. - Classifier: more authorities (Finansinspektionen, Arbetsförmedlingen, Pensionsmyndigheten, ...) and fee words (registreringsavgift, tillsynsavgift, ...) so fee lines stop becoming suppliers. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * fix(parties): scope the suggestion repair to keys the new ledger_key no longer produces Superagent flagged the repair DELETE as global. It now only removes untouched pipeline suggestions that no posted voucher of the company maps to under the new function; suggestions whose key is unchanged stay. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> --------- Co-authored-by: Jakob Wennberg <311770904+jakobwennberg-oss@users.noreply.github.com> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
101 lines
4.3 KiB
TypeScript
101 lines
4.3 KiB
TypeScript
import { normalizeCounterpartyName } from '@/lib/bookkeeping/counterparty-templates'
|
|
|
|
/**
|
|
* Legibility key for a voucher description: the identity string an observed
|
|
* party is grouped and displayed by.
|
|
*
|
|
* Built on top of normalizeCounterpartyName() (mirrored in SQL by
|
|
* normalize_counterparty_key) and adds the stages the AP registers of Fortnox,
|
|
* Visma and BL make necessary: "Levfakt BEIJER BYGGMATERIAL AB (2089)" and
|
|
* "Levfakt Beijer Byggmaterial AB, 097" must land on one key,
|
|
* "Leverantörsfaktura från 18 Loopia" on "loopia".
|
|
*
|
|
* Mirrored in SQL by public.ledger_key() (migration 20260902170000) and pinned
|
|
* by tests/pg/observed-parties-rpc.pg.test.ts. Change both or neither.
|
|
*
|
|
* "inköp" is deliberately not a stripped prefix: it turns the generic
|
|
* "inköp av varor" into a vendor-looking "varor" (measured 2026-07-27).
|
|
*/
|
|
const AP_PREFIX = /^(levfakt|levfkt|leverantörsfaktura från|leverantörsfaktura|levbet|faktura|kvitto|utgift)\s+/
|
|
const LEADING_SUPPLIER_NUMBER = /^\d{1,5}\s+/
|
|
const TRAILING_SHORT_DIGITS = /(\s+\d{1,3})+$/
|
|
const BANK_METHOD = /(kortköp\/uttag|kortkp\/uttag|överföring via internet|bg-bet\.? via internet|pg-bet\.? via internet|bg-bet\.?|autogiro)/gi
|
|
const GIRO_REFERENCE = /\b(bg|pg)\s*\d{5,}\b/gi
|
|
const CARD_REFERENCE = /\bk\d{3,6}\b/gi
|
|
const LONG_DIGITS = /\b\d{6,}\b/g
|
|
|
|
/**
|
|
* What our own booking flows write: "<counterpart> · <note>", bank method
|
|
* tokens and long references. Keep the head before " · " (or, when the head
|
|
* has no letters, the text after it up to the first comma), drop the tokens.
|
|
* Mirrors the pre-clean in the SQL ledger_key (20260904002000).
|
|
*/
|
|
export function preClean(raw: string): string {
|
|
let pre = raw
|
|
const sep = pre.indexOf(' · ')
|
|
if (sep >= 0) {
|
|
let head = pre.slice(0, sep)
|
|
if (!/\p{L}/u.test(head)) head = pre.slice(sep + 3).split(',')[0] ?? ''
|
|
pre = head
|
|
}
|
|
return pre.replace(BANK_METHOD, ' ').replace(GIRO_REFERENCE, ' ').replace(CARD_REFERENCE, ' ').replace(LONG_DIGITS, ' ')
|
|
}
|
|
|
|
export function ledgerKey(raw: string | null | undefined): string {
|
|
const k = normalizeCounterpartyName(preClean(raw ?? ''))
|
|
if (!k) return ''
|
|
const stripped = k
|
|
.replace(AP_PREFIX, '')
|
|
.replace(LEADING_SUPPLIER_NUMBER, '')
|
|
.replace(TRAILING_SHORT_DIGITS, '')
|
|
.replace(/\s+/g, ' ')
|
|
.trim()
|
|
return stripped === '' ? k : stripped
|
|
}
|
|
|
|
const CORE_AP_PREFIX = /^(levfakt|levfkt|lev\.?fakt\.?|leverantörsfaktura från|leverantörsfaktura|levbet\.?|kvitto|faktura|utgift|inköp)\s+/
|
|
const CORE_LEGAL_FORM = /\b(ab|aktiebolag|hb|kb|sverige|sweden|ltd|limited|oy|gmbh|inc|sarl|publ|filial)\b/g
|
|
|
|
/**
|
|
* The "core" of a key: what is left when AP prefixes, digit runs and legal
|
|
* form suffixes are gone. Two keys with one core are the same trade name,
|
|
* which is NOT the same party (Fortnox AB and Fortnox Finans AB share one),
|
|
* so the core only ever ranks or annotates candidates; it never merges.
|
|
* Measured on the document-anchored gold set 2026-09-02: pair precision
|
|
* 0.909, recall 0.776 (scripts/parties/README.md).
|
|
*/
|
|
export function coreKey(key: string): string {
|
|
return key
|
|
.toLowerCase()
|
|
.replace(CORE_AP_PREFIX, '')
|
|
.replace(/\b\d+\b/g, '')
|
|
.replace(CORE_LEGAL_FORM, '')
|
|
.replace(/[^a-zåäöé ]+/g, ' ')
|
|
.split(/\s+/)
|
|
.filter(Boolean)
|
|
.join(' ')
|
|
}
|
|
|
|
const DISPLAY_PREFIX =
|
|
/^(levfakt|levfkt|lev\.?fakt\.?|leverantörsfaktura från\s+\d*|leverantörsfaktura|levbet\.?|kundbet\.?|kundfaktura|kundfakt|inbetalning från|inbetalning|utbetalning till|utbetalning|betalning till|betalning|faktura från|faktura|kvitto|utgift)\s+/i
|
|
const DISPLAY_SUFFIX = /(\s*[,(]\s*\d{1,6}\s*\)?|\s+\d{1,4})+$/
|
|
|
|
/**
|
|
* A display name from raw voucher text: the AP/AR prefix and the supplier
|
|
* number go, the casing and the legal form stay. "Levfakt BEIJER
|
|
* BYGGMATERIAL AB (2089)" reads "BEIJER BYGGMATERIAL AB"; "Kundbet Acme
|
|
* Konsult AB" reads "Acme Konsult AB". Used only when no document carries
|
|
* a printed name; nothing here is generated, only removed.
|
|
*/
|
|
export function displayNameFromVoucherText(raw: string): string {
|
|
const cleaned = preClean(raw)
|
|
.trim()
|
|
.replace(DISPLAY_PREFIX, '')
|
|
.replace(/^\d{6,10}[,\s]+/, '')
|
|
.replace(DISPLAY_SUFFIX, '')
|
|
.replace(/\s+/g, ' ')
|
|
.replace(/^[,\s]+|[,\s]+$/g, '')
|
|
.trim()
|
|
return cleaned.length >= 2 ? cleaned : raw.trim()
|
|
}
|