77 lines
3.2 KiB
TypeScript
77 lines
3.2 KiB
TypeScript
/**
|
|
* Payslip bulk-upload auto-match (AK-P17b). Pure function — no DB, no IO — so it
|
|
* is unit-testable in isolation. Given an uploaded file name and the tenant's
|
|
* employees, it classifies the match confidence:
|
|
* - `hard`: the file name contains the employee's personal number as a token.
|
|
* - `soft`: the file name contains the employee's (last) name but no personal
|
|
* number hit, OR more than one employee matches by name.
|
|
* - `none`: nothing matched.
|
|
* The admin confirms/overrides before any payslip is created. This module maps a
|
|
* file name to a single best classification plus the ambiguous candidate set.
|
|
*/
|
|
|
|
export type PayslipMatchConfidence = "hard" | "soft" | "none";
|
|
|
|
export interface PayslipMatchEmployee {
|
|
readonly userId: string;
|
|
readonly personalNumber: string | null;
|
|
/** Display name (e.g. "Alex Doe"); matched token-wise, case-insensitive. */
|
|
readonly name: string;
|
|
}
|
|
|
|
export interface PayslipMatchResult {
|
|
readonly fileName: string;
|
|
readonly confidence: PayslipMatchConfidence;
|
|
/** The single matched employee when unambiguous; null for none/ambiguous. */
|
|
readonly employeeUserId: string | null;
|
|
/** All plausible candidates (length > 1 ⇒ ambiguous ⇒ soft). */
|
|
readonly candidateUserIds: readonly string[];
|
|
}
|
|
|
|
/** Lowercased alphanumeric tokens of length ≥ 2 (drops the file extension). */
|
|
function tokenize(value: string): string[] {
|
|
const withoutExt = value.replace(/\.[a-z0-9]{1,5}$/i, "");
|
|
return withoutExt
|
|
.toLowerCase()
|
|
.split(/[^a-z0-9]+/i)
|
|
.filter((token) => token.length >= 2);
|
|
}
|
|
|
|
/** Normalised digits of a personal number (so "PN-0042" matches "0042"). */
|
|
function digitsOf(value: string): string {
|
|
return value.replace(/\D+/g, "");
|
|
}
|
|
|
|
export function matchPayslipFile(fileName: string, employees: readonly PayslipMatchEmployee[]): PayslipMatchResult {
|
|
const tokens = tokenize(fileName);
|
|
const tokenSet = new Set(tokens);
|
|
const haystackDigits = digitsOf(fileName);
|
|
|
|
// 1. Hard match on personal number (unique digit token in the file name).
|
|
const hardHits = employees.filter((employee) => {
|
|
const pn = employee.personalNumber ? digitsOf(employee.personalNumber) : "";
|
|
return pn.length >= 2 && haystackDigits.includes(pn);
|
|
});
|
|
if (hardHits.length === 1) {
|
|
return { fileName, confidence: "hard", employeeUserId: hardHits[0]!.userId, candidateUserIds: [hardHits[0]!.userId] };
|
|
}
|
|
if (hardHits.length > 1) {
|
|
// Multiple personal-number hits ⇒ ambiguous, downgrade to soft.
|
|
return { fileName, confidence: "soft", employeeUserId: null, candidateUserIds: hardHits.map((e) => e.userId) };
|
|
}
|
|
|
|
// 2. Soft match on name tokens (every name token present in the file name).
|
|
const nameHits = employees.filter((employee) => {
|
|
const nameTokens = tokenize(employee.name);
|
|
return nameTokens.length > 0 && nameTokens.every((token) => tokenSet.has(token));
|
|
});
|
|
if (nameHits.length === 1) {
|
|
return { fileName, confidence: "soft", employeeUserId: nameHits[0]!.userId, candidateUserIds: [nameHits[0]!.userId] };
|
|
}
|
|
if (nameHits.length > 1) {
|
|
return { fileName, confidence: "soft", employeeUserId: null, candidateUserIds: nameHits.map((e) => e.userId) };
|
|
}
|
|
|
|
return { fileName, confidence: "none", employeeUserId: null, candidateUserIds: [] };
|
|
}
|