Files
itinova-budget-si/scripts/diagnosePayrollExtraction.ts

29 lines
1.5 KiB
TypeScript

import fs from "node:fs";
import { extractPayrollBulletins } from "../server/payrollParser";
import { extractPayrollPdfText } from "../server/payrollPdfText";
const filePath = process.argv[2];
if (!filePath) throw new Error("Chemin PDF requis");
const text = await extractPayrollPdfText(fs.readFileSync(filePath));
const extraction = extractPayrollBulletins(text);
const lines = text.split(/\r?\n/);
const redactStructure = (line: string) => line.replace(/[A-Za-zÀ-ÿ]/g, "x").replace(/\d/g, "#").slice(0, 240);
const prohibitedPosteContent = extraction.bulletins.some((bulletin) => /iban|rib|n°\s*ss|adresse|fr\d{2}/i.test(bulletin.poste));
console.log(JSON.stringify({
bulletins: extraction.bulletins.length,
statutExtraction: extraction.statutExtraction,
hasExtractionWarning: Boolean(extraction.erreurExtraction),
containsProhibitedPosteContent: prohibitedPosteContent,
preservedPageSeparators: (text.match(/\f/g) ?? []).length,
bulletinMarkers: (text.match(/##BULLETIN##/g) ?? []).length,
matriculeMatchCount: (text.match(/Matricule\s*:\s*(?:M|N|N°)?\s*\d+/g) ?? []).length,
emploiSameLineCount: lines.filter((line) => /Emploi\s*:.*(?:Monsieur|Madame)\b/i.test(line)).length,
baseLineCount: lines.filter((line) => line.trim().startsWith("Salaire de base")).length,
grossLineCount: lines.filter((line) => line.trim().startsWith("Salaire brut")).length,
emploiVicinityStructures: lines
.flatMap((line, index) => /Emploi\s*:/i.test(line) ? lines.slice(index, index + 4) : [])
.slice(0, 8)
.map(redactStructure),
}));