rhl-lieferscheine/lib/search/pathMapping.js

/**
 * Map Qsirch items back into our NAS convention:
 *
 * Qsirch fields observed:
 * - item.path = "Niederlassungen/NL20/2025/12/18"   (directory)
 * - item.name = "Stapel_Seiten-4_Zeit-141039"      (basename without extension)
 * - item.extension = "pdf"
 *
 * We strictly accept only:
 *   <prefix>/<branch>/<year>/<month>/<day>
 *
 * And we build:
 * - filename: "<name>.<extension>"
 * - relativePath: "<branch>/<year>/<month>/<day>/<filename>"
 *
 * Any unexpected shape is rejected (returns null) as defense-in-depth.
 */

const BRANCH_RE = /^NL\d+$/;
const YEAR_RE = /^\d{4}$/;
const MONTH_RE = /^(0[1-9]|1[0-2])$/;
const DAY_RE = /^(0[1-9]|[12]\d|3[01])$/;

function stripLeadingSlash(p) {
	const s = String(p || "");
	return s.startsWith("/") ? s.slice(1) : s;
}

function normalizePrefix(prefix) {
	let p = stripLeadingSlash(prefix);
	p = p.trim();
	if (p.endsWith("/")) p = p.replace(/\/+$/, "");
	return p;
}

function stripPrefix(pathValue, prefix) {
	const p = stripLeadingSlash(pathValue).trim();
	const pref = normalizePrefix(prefix);

	if (!pref) return p;

	if (p === pref) return "";
	if (p.startsWith(`${pref}/`)) return p.slice(pref.length + 1);

	return null;
}

function toFilename(baseName, ext) {
	const safeBase = String(baseName || "").trim();
	const safeExt = String(ext || "").trim();

	if (!safeBase) return null;
	if (!safeExt) return null;

	const lowerExt = safeExt.toLowerCase();

	// We only support PDFs for the Lieferscheine workflow.
	if (lowerExt !== "pdf") return null;

	// Qsirch usually returns name without extension. If it already ends with ".pdf",
	// keep it as-is to avoid double extensions.
	if (safeBase.toLowerCase().endsWith(`.${lowerExt}`)) {
		return safeBase;
	}

	return `${safeBase}.${safeExt}`;
}

/**
 * @param {any} item - Qsirch item
 * @param {{ pathPrefix: string }} options
 * @returns {null | {
 *   branch: string,
 *   date: string,
 *   year: string,
 *   month: string,
 *   day: string,
 *   filename: string,
 *   relativePath: string
 * }}
 */
export function mapQsirchItemToSearchItem(item, { pathPrefix }) {
	const dirPath = typeof item?.path === "string" ? item.path : null;
	const name = typeof item?.name === "string" ? item.name : null;
	const ext = typeof item?.extension === "string" ? item.extension : null;

	if (!dirPath || !name || !ext) return null;

	const stripped = stripPrefix(dirPath, pathPrefix);
	if (stripped === null) return null;

	const parts = stripped.split("/").filter(Boolean);

	// Must be exactly NLxx/YYYY/MM/DD
	if (parts.length !== 4) return null;

	const [branch, year, month, day] = parts;

	if (!BRANCH_RE.test(branch)) return null;
	if (!YEAR_RE.test(year)) return null;
	if (!MONTH_RE.test(month)) return null;
	if (!DAY_RE.test(day)) return null;

	const filename = toFilename(name, ext);
	if (!filename) return null;

	const date = `${year}-${month}-${day}`;
	const relativePath = `${branch}/${year}/${month}/${day}/${filename}`;

	return { branch, date, year, month, day, filename, relativePath };
}