spb/websensor

1import { XMLParser } from "fast-xml-parser";23export const xmlParser = new XMLParser({4  ignoreAttributes: false,5  attributeNamePrefix: "@_",6  textNodeName: "#text",7  cdataPropName: "__cdata",8  trimValues: true,9  parseTagValue: false,10  parseAttributeValue: false,11  processEntities: true,12  htmlEntities: true,13  removeNSPrefix: false,14});1516export function parseXml(text: string): Record<string, unknown> {17  // Strip BOM and leading junk some feeds emit before the prolog.18  const cleaned = text.replace(/^/, "").replace(/^[^<]+/, "");19  return xmlParser.parse(cleaned) as Record<string, unknown>;20}2122export function asArray<T>(v: T | T[] | undefined | null): T[] {23  if (v === undefined || v === null) return [];24  return Array.isArray(v) ? v : [v];25}2627/** Text of an XML node that may be string, cdata object or {#text}. */28export function textOf(v: unknown): string {29  if (v === undefined || v === null) return "";30  if (typeof v === "string") return v.trim();31  if (typeof v === "number" || typeof v === "boolean") return String(v);32  if (typeof v === "object") {33    const o = v as Record<string, unknown>;34    if (typeof o.__cdata === "string") return o.__cdata.trim();35    if (typeof o["#text"] === "string") return (o["#text"] as string).trim();36    if (o.__cdata && typeof o.__cdata === "object") return textOf(o.__cdata);37    if (typeof o["@_href"] === "string") return o["@_href"].trim();38  }39  return "";40}4142export function stripHtml(s: string): string {43  return s44    .replace(/<script[\s\S]*?<\/script>/gi, " ")45    .replace(/<style[\s\S]*?<\/style>/gi, " ")46    .replace(/<[^>]+>/g, " ")47    .replace(/&nbsp;/g, " ")48    .replace(/&amp;/g, "&")49    .replace(/&lt;/g, "<")50    .replace(/&gt;/g, ">")51    .replace(/&quot;/g, '"')52    .replace(/&#39;|&apos;/g, "'")53    .replace(/\s+/g, " ")54    .trim();55}5657export function parseDate(v: unknown): Date | null {58  const s = textOf(v);59  if (!s) return null;60  const d = new Date(s);61  if (!Number.isNaN(d.getTime())) return d;62  // RFC 822 variants with odd zones like "PST"/"EDT" are handled by Date; try trimming63  const d2 = new Date(s.replace(/\s+[A-Z]{3,4}$/, ""));64  return Number.isNaN(d2.getTime()) ? null : d2;65}66