mini-dom-parser/index.js

Un DOMParser minimalista sin dependencias, compatible con xmldom para casos simples.

// ---------------- DOM CORE ----------------
class Node {
  constructor(nodeType, nodeName, nodeValue = null) {
    this.nodeType = nodeType; // 1 = element, 3 = text, 9 = document
    this.nodeName = nodeName;
    this.nodeValue = nodeValue;
    this.childNodes = [];
    this.attributes = {};
    this.parentNode = null;
  }

  appendChild(node) {
    node.parentNode = this;
    this.childNodes.push(node);
  }

  getElementsByTagName(tagName) {
    let results = [];
    function traverse(node) {
      if (node.nodeType === 1 && node.nodeName === tagName) results.push(node);
      for (const child of node.childNodes) traverse(child);
    }
    traverse(this);
    return results;
  }
}

class Document extends Node {
  constructor() {
    super(9, "#document");
  }
}

class DOMParser {
  parseFromString(str, mimeType) {
    if (!mimeType || !mimeType.includes("xml")) {
      throw new Error("Unsupported mimeType: " + mimeType);
    }

    const doc = new Document();
    str = str.replace(/<\?xml.*?\?>/, "");

    const tagRegex = /<([^!?\/\s>]+)([^>]*)>|<\/([^>]+)>|([^<]+)/g;
    let stack = [doc];
    let match;

    while ((match = tagRegex.exec(str))) {
      if (match[1]) {
        const nodeName = match[1].includes(":") ? match[1].split(":")[1] : match[1];
        const node = new Node(1, nodeName);
        const attrs = match[2]?.trim();
        if (attrs) {
          const attrRegex = /(\S+)=["']([^"']+)["']/g;
          let attr;
          while ((attr = attrRegex.exec(attrs))) {
            node.attributes[attr[1]] = attr[2];
          }
        }
        stack[stack.length - 1].appendChild(node);
        stack.push(node);
      } else if (match[3]) {
        if (stack.length > 1) stack.pop();
      } else if (match[4] && match[4].trim()) {
        const textNode = new Node(3, "#text", match[4].trim());
        stack[stack.length - 1].appendChild(textNode);
      }
    }

    return doc;
  }
}

// ---------------- XML → JSON HELPER ----------------
function nodeToJson(node) {
  if (node.nodeType === 3) return node.nodeValue;

  if (node.nodeType === 1) {
    let obj = {};
    for (const child of node.childNodes) {
      const name = child.nodeType === 1
        ? child.nodeName
        : "#text";

      const value = nodeToJson(child);
      if (name === "#text") {
        if (!obj) return value;
        if (!obj.value) obj.value = value;
        else obj.value += " " + value;
      } else {
        if (!obj[name]) obj[name] = value;
        else {
          if (!Array.isArray(obj[name])) obj[name] = [obj[name]];
          obj[name].push(value);
        }
      }
    }

    // simplificar nodo de solo texto
    if (Object.keys(obj).length === 1 && obj.value !== undefined) return obj.value;

    return obj;
  }

  if (node.nodeType === 9) {
    let result = {};
    for (const child of node.childNodes) result[child.nodeName] = nodeToJson(child);
    return result;
  }

  return null;
}

function xmlToJson(xmlString) {
  const parser = new DOMParser();
  const doc = parser.parseFromString(xmlString, "text/xml");
  return nodeToJson(doc);
}

export { DOMParser, xmlToJson };