const { createReadStream } = require("fs");
const { Parser } = require("htmlparser2");
const { pipeline, Transform } = require("stream");
const { isArray, last } = require("lodash");

let i = 0;

class XML extends Transform {
  /**
   * @constructor
   * @param {string} tagName    Target tag name
   * @param {number} tagDepth   Target tag depath
   * @param {object} options    All transform stream options
   */
  constructor(tagName, tagDepth, options = {}) {
    super({
      ...options,
      objectMode: true
    });

    let depth = 0;
    let tree;

    this._parser = new Parser(
      {
        onopentag: (name, attrs) => {
          if (!tree && name === tagName && depth === tagDepth) {
            tree = [];
          }
          if (tree) {
            const parent = last(tree);
            const child = { _attributes: attrs };
            if (parent) {
              if (!parent[name]) {
                parent[name] = child;
              } else if (isArray(parent[name])) {
                parent[name].push(child);
              } else {
                parent[name] = [parent[name], child];
              }
            }
            tree.push(child);
          }
          depth++;
        },
        ontext: text => {
          if (tree) {
            last(tree)._text = text;
          }
        },
        onclosetag: () => {
          if (tree) {
            const node = tree.pop();
            if (tree.length <= 0) {
              tree = undefined;
              this.push(node);
            }
          }
          depth--;
        },
        onend: () => {
          this._close();
        },
        onerror: err => {
          this.emit("error", err);
        }
      },
      { decodeEntities: true }
    );
  }

  _transform(chunk, encoding, callback) {
    this._parser.write(chunk.toString("utf8"));
    callback();
  }

  _final(callback) {
    this._close = callback;
    this._parser.end();
  }
}

const file = "big.xml";
const encoding = "utf8";

async function foo() {
  await new Promise((resolve, reject) => {
    pipeline(
      //
      createReadStream(file, { encoding }),
      //
      new XML("page", 1),
      //
      err => (err ? reject(err) : resolve())
    )
      // .on("data", data => console.log(data))
      .on("data", () => i++);
  });
}

console.log(new Date());
foo()
  .catch(err => console.error(err))
  .then(() => console.log(i))
  .then(() => console.log(new Date()));