Test HtmlParser2

test.htmlparser2.js

import Fs from "fs";
import {Parser} from "htmlparser2";
import MemWatch from "memwatch-next";
import humanFormat from "human-format";

process.on("unhandledRejection", function (reason, p) {
  throw reason;
});

let timeScale = new humanFormat.Scale({
  seconds: 1,
  minutes: 60,
  hours: 3600,
});

let source = Fs.createReadStream("./huge.html");
let pCount = 0;

let parser = new Parser({
  onopentag: function (name, attrs) {
    if (name == "p") {
      pCount += 1;
    }
  },
  onend: function () {
    console.log("Total pCount =", pCount);
    setImmediate(final);
  }
}, {
    decodeEntities: true
  }
);

let maxUsage = 0;
let hd = new MemWatch.HeapDiff();

MemWatch.on("stats", function (stats) {
  if (stats.current_base > maxUsage) {
    maxUsage = stats.current_base;
  }
});

let startDate = new Date();
console.log(`started`);

source.pipe(parser);

let final = function () {
  let stopDate = new Date();
  let deltaTime = humanFormat((stopDate - startDate) / 1000, {scale: timeScale});
  console.log(`stopped (${deltaTime})`);

  MemWatch.gc();
  let diff = hd.end();
  console.log("Mem before:", diff.before.size);
  console.log("Mem after:", diff.after.size);
  console.log("Mem max:", humanFormat(maxUsage).toLowerCase() + "b");
};