Extract links from a webpage the streaming way

extractLinksStream.js

var request = require('request');

var extractLinksStream = extractStream.bind(null, /<a.*?href="(.*?)".*?>(.*?)<\/a>/gi);

request(process.argv[2]).pipe(extractLinksStream(function(found) {
  var hrefs = found.map(function(item) { return item[1]; }); // Extract 1st capture
  console.log(hrefs);
}));


/* ----------------------------------------------- */
var concat = require('concat-stream');

function extractStream(re, callback) {
  return concat(function(buf) {
    callback(extract(buf.toString(), re));
  });
}

function extract(str, re) {
  var found = [];
  var matches;
  while(matches = re.exec(str)) {
    delete matches.index;
    delete matches.input;
    found.push(matches);
  };
  return found;
}