cities.js

// population of all cities over 1 million people using the cities1000 dataset from http://geonames.org

var fs = require('fs');
var cities = fs.createReadStream(__dirname + '/cities1000.txt');
var split = require('split');
var through = require('through');
var fields = require('cities1000').fields;
var popIndex = fields.indexOf('population');

var sum = 0;
cities.pipe(split()).pipe(through(write, end))

function write (line) {
    if (line === '') return;
    var row = line.split('\t');
    var n = Number(row[popIndex]);
    if (n >= 1000 * 1000) {
        sum += n;
        process.stdout.write(sum + '      \r');
    }
}

function end () {
    console.log('\nTOTAL: ' + sum);
}

citylots.js

// san francisco city lot streaming json parser

var JSONStream = require('JSONStream');
var fs = require('fs');
var parser = JSONStream.parse([ 'features', true ]);

parser.on('data', function (row) {
    console.log(
        row.properties.TO_ST,
        row.properties.FROM_ST,
        row.properties.STREET,
        row.geometry.coordinates
    );
});
fs.createReadStream('citylots.json').pipe(parser);

scrape.js

// scrape the memory alpha Data page for chronology data

var request = require('request');
var trumpet = require('trumpet');
var concat = require('concat-stream');
var tr = trumpet();

var data = {};
var latest;

tr.selectAll('dt a', function (dt) {
    dt.createReadStream().pipe(concat(function (body) {
        latest = body.toString('utf8');
    }));
});

tr.selectAll('dd', function (dd) {
    dd.createReadStream().pipe(concat(function (body) {
        data[latest] = body.toString('utf8').trim();
    }));
});

tr.on('end', function () {
    console.log(JSON.stringify(data, null, 2));
});


request('http://en.memory-alpha.org/wiki/Data').pipe(tr);