HTTP, Yeah You Know Me!

Project Overview

In this project we’ll begin to introduce HTTP, the protocol that runs the web, and build a functioning web server to put that understanding into action.

Learning Goals

  • Practice breaking a workflow into a system of coordinating components
  • Practice using TDD at the unit, integration, and acceptance levels
  • Understand how the HTTP request/response cycle works
  • Practice implementing basic HTTP requests and responses

Background

The internet, which for most people is the web…how does that work?

HTTP (HyperText Transfer Protocol) is the protocol used for sending data from your browser to a web server then getting data back from the server. As protocols go, it’s actually a very simple one.

HTTP with Penpals

Imagine that you’re requesting information from a penpal (old school with paper, envelopes, etc). The protocol would go something like this:

  • You write a letter requesting information
  • You wrap that letter in an envelope
  • You add an address that uniquely identifies the destination of the letter
  • You hand the sealed enveloper to your mail person
  • It travels through a network of people, machines, trucks, planes, etc
  • Assuming the address is correct, it arrives at your penpal’s mailbox
  • Your penpal opens the envelope and reads the letter
  • Assuming they understand your question, your penpal writes a letter of their own back to you
  • They wrap it in an envelope and add an address that uniquely identifies you (which they got from the return address on your envelope)
  • They hand their letter to their mail person, it travels through a series of machines and people, and eventually arrives back at your mailbox
  • You open the envelope and do what you see fit with the information contained in there.

HTTP in Actuality

Metaphor aside, let’s run through the protocol as executed by computers:

  • You open your browser and type in a web address like http://turing.io and hit enter. The URL (or “address”) that you entered is the core of the letter.
  • The browser takes this address and builds a request, the envelope. It uniquely identifies the machine (or server) out there on the internet that the message is intended for. It includes a return address and other information about the requestor.
  • The request is handed off to your Internet Service Provider (ISP) (like CenturyLink or Comcast) and they send it through a series of wires and fiber optic cables towards the server
  • The request arrives at the server. The server reads the precisely formatted request to figure out (a) who made the request and (b) what they requested
  • The server fetches or calculates the requested information and prepares a response. The response wraps the requested information in an envelope that has the destination address on it (your machine).
  • The server hands the response off to their ISP and it goes through the internet to arrive at your computer
  • Your browser receives that response, unwraps it, and displays the data on your machine.

That’s HTTP. You can read more on wikipedia article or the IETF specification.

The Request

Here is what an actual request looks like. Note that it’s just a single highly-formatted string:

GET / HTTP/1.1
Host: 127.0.0.1:9292
Connection: keep-alive
Cache-Control: max-age=0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8
Upgrade-Insecure-Requests: 1
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_10_5) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/46.0.2490.80 Safari/537.36
Accept-Encoding: gzip, deflate, sdch
Accept-Language: en-US,en;q=0.8

The parts we’re most interested in are:

  • The first line, GET / HTTP/1.1, which specifies the verb, path, and protocol which we’ll pick apart later
  • Host which is where the request is sent to
  • Accept which specifies what format of data the client wants back in the response

With those pieces of information a typical server can generate a response.

The Response

The Server generates and transmits a response that looks like this:

http/1.1 200 ok
date: Sun,  1 Nov 2015 17:25:48 -0700
server: ruby
content-type: text/html; charset=iso-8859-1
content-length: 27

The response body goes here

The parts we’re most interested in are:

  • The first line, HTTP/1.1 200 ok, which has the protocol and the response code
  • The unmarked lines at the end which make up the body of the response
  • content-length which tells the client when to stop listening

Tooling

While working on this project you’re going to need to make a lot of HTTP requests. There are many tools that can help you with that, but we recommend you use the following:

  1. Web Browser – if you need to make HTTP GET requests you can use the browser, but it’s the weakest of these three tools.
  2. Postman – a Chrome extension which gives you amazing control and the ability to make any kind of request. Use this for your manual testing and experimentation.
  3. Faraday – a Ruby library for making requests and parsing responses. Use this for your automated testing, basically like a scripted version of request/reponse cycles you could do with Postman.

Experiment

Ruby has handy built-in libraries for dealing with most of the low-level networking details about running a server. Let’s write a short program that can start up, listen for a request, print that request out to the screen, then shut down.

First, we need to “open a port” which basically means “tell the computer that network requests identified addressed for a specific port should belong to this program”.

On your computer there are dozens of programs that are using the network connection at any given time. If the messages in and out of those programs were all happening through the same channel then it’d be confusing which message belongs to which program. Think of the port like a mailbox in an apartment building: all the residents (aka programs) share the same street address (your computer) but each have their own mailbox (or port).

Let’s start our server instance and have it listen on port 9292:

require 'socket'
tcp_server = TCPServer.new(9292)
client = tcp_server.accept

We can read the request from the client object which is what we call an IO stream. Here’s a snippet to keep reading from that stream until the input is a blank line and store all the request lines in an array request_lines:

puts "Ready for a request"
request_lines = []
while line = client.gets and !line.chomp.empty?
  request_lines << line.chomp
end

Note that when the program runs it’ll hang on that gets method call waiting for a request to come in. When it arrives it’ll get read and stored into request_lines, then lets print it to the console for debugging:

puts "Got this request:"
puts request_lines.inspect

Then it’s time to build a response. For this example let’s just print out the request data as the response:

puts "Sending response."
response = "<pre>" + request_lines.join("\n") + "</pre>"
output = "<html><head></head><body>#{response}</body></html>"
headers = ["http/1.1 200 ok",
          "date: #{Time.now.strftime('%a, %e %b %Y %H:%M:%S %z')}",
          "server: ruby",
          "content-type: text/html; charset=iso-8859-1",
          "content-length: #{output.length}\r\n\r\n"].join("\r\n")
client.puts headers
client.puts output

And close up the server:

puts ["Wrote this response:", headers, output].join("\n")
client.close
puts "\nResponse complete, exiting."

Save that file and run it. Open your web browser and enter the address http://127.0.0.1:9292. If everything worked then your browser should show all the details of your request. Flip over to the terminal where your ruby program was running and you should see the request outputted to the terminal.

You just built a web server.

Having trouble? Check out the whole file here.

The Project

You’re going to build a web application capable of:

  • Receiving a request from a user
  • Comprehending the request’s intent and source
  • Generating a response
  • Sending the response to the user

Iteration 0 - Hello, World

Build a web application/server that:

  • listens on port 9292
  • responds to HTTP requests
  • responds with a valid HTML response that displays the words Hello, World! (0) where the 0 increments each request until the server is restarted

Iteration 1 - Outputting Diagnostics

Let’s start to rip apart that request and output it in your response. In the body of your response, include a block of HTML like this including the actual information from the request:

<pre>
Verb: POST
Path: /
Protocol: HTTP/1.1
Host: 127.0.0.1
Port: 9292
Origin: 127.0.0.1
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8
</pre>

Keep the code that outputs this block at the bottom of all your future outputs to help with your debugging.

Iteration 2 - Supporting Paths

Now let’s react to the path that the user specifies:

  • If they request the root, aka /, respond with just the debug info from Iteration 1.
  • If they request /hello, respond with “Hello, World! (0)” where the 0 increments each time the path is requested, but not when any other path is requested.
  • If they request /datetime, respond with today’s date and time in this format: 11:07AM on Sunday, November 1, 2015.
  • If they request /shutdown, respond with “Total Requests: 12” where 12 is the aggregate of all requests. It also causes the server to exit / stop serving requests.

Iteration 3 - Supporting Parameters

Often we want to supply some data with a request. For instance, if you were submitting a search, that’d typically be a GET request that has a parameter. When we use parameters in GET requests they are embedded in the URL like this:

http://host:port/path?param=value&param2=value2

You know your computer has a dictionary built in, right? It’s stored in a special file on your machine located at /usr/share/dict/words. Let’s use this information to write an “endpoint” that works like this:

  • The path is /word_search
  • The verb will always be a GET
  • The parameter will be named word
  • The value will be a possible word fragment

In your HTML response page, output one of these:

  • WORD is a known word
  • WORD is not a known word

Where WORD is the parameter from the URL.

Iteration 4 - Verbs & Parameters

The path is the main way that the user specifies what they’re requesting, but the secondary tool is the verb. There are several official verbs, but the only two typical servers use are GET and POST.

We use GET to fetch information. We typically use POST to send information to the server. When we submit parameters in a POST they’re in the body of the request rather than in the URL.

Changing the verb and submitting parameters in the body instead of the parameters for a POST request can both be done in Postman.

Let’s practice applying these techniques by building a simple guessing game that can be played via our HTTP server.

The game will work like this:

  1. When a player starts a new game, the server picks a random number between 0 and 100.
  2. The player can make a new guess by sending a POST request containing the number they want to guess.
  3. When the player requests the game path, the server should show some information about the game including how many guesses have been made, what the most recent guess was, and whether it was too high, too low, or correct.

The HTTP endpoints to support this game will look like this:

POST to /start_game

This request begins a game. The response says Good luck! and starts a game.

GET to /game

A request to this verb/path combo tells us:

  • a) how many guesses have been taken.
  • b) if a guess has been made, it tells what the guess was and whether it was too high, too low, or correct

POST to /game

This is how we make a guess. The request includes a parameter named guess. The server stores the guess and sends the user a redirect response, causing the client to make a GET to /game.

Handling POST Requests

So far all of the requests we’ve dealt with have been using HTTP’s GET verb. A GET request signifies fetching rather than sending information. POST requests, on the other hand, signify sending information. In this section we’ll be using post requests to allow users to send us data to interact with the simple HTTP game.

So what do we need to do differently when handling POST requests vs. GETs? Fortunately, most of the other parsing we’ve done will remain the same – headers, paths, parameters, etc. When reading a POST request, however, we also need to give attention to the request body.

Remember that in an HTTP request the body is separated from the headers with a blank line – in our basic parsing so far, encountering this blank line is how we know when we’re done reading headers. Everything after the blank line, then, is the body. To read the body from the request, we need to look at a specific header, called Content-Length which is included with the request. The Content-Length header tells us the number of bytes (roughly equivalent with the number of characters) that are included in the request body.

Once we find this number, we can use a special socket method read, to read that specific number of bytes.

In short, then our process for reading the body of a post request looks like this:

  1. Read the request headers by reading lines until we encounter a blank one (same as before)
  2. Find the Content-Length header and see how many bytes it says are included in the request.
  3. User Socket#read to read that number of bytes from the socket, thus reading the request body.

For this iteration you’ll need to update your request handling to use this process to read the request body for any POST requests your server receives.

Sending a Redirect

A Redirect is a special kind of HTTP response. It indicates to an HTTP client that the resource they requested should be fetched from a different location. A redirect is HTTP’s mechanism of telling a client (often a web browser) to “go over there.” You’ve seen this on the web whenever you submit a web form and your browser automatically loads a new page. Redirects are often used in response to POST requests.

To respond with a redirect, you need to send 2 things:

  1. A 3xx status code – in our case 302 will be the standard status code for redirecting
  2. A special header called Location – the Location header indicates the new URL the browser should visit. For example the header Location: http://google.com would tell a web browser to navigate to google’s homepage.

Here’s what the headers for an example redirect response would look like:

$ curl -I google.com
HTTP/1.1 301 Moved Permanently
Location: http://www.google.com/
Content-Type: text/html; charset=UTF-8
Date: Fri, 26 Feb 2016 01:55:24 GMT
Expires: Sun, 27 Mar 2016 01:55:24 GMT
Cache-Control: public, max-age=2592000
Server: gws
Content-Length: 219
X-XSS-Protection: 1; mode=block
X-Frame-Options: SAMEORIGIN

Iteration 5 - Response Codes

We use the HTTP response code as a short hand way to explain the result of the request. Here are the most common HTTP status codes:

  • 200 OK
  • 301 Moved Permanently
  • 401 Unauthorized
  • 403 Forbidden
  • 404 Not Found
  • 500 Internal Server Error

Let’s modify your game from Iteration 4 to use status codes:

  • Most requests, unless listed below, should respond with a 200.
  • When you submit the POST to /start_game and there is no game in progress, it should start one and respond with a 301 redirect.
  • When you submit the POST to /start_game but there is already a game in progress, it should respond with 403.
  • If an unknown path is requested, like /fofamalou, the server responds with a 404.
  • If the server generates an error, then it responds with a 500. Within the response let’s present the whole stack trace. Since you don’t write bugs, create an /force_error endpoint which just raises a SystemError exception.

Extensions

1. HTTP-Accept

The HTTP-Accept parameter is used to specify what kind of data the client wants in response. Modify your /word_search path so that if the HTTP-Accept starts with application/json then they are sent a JSON body like the following.

A search for pizza returns this JSON:

{"word":"pizza","is_word":true}

A search for pizz returns JSON with possible matches like this:

{"word":"pizza","is_word":true,"possible_matches":["pizza","pizzeria","pizzicato"]}

2. Threading

A Thread represents an independent flow (thread) of execution. Threads allow us to model concurrent processes that need to occur at the same time, independent of one another.

You can read more about threads in the ruby docs, but the basic usage looks like this:

t = Thread.new do
  puts "hi from the other thread"
end
hi from other thread
=> #<Thread:0x007f81d3bddd08@(pry):65 sleep>

If a thread returns a value, we can capture it by calling #value:

pry(main)> t = Thread.new { 1 + 1 }
=> #<Thread:0x007f81d4398070@(pry):66 dead>
pry(main)> t.value
=> 2

pry(main)> t = Thread.new { sleep(3); 1 + 1 }
=> #<Thread:0x007f81d42ec770@(pry):68 sleep>
pry(main)> t.value #hangs
=> 2

Note that if the thread is busy doing some work, calling #value will hang until that thread is finished.

With these tools at your disposal, enhance your server handler so that each request/response gets handled in its own thread. If done correctly, this will allow you to handle multiple requests at a time.

To demonstrate this functionality, additionally add a new GET endpoint /sleepy which sleeps for 3 seconds and then returns a 200 OK response with the body "yawn...". You should be able to handle multiple /sleepy requests in a row without the second one waiting for the first one to complete.

Evaluation Rubric

The project will be assessed with the following rubric:

1. Fundamental Ruby & Style

  • 4: Application demonstrates excellent knowledge of Ruby syntax, style, and refactoring
  • 3: Application shows strong effort towards organization, content, and refactoring
  • 2: Application runs but the code has long methods, unnecessary or poorly named variables, and needs significant refactoring
  • 1: Application generates syntax error or crashes during execution

2. Encapsulation / Breaking Logic into Components

  • 4: Application is expertly divided into logical components each with a clear, single responsibility
  • 3: Application effectively breaks logical components apart but breaks the principle of SRP
  • 2: Application shows some effort to break logic into components, but the divisions are inconsistent or unclear
  • 1: Application logic shows poor decomposition with too much logic mashed together

3. Test-Driven Development

  • 4: Application is broken into components which are well tested in both isolation and integration
  • 3: Application uses tests to exercise core functionality and some edge cases, but fails to break out component objects/tests.
  • 2: Application uses tests to exercise core functionality but leaves many common edge cases untested.
  • 1: Application does not demonstrate strong use of TDD

4. Functional Expectations

  • 4: Application implements all five iterations and at least one extension
  • 3: Application implements iterations 0 - 4
  • 2: Application implements iterations 0 - 3
  • 1: Application implements through interation 2 or less

Echo/Foxtrot Note: Requirements listed are for Foxtrot pairs – Echo requirements are shifted by 1 additional iteration – So a 3 for Echo requires completing Iterations 0-5

Addendum Content

There is content from previous versions not germane to the assignment above available here.