Project 5: Web Crawler

Interact with web services via HTTP

Description

This assignment is intended to familiarize you with the HTTP protocol. HTTP is (arguably) the most important application level protocol on the Internet today: the Web runs on HTTP, and increasingly other applications use HTTP as well (including Bittorrent, streaming video, Facebook and Twitter’s social APIs, etc.).

Your goal in this assignment is to implement a web crawler that gathers data from a fake social networking website that we have set up for you. The site is available here: Fakebook.

What is a Web Crawler?

A web crawler (sometimes known as a robot, a spider, or a scraper) is a piece of software that automatically gathers and traverses documents on the web. For example, lets say you have a crawler and you tell it to start at www.wikipedia.com. The software will first download the Wikipedia homepage, then it will parse the HTML and locate all hyperlinks (i.e. anchor tags) embedded in the page. The crawler then downloads all the HTML pages specified by the URLs on the homepage, and parses them looking for more hyperlinks. This process continues until all of the pages on Wikipedia are downloaded and parsed.

Web crawlers are a fundamental component of today’s web. For example, Googlebot is Google’s web crawler. Googlebot is constantly scouring the web, downloading pages in search of new and updated content. All of this data forms the backbone of Google’s search engine infrastructure.

Fakebook

We have set up a fake social network for this project called Fakebook. Fakebook is a very simple website that consists of the following pages:

  • Homepage: The Fakebook homepage displays some welcome text, as well as links to several random Fakebook users' personal profiles.
  • Personal Profiles: Each Fakebook user has a profile page that includes their name, some basic demographic information, as well as a link to their list of friends.
  • Friends List: Each Fakebook user is friends with one or more other Fakebook users. This page lists the user’s friends and has links to their personal profiles.

To browse Fakebook, you must first login with a username and password. We will email each student to give them a unique username and password.

WARNING: DO NOT TEST YOUR CRAWLERS ON PUBLIC WEBSITES

Many web server administrators view crawlers as a nuisance, and they get very mad if they see strange crawlers traversing their sites. Only test your crawler against Fakebook, do not test it against any other websites.

High-level Requirements

Your goal is to collect 5 secret flags that have been hidden somewhere on the Fakebook website. The flags are unique for each student, and the pages that contain the flags will be different for each student. Since you have no idea what pages the secret flags will appear on, and the Fakebook site is very large (tens of thousands of pages), your only option is to write a web crawler that will traverse Fakebook and locate your flags.

Your web crawler must execute on the command line using the following syntax:

$ ./webcrawler [username] [password]

username and password are used by your crawler to log-in to Fakebook. You may assume that the root page for Fakebook is available at https://fakebook.3700.network/fakebook/. You may also assume that the log-in form for Fakebook is available at https://fakebook.3700.network/accounts/login/?next=/fakebook/.

Your web crawler should print exactly fives lines of output to STDOUT: the five secret flags discovered during the crawl of Fakebook. If your program encounters an unrecoverable error, it may print an error message before terminating.

Secret flags may be hidden on any page on Fakebook, and their exact location on each page may be different. Each secret flag is a 64 character long sequences of random alphanumerics. All secret flags will appear in the following format (which makes them easy to identify):

<h2 class='secret_flag' style="color:red">FLAG: 64-characters-of-random-alphanumerics</h2>

Sockets, Ports, and TLS

Fakebook uses HTTPS, which means that the full protocol stack is HTTP over TLS over TCP. Thus, in this project, your web crawler will need to connect to Fakebook using a TCP socket wrapped in TLS. Note that in HTTPS, the TCP socket gets wrapped in TLS immediately after connection, before any HTTP protocol messages are sent. This is unlike the FTPS protocol, where TLS wrapping occurs after some FTP protocol messages are sent in plaintext.

By convention, HTTP uses TCP port 80 and HTTPS uses port 443. Thus, in this project, you will be connecting to Fakebook on port 443.

HTTP and (Il)Legal Libraries

Part of the challenge of this assignment is that all HTTP request and response code must be written by the student, from scratch. In other words, you need to implement the ability to send HTTP/1.1 messages and parse HTTP responses. Students may use any available libraries to create socket connections, parse URLs, and parse HTML. However, you may not use any libraries/modules/etc. that implement HTTP or manage cookies for you.

For example, if you were to write your crawler in Python, the following modules would all be allowed: socket, parseurl, html, html.parse, and xml. However, the following modules would not be allowed: urllib, urllib2, httplib, requests, pycurl, and cookielib.

Similarly, if you were to write your crawler in Java, it would not be legal to use java.net.CookieHandler, java.net.CookieManager, java.net.HttpCookie, java.net.HttpUrlConnection, java.net.URLConnection, URL.openConnection(), URL.openStream(), or URL.getContent().

If students have any questions about the legality of any libraries please post them to Piazza. It is much safer to ask ahead of time, rather than turn in code that uses a questionable library and receive points off for the assignment after the fact.

Implementation Details and Hints

In this assignment, your crawler must implement HTTP/1.1 (not 0.9 or 1.0). This means that there are certain HTTP headers like Host that you must include in your requests (i.e., they are required for all HTTP/1.1 requests). We encourage you to implement Connection: Keep-Alive (i.e., pipelining) to improve your crawler’s performance (and lighten the load on our server), but this is not required, and it is tricky to get correct. We also encourage students to implement Accept-Encoding: gzip (i.e., compressed HTTP responses), since this will also improve performance for everyone, but this is also not required. If you want to get crazy, you can definitely speed up your crawler by using multithreading or multiprocessing, but again this is not required functionality.

WARNING: PLEASE BE KIND AND LIMIT YOUR PARALLELISM
While we do our best to keep our web servers well-provisioned so they can handle the load generated by a full class of students, the servers can easily crumble in the face of crawlers that are massively parallel, i.e., send many requests in parallel at the same time. If you plan to implement parallelism (e.g., through async IO, threads, or multiprocessing) we ask that you limit your parallelism to five (5) requests at a time. This will enable you to build and test your parallel implementation while still ensuring that the web server has enough bandwidth left over to serve other students.

One of the key differences between HTTP/1.0 and HTTP/1.1 is that the latter supports chunked encoding. HTTP/1.1 servers may break up large responses into chunks, and it is the client’s responsibility to reconstruct the data by combining the chunks. Our server may return chunked responses, which means your client must be able to reconstruct them. To aid in debugging, you might consider using HTTP/1.0 for your initial implementation; once you have a working 1.0 implementation, you can switch to 1.1 and add support for chunked responses.

In order to build a successful web crawler, you will need to handle several different aspects of the HTTP protocol:

  • HTTP GET - These requests are necessary for downloading HTML pages.
  • HTTP POST - You will need to implement HTTP POST so that your code can login to Fakebook. As shown above, you will pass a username and password to your crawler on the command line. The crawler will then use these values as parameters in an HTTP POST in order to log-in to Fakebook.
  • Cookie Management - Fakebook uses cookies to track whether clients are logged in to the site. If your crawler successfully logs in to Fakebook using an HTTP POST, Fakebook will return a session cookie to your crawler. Your crawler should store this cookie, and submit it along with each HTTP GET request as it crawls Fakebook. If your crawler fails to handle cookies properly, then your software will not be able to successfully crawl Fakebook.

In addition to crawling Fakebook, your web crawler must be able to correctly handle HTTP status codes. Obviously, you need to handle 200, since that means everything is okay. Your code must also handle:

  • 302 - Found: This is as an HTTP redirect. Your crawler should try the request again using the new URL given by the server in the Location header.
  • 403 - Forbidden and 404 - Not Found: Our web server may return these codes in order to trip up your crawler. In this case, your crawler should abandon the URL that generated the error code.
  • 500 - Internal Server Error: Our web server may randomly return this error code to your crawler. In this case, your crawler should re-try the request for the URL until the request is successful.

I highly recommend the HTTP Made Really Easy tutorial as a starting place for students to learn about the HTTP protocol. Furthermore, the developer tools built-in to Chrome and Firefox are both excellent for inspecting and understanding HTTP requests.

In addition to HTTP-specific issues, there are a few key things that all web crawlers must do in order function:

  • Track the Frontier: As your crawler traverses Fakebook it will observe many URLs. Typically, these uncrawled URLs are stored in a queue, stack, or list until the crawler is ready to visit them. These uncrawled URLs are known as the frontier.
  • Watch Out for Loops: Your crawler needs to keep track of where it has been, i.e., the URLs that it has already crawled. Obviously, it isn’t efficient to revisit the same pages over and over again. If your crawler does not keep track of where it has been, it will almost certainly enter an infinite loop. For example, if users A and B are friends on Fakebook, then that means A’s page links to B, and B’s page links to A. Unless the crawler is smart, it will ping-pong back and forth going A->B, B->A, A->B, B->A, …, etc.
  • Only Crawl The Target Domain: Web pages may include links that point to arbitrary domains (e.g., a link on google.com that points to cnn.com). Your crawler should only traverse URLs that point to pages on fakebook.3700.network. For example, it would be valid to crawl https://fakebook.3700.network/fakebook/018912/, but it would not be valid to crawl https://www.facebook.com/018912/. Your code should check to make sure that each URL has a valid domain (i.e., the domain is fakebook.3700.network) before you attempt to visit it.

Logging in to Fakebook

In order to write code that can successfully log-in to Fakebook, you will need to reverse engineer the HTML form on the log-in page. Students should carefully inspect the form’s code, since it may not be as simple as it initially appears. The key acronym you should be on the lookout for is CSRF.

Language

You can write your code in whatever language you choose, as long as your code compiles and runs on unmodified Khoury College Linux machines on the command line. Do not use libraries that are not installed by default on the Khoury College Linux machines, or that are disallowed for this project. You may use IDEs (e.g., Eclipse) during development, but do not turn in your IDE project without a Makefile. Make sure you code has no dependencies on your IDE.

Submitting Your Project

To turn-in your project, you must submit the following four things:

  1. The thoroughly documented source code for your crawler.
  2. A Makefile that compiles your code.
  3. A plain-text (no Word or PDF) README.md file. In this file, you should briefly describe your high-level approach, any challenges you faced, and an overview of how you tested your code.
  4. A file called secret_flags. This file should contain the secret flags of all group members, one per line, in plain ASCII. For example, a group of two should have a file with exactly ten lines in it.

Your README.md, Makefile, secret_flags file, source code, etc. should all be placed in the root of a compressed archive (e.g., a .zip or .tar.gz) and then uploaded to Gradescope. Alternatively, you can check these items in to Github and then instruct Gradescope to clone your Github repository.

Double Checking Your Submission

To try and make sure that your submission is (1) complete and (2) will work with our grading scripts, we provide a simple script that checks the formatting of your submission. You can download the script here and invoke it using the following command:

$ ./crawler_fmt_chk.py [path to your project directory]

Note that you may need to chmod +x crawler_fmt_chk.py to make the script executable.

This script will attempt to make sure that the correct files (e.g., README.md, secret_flags, and Makefile) are available in the given directory, that your secret_key file contains at least ten 64-byte keys, that your Makefile will run without errors (or is empty), and that after running the Makefile a program named webcrawler exists in the directory. The script will also try to determine if your files use Windows-style line endings (\r\n) as opposed to Unix-style line endings (\n). If your files are Windows-encoded, you should convert them to Unix-encoding using the dos2unix utility before turning in.

Grading

This project is worth 10% of your final grade. You will receive full credit if 1) your code compiles, runs, and produces the expected output, 2) you have not used any illegal libraries, and 3) you successfully submit the secret flags of all group members. All student code will be scanned by plagiarism detection software to ensure that students are not copying code from the Internet or each other.