1 Concept

In this MP, you will create a web service that works together with other students’ web services to create a collaborative distributed infinite maze.

The maze is organized around tiles or segments, 7×7 mini-mazes. Each tile has (up to) four exits, one on the center of each of its four sides. The infinite maze is generated by requesting new tiles as soon as a given tile is exited. We have an early proof of concept which doesn’t use servers, instead re-using the same tile every time, which we hope will help the idea make more sense.

Some of the points in this MP are handled by automated tests. The remaining points are handled by your code correctly interacting with others’ code in an in-class demo.

2 Technical Overview

• There are two web servers
  1. A middleware we fully write for you (provided in mazeview.py)
  2. A maze generator you write (we provide code to connect it to the middleware)
• Each of your maze generators will
  1. Define two API endpoints: one generates the same maze every time, the other generates a random maze tile
  2. Support registering itself with various middlewares (this is provided for you)

3 Initial Files

Initial files are available in mp9.zip, including two files you’ll edit:

• mazegen.py is an aiohttp web service.
• mazelib.py is an optional second file where we recommend you put your maze generation code.

The Makefile has a variety of targets:

• make run starts mazegen.py on port 5000
• make view starts both mazegen.py on port 5000 and mazeview.py on port 34009
• make test uses pytest to check your code’s correctness
• make background starts mazegen.py on port 5000 as a daemon process; make stop kills that daemon.

4 Specification

In mazegen.py, implement a web service with the following properties:

• GET /static returns the same 7×7 maze tile with 4 exists each time it is invoked. We recommend hand-crafting a maze that you think looks cool.

  Mazes tiles are returned as JSON arrays of seven seven-character strings; each character is a hexadecimal digit indicating which walls of that cell are present.

  A hexadecimal digit has four bits.

  • bit 1 is 1 if there’s a wall to the left of the cell;
  • bit 2 is 1 if there’s a wall to the bottom of the cell;
  • bit 4 is 1 if there’s a wall to the right of the cell;
  • bit 8 is 1 if there’s a wall to the top of the cell.

  0 1 2 3 4 5 6 7 8 9 A B C D E F

  An empty cell with four exists is encoded as

  [ "988088c"
  , "1000004"
  , "1000004"
  , "0000000"
  , "1000004"
  , "1000004"
  , "3220226"
  ]

  The walls must be two-sided; for example, if one cell has a wall on it’s right then the cell to the right of it must have a wall on its left.

• GET /dynamic/0, GET /dynamic/1, …, GET /dynamic/F return random 7×7 mazes with the specified set of exits.

  These are in the same JSON format as static maze tiles.

  Its exists are indicated with the same hexadecimal code as cells: 0 has all four exits open, 1 has the left exit closed but the other three open, and so on.

  The maze must be solvable; that is, there must be a path from each open exit to each other open exit.

5 Tips

We have a separate page on maze generation.

6 Testing your code

If you run your mazegen.py and visit it in the browser, you will see a visualization of mazes and a field for registering your maze generator with a middleware.

If you run python3 -m pytest you’ll see our estimation of the correctness of your solution.

Automated tests will also be run when you submit your code on the secure upload site.

7 In-lecture deployment on November 21

On November 21 we will run all maze generators as part of an in-lecture check-off for the other half of your MP9 grade.

You can pre-test that your code works by using make view to run our example middleware and make sure you can register your maze generator with the middle ware and explore the resulting maze.

On November 21, your maze generator must be running on your course virtual machine. The make background target is provided to make that easier to achieve.