A2: Adventure
In this assignment and the next, you will develop a text adventure game (TAG), also known as interactive fiction. The characteristic elements of TAGs include gameplay driven by exploration and puzzle-solving, and a text-based interface in which users type natural-language commands and the game responds with text. The seminal work in this genre is the Colossal Cave Adventure, which you can play online.
In A2, every team will build the same, specified part of the game: exploration of a map. In A3, your team will have the opportunity to make your game unique by extending your A2 solution with functionality of your own choice.
You will actually implement not just a single game, but a game engine that could be used to play many adventures. The game engine is an OCaml program that implements the gameplay and user interface. An adventure is a data file that is input by the game engine and describes a particular gaming experience: exploring a cave, hitchhiking on a spaceship, finding the missing pages of a powerful magical book, etc. This factoring of responsibility between the engine and input file is known as data driven design in games.
What you’ll do: Implement and test a couple OCaml modules; and work with a development team.
Objectives:
- Design your own data types.
- Work with lists and trees.
- Use pattern matching and higher-order functions.
- Read information from files, and interact with the user.
- Learn about JSON, a widely-used data format.
Table of contents:
- Step 1: Boot Up Your Team
- Step 2: Get Git Going and Explore the Release Code
- Step 3: JSON Tutorial
- Step 4: Load Adventure Files
- Step 5: Parse Commands
- Step 6: State Transitions
- Step 7: Interface
- Scope
- Documentation
- Submission
Step 1: Boot Up Your Team
Your team has been formed for you in CMS. Here’s what you should do to boot up your team.
-
Set up an initial meeting to get to know one another. At that meeting, create your Team Expectations Agreement. Give everyone a day to think about it, make any amendments necessary, then upload the result to CMS.
-
At the same time, by yourself, complete the Coping with Hitchhikers assignment.
-
Set up a messaging tool. You’re going to need something better than just (e.g.) a massive Facebook Messenger thread. For example, consider creating a Slack workspace for your team.
-
Read the Team Policies document. As a team, choose roles for this assignment. Those roles will rotate in future assignments.
As you proceed with teamwork on this assignment, it might be tempting to somehow divide up the steps and work on them in parallel. That is not the intent. Each person on the team needs to learn the entire content of the assignment to be successful in the rest of the course, including exams.
So instead we recommend the following approach for each step that involves developing code:
-
Read the step and discuss it as a team to make sure everyone feels that they understand what the task is.
-
Everyone work on their own for awhile to sketch out an approach to a solution.
-
Come back together to compare your ideas and decide as a team how to proceed.
-
Now you can parallelize. Have two people implement the code for the step using test-driven development and pair programming. Have another one or two people independently develop their own test cases without seeing the code that the pair is writing.
-
Get all the test cases to pass. Move on to the next step. Make sure to mix it up so that different people get a chance to develop code and tests.
Step 2: Get Git Going and Explore the Release Code
Create a git repo for your team’s collaboration on this assignment. Make sure the repo is private. Any one of you can create it on Cornell github, then go to Settings->Collaborators, and add your teammates. Add the release code to your repo. Refer back to the instructions in A1 if you need help with that.
This is the first assignment in which the release code contains interface (.mli
)
files. The FAQ has been updated to address those.
Now that we are using modules and creating larger programs, working in
utop becomes a little more difficult; your normal mode of interaction
with OCaml is necessarily going to shift away from utop and toward VS
Code and the command line. Nonetheless we do provide a simple make
target that will open utop with all your code available for use. That
command first builds all your code, then loads utop and runs all the
commands in .ocamlinit
, which is a file provided in the release code.
Note that after changing any code, you must exit utop and re-run make
for your changes to be reflected in utop.
Here is a summary of all the Makefile targets:
make
: rebuild your code and launch utop with that code availablemake build
: just rebuild your codemake test
,make check
,make finalcheck
,make clean
: as usualmake docs
: has changed; see the Documentation section at the end of this handoutmake play
: launch your game interfacemake zip
: create a ZIP file for CMS submission
The latter two targets won’t be used until much later in the assignment; we discuss them below where they become relevant.
Feel free to browse through the release code at this point, but don’t worry about familiarizing yourself with all of it yet. The steps of the assignment, below, will take you through them in a guided order.
Step 3: JSON Tutorial
The adventure files that your game engine will input are formatted in JSON, the widely-used JavaScript Object Notation. If you’ve never used JSON before, read the brief overview of it on the JSON webpage.
In OCaml, you can use the Yojson library’s Basic module for
parsing JSON. In the release code, we provide a small tutorial on Yojson in the
file json_tutorial.ml
, which uses the file cornell.json
as an example input.
Although the Yojson library is large and provides a lot of functionality, all
that you need to know for this assignment is covered by the tutorial.
Your task: read the tutorial together with your team, following along and entering lines in utop to experience it firsthand, then answer the following questions as a team to check your understanding:
-
What is a polymorphic variant? How does it differ from a parameterized variant?
-
What are the two functions you can use to input JSON from a file or from a string?
-
What is the OCaml data structure that corresponds to a JSON object? What OCaml library provides useful functions for that data structure?
-
What is the Yojson function you would use to extract a string from a
json
value? What would happen if that value were not actually a string? -
What is the
{| ... |}
syntax in OCaml? Why is it useful?
(There’s no need to turn in your answers.)
Caution: There is a chapter on JSON in Real World OCaml, but you
should ignore it. The features used in that chapter are more complicated than
you need, and will be more confusing than helpful for this assignment. The
ATDgen library and tool at the end of that chapter are not permitted for use on
this assignment, because using them would preclude some of the list and tree
processing that we want you to learn from this assignment. Note that the Core
library used in that book is not supported in this course and will cause your
code to fail make check
.
Step 4: Load Adventure Files
The gameplay of TAGs is based on an adventurer moving between rooms. Rooms might represent actual rooms, or they might be more abstract—for example, a room might be an interesting location in a forest. Rooms have named exits through which the adventurer may move to other rooms. The human player’s goal is to explore the rooms.
Adventure files are formatted in JSON. We provide a couple example adventure
files (lonely_room.json
, ho_plaza.json
) in the release code. Take a look
now to familiarize yourself with them. An adventure file contains these entries:
- The rooms. Each room contains these entries:
- an identifier,
- a description of the room, and
- the exits from the room. An exit itself contains two entries:
- the name of the exit, and
- the identifier of the room to which it leads.
- The identifier of the starting room, where the adventurer begins.
Note that JSON strings are case sensitive and may contain whitespace. Unfortunately, JSON does not support multiline strings. That’s why the descriptions in one of those examples necessarily violate the 80-column limit.
Your task: Implement and test the Adventure
compilation unit provided in
the release code. Remember to use test-driven development: write a unit test
that fails, then write code to make the test pass; keep doing that until you are
convinced that your unit tests are sufficient to demonstrate that your code is
correct and complete. All your tests should be in the file test.ml
, which
is provided in the release code. The make test
target will run your test suite
from that file.
The Adventure
documentation mentions set-like lists.
A set-like list is a list in which no element appears more than once, and
in which order is irrelevant. So [1;2;3]
and [3;2;1]
are both set-like
lists and are considered equivalent, but [1;1;2;3]
is not a set-like list.
The starter code provided in test.ml
contains a couple helper functions
for tests involving set-like lists. Tip: make sure that, anywhere
a function specification says it returns a set-like list, you remove any
duplicates that might be in the list. Otherwise, it is not a set-like list
and will fail our test cases.
The Adventure
documentation also mentions valid JSON adventure
representations. A JSON representation of an adventure is valid if and only if:
-
The JSON complies with the description given above, as well as the examples provided in the release code. More precisely, the JSON must match the schema provided in
schema.json
in the release code. The schema specifies what the required components of the JSON are, as well as their names and JSON types. Using a JSON schema validator, you can check the well-formedness of any JSON again the schema. That could be useful in developing your own unit test cases. -
Every room has a unique identifier.
-
Every exit from a given room has a unique name.
-
Exit names contain only alphanumeric (A-Z, a-z, 0-9) and space characters (only ASCII character code 32; not tabs or newlines, etc.).
-
No exit name contains any leading or trailing whitespace. Internally only a single space is permitted between each word (i.e., consecutive sequence of non-space characters).
-
The target of every exit actually exists. That is, the room identifier to which the exit purportedly leads is, in fact, the identifier of a room in the file.
-
The starting room actually exists.
Note that validity is a precondition, not postcondition, of Adventure.from_json
,
therefore your implementation is not required to check for validity. Consequently,
in grading your submission we will never pass invalid adventures to that function.
(If we did, your function would be free to do anything it wanted, including
set our grading computers on fire.)
Furthermore, although it is not technically part of the definition of “valid”, we promise that in our testing of your submission the adventures we use will not be huge. There will be at most on the order of magnitude of 100 rooms, and each room will have at most on the order of magnitude of 100 exits.
This is the stopping point for a satisfactory solution.
Step 5: Parse Commands
The interface to a TAG is based on the player issuing text commands to
a prompt; the game replies with more text and a new prompt, and so on.
Thus, the interface is a kind of read-eval-print-loop (REPL), much like
utop
. For this assignment, commands will be phrases of the form <verb> <object>
.
Verbs are always a single word, whereas objects might consist of multiple words
separated by spaces.
There are only two verbs your engine needs to support:
-
go: The player moves from one room to another by with the verb “go” followed by the name of an exit.
-
quit: The player exits the game engine with this verb, which takes no object.
Commands are case sensitive, as are exit names. So whereas go clock tower
would move the player from Ho Plaza to McGraw Tower in the sample adventure file,
neither GO clock tower
nor go Clock Tower
would.
Your task: Implement and test the Command
compilation unit. The (non-deprecated)
functions in the standard library String
module are perfectly
adequate for the work you need to do. Hint: investigate String.split_on_char
.
Step 6: State Transitions
As the player progresses through an adventurer, some information does not change: the rooms, their exits, and so forth. But other information does change: the player’s current room, and the set of rooms the player has visited. In this assignment we’ll keep track of the latter kind of information as part of the game state. In an imperative language, the game state would be a mutable variable that is changed by functions that implement the game. But in a functional language, the game state must instead be an immutable value. Which leads to the question: how to represent changes?
Looking back at A1’s step
function, we can spot an answer: functions can take in
an old state and return a new state. That’s exactly the solution we’ll use in this
assignment. In particular, when the player attempts to move the adventurer from
one room to another, the function that implements that movement will take in the
current state of the game, and return a new state in which the adventurer has moved.
Or perhaps the movement will turn out to be impossible, in which case the state
will not change.
Your task: Implement and test the State
compilation unit. Note carefully that
the State.go
function’s specification does not permit it to print, which is intended
to guide you toward an idiomatic and functional implementation.
This is the stopping point for a good solution.
Step 7: Interface
At last, it’s time to build the user interface and make the game playable. The requirements for the interface are relatively minimal:
-
When the engine starts, the interface prompts for the name of an adventure file to play. You may not hardcode the adventure file, nor assume anything about its name (e.g., that it ends in
.json
, or that it is in the current directory), nor even that it exists. -
Before prompting for a command, the interface always prints the description of the room in which the player is currently located. The interface does not need to print information about the exit names. (Indeed, level designers might prefer that it not. The
ho_plaza.json
example adventure includes an Easter egg based on that.) -
If the player attempts to move illegally (that is, to an exit that does not exist in the current room), then the interface displays an error message of your choice, then prompts for a new command.
-
If the player issues the quit command, the interface prints a farewell message of your choice, then terminates without any exceptions or error messages from the operating system. That can be implemented simply by allowing all functions to return, or with the expression
exit 0
. (The functionPervasives.exit
terminates the running process, and the0
return code indicates a normal termination.) -
If the player issues a command that cannot be understood, the interface prints an error message of your choice, then prompts for a new command.
We leave the rest of the design of the user interface up to your own creativity. In grading, we will not be strictly comparing your user interface’s text output against expected output, so you have freedom in designing the interface.
The Makefile contains a new target, make play
, that will build your
game engine and launch the interface.
Your task: Implement the Main
compilation unit. Your user interface
must be implemented entirely within main.ml
. It may not be implemented in
state.ml
. As the specification of State.go
says, that function may not have
any side effects, especially not printing.
All the console I/O functions you need are in the Pervasives
module. Its read_line
function is what you should use for
input. You’re welcome to investigate the Printf
and Scanf
modules,
but they are overkill for this assignment. You will likely find the
String.concat
function useful in manipulating object phrases.
The Main
compilation unit is the only part of this assignment for which you
are not required to write unit tests. Instead, you may interactively playtest
your interface.
Scope
Here’s what we consider a satisfactory, good, and excellent solution:
-
Satisfactory: The ZIP file is properly constructed with
make zip
. The solution passesmake check
. TheAdventure
compilation unit is completed. -
Good: The
Command
andState
compilation units are also completed. -
Excellent: The user interface in
Main
is also completed.
“Completed” means implemented, tested, and documented.
Looking ahead to A3: Your A2 grade will be based on what you complete by the time A2 is due. A3 will then ask you to add new functionality. Any functionality that you leave incomplete in A2 will become part of the Satisfactory scope of A3.
Documentation
We are now working with compilation units, which have two pieces: interfaces (.mli
)
and implementations (.ml
). The documentation we produce will now also have two
pieces:
-
documentation for clients of our code base—the people who only need to understand the functions and other names exposed through the interface files, and
-
documentation for maintainers of our code base—the people who need to understand all of the code, including the implementation files.
So the make docs
command now produces two directories of documentation, doc.public
and doc.private
. The “public” documentation is for clients; the “private”, for
maintainers.
The public documentation includes only the names exposed through the interface files. Since those files were provided to you, they have been fully documented already. The only documentation you need to add is for any additional names you might choose to expose through the interfaces. (Though typically you won’t need to do that, unless it’s because you want to make a function publicly available for testing purposes.) You are also free to improve the documentation if you wish.
The private documentation includes all the names, including any helper functions you add to the implementation files. Those you will need to document yourself. The graders will be assessing this private documentation, so if you want to “Meet Expectations”, you need to ensure that all names in it have documentation comments.
If a name is documented in both the .mli
and .ml
files, then in the public
documentation it will contain only the comment from the .mli
file; whereas
in the private documentation, it will contain both the comments from the .ml
file and .mli
files merged together. That means you can add information to
comments for maintainers, but clients won’t see it.
Submission
Record your team’s NetIDs in authors.mli
, and set the hours_worked
variable at the end of authors.ml
.
Run make zip
to construct the ZIP file you need to submit on CMS. Our
autograder needs to be able to find the files you submit inside that ZIP
without any human assistance, so do not use your operating system’s
graphical file browser to construct the ZIP file. Use only the make
zip
command we have taken the trouble to provide. Any mal-constructed
ZIP files will receive a penalty of 20 points. If CMS says your ZIP
file is too large, it’s probably because you did not use make zip
to
construct it; the file size limit in CMS is plenty large for properly
constructed ZIP files.
Ensure that your solution passes make finalcheck
. Submit your
adv.zip
on CMS. Double-check before the deadline that you
have submitted the intended versions of your file.
Congratulations! You’ve had an Adventure!
Acknowledgement: Adapted from Prof. John Estell (Ohio Northern University).