project2b

Project 2b: OCaml Higher Order Functions and Data

Due: October 12, 2021 at 11:59 PM (late October 13, 10% penalty)

Points: 65 public, 35 semipublic

This is an individual assignment. You must work on this project alone.

Overview

The goal of this project is to increase your familiarity with programming in OCaml and give you practice using higher order functions and user-defined types. You will have to write a number of small functions, the specifications of which are given below.

Ground Rules

In addition to writing your own code, you may use library functions found in the Stdlib module and the functions provided in funs.ml. You may not (under threat of a grading penalty) use the List module, any submodules of Stdlib (such as Hashtbl, Set, etc.), or any imperative features of OCaml (e.g., mutable references and arrays). You are allowed to use the @ (list append) operator.

Testing & Submitting

Submit by running gradescope-submit. ALternatively, upload higher.ml and data.ml to the assignment on gradescope.com.

To test locally, run dune runtest -f. Besides the provided public tests, you will also find the file student.ml on test/student/, where you'll be able to add OUnit tests of your own. More detailed information about writing tests can be found here. Here are the timestamps for the topics covered in the video:

• Installing necessary software: 00:46
• How to build and test: 01:14
• List all available tests: 04:40
• Running a specific test: 05:05
• Testing inside of utop: 09:00
• Understanding test cases: 16:00
• Writing your own test cases: 19:20

You can interactively test your code by doing dune utop src, which will include your source files. (As usual, all of your commands in utop need to end with two semicolons (i.e. ;;), otherwise it will appear that your terminal is hanging.)

Property-based Tests

This project also includes Property-based tests (PBT) for you to use. Property based testing enables pretty good test coverage without having to write a lot of individual tests. In particular, you specify a property that your code should have on many possible inputs, not just a particular one, and the property tester will generate lots of random inputs against which to test the property.

The property-based tests (PBTs) we've provided are not meant to test your code entirely, but rather to give you a little help. You should still write your own tests. You can write typical unit tests (and/or test using utop or the ocaml top level), or if you like you can write more PBTs. You can find the provided PBTs in test/pbt, along with comments and instructions on how to expand them. We'll cover PBTs in more detail later in the class, so don't feel obligated to play with these now.

Notes

• In this project, we've changed the way we give you examples. Instead of giving you example cases and expected output, we've given you OCaml code that you run in utop ot the ocaml top level. Make sure to ignore any failures when running examples for code where order of the result doesn't matter.
• Some of the examples given use OUnit2.assert_raises to handle exceptions and failures. While the normal assert works fine in utop, assert_raises requires the OUnit2 module to be opened.
• Unlike most other languages, = in OCaml is the operator for structural equality whereas == is the operator for physical equality. All functions in this project (and in this class, unless ever specified otherwise) are concerned with structural equality.
• At a few points in this project, you will need to raise an Invalid_argument exception. Use the invalid_arg function to do so:

  invalid_arg "something went wrong"

  Use the error message that the function specifies as the argument.

Part 1: Higher Order Functions

Write the following functions in higher.ml using map, fold, or fold_right as defined in the file funs.ml. You must use map, fold, or fold_right to complete these functions, so no functions in higher.ml should be defined using the rec keyword. You will lose points if this rule is not followed. Use the other provided functions in funs.ml to make completing the functions easier.

Some of these functions will require just map or fold, but some will require a combination of the two. The map/reduce design pattern may come in handy: Map over a list to convert it to a new list which you then process a second time using fold. The idea is that you first process the list using map, and then reduce the resulting list using fold.

contains_elem lst e

• Type: 'a list -> 'a -> bool
• Description: Returns true if e is present in the list lst, and false if it is not.
• Examples:

  assert(contains_elem [] 1 = false);;
  assert(contains_elem [1;2;3] 4 = false);;
  assert(contains_elem [1;2;3;3;2;4] 2 = true);;

is_present lst x

• Type: 'a list -> 'a -> int list
• Description: Returns a list of the same length as lst which has a 1 at each position in which the corresponding position in lst is equal to x, and a 0 otherwise.
• Examples:

  assert(is_present [1;2;3] 1 = [1;0;0]);;
  assert(is_present [1;1;0] 0 = [0;0;1]);;
  assert(is_present [2;0;2] 2 = [1;0;1]);;

count_occ lst target

• Type: 'a list -> 'a -> int
• Description: Returns how many elements in lst are equal to target.
• Examples:

  assert(count_occ [] 1 = 0);;
  assert(count_occ [1] 1 = 1);;
  assert(count_occ [1; 2; 2; 1; 3] 1 = 2);;

uniq lst

• Type: 'a list -> 'a list
• Description: Given a list, returns a list with all duplicate elements removed. Order does not matter, in the output list.
• Examples:

  assert(uniq [] = []);;
  assert(uniq [1] = [1]);;
  assert(uniq [1; 2; 2; 1; 3] = [2; 1; 3]);;

assoc_list lst

• Type: 'a list -> ('a * int) list
• Description: Given a list, returns a list of pairs where the first integer represents the element of the list and the second integer represents the number of occurrences of that element in the list. This associative list should not contain duplicates. Order does not matter, in the output list.
• Examples:

  assert(assoc_list [] = []);;
  assert(assoc_list [1] = [(1,1)]);;
  assert(assoc_list [1; 2; 2; 1; 3] = [(1, 2); (2, 2); (3, 1)]);;

ap fns args

• Type: ('a -> 'b) list -> 'a list -> 'b list
• Description: Applies each function in fns to each argument in args in order, collecting all results in a single list.
• Examples:

  assert(ap [] [1;2;3;4] = []);;
  assert(ap [succ] [] = []);;
  assert(ap [(fun x -> x^"?"); (fun x -> x^"!")] ["foo";"bar"] = ["foo?";"bar?";"foo!";"bar!"]);;
  assert(ap [pred;succ] [1;2] = [0;1;2;3]);;
  assert(ap [int_of_float;fun x -> (int_of_float x)*2] [1.0;2.0;3.0] = [1; 2; 3; 2; 4; 6]);;

(Here, succ, pred, and int_of_float are standard library functions. The (^) function is string concatenation.)

Part 2: Three-Way Search Tree

Solutions to the problems in this and the following sections should be implemented in data.ml.

Here, you will write functions that will operate on a 3-Way Search Tree (3WST). Here is an example of a 3WST, and its type definition int_tree follows.

type int_tree =
  | IntLeaf
  | IntNode of int * int option * int_tree * int_tree * int_tree 

let empty_int_tree = IntLeaf

According to this definition, an int_tree is either: empty (just a leaf IntLeaf), or a node IntNode containing 2 integers and three branches. The first integer is sure to be there, but the second may not be -- it's a int option. When both integers are present (i.e., the second is Some x for some x), the left integer must be strictly smaller than the right integer. When only the one integer is present, all branches should be IntLeaf. When both integers are present, the the first branch is an int_tree containing integers that are strictly smaller than the first integer; the second branch consists an int_tree of integers in between the first and second integer, and the third is an int_tree with integers strictly larger than the second integer. A consequence of these invariants is that a int_tree may not contain duplicate integers.

Stepping back, this is similar to a binary search tree (BST), almost like two BSTs smushed together.

int_insert x t

As is typical in OCaml, 3WSTs are immutable: Once created a int_tree value cannot be changed. To insert an element into a tree, create a new tree that has the same contents as the old, but with the new element added.

• Type: int -> int_tree -> int_tree
• Description: Given an integer x and a 3WST t, this function should return the 3WST that results from inserting x in t. If x is already present in t, then t will be returned unchanged.
• Hint: When t is a node with only one integer, adding to that node means returning a node in which the second integer is non-None: If the integer you are adding is less than the current integer, the new node will put the old integer second, and the new one first.
• Examples:

  let t0 = int_insert 5 IntLeaf;;
  let t1 = int_insert 5 t0;;
  let t2 = int_insert 1 t1;;
  let t3 = int_insert 6 t2;;
  let t4 = int_insert 0 t3;;
  
  assert(t0 = IntNode (5, None, IntLeaf, IntLeaf, IntLeaf));;
  assert(t1 = IntNode (5, None, IntLeaf, IntLeaf, IntLeaf));;
  assert(t2 = IntNode (1, Some 5, IntLeaf, IntLeaf, IntLeaf));;
  assert(t3 = IntNode (1, Some 5, IntLeaf, IntLeaf, IntNode(6, None, IntLeaf, IntLeaf, IntLeaf)));;
  assert(t4 = IntNode (1, Some 5, IntNode(0, None, IntLeaf, IntLeaf, IntLeaf), IntLeaf, IntNode(6, None, IntLeaf, IntLeaf, IntLeaf)));;

int_mem x t

• Type: int -> int_tree -> bool

• Description: Given an integer x and an int tree t, this function should return true if x can be found as the data in any node of t and false otherwise.

• Examples (Using t4 above):

  assert(int_mem 5 t4 = true);;
  assert(int_mem 10 t4 = false);;

int_size t

• Type: int_tree -> int
• Description: Returns the number of values in the nodes int tree t. A node with 2 integers is counted as size 2.
• Examples:

  assert(int_size empty_int_tree = 0);;
  assert(int_size t4 = 4);;

int_max t

• Type: int_tree -> int
• Description: Returns the maximum element in tree t. Raises exception Invalid_argument("int_max") on an empty tree. This function should be O(height of the tree).
• Examples:

  assert(int_max t4 = 6);;

Part 3: Three-Way Search Tree-based Map

In this part, you will extend the Three-Way Search Tree to implement a map from a key of type int to a value of type 'a. Below is the type definition for the map.

type 'a tree_map =
  | MapLeaf
  | MapNode of (int * 'a) * (int * 'a) option * 'a tree_map * 'a tree_map * 'a tree_map

let empty_tree_map = MapLeaf

This is basically the same as your 3WST, except that instead of two integers (one of which is an option), you have two pairs, where each pair is an integer and the corresponding value in the map. This value is polymorphic (aka generic) and can have any type 'a (which will be fixed to a particular type for a particular map). Implementing the tree map is similar to implementing the 3WST. Instead of inserting an integer into an int_tree, you are inserting the integer and the value it maps to. You should be able to mostly reuse most your code from Part 2. If you find you are writing more than just a small bit of new code, you are probably doing something wrong.

map_put k v t

• Type: int -> 'a -> 'a tree_map -> 'a tree_map
• Description: Given an integer key k and a value v and a map t, return the result of inserting the mapping of k to v in t. If a mapping for k already exists in t, raise the exception Invalid_argument("map_put").
• Examples:

    let t0 = empty_tree_map;;
    let t1 = map_put 5 "Hello" t0;;
    let t2 = map_put 2 "Goodbye" t1;;
    let t3 = map_put 4 "World" t2;; 
  
    assert(t0 = MapLeaf);;
    assert(t1 = MapNode ((5, "Hello"), None, MapLeaf, MapLeaf, MapLeaf));;
    assert(t2 = MapNode ((2, "Goodbye"), Some (5, "Hello"), MapLeaf, MapLeaf, MapLeaf));;
    assert(t3 = MapNode ((2, "Goodbye"), Some (5, "Hello"), MapLeaf, MapNode ((4, "World"), None, MapLeaf, MapLeaf, MapLeaf), MapLeaf));;
  
    OUnit2.assert_raises (Invalid_argument "map_put") (fun () -> map_put 4 "Mars" t3);;

map_contains k t

• Type: int -> 'a tree_map -> bool
• Description: Given a key k returns true if there is a value associated with the key k in t, and false otherwise.
• Examples:

    let t0 = empty_tree_map;;
    let t1 = map_put 1 "Ace" t0;;
    let t2 = map_put 2 "King" t1;;
    let t3 = map_put 7 "Howdy" t2;;
  
    assert(map_contains 1 t1 = true);;
    assert(map_contains 2 t1 = false);;
    assert(map_contains 7 t3 = true);;
    assert(map_contains 13 t3 = false);;  

map_get k t

• Type: int -> 'a tree_map -> 'a
• Description: Given a key k returns the value mapped from k in t, if it exists. If it doesn't, raise the exception Invalid_argument("map_get").
• Examples:

    let t0 = empty_tree_map;;
    let t1 = map_put 3 "Hello" t0;;
    let t2 = map_put 2 "Goodbye" t1;;
    let t3 = map_put 4 "World" t2;;
  
    assert(map_get 3 t2 = "Hello");;
    assert(map_get 2 t2 = "Goodbye");;
    assert(map_get 4 t3 = "World");;
  
    OUnit2.assert_raises (Invalid_argument "map_get") (fun () -> map_get 5 t3);;
    OUnit2.assert_raises (Invalid_argument "map_get") (fun () -> map_get 2 t1);;

Part 4: Variable Lookup

For this part of the project, you will be trying to implement a variable lookup table that handles nested scopes, and shadowing. Consider the following code snippet in C that demonstrates the function of block scopes, and shows how variables defined in inner scopes can shadow those in outer scopes.

{
  int a = 30;
  int b = 40;
  // POINT A
  {
    int a = 20;
    int c = 10;
    // POINT B
  }
  // POINT C
}

The inner braces create a new scope for variable bindings. At POINT A (marked in a comment), a is bound to 30, but at that point a new scope is opened, and a new binding for a is introduced. This binding shadows the original binding, so that any reference to a, e.g., at POINT B, will return 20, not 30. Once POINT C is reached, the inner binding of a is dropped and the original binding is visible again; c's binding is dropped too. Note that you cannot shadow a variable within a scope, only when a new scope is opened. If you're a little rusty on block scoping, you can check out this link or play around with scopes and print statements in C using gcc.

For this part of the project, you will design your own data type lookup_table in data.ml. You want to design a type that can be used to build all of the following functions, which implement map from variables (represented as strings) to values (always ints), within a regime of blocked scopes. We will only test your table through the functions that you implement (specified below). There are many different ways to solve this portion of the project! Pick one that makes sense to you.

type lookup_table

• Description: This is not a function. Rather, it is a type that you'll have to define based on what you think is necessary for the below requirements.
• What you need to do: Change the type definition line in the data.ml file to the type that you choose to represent a lookup_table.

empty_table

• Type: lookup_table
• Description: This is a value of type lookup_table that represents an empty lookup table, i.e., with no scopes and no mappings. The implementation of this will depend on how you define the lookup table above.

push_scope table

• Type: lookup_table -> lookup_table
• Description: Returns a lookup_table that has an added inner scope.
• Examples:

  let t = empty_table in
  push_scope t;;

pop_scope table

• Type: lookup_table -> lookup_table
• Description: Returns a lookup_table that has removed its most recent scope. All of the bindings previous to the dropped scope should remain intact. If no scopes are left, throw a failure with failwith "No scopes remain!"
• Examples:

  let t = empty_table;;
  let t1 = push_scope t;;
  let t2 = pop_scope t1;;
  
  OUnit2.assert_raises (Failure "No scopes remain!") (fun () -> pop_scope t2);;

add_var name value table

• Type: string -> int -> lookup_table -> lookup_table
• Description: Updates the most recent scope in table to include a new variable named name with value value. If no scopes exist in table, raise a failure with failwith "There are no scopes to add a variable to!" If a binding for variable name already exists in the current scope then raise a failure with failwith "Duplicate variable binding in scope!".
• Examples:

  let t = empty_table;;
  let t1 = add_var "hello" 3 (push_scope t);;
  let t2 = add_var "hi" 5 t1;;
  
  OUnit2.assert_raises (Failure "Duplicate variable binding in scope!") (fun () -> add_var "hello" 6 t2);;
  OUnit2.assert_raises (Failure "There are no scopes to add a variable to!") (fun () -> add_var "oops" 1 empty_table);;

lookup name table

• Type: string -> lookup_table -> int
• Description: Returns the value associated with name in table in the current environment. If multiple values have been assigned to the same variable name, the most recently bound value in the current scope should be returned. If no variable named name is in table, raise a failure with failwith "Variable not found!".
• Examples:

  let t1 = add_var "hello" 3 (push_scope empty_table);;
  assert(lookup "hello" t1 = 3);;
  let t2 = add_var "hi" 5 t1;;
  assert(lookup "hello" t2 = 3);;
  let t3 = push_scope t2;;
  let t4 = add_var "hello" 6 t3;;
  assert(lookup "hello" t4 = 6);;
  let t5 = pop_scope t4;;
  assert(lookup "hello" t5 = 3);;
  
  OUnit2.assert_raises (Failure "Variable not found!") (fun () -> lookup "hello" empty_table);;