Download hw07.zip. Inside the archive, you will find a file called hw07.scm, along with a copy of the ok
autograder.
Submission: When you are done, submit with python3 ok --submit
. You may submit more than once before the deadline; only the final submission will be scored. Check that you have successfully submitted your code on okpy.org. See Lab 0 for more instructions on submitting assignments.
Using Ok: If you have any questions about using Ok, please refer to this guide.
Readings: You might find the following references useful:
Grading: Homework is graded based on correctness. Each incorrect problem will decrease the total score by one point. There is a homework recovery policy as stated in the syllabus. This homework is out of 2 points.
Scheme is a famous functional programming language from the 1970s. It is a dialect of Lisp (which stands for LISt Processing). The first observation most people make is the unique syntax, which uses a prefix notation and (often many) nested parentheses (see http://xkcd.com/297/). Scheme features first-class functions and optimized tail-recursion, which were relatively new features at the time.
You may find it useful to try code.cs61a.org/scheme when working through problems, as it can draw environment and box-and-pointer diagrams and it lets you walk your code step-by-step (similar to Python Tutor). Don’t forget to submit your code through Ok though!
As you’re writing your code, you can debug using the Scheme Editor. In your scheme
folder you will find a new editor. To run this editor, run python3 editor
. This should pop up a window in your browser; if it does not, please navigate to localhost:31415 and you should see it.
Make sure to run python3 ok
in a separate tab or window so that the editor keeps running.
If you find that your code works in the online editor but not in your own interpreter, it’s possible you have a bug in code from an earlier part that you’ll have to track down. Every once in a while there’s a bug that our tests don’t catch, and if you find one you should let us know!
These videos may provide some helpful direction for tackling the coding problems on this assignment.
To see these videos, you should be logged into your berkeley.edu email.
In the following problems, you will explore creating two separate implementations for the same abstraction.
A keyword list is the Scheme analogue of a dict
in Python, with a few key differences:
The kwlist
abstraction keeps a mapping of keys
and values
. To create a kwlist
, call the constructor (make-kwlist keys values)
where keys
is a Scheme list of symbols and values
is a Scheme list of any type. This returns some abstracted item lst
that we can call the following methods to either retrieve or add items:
scm> (define lst (make-kwlist '(x y z) '(7 8 9)) ; create the keyword list
lst
scm> (get-first-from-kwlist lst 'x) ; get an item
7
scm> (define lst (add-to-kwlist lst 'a 10)) ; add a new item
lst
scm> (get-first-from-kwlist lst 'a) ; get the new item.
10
First, implement abstractions for kwlist
in two ways, with the following example: (kwlist '(x y z) '(7 8 9))
kwlist1
, which stores a keyword list in the following manner: ((key1 key2 key3 ...) (value1 value2 value3 ...)
. With the example above, this should look like ((x y z) (7 8 9))
.kwlist2
, which stores a keyword list in the following manner: ((key1 value1) (key2 value2) ...)
. With the example above, this should look like ((x 7) (y 8) (z 9))
.Specifically, implement constructors and selectors for kwlist1
and kwlist2
.
make-kwlist1
and make-kwlist2
, should take in Scheme lists for both keys
and values
, and construct the abstraction as above.get-keys-kwlist1
, get-keys-kwlist2
, get-values-kwlist1
, and get-values-kwlist1
, should take in a kwlist1
or kwlist2
and return their keys and values respectively. Note that because you are currently creating the implementation, you are “under the abstraction barrier;” feel free to refer to specific details of the structure of kwlist1
and kwlist2
.Hint: The
map
function may prove to be useful, but is not required. You may also use thecadr
function, which is defined for you in the file.
scm> (define ex-lst1 (make-kwlist1 '(a b c) '(1 2 3)))
ex-list
scm> (get-keys-kwlist1 ex-lst1)
(a b c)
scm> (get-values-kwlist1 ex-lst1)
(1 2 3)
scm> (define ex-lst2 (make-kwlist2 '(a b c) '(1 2 3)))
ex-list
scm> (get-keys-kwlist2 ex-lst)
(a b c)
scm> (get-values-kwlist2 ex-lst)
(1 2 3)
(define (make-kwlist1 keys values)
'YOUR-CODE-HERE
)
(define (get-keys-kwlist1 kwlist)
'YOUR-CODE-HERE
)
(define (get-values-kwlist1 kwlist)
'YOUR-CODE-HERE
)
(define (make-kwlist2 keys values)
'YOUR-CODE-HERE
)
(define (get-keys-kwlist2 kwlist)
'YOUR-CODE-HERE
)
(define (get-values-kwlist2 kwlist)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q kwlist_construct
Important: For the following questions, your implementations should be invariant with respect to the abstraction used; that is, it should work regardless of whether
kwlist1
orkwlist2
is used. Specifically, in the tests, we will define the abstractionkwlist
as eitherkwlist1
orkwlist2
:scm> (define make-kwlist make-kwlist1) scm> (define get-keys-kwlist get-keys-kwlist1) scm> (define get-values-kwlist get-values-kwlist1) ; tests here... scm> (define make-kwlist make-kwlist2) scm> (define get-keys-kwlist get-keys-kwlist2) scm> (define get-values-kwlist get-values-kwlist2) ; tests here...
You should refer to the above
kwlist
procedures, notkwlist1
orkwlist2
’s procedures in your implementation.
Now, implement add-to-kwlist
, which implements support for adding a new (key
, value
) pair to any implementation of a kwlist
. Specifically, add-to-kwlist
takes in a kwlist
, a key
, and a value
as input, and returns a new kwlist
with updated keys and values. Note that kwlist
s are ordered; that is, a pair p1
that was added to a kwlist
before a different pair p2
should appear earlier in the kwlist
.
Hint: The
append
method may be useful here. To make your implementation work with both abstractions, be sure to use methods ending inkwlist
, notkwlist1
orkwlist2
.
scm> (define ex-lst (make-kwlist '(a b c) '(1 2 3)))
ex-lst
scm> (get-keys-kwlist ex-lst)
(a b c)
scm> (get-values-kwlist ex-lst)
(1 2 3)
scm> (define ex-lst (add-to-kwlist ex-lst 'd '4))
ex-lst
scm> (get-keys-kwlist ex-lst) ; note that new items are at the end of the list!
(a b c d)
scm> (get-values-kwlist ex-lst) ; here too!
(1 2 3 4)
(define (add-to-kwlist kwlist key value)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q kwlist_add
Now, implement get-first-from-kwlist
, which implements support for getting the first value bound to a key
in kwlist
. If key
is not present in the list, the function should return nil
to indicate that there were no valid keys found.
Hint: Consider using
let
to temporarily bind names to values. To make your implementation work with both abstractions, be sure to use methods ending inkwlist
, notkwlist1
orkwlist2
.
scm> (define ex-lst (make-kwlist '(a b c) '(1 2 3)))
ex-lst
scm> (get-first-from-kwlist ex-lst 'b)
2
scm> (get-first-from-kwlist ex-lst 'd) ; if not found, return nil
()
scm> (define ex-lst (add-to-kwlist ex-lst 'd '4))
ex-lst
scm> (get-first-from-kwlist ex-lst 'b)
2
scm> (get-first-from-kwlist ex-lst 'd)
4
scm> (define ex-lst (add-to-kwlist ex-lst 'd '5))
ex-lst
scm> (get-first-from-kwlist ex-lst 'b)
2
scm> (get-first-from-kwlist ex-lst 'd) ; return the *first* occurrence
4
(define (get-first-from-kwlist kwlist key)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q kwlist_get
Note that the following question is separate from the previous questions.
Implement prune-expr
, a procedure that takes in an expression, which is represented as a list, and returns the same expression with every other argument included. The operator should not be modified.
Hint: You may find it helpful to write a helper function that prunes a list.
The behavior of prune-expr
is specified by the following doctests:
scm> (prune-expr '(+ 10 20))
(+ 10)
scm> (prune-expr '(+ 10 20 30))
(+ 10 30)
scm> (eval (prune-expr '(+ 10 20 30)))
40
(define (prune-expr expr)
(define (prune-helper lst)
'YOUR-CODE-HERE
)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q prune-expr
Recall that curry
ing transforms a multiple argument function into a series of higher-order, one argument functions. In the next set of questions, you will be creating functions that can automatically curry a function of any length using the notion that programs are data!
Implement the function curry-cook
, which takes in a Scheme list formals
and a quoted expression body
. curry-cook
should generate a program as a list which is a curried version of a lambda function. The outputted program should be a curried version of a lambda function with formal arguments equal to formals
, and a function body equal to body
. You may assume that all functions passed in will have more than 0 formals
; otherwise, it would not be curry-able!
For example, if you wanted to curry the function (lambda (x y) (+ x y))
, you would set formals
equal to '(x y)
, the body
equal to '(+ x y)
, and make a call to curry-cook
: (curry-cook '(x y) '(+ x y))
.
scm> (curry-cook '(a) 'a)
(lambda (a) a)
scm> (curry-cook '(x y) '(+ x y))
(lambda (x) (lambda (y) (+ x y)))
(define (curry-cook formals body)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q curry_cook
Now that you have a function that creates lambda programs as lists, create a function which is able to evaluate lambda functions using a series of arguments. Specifically, implement the function curry-consume
, which takes in a curried lambda function curries
(not a list), and apply
s the function to a list of arguments args
. Similarly to the previous question, you may make several assumptions:
curries
is an n
-curried function, then there will be at most n
arguments in args
.curries
has been fully apply
‘d with relevant arguments; in this case, curries
now contains a value representing the output of the lambda function. Return it.Note that there can be fewer args
than formals
for the corresponding lambda function curries
! In the case that there are fewer arguments, curry-consume
should return a curried lambda function, which is the result of partially apply
ing curries
up to the number of args
provdied.
scm> (define three-curry (curry-cook '(x y z) '(+ x (* y z))))
three-curry
scm> three-curry
(lambda (x) (lambda (y) (lambda (z) (+ x (* y z)))))
scm> (define three-curry-fn (eval three-curry)) ; three-curry-fn is a lambda function derived from the program
three-curry-fn
scm> (define eat-two (curry-consume three-curry-fn '(1 2))) ; pass in only two arguments, return should be a one-arg lambda function!
eat-two
scm> eat-two
(lambda (z) (+ x (* y z)))
scm> (eat-two 3) ; pass in the last argument; 1 + (2 * 3)
7
scm> (curry-consume three-curry-fn '(1 2 3)) ; all three arguments at once
7
(define (curry-consume curries args)
'YOUR-CODE-HERE
)
Use Ok to test your code:
python3 ok -q curry_consume
Homework assignments will also contain prior exam-level questions for you to take a look at. These questions have no submission component; feel free to attempt them if you’d like a challenge!