1. Lists CSC 358/458 4.3.2006

2. Outline Lab #1 / Homework #1 Lists A question List Internals of Lists List operations Extended Example

3. Lab #1 (defun add-car (lst1 lst2) (+ (car lst1) (car lst2))) (defun deal-hand (size deck) (let ((hand '())) (dotimes (i size) (push (pop deck) hand)) hand)) (defun count-face-cards (hand) (let ((counter 0)) (dolist (card hand) (if (face-cardp card) (incf counter))) counter)) (defun face-cardp (card) (not (numberp (card-rank card))))

4. Lab #1 (Extra credit) (defun average-face-cards (trials size) (let ((deck (make-deck)) (count 0)) (dotimes (k trials) (setf deck (shuffle-deck deck)) (incf count (count-face-cards (deal-hand size deck)))) (/ (float count) trials)))

5. Homework #1 ;; Cheating version (using case) (defun card-blackp (card) (case (card-suit card) ((H D) nil) ((S C) t))) ;; Non-cheating version (using cond) (defun card-blackp (card) (let ((suit (card-suit card))) (cond ((or (eq suit 'H) (eq suit 'D)) nil) (t t)))) ;; Functional version with boolean return value (defun card-blackp (card) (let ((suit (card-suit card))) (or (eq suit 'S) (eq suit 'C))))

6. Homework #1 (defun not-twenty-onep (card1 card2) (let ((rank1 (card-rank card1)) (rank2 (card-rank card2))) (if (and (= rank1 1) (= rank2 1)) ;; special case 12 (let ((value1 (card-points-bj rank1)) (value2 (card-points-bj rank2))) (let ((value-sum (+ value1 value2))) (if (= value-sum 21) nil value-sum)))))) (defun card-points-bj (rank) (case rank ((k q j) 10) (1 11) (otherwise rank)))

7. Homework #1 (defun card-points-skat (rank) (case rank (1 11) (10 10) (k 4) (q 3) (j 2) (otherwise 0))) (defun skat-trick (trick) (let ((score 0)) (dolist (card trick) (incf score (card-points-skat (card-rank card)))) score))

8. Homework #1 (XC) (defun can-buildp (hand center) (dolist (hand-card hand) (dolist (center-card center) (if (can-build-cards hand-card center-card hand) (return-from can-buildp (list hand-card center-card))))))) (defun can-build-cards (hand-card center-card hand) (let ((hand-value (card-points-cassino (card-rank hand-card))) (center-value (card-points-cassino (card-rank center-card)))) (let ((build-value (+ hand-value center-value))) (dolist (card hand) (if (= build-value (card-points-cassino (card-rank card))) (return-from can-build-cards t)))))) (defun card-points-cassino (rank) (case rank (k 13) (q 12) (j 11) (otherwise rank)))

9. Homework #1 (defun can-buildp (hand center) (let ((ans nil)) (dolist (hand-card hand) (dolist (center-card center) (if (can-build-cards hand-card center-card hand) (setf ans (list hand-card center-card))))) ans))

10. Last week How to deal with card games with different point values? One possibility lots of named functions ick Better approach be data-driven let the functions be defined by the data Problem don't want an extra argument to a comparison operator suppose I want to supply the comparison to a sort function?

11. Lisp solution Create an anonymous function using the data use this function for comparing Benefits can create as many as needed symbol table not filled with unnecessary stuff

12. card-comparator (defvar *default-card-values* '((K 13) (Q 12) (J 11))) (defvar *ace-high-values* '((1 14) (K 13) (Q 12) (J 11))) (defvar *skat-card-values* '((1 14) (10 13) (K 12) (Q 11) (J 10))) (defun rank-to-number2 (rank card-values) (let ((value-pair (assoc rank card-values))) (if value-pair (cadr value-pair) rank))) (defun card-comparator (&optional (card-values *default-card- values*)) #'(lambda (card1 card2) (let ((rank1 (card-rank card1)) (rank2 (card-rank card2))) (< (rank-to-number2 rank1 card-values) (rank-to-number2 rank2 card-values)))))

13. Some syntax: lambda lambda a lot more about this next week this is how we define an anonymous function defun is just a convenience (setf (function foo) #'(lambda (x y) (+ x y))) the same as (defun foo (x y) (+ x y)) lambda is fundamental

14. Optional arguments More about this next week I can specify in the parameter list that some arguments are optional and then supply default values Optional arguments must be at the end for obvious reasons

15. Closure Note the function I define using lambda refers to a parameter defined in the enclosing scope card-values the function is returned to the caller the enclosing scope disappears can I do this? Yes! technical term = "lexical closure" more about this next week

16. Lists

17. A Question Why can’t I really deal from my deck? Why does this function not change the value of its argument? (defun mypop (lst) (pop lst)) > (setf l1 '(a b c)) > (mypop l1) A > l1 (A B C)

18. Lists sequences of items atoms or lists car first element cdr rest cons builds a list

19. Other List Builders list a list containing the arguments append a list combining the arguments

20. Internals (a b c)

21. Nested List

22. Improper List

23. What Does It Do?* (setf l1 '(a (b c))) (setf l2 '((d) e)) (car l2) (cdr l1) (cons 'b l1) (list l2 'd) (append l1 l2)

24. What does it do? (pop l1) special form short for (setf l1 (cdr l1)) (defun mypop (lst) (pop lst)) (defun my pop (lst) (setf lst (cdr lst))) (mypop l1) l1 doesn't change why should it? it is still bound to the same thing

25. How can we achieve this? Global variable ick Game state structure pass in and out contents manipulated better

26. Example: Blackjack (defun make-bj-game (player-count) (list (shuffle-deck (make-deck)) (make-list player-count :initial-element nil))) (defun player-count (game) (length (hands game))) (defun deck (game) (cadr game)) (defun hands (game) (caddr game)) (defun hand (player game) (nth player (hands game)))

27. New function make-list note the function call :initial-element keyword argument we'll see this again later

28. Shared Structure It is easy to create lists that share structure (setf c (append a b)) Not a problem unless a or b is destructively modified

29. List Functions Accessors navigating around in lists Operations manipulate the whole list Destructive operations modify the list

30. List Accessors car, cdr positional first, second, third (nth i lst) (nthcdr i lst) end of the list last actually returns last cdr

31. List Operations Many! Mapping functions Boolean operations Finding Modifying (copy) Reduce

32. Mapping Functions that apply a function over and over across a list (mapcar #’fn lst) applies the function to each element in the list returns a list of the results Also works with multiple lists (mapcar #’fn lst1 lst2)

33. Example Matt's question What if I want to add 5 to everything in a list? (mapcar #'(lambda (x) (+ x 5)) '(1 2 3))

34. Other Mapping Functions maplist applies to successive cdrs mapcan joins results with append mapc, mapl like mapcar and maplist but doesn't return results

35. Boolean Operations logical summary of a list relative to a function (some #’fn lst) true if any element of lst returns true when the function is applied every false if any element returns false notany notevery

36. Example flushp is a poker hand a flush? 5-card hand, please logic all cards the same suit simple method all spades or all clubs or... better way?

37. Finding Operations (member elem lst) if elem is in list, returns the cdr headed by elem (position elem lst) if elem is in list, return position index both return nil is absent

38. Modifications (remove elem lst) returns a copy of lst with elem removed not “destructive” Also "if" version remove-if has a test instead of an element (substitute old new lst) returns a copy of lst with old replaced by new

39. Reduce “reduce” list into a single value (reduce #’fn lst) applies fn to first two elements applies fn to that result and next element etc. (reduce #’+ ‘(1 2 3 4)) => 10

40. Example count-face-cards mapcar and reduce Original version (defun count-face-cards (hand) (let ((counter 0)) (dolist (card hand) (if (face-cardp card) (incf counter))) counter)) (defun face-cardp (card) (not (numberp (card-rank card))))

41. Destructive Operations I Most Lisp operations return “new” lists remove (setf lst ‘(b a b a b)) (remove ‘a lst) => (b b b) lst => (b a b a b) Use the result of the function New cons cells created not always what you want

42. Destructive Operations II Alter the contents of cons cells rplaca = replace the car rplacd = replace the cdr nconc = destructive append delete = destructive remove nsubst = destructive subst Others “n” means “non-consing”

43. Generalized Variables setf can be similarly used (setf (car lst) val) (setf (cdr lst) val) Main reason to use efficiency creating new cons cells creating garbage

44. Problems with Destructive Operations Shared structure Effects hard to predict Not necessary Main reason to use efficiency but remember Knuth “Premature optimization is the root of all evil.”

45. Example Look at deck after shuffle-deck

46. defsetf One principle of Lisp is that the programmer has the same power as the language designer You can create your own generalized variables use defsetf

47. Example (defun 3rd (lst) (nth 2 lst)) (defun set-3rd (lst value) (setf (nth 2 lst) value)) (defsetf 3rd set-3rd) (setf lst '( a b c d)) (setf (3rd lst) 'q) lst => (a b q d) (setf (3rd lst))

48. Association List Allows lookup of values associated with keys ((key1. val1) (key2. val2)... (keyn. valn)) OK for short associations large associations a hash table is better (assoc key alist) returns the pair (rassoc val alist) returns the pair

49. Keyword parameters what if the value is not an atom? (rassoc '(21 M) '((John 21 M) (Jane 19 F) (Joe 25 M))) doesn’t work because rassoc uses eq extra arguments to rassoc :test specifies which function to use instead of default (rassoc '(21 M) '((John 21 M) (Jane 19 F) (Joe 25 M)) :test #'equal)

50. Keyword Parameters Many functions have these Most common :test :key specifies a function to apply before testing (rassoc '21 '((John 21 M) (Jane 19 F) (Joe 25 M)) :key #'car) :start :end

51. Equality Operators = numerical equality only eq reference equality eql eq numbers and characters equal tests lists element by element can be expensive! Which one to use? if you know your data type, be specific (equal for lists, = for numbers, eql for everything else) if you really want reference equality, use eq

52. Defining Keyword Parameters (defun foo (a b &key c (d 5))...) 5 is the default value for d can be called (foo 1 2) (foo 1 2 :c 1 :d 3)

53. Example sorting hands by rank by suit allow different possibilities (sort-hand hand :test...)

54. Sequences Many of the list function also work on vectors lisp 1-D arrays The data type that covers both lists and vectors is called the "sequence" functions that take a sequence as an argument will operate on either

55. Extended Example blackjack