|
| 1 | +#lang racket |
| 2 | + |
| 3 | +(require data/heap) |
| 4 | + |
| 5 | +(struct branch (freq left right)) |
| 6 | +(struct leaf (sym freq)) |
| 7 | + |
| 8 | +(define (weight node) |
| 9 | + (if (branch? node) |
| 10 | + (branch-freq node) |
| 11 | + (leaf-freq node))) |
| 12 | + |
| 13 | +(define (make-huffman-tree leaves) |
| 14 | + (define (combine x y) |
| 15 | + (branch (+ (weight x) (weight y)) x y)) |
| 16 | + |
| 17 | + (define hp (make-heap |
| 18 | + (lambda (x y) |
| 19 | + (<= (weight x) (weight y))))) |
| 20 | + |
| 21 | + (heap-add-all! hp leaves) |
| 22 | + |
| 23 | + (define (merge) |
| 24 | + (if (= (heap-count hp) 1) |
| 25 | +; if there's only one node in the heap then return that element |
| 26 | + (heap-min hp) |
| 27 | + |
| 28 | +; else merge the two smallest nodes |
| 29 | + (let ((a (heap-min hp))) |
| 30 | +; get the smallest node and remove it |
| 31 | + (heap-remove-min! hp) |
| 32 | + |
| 33 | + (let ((b (heap-min hp))) |
| 34 | +; get the second smallest node and remove it |
| 35 | + (heap-remove-min! hp) |
| 36 | + |
| 37 | +; add the combined node to the heap |
| 38 | + (heap-add! hp (combine a b)) |
| 39 | + (merge))))) |
| 40 | + (merge)) |
| 41 | + |
| 42 | +(define (make-codebook tree) |
| 43 | + (define ht (make-hash)) |
| 44 | + |
| 45 | + (define (dfs node path) |
| 46 | + (if (leaf? node) |
| 47 | +; if node is a leaf then add the symbol to the codebook |
| 48 | + (hash-set! ht (leaf-sym node) path) |
| 49 | + (begin (dfs (branch-left node) (append path (list 0))) |
| 50 | + (dfs (branch-right node) (append path (list 1)))))) |
| 51 | + |
| 52 | + (dfs tree null) |
| 53 | + ht) |
| 54 | + |
| 55 | +(define (encode codebook msg) |
| 56 | + (if (null? msg) |
| 57 | + null |
| 58 | + (append (hash-ref codebook (car msg)) |
| 59 | + (encode codebook (cdr msg))))) |
| 60 | + |
| 61 | +(define (decode tree msg) |
| 62 | + (define (dfs node bits) |
| 63 | + (if (leaf? node) |
| 64 | + (cons (leaf-sym node) (decode tree bits)) |
| 65 | + (if (= (car bits) 0) |
| 66 | + (dfs (branch-left node) (cdr bits)) |
| 67 | + (dfs (branch-right node) (cdr bits))))) |
| 68 | + |
| 69 | + (if (null? msg) |
| 70 | + null |
| 71 | + (dfs tree msg))) |
| 72 | + |
| 73 | +(define (generate-leaves msg) |
| 74 | + (define ht (make-hash)) |
| 75 | + (for ([ch msg]) |
| 76 | + (hash-update! ht ch add1 (lambda () 0))) |
| 77 | + (for/list ([(k v) (in-hash ht)]) |
| 78 | + (leaf k v))) |
| 79 | + |
| 80 | +; test |
| 81 | +(define message (string->list "bibbity bobbity")) |
| 82 | +(define leaves (generate-leaves message)) |
| 83 | +(define huffman-tree (make-huffman-tree leaves)) |
| 84 | +(define codebook (make-codebook huffman-tree)) |
| 85 | + |
| 86 | +(define encoded (encode codebook message)) |
| 87 | +(define decoded (list->string (decode huffman-tree encoded))) |
| 88 | + |
| 89 | +(println encoded) |
| 90 | +(println decoded) |
0 commit comments