Solution #f3599802-995a-41e2-acd1-e92a052ec12f

1def solve(input):
2    text = input.get("text", "")
3    n = len(text)
4    if n == 0:
5        return {"encoded_length": 0, "decoded": ""}
6
7    from array import array
8
9    counts = {}
10    get = counts.get
11    for ch in text:
12        counts[ch] = get(ch, 0) + 1
13
14    m = len(counts)
15    if m == 1:
16        return {"encoded_length": n, "decoded": text}
17
18    freqs = array('Q', counts.values())
19    freqs = array('Q', sorted(freqs))
20
21    q = array('Q', [0]) * (m - 1)
22    i = j = k = 0
23    total = 0
24
25    while k < m - 1:
26        if i < m and (j >= k or freqs[i] <= q[j]):
27            a = freqs[i]
28            i += 1
29        else:
30            a = q[j]
31            j += 1
32
33        if i < m and (j >= k or freqs[i] <= q[j]):
34            b = freqs[i]
35            i += 1
36        else:
37            b = q[j]
38            j += 1
39
40        s = a + b
41        q[k] = s
42        total += s
43        k += 1
44
45    return {"encoded_length": total, "decoded": text}

1def solve(input):
2    text = input.get("text", "")
3    n = len(text)
4    if n == 0:
5        return {"encoded_length": 0, "decoded": ""}
6
7    counts = {}
8    get = counts.get
9    for ch in text:
10        counts[ch] = get(ch, 0) + 1
11
12    freqs = list(counts.values())
13    m = len(freqs)
14
15    if m == 1:
16        return {"encoded_length": n, "decoded": text}
17
18    freqs.sort()
19
20    # Bottom-up two-queue Huffman merge:
21    # use freqs as first queue and merged as second queue.
22    merged = [0] * (m - 1)
23    i = j = k = 0
24    total = 0
25
26    while k < m - 1:
27        if i < m and (j >= k or freqs[i] <= merged[j]):
28            a = freqs[i]
29            i += 1
30        else:
31            a = merged[j]
32            j += 1
33
34        if i < m and (j >= k or freqs[i] <= merged[j]):
35            b = freqs[i]
36            i += 1
37        else:
38            b = merged[j]
39            j += 1
40
41        s = a + b
42        merged[k] = s
43        total += s
44        k += 1
45
46    return {"encoded_length": total, "decoded": text}