Solution #fd0fb3a5-5085-4e55-8c7e-06311ad1ad15

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    if len(counts) == 1:
13        return {"encoded_length": n, "decoded": text}
14
15    try:
16        from bisect import insort
17    except:
18        def insort(a, x):
19            lo, hi = 0, len(a)
20            while lo < hi:
21                mid = (lo + hi) >> 1
22                if a[mid] < x:
23                    lo = mid + 1
24                else:
25                    hi = mid
26            a.insert(lo, x)
27
28    weights = []
29    for w in counts.values():
30        insort(weights, w)
31
32    total = 0
33    while len(weights) > 1:
34        a = weights.pop(0)
35        b = weights.pop(0)
36        s = a + b
37        total += s
38        insort(weights, s)
39
40    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}