|
1 |
| -from collections import defaultdict |
2 |
| -from algorithm.base import Algorithm |
3 |
| - |
4 |
| -class BoyerMooreHorspool(Algorithm): |
5 |
| - |
6 |
| - def __init__(self, reference): |
7 |
| - spec_characters = '\tEmCGH7' |
8 |
| - reference = reference.translate(reference.maketrans('', '', spec_characters)) |
9 |
| - self.reference = reference |
10 |
| - |
11 |
| - @property |
12 |
| - def name(self): |
13 |
| - return 'Boyer-Moore-Horspool' |
14 |
| - |
15 |
| - def set_candidate(self, candidate, **params): |
16 |
| - self.candidate = candidate |
17 |
| - |
18 |
| - def set_skip_table(self): |
19 |
| - len_reference = len(self.reference) |
20 |
| - len_candidate = len(self.candidate) |
21 |
| - if len_candidate > len_reference: |
22 |
| - print("Error: len candidate > len references {} > {}.".format(len_candidate, len_reference)) |
23 |
| - return -1 |
24 |
| - |
25 |
| - self.table_skip = defaultdict(lambda: len_candidate) |
26 |
| - |
27 |
| - for offset in range(len_candidate - 1): |
28 |
| - self.table_skip[ord(self.candidate[offset])] = len_candidate - offset - 1 |
29 |
| - |
30 |
| - def search(self, multiple_search=False) -> list: |
31 |
| - self.set_skip_table() |
32 |
| - |
33 |
| - len_reference = len(self.reference) |
34 |
| - len_candidate = len(self.candidate) |
35 |
| - |
36 |
| - offset_lst = [] |
37 |
| - offset = len_candidate - 1 |
38 |
| - |
39 |
| - while offset < len_reference: |
40 |
| - j = len_candidate - 1 |
41 |
| - i = offset |
42 |
| - while j >= 0 and self.reference[i] == self.candidate[j]: |
43 |
| - j -= 1 |
44 |
| - i -= 1 |
45 |
| - if j == -1: |
46 |
| - offset_lst.append(i + 1) |
47 |
| - offset += self.table_skip[ord(self.reference[offset])] |
48 |
| - |
49 |
| - return offset_lst |
| 1 | +from collections import defaultdict |
| 2 | +from algorithm.base import Algorithm |
| 3 | +from string import ascii_letters |
| 4 | + |
| 5 | + |
| 6 | +class BoyerMooreHorspool(Algorithm): |
| 7 | + |
| 8 | + def __init__(self, reference): |
| 9 | + # reference = reference.translate(reference.maketrans('', '', ascii_letters)) |
| 10 | + self.reference = reference |
| 11 | + |
| 12 | + @property |
| 13 | + def name(self): |
| 14 | + return 'Boyer-Moore-Horspool' |
| 15 | + |
| 16 | + def set_candidate(self, candidate, **params): |
| 17 | + self.candidate = candidate |
| 18 | + |
| 19 | + def set_skip_table(self): |
| 20 | + len_reference = len(self.reference) |
| 21 | + len_candidate = len(self.candidate) |
| 22 | + if len_candidate > len_reference: |
| 23 | + print("Error: len candidate > len references {} > {}.".format(len_candidate, len_reference)) |
| 24 | + return -1 |
| 25 | + |
| 26 | + self.table_skip = defaultdict(lambda: len_candidate) |
| 27 | + |
| 28 | + for offset in range(len_candidate - 1): |
| 29 | + self.table_skip[ord(self.candidate[offset])] = len_candidate - offset - 1 |
| 30 | + |
| 31 | + def search(self, multiple_search=False) -> list: |
| 32 | + self.set_skip_table() |
| 33 | + |
| 34 | + len_reference = len(self.reference) |
| 35 | + len_candidate = len(self.candidate) |
| 36 | + |
| 37 | + offset_lst = [] |
| 38 | + offset = len_candidate - 1 |
| 39 | + |
| 40 | + while offset < len_reference: |
| 41 | + j = len_candidate - 1 |
| 42 | + i = offset |
| 43 | + while j >= 0 and self.reference[i] == self.candidate[j]: |
| 44 | + j -= 1 |
| 45 | + i -= 1 |
| 46 | + if j == -1: |
| 47 | + offset_lst.append(i + 1) |
| 48 | + offset += self.table_skip[ord(self.reference[offset])] |
| 49 | + |
| 50 | + return offset_lst |
0 commit comments