|
| 1 | +#!/usr/bin/python |
| 2 | +# -*- coding: UTF-8 -*- |
| 3 | + |
| 4 | +from time import time |
| 5 | + |
| 6 | + |
| 7 | +def bf(main, pattern): |
| 8 | + """ |
| 9 | + 字符串匹配,bf暴搜 |
| 10 | + :param main: 主串 |
| 11 | + :param pattern: 模式串 |
| 12 | + :return: |
| 13 | + """ |
| 14 | + n = len(main) |
| 15 | + m = len(pattern) |
| 16 | + |
| 17 | + if n <= m: |
| 18 | + return 0 if pattern == main else -1 |
| 19 | + |
| 20 | + for i in range(n-m+1): |
| 21 | + for j in range(m): |
| 22 | + if main[i+j] == pattern[j]: |
| 23 | + if j == m-1: |
| 24 | + return i |
| 25 | + else: |
| 26 | + continue |
| 27 | + else: |
| 28 | + break |
| 29 | + return -1 |
| 30 | + |
| 31 | + |
| 32 | +def simple_hash(s, start, end): |
| 33 | + """ |
| 34 | + 计算子串的哈希值 |
| 35 | + 每个字符取acs-ii码后求和 |
| 36 | + :param s: |
| 37 | + :param start: |
| 38 | + :param end: |
| 39 | + :return: |
| 40 | + """ |
| 41 | + assert start <= end |
| 42 | + |
| 43 | + ret = 0 |
| 44 | + for c in s[start: end+1]: |
| 45 | + ret += ord(c) |
| 46 | + return ret |
| 47 | + |
| 48 | + |
| 49 | +def rk(main, pattern): |
| 50 | + n = len(main) |
| 51 | + m = len(pattern) |
| 52 | + |
| 53 | + if n <= m: |
| 54 | + return 0 if pattern == main else -1 |
| 55 | + |
| 56 | + # 子串哈希值表 |
| 57 | + hash_memo = [None] * (n-m+1) |
| 58 | + hash_memo[0] = simple_hash(main, 0, m-1) |
| 59 | + for i in range(1, n-m+1): |
| 60 | + hash_memo[i] = hash_memo[i-1] - simple_hash(main, i-1, i-1) + simple_hash(main, i+m-1, i+m-1) |
| 61 | + |
| 62 | + # 模式串哈希值 |
| 63 | + hash_p = simple_hash(pattern, 0, m-1) |
| 64 | + |
| 65 | + for i, h in enumerate(hash_memo): |
| 66 | + # 可能存在哈希冲突 |
| 67 | + if h == hash_p: |
| 68 | + if pattern == main[i:i+m]: |
| 69 | + return i |
| 70 | + else: |
| 71 | + continue |
| 72 | + return -1 |
| 73 | + |
| 74 | + |
| 75 | +if __name__ == '__main__': |
| 76 | + m_str = 'a'*10000 |
| 77 | + p_str = 'a'*200+'b' |
| 78 | + |
| 79 | + print('--- time consume ---') |
| 80 | + t = time() |
| 81 | + print('[bf] result:', bf(m_str, p_str)) |
| 82 | + print('[bf] time cost: {0:.5}s'.format(time()-t)) |
| 83 | + |
| 84 | + t = time() |
| 85 | + print('[rk] result:', rk(m_str, p_str)) |
| 86 | + print('[rk] time cost: {0:.5}s'.format(time()-t)) |
| 87 | + |
| 88 | + print('') |
| 89 | + print('--- search ---') |
| 90 | + m_str = 'thequickbrownfoxjumpsoverthelazydog' |
| 91 | + p_str = 'jump' |
| 92 | + print('[bf] result:', bf(m_str, p_str)) |
| 93 | + print('[rk] result:', rk(m_str, p_str)) |
0 commit comments