BM字符串匹配算法

/**
 * @file string_BM.cpp
 * @author Invisiphatom (ethancao16770@gmail.com)
 * @brief BM algorithm
 * @version 0.1
 * @date 2023-04-12
 * 
 * @copyright Copyright (c) 2023 Invisiphatom
 * All Rights Reserved.
 * 
 */
#include <iostream>
#include <string>
#include <vector>

using namespace std;
const int TABLE_SIZE = 256;     // set string's ANSIC range

int get_suffix_len(const string P, const int end) {
    int suffix_end = P.size() - 1;
    int midstr_end = end;
    int suffix_len = 0;
    while (midstr_end >= 0 && P[midstr_end] == P[suffix_end]) {
        suffix_len++;
        midstr_end--;
        suffix_end--;
    }
    return suffix_len;
}

bool judge_is_prefix(const string P, const int start) {
    int prefix_start = 0;
    int midstr_start = start;
    while (midstr_start < P.size()) {
        if (P[midstr_start] != P[prefix_start]) return false;
        midstr_start++;
        prefix_start++;
    }
    return true;
}

vector<int> build_bad_table(const string P) {
    vector<int> bad_table(TABLE_SIZE, P.size());
    // NOTE {P.size() - 1} -> The last element is excluded
    for (int pi = 0; pi < P.size() - 1; pi++)
        bad_table[P[pi]] = P.size() - (pi + 1);
    return bad_table;
}

vector<int> build_good_table(const string P) {
    vector<int> good_table(P.size(), 0);

    for (int pi = P.size() - 1, match_len = 0; pi != 0; pi--) {
        // reset match_len dynamically according to different pi
        if (judge_is_prefix(P, pi)) match_len = P.size() - (pi + 1);
        // (P.size() + (P.size() - (pi + 1))) -> 过度滑动， match_len -> 缓解过度滑动
        good_table[pi - 1] = (P.size() + (P.size() - (pi + 1))) - match_len;
    }

    // 顺序扫描，将中间字符串与相应后缀之间配对起来
    // NOTE {P.size() - 1} -> The last element is excluded
    for (int pi = 0; pi < P.size() - 1; pi++) {
        int suffix_len = get_suffix_len(P, pi);
        if (suffix_len != 0)
            good_table[P.size() - 1 - suffix_len] = P.size() - (pi - suffix_len + 1);
    }

    return good_table;
}

void BM_matcher(const string T, const string P) {
    if (P.size() > T.size()) return;

    auto bad_table = build_bad_table(P);
    auto good_table = build_good_table(P);

    auto ti = P.size() - 1;
    auto pi = P.size() - 1;
    while (ti < T.size()) {
        while (T[ti] == P[pi]) {
            if (pi == 0) {
                cout << "Match in position:  " << ti << endl;
                break;
            }
            ti--;
            pi--;
        }
        ti += max(bad_table[T[ti]], good_table[pi]);
        pi = P.size() - 1;
    }
}

int main() {
    string T = "HERE IS A SIMPLE EXAMPLE";
    string P = "EXAMPLE";
    BM_matcher(T, P);
    return 0;
}

参考文章：

不用找了，学习BM算法，这篇就够了（思路+详注代码）-CSDN博客

字符串匹配算法(二)：BM(BoyerMoore)算法、坏字符规则，好后缀规则_bm坏字符规则-CSDN博客