ACM字符串模板

字符串

0x01 KMP与扩展KMP

1. KMP

struct KMP {
    int n;
    string p;
    vector<int> nxt;
    KMP(string &s) : n(s.size() - 1), p(s), nxt(n + 1) {
        for (int i = 2, j = 0; i <= n; i ++ ) {
            while (j && p[i] != p[j + 1]) j = nxt[j];
            if (p[i] == p[j + 1]) j ++ ;
            nxt[i] = j;
        }
    }
    int board(int x) {
        return nxt[x];
    }
    vector<int> match(string &s, bool first_only = false) {
        vector<int> start_pos;
        for (int i = 1, j = 0; i < s.size(); i ++ ) {
            while (j && s[i] != p[j + 1]) j = nxt[j];
            if (s[i] == p[j + 1]) j ++ ;
            if (j == n) {
                start_pos.push_back(i - j + 1);
                if (first_only) return start_pos;
                j = nxt[j];
            }
        }
        return start_pos;
    }
    // 循环周期
    vector<int> periodic() {
        vector<int> ret;
        int now = n;
        while (now) {
            now = nxt[now];
            ret.push_back(n - now);
        }
        return ret;
    }
    // 循环节
    vector<int> periodic_loop() {
        vector<int> ret;
        for (int x : periodic()) {
            if (n % x == 0) {
                ret.push_back(x);
            }
        }
        return ret;
    }
    int min_periodic_loop() {
        return periodic_loop()[0];
    }
};

2. 扩展KMP

// nxt[i] = LCP(T[i,lent],T)
// extend[i] = LCP(S[i,lens],T)
struct EXKMP {
    int n;
    string p;
    vector<int> nxt;
    EXKMP(string &str) : n(str.size() - 1), p(str), nxt(exkmp(p)) {}
    vector<int> exkmp(string &s) {
        vector<int> extend(s.size());
        extend[0] = 0;
        for (int i = 1, st = 0, ed = 0; i < s.size(); i ++ ) {
            extend[i] = i <= ed ? min(nxt[i - st + 1], ed - i + 1) : 0;
            while (i + extend[i] < s.size() && extend[i] < n && s[i + extend[i]] == p[extend[i] + 1]) {
                extend[i] ++ ;
            }
            if (i + extend[i] - 1 >= ed && i != 1) {
                st = i, ed = i + extend[i] - 1;
            }
        }
        return extend;
    }
};

3. 二维循环周期

void KMP2(vector<string> &s) {
    int p = n, q = m;
    vector<int> row(m + 1);
    for (int k = 1; k <= n; k ++ ) {
        vector<int> ne(m + 1);
        for (int i = 2, j = 0; i <= m; i ++ ) {
            while (j && s[k][i] != s[k][j + 1]) j = ne[j];
            if (s[k][i] == s[k][j + 1]) j ++ ;
            ne[i] = j;
        }
        for (int j = ne[m]; j; j = ne[j]) {
            row[m - j] ++ ;
        }
    }
    for (int i = 1; i <= m; i ++ ) {
        if (row[i] == n) {
            q = i;
            break;
        }
    }

    vector<int> col(n + 1);
    for (int k = 1; k <= m; k ++ ) {
        vector<int> ne(n + 1);
        for (int i = 2, j = 0; i <= n; i ++ ) {
            while (j && s[i][k] != s[j + 1][k]) j = ne[j];
            if (s[i][k] == s[j + 1][k]) j ++ ;
            ne[i] = j;
        }
        for (int j = ne[n]; j; j = ne[j]) {
            col[n - j] ++ ;
        }
    }
    for (int i = 1; i <= n; i ++ ) {
        if (col[i] == m) {
            p = i;
            break;
        }
    }

    cout << p * q << "\n";
}

0x02 字符串哈希

1. 一维哈希

using ULL = unsigned long long;

ULL Prime_Pool[] = {1998585857ul,23333333333ul};
ULL Seed_Pool[]={911,146527,19260817,91815541};
ULL Mod_Pool[]={29123,998244353,1000000009,4294967291ull};

constexpr int P1 = 1e9 + 7, P2 = 1e9 + 9;
struct Hashv {
    int h1, h2;
    Hashv(int base1 = 0, int base2 = 0) : h1(base1), h2(base2) {}
    i64 val() { return 1ll * h1 * P2 + h2; }
    Hashv &operator*=(const Hashv &rhs) {
        h1 = (i64)h1 * rhs.h1 % P1;
        h2 = (i64)h2 * rhs.h2 % P2;
        return *this;
    }
    Hashv &operator+=(const Hashv &rhs) {
        h1 += rhs.h1; if (h1 >= P1) h1 -= P1;
        h2 += rhs.h2; if (h2 >= P2) h2 -= P2;
        return *this;
    }
    Hashv &operator-=(const Hashv &rhs) {
        h1 -= rhs.h1; if (h1 < 0) h1 += P1;
        h2 -= rhs.h2; if (h2 < 0) h2 += P2;
        return *this;
    }
    friend Hashv operator*(const Hashv &lhs, const Hashv &rhs) {
        Hashv res = lhs;
        res *= rhs;
        return res;
    }
    friend Hashv operator+(const Hashv &lhs, const Hashv &rhs) {
        Hashv res = lhs;
        res += rhs;
        return res;
    }
    friend Hashv operator-(const Hashv &lhs, const Hashv &rhs) {
        Hashv res = lhs;
        res -= rhs;
        return res;
    }
    friend bool operator==(const Hashv &lhs, const Hashv &rhs) {
        return lhs.h1 == rhs.h1 && lhs.h2 == rhs.h2;
    }
    friend bool operator<(const Hashv &lhs, const Hashv &rhs) {
        return lhs.h1 != rhs.h1 ? lhs.h1 < rhs.h1 : lhs.h2 < rhs.h2;
    }
};

const Hashv base(131, 233);
// using Hashv = unsigned long long;
// const Hashv base = 233;

struct StringHash {
    const int n;
    vector<Hashv> h1, h2, power;
    StringHash(string &s) : n(s.size() - 1), h1(n + 2), h2(n + 2), power(n + 2) {
        for (int i = 1; i <= n; i ++ ) h1[i] = h1[i - 1] * base + s[i];
        for (int i = n; i >= 1; i -- ) h2[i] = h2[i + 1] * base + s[i];
        power[0] = Hashv(1, 1);
        // power[0] = 1;
        for (int i = 1; i <= n; i ++ ) power[i] = power[i - 1] * base;
    }
    Hashv hash1(int l, int r) {
        return h1[r] - h1[l - 1] * power[r - l + 1];
    }
    Hashv hash2(int l, int r) {
        return h2[l] - h2[r + 1] * power[r - l + 1];
    }
};

2. 二维哈希

signed main(signed argc, char const *argv[])
{
    ios::sync_with_stdio(false);
    cin.tie(nullptr);

    int n, m, a, b;
    cin >> n >> m >> a >> b;

    vector<string> s(n + 1);
    for (int i = 1; i <= n; i ++ ) {
        cin >> s[i];
        s[i] = " " + s[i];
    }
    
    vector<Hashv> pow1(max(n, m) + 1);
    vector<Hashv> pow2(max(n, m) + 1);
    pow1[0] = pow2[0] = 1;
    for (int i = 1; i < pow1.size(); i ++ ) {
        pow1[i] = pow1[i - 1] * base1;
        pow2[i] = pow2[i - 1] * base2;
    }

    auto get_hash = [&](vector<vector<Hashv>> &h, vector<string> &s) {
        int n = h.size() - 1, m = h[0].size() - 1;
        for (int i = 1; i <= n; i ++ ) {
            for (int j = 1; j <= m; j ++ ) {
                h[i][j] = h[i][j - 1] * base1 + s[i][j];
            }
        }
        for (int j = 1; j <= m; j ++ ) {
            for (int i = 1; i <= n; i ++ ) {
                h[i][j] = h[i - 1][j] * base2 + h[i][j];
            }
        }
    };

    vector h(n + 1, vector<Hashv>(m + 1));
    get_hash(h, s);
    unordered_set<Hashv> S;
    for (int i = a; i <= n; i ++ ) {
        for (int j = b; j <= m; j ++ ) {
            Hashv hs = h[i][j] - h[i - a][j] * pow2[a] - h[i][j - b] * pow1[b] + h[i - a][j - b] * pow2[a] * pow1[b];
            S.insert(hs);
        }
    }
    int q;
    cin >> q;
    while (q -- ) {
        vector<string> p(a + 1);
        for (int i = 1; i <= a; i ++ ) {
            cin >> p[i];
            p[i] = " " + p[i];
        }
        vector hp(a + 1, vector<Hashv>(b + 1));
        get_hash(hp, p);
        if (S.count(hp[a][b])) {
            cout << "1\n";
        } else {
            cout << "0\n";
        }
    }
    return 0;
}

0x03 Trie树、Border树、AC自动机

Border树的性质：

1. 每个前缀的所有 $Border$：节点 $i$ 到根的链
2. 哪些前缀有长度为 $x$ 的 $Border$：$x$ 的子树
3. 求两个前缀的公共 $Border$ 等价与求 LCA

AC自动机 = Trie树 + Board树
概念上 AC自动机 和 Trie图 是两个不同的概念，但一般 AC自动机 在建立的过程中顺便就把 Trie图 建立好了，因此实际应用中不做区分。

constexpr int N = 1000010;
struct ACAM {
    struct Node {
        int s[26], fail;
    }tr[N];
    int idx, root;
    ACAM() { clear(); }
    void clear() {
        memset(tr[0].s, 0, sizeof tr[0].s);
        tr[0].fail = root = idx = 0;
    }
    int new_node() {
        idx ++ ;
        memset(tr[idx].s, 0, sizeof tr[idx].s);
        tr[idx].fail = 0;
        return idx;
    }
    void insert(string &s, int i) {
        int p = root;
        for (auto c : s) {
            int u = c - 'a';
            if (!tr[p].s[u]) tr[p].s[u] = new_node();
            p = tr[p].s[u];
        }
    }
    void build() {
        queue<int> q;
        for (int i = 0; i < 26; i ++ ) {
            if (tr[0].s[i]) {
                q.push(tr[0].s[i]);
            }
        }
        while (q.size()) {
            int u = q.front();
            q.pop();
            for (int i = 0; i < 26; i ++ ) {
                int v = tr[u].s[i];
                if (v == 0) {
                    tr[u].s[i] = tr[tr[u].fail].s[i];
                } else {
                    tr[v].fail = tr[tr[u].fail].s[i];
                    q.push(v);
                }
            }
        }
    }
}AC;

0x04 Manacher

回文串性质：

1. 不同回文串的个数有 $O(n)$ 个。
2. 回文串 $S$ 的回文前缀（后缀）等价于 Border 。

vector<int> manacher(string &s)
{
    string b = "$|";
    for (auto c : s) {
        b += c;
        b += '|';
    }
 
    vector<int> len(b.size());
    int maxright = 1, mid = 1;
    int ans = 0;
    for (int i = 1; i < b.size(); i ++ ) {
        if (i < maxright) {
            len[i] = min(len[2 * mid - i], maxright - i);
        } else {
            len[i] = 1;
        }
        if (i < maxright && i - len[i] < mid) {
            ans = max(ans, (i - mid) * 2);
        }
        while (b[i - len[i]] == b[i + len[i]]) {
            len[i] ++ ;
        }
        if (i + len[i] > maxright) {
            maxright = i + len[i];
            mid = i;
        }
    }

    return len;
}

0x05 PAM

PAM基础

PAM 基础知识：

1. 节点：至多有 $N$ 个，每个节点表示一种回文串，长度为 $len$。
2. 后继边：S(v) = ch S(u) ch，len[v] = len[u] + 2 。
3. 失配边：S(v) 是 S(u) 的最大 Border，即最长回文子串。

struct PAM {
    static constexpr int ALPHABET_SIZE = 26, N = 1000010;
    struct Node {
        // basic
        int s[ALPHABET_SIZE];
        int fail, len, num; // num记录以该点为结尾的回文串的数量
        // extension
        int cnt; // 节点代表的回文串出现的次数
    }tr[N];
    int idx, last;
    string s;
    PAM() {
        // 1：奇数根  2：偶数根
        tr[0].fail = 1, tr[0].len = 0;
        tr[1].fail = 0, tr[1].len = -1;
        idx = 1, last = 0;
        memset(tr[0].s, 0, sizeof tr[0].s);
        memset(tr[1].s, 0, sizeof tr[1].s);
    }
    int get_fail(int x, int i) {
        while (s[i - tr[x].len - 1] != s[i]) {
            x = tr[x].fail;
        }
        return x;
    }
    int new_node(int p) {
        int q = ++ idx;
        memset(tr[q].s, 0, sizeof tr[q].s);
        tr[q].fail = tr[q].cnt = 0;
        tr[q].len = tr[p].len + 2;
        return q;
    }
    void insert(string &str) {
        s = str;
        for (int i = 1; i < s.size(); i ++ ) {
            int u = s[i] - 'a';
            int p = get_fail(last, i);
            if (!tr[p].s[u]) {
                int q = new_node(p);
                tr[q].fail = tr[get_fail(tr[p].fail, i)].s[u];
                tr[q].num = tr[tr[q].fail].num + 1;
                tr[p].s[u] = q;
            }
            last = tr[p].s[u];
            tr[last].cnt ++ ;
        }
    }
    // 拓扑更新cnt
    void build() {
        for (int i = idx; i >= 2; i -- ) {
            tr[tr[i].fail].cnt += tr[i].cnt;
        }
        tr[0].cnt = tr[1].cnt = 0;
    }
}pam;

广义PAM

0x06 后缀数组

倍增求SA

struct ST {
    int n, m;
    vector<vector<int>> st;
    ST() {}
    ST(vector<int> &w) : n(w.size()), m(31), st(n, vector<int>(m)) {
        for (int j = 0; j < m; j ++ ) {
            for (int i = 0; i + (1 << j) - 1 < n; i ++ ) {
                if (j == 0) {
                    st[i][j] = w[i];
                } else {
                    st[i][j] = max(st[i][j - 1], st[i + (1 << j - 1)][j - 1]);
                }
            }
        }
    }
    int query(int l, int r) {
        int len = r - l + 1;
        int k = log(len) / log(2);
        return max(st[l][k], st[r - (1 << k) + 1][k]);
    }
};
constexpr int maxm = 127;
struct SA {
    int n, m;
    string s;
    vector<int> sa, rk, height;
    vector<int> x, y, c;
    ST st;
    void get_sa() {
        for (int i = 1; i <= n; i ++ ) c[x[i] = s[i]] ++ ;
        for (int i = 2; i <= m; i ++ ) c[i] += c[i - 1];
        for (int i = n; i >= 1; i -- ) sa[c[x[i]] -- ] = i;
        for (int k = 1; k <= n; k *= 2) {
            int num = 0;
            for (int i = n - k + 1; i <= n; i ++ ) y[ ++ num] = i;
            for (int i = 1; i <= n; i ++ ) { // 此处按顺序枚举所有后缀
                if (sa[i] > k) { // 第i个后缀的第二关键字为第i+k个后缀的第一关键字
                    y[ ++ num] = sa[i] - k; // y记录的是按照第二关键字排序的结果
                }
            }
            for (int i = 1; i <= m; i ++ ) c[i] = 0;
            for (int i = 1; i <= n; i ++ ) c[x[i]] ++ ;
            for (int i = 1; i <= m; i ++ ) c[i] += c[i - 1];
            for (int i = n; i; i -- ) sa[c[x[y[i]]] -- ] = y[i], y[i] = 0;
            swap(x, y); // 此处y记录的为x的信息
            x[sa[1]] = 1, num = 1;
            for (int i = 2; i <= n; i ++ ) {
                x[sa[i]] = (y[sa[i]] == y[sa[i - 1]] && y[sa[i] + k] == y[sa[i - 1] + k]) ? num : ++ num;
            }
            if (num == n) break;
            m = num;
        }
    }
    void get_height() { // 表示为排名为i和排名为i-1的后缀的最长公共前缀
        for (int i = 1; i <= n; i ++ ) rk[sa[i]] = i;
        for (int i = 1, k = 0; i <= n; i ++ ) { // 按rank枚举
            if (rk[i] == 1) continue;
            // h[i] = height[rk[i]]，即位与第i个的后缀与排名在它之前的第一个后缀的最长公共前缀
            if (k) k -- ; // 由于h[i]>=h[i-1]-1，因此从k-1开始枚举
            int j = sa[rk[i] - 1];
            while (i + k <= n && j + k <= n && s[i + k] == s[j + k]) k ++ ;
            height[rk[i]] = k;
        }
    }
    SA(string &str) : s(str), n(str.size() - 1), m(maxm), sa(n + m), rk(n + m), height(n + m), x(n + m), y(n + m), c(n + m) {
        get_sa();
        get_height();
        st = ST(height);
    }
    int lcp(int x, int y) { // 得到两个后缀的lcp
        int rkx = rk[x];
        int rky = rk[y];
        if (rkx > rky) swap(rkx, rky);
        rkx ++ ;
        int lcp = st.query(rkx, rky);
        return lcp;
    }
    i64 get_distinct_substrings() { // sum(n - sa[i] + 1 - height[i])
        i64 ans = 0;
        for (int i = 1; i <= n; i ++ ) {
            ans += n - sa[i] + 1 - height[i];
        }
        return ans;
    }
};

SAIS

#include <bits/stdc++.h>
using namespace std;
using i64 = long long;
const int MAXN=1e6+10;
int n,m;
char s[MAXN];
template <size_t size>
struct SuffixArray {
    bool t[size<<1];
    int sa[size],ht[size],rk[size];
    inline bool islms(const int i, const bool *t){
        return i>0 && t[i]&&!t[i-1];
    }
    template<class T>
    void sort(T s,int *sa, const int len, const int sigma, const int sz, bool *t, int *b,int *cb, int *p) {
        memset(b, 0, sigma*sizeof(int));
        memset(sa,-1,len*sizeof(int));
        for (int i=0;i<len;i++)b[s[i]]++;
        cb[0]=b[0];
        for(int i=1;i<sigma;i++)cb[i]=cb[i-1]+b[i];
        for(int i=sz-1;i>=0;--i)sa[--cb[s[p[i]]]]=p[i];
        for(int i=1;i<sigma;i++)cb[i]=cb[i-1]+b[i-1];
        for (int i=0;i<len;i++)if(sa[i]>0&&!t[sa[i]-1])
            sa[cb[s[sa[i]-1]]++]=sa[i]-1;
        cb[0]=b[0];
        for(int i=1;i<sigma;i++)cb[i]=cb[i-1]+b[i];
        for(int i=len-1;i>=0;--i)if(sa[i]>0 && t[sa[i]-1])
            sa[--cb[s[sa[i]-1]]]=sa[i]-1;
    }
    template<class T>
    void sais(T s,int *sa, const int len, bool *t, int *b, int *b1, const int sigma) {
        int i,j,x,p=-1,cnt=0,sz=0,*cb=b+sigma;
        for(t[len-1]=1,i=len-2;i>=0;--i)t[i]=s[i]<s[i+1]||(s[i]==s[i+1]&&t[i+1]);
        for(i=1;i<len;i++)if(t[i]&&!t[i-1])b1[sz++]=i;
        sort(s,sa,len,sigma,sz,t,b,cb,b1);
        for(i=sz=0;i<len;i++)if(islms(sa[i],t))sa[sz++]=sa[i];
        for(i=sz;i<len;i++)sa[i]=-1;
        for(i=0;i<sz;i++){
            for(x=sa[i],j=0;j<len;j++){
                if(p==-1||s[x+j]!=s[p+j]||t[x+j]!=t[p+j]){
                    cnt++,p=x;break;
                }
                else if(j>0&&(islms(x+j,t)||islms(p+j,t)))break;
            }
            sa[sz+(x>>=1)]=cnt-1;
        }
        for(i=j=len-1;i>=sz;i--)if(sa[i]>=0)sa[j--]=sa[i];
        int *s1=sa+len-sz,*b2=b1+sz;
        if(cnt<sz)sais(s1,sa,sz,t+len,b,b1+sz,cnt);
        else for(i=0;i<sz;i++)  sa[s1[i]]=i;
         for(i=0;i<sz;i++)b2[i]=b1[sa[i]];
        sort(s,sa,len,sigma,sz,t,b,cb,b2);
    }
    template<class T>
    void getHeight(T s,int n){
        int j=0,k=0;
        for(int i=0;i<n;ht[rk[i++]]=k)
            for(k?k--:0,j=sa[rk[i]-1];s[i+k]==s[j+k];k++);
        for (int i = 1; i <= n; i ++ ) sa[i] ++ ;
    }
    template<class T>
    void init(T s,const int len,  const int sigma){
        sais(s,sa,len+1,t,rk,ht,sigma),rk[0]=0;
        for(int i=1;i<=len;i++)rk[sa[i]]=i;
    }
};
SuffixArray<MAXN> SA;
signed main()
{
    scanf("%s", s + 1);
    n = strlen(s + 1);
    m = 127;
    SA.init(s + 1, n, m);
    SA.getHeight(s + 1, n);
    for (int i = 1; i <= n; i ++ ) printf("%d ", SA.sa[i]);
    puts("");
    for (int i = 1; i <= n; i ++ ) printf("%d ", SA.ht[i]);
    puts("");
    return 0;
}

0x07 SAM

SAM 基础

Right 集合： 表示 s 状态代表子串的出现位置右端点，性质如下：

1. 所有节点的 Right 集合 互不相等。
2. Right 集合 之间只存在包含关系。

SC树（后缀链接树）： Right 集合 包含关系形成的树，性质如下：

1. 每个前缀所在状态两两不同。
2. 共有 |S| 个节点代表每个前缀，因此最多有 |S| 个节点，加之 Right 集合的性质，因此 SAM 的节点个数为 $O(2n)$ 。
3. 任意串 w 的后缀全部位于 s(w) 的后缀链接树上。
4. 每个状态的 Right 集合 等价于他在后缀链接树子树的叶子节点集合。

constexpr int ALPHABET_SIZE = 26, N = 1000010;
#ifdef RIGHT
struct Chairman_Tree {
    struct Node { int l, r, val; }tr[N * 30];
    int root[N * 2];
    int idx = 0;
    void init() {
        memset(root, 0, sizeof root);
        idx = 0;
    }
    int build(int l, int r) {
        int q = ++ idx;
        tr[q].val = 0;
        if (l == r) return q;
        int mid = l + r >> 1;
        tr[q].l = build(l, mid), tr[q].r = build(mid + 1, r);
        return q;
    }
    int update(int p, int l, int r, int x, int v) {
        assert(idx < N * 50);
        int q = ++ idx;
        tr[q].val = tr[p].val + v;
        if (l == r) return q;
        int mid = l + r >> 1;
        if (x <= mid) {
            tr[q].l = update(tr[p].l, l, mid, x, v);
            tr[q].r = tr[p].r;
        } else {
            tr[q].l = tr[p].l;
            tr[q].r = update(tr[p].r, mid + 1, r, x, v);
        }
        return q;
    }
    int query(int p, int q, int l, int r, int L, int R) {
        if (l > R || r < L) return 0;
        if (L <= l && R >= r) return tr[q].val - tr[p].val;
        int mid = l + r >> 1;
        return query(tr[p].l, tr[q].l, l, mid, L, R) + query(tr[p].r, tr[q].r, mid + 1, r, L, R);
    }
}tree;
#endif
int ans[N]; // ans表示长度为len的串的最大数量
struct SuffixAutomaton {
    // basic
    int n;
    string s;
    struct Node {
        int s[ALPHABET_SIZE];
        int fa, len;
    }tr[2 * N];
    int last, idx;
    // extension
    int cntA[N * 2], A[N * 2];
    int num[N * 2]; // 记录right集合的大小
#ifdef RIGHT
    vector<int> g[N * 2];
    int L[N * 2], R[N * 2], dfn;
    int pos[N * 2]; // 记录dfs序映射的点
    int end_pos[N * 2];
#endif
    SuffixAutomaton(){ clear(); }
    void clear() {
        last = idx = 1;
        tr[1].fa = tr[1].len = 0;
        fill(tr[1].s, tr[1].s + 26, 0);
    }
    void init(string &str) {
        n = str.size();
        s = str;
        for (auto c : s) extend(c - 'a');
    }
    void extend(int c) {
        int p = last;
        int np = last = ++ idx;
        tr[np].len = tr[p].len + 1;
        memset(tr[np].s, 0, sizeof tr[np].s);
        while (p && !tr[p].s[c]) tr[p].s[c] = np, p = tr[p].fa;
        if (!p) {
            tr[np].fa = 1;
        } else {
            int q = tr[p].s[c];
            if (tr[q].len == tr[p].len + 1) {
                tr[np].fa = q;
            } else {
                int nq = ++ idx;
                tr[nq] = tr[q];
                tr[nq].len = tr[p].len + 1;
                tr[q].fa = tr[np].fa = nq;
                while (tr[p].s[c] == q) tr[p].s[c] = nq, p = tr[p].fa;
            }
        }
    }
    void build() {
        memset(cntA, 0, sizeof cntA);
        memset(num, 0, sizeof num);
        for (int i = 1; i <= idx; i ++ ) cntA[tr[i].len] ++ ;
        for (int i = 1; i <= idx; i ++ ) cntA[i] += cntA[i - 1];
        for (int i = idx; i >= 1; i -- ) A[cntA[tr[i].len] -- ] = i;
        // 更行主串节点
        int temp = 1;
        for (int i = 0; i < n; i ++ ){
            num[temp = tr[temp].s[s[i] - 'a']] = 1;
        }
        /*拓扑更新*/
        for (int i = idx; i >= 1; i -- ){
            //basic
            int x = A[i];
            num[tr[x].fa] += num[x];
            //special
            ans[tr[x].len] = max(ans[tr[x].len], num[x]);
        }
        //special
        for (int i = tr[last].len; i > 1; i -- ){
            ans[i - 1] = max(ans[i - 1], ans[i]);
        }
    }
#ifdef RIGHT
    int get_right_between(int u, int l, int r) {
        return tree.query(tree.root[L[u] - 1], tree.root[R[u]], 1, n, l, r);
    }
    void dfs(int u, int father) {
        L[u] = ++ dfn;
        pos[dfn] = u;
        for (auto v : g[u]) {
            dfs(v, u);
            num[u] += num[v];
        }
        R[u] = dfn;
    }
    void build_sc_tree() {
        int temp = 1;
        for (int i = 0; i < n; i ++ ) {
            temp = tr[temp].s[s[i] - 'a'];
            end_pos[temp] = i + 1;
            num[temp] = 1;
        }
        for (int i = 2; i <= idx; i ++ ) {
            g[tr[i].fa].push_back(i);
        }
    }
    void build_chairman_tree() {
        tree.root[0] = tree.build(1, n);
        for (int i = 1; i <= idx; i ++ ) {
            int u = pos[i];
            if (end_pos[u]) {
                int x = end_pos[u];
                tree.root[i] = tree.update(tree.root[i - 1], 1, n, x, 1);
            } else {
                tree.root[i] = tree.root[i - 1];
            }
        }
    }
    void extract_right() {
        build_sc_tree();
        dfn = 0;
        dfs(1, 0);
        build_chairman_tree();
    }
#endif
}sam;

更新更加灵活的线段树合并模板

constexpr int ALPHABET_SIZE = 26, N = 100010;
#ifdef RIGHT
struct SegmentTree {
    struct Node { 
        int l, r, val;
        void clear() { 
            l = r = val = 0; 
        }
    }tr[N * 50];
    int root[N * 2], idx = 0;
    void init() {
        memset(root, 0, sizeof root);
        tr[0].clear();
        idx = 0;
    }
    void pushup(int u) {
        tr[u].val = tr[tr[u].l].val + tr[tr[u].r].val;
    }
    int modify(int u, int l, int r, int x) {
        if (!u) {
            u = ++ idx;
            tr[u].clear();
        }
        if (l == r) {
            tr[u].val = 1;
            tr[u].maxv = tr[u].minv = x;
            return u;
        }
        int mid = l + r >> 1;
        if (x <= mid) tr[u].l = modify(tr[u].l, l, mid, x);
        else tr[u].r = modify(tr[u].r, mid + 1, r, x);
        pushup(u);
        return u;
    }
    int merge(int x, int y, int l, int r) {
        if (!x || !y) return x + y;
        int u = ++ idx;
        tr[u].clear();
        if (l == r) return u;
        int mid = l + r >> 1;
        tr[u].l = merge(tr[x].l, tr[y].l, l, mid);
        tr[u].r = merge(tr[x].r, tr[y].r, mid + 1, r);
        pushup(u);
        return u;
    }
    int query(int u, int l, int r, int L, int R) {
        if (R < l || L > r) return 0;
        if (l >= L && r <= R) return tr[u].val;
        int mid = l + r >> 1;
        return query(tr[u].l, l, mid, L, R) + query(tr[u].r, mid + 1, r, L, R);
    }
}tree;
#endif
int ans[N]; // ans表示长度为len的串的最大数量
int n;
string s;
struct SuffixAutomaton {
    // basic
    int n;
    struct Node {
        int s[ALPHABET_SIZE];
        int fa, len;
    }tr[2 * N];
    int last, idx;
    // extension
    int cntA[N * 2], A[N * 2];
    int num[N * 2]; // 记录right集合的大小
#ifdef RIGHT
    vector<int> g[N * 2];
    int end_pos[N * 2];
#endif
    SuffixAutomaton(){ clear(); }
    void clear() {
        for (int i = 1; i <= idx; i ++ ) {
            g[i].clear();
            end_pos[i] = 0;
        }
        last = idx = 1;
        tr[1].fa = tr[1].len = 0;
        fill(tr[1].s, tr[1].s + 26, 0);
    }
    void init(string &s) {
        n = s.size();
        for (auto c : s) extend(c - 'a');
    }
    void extend(int c) {
        int p = last;
        int np = last = ++ idx;
        tr[np].len = tr[p].len + 1;
        memset(tr[np].s, 0, sizeof tr[np].s);
        while (p && !tr[p].s[c]) tr[p].s[c] = np, p = tr[p].fa;
        if (!p) {
            tr[np].fa = 1;
        } else {
            int q = tr[p].s[c];
            if (tr[q].len == tr[p].len + 1) {
                tr[np].fa = q;
            } else {
                int nq = ++ idx;
                tr[nq] = tr[q];
                tr[nq].len = tr[p].len + 1;
                tr[q].fa = tr[np].fa = nq;
                while (tr[p].s[c] == q) tr[p].s[c] = nq, p = tr[p].fa;
            }
        }
    }
    void build() {
        memset(cntA, 0, sizeof cntA);
        memset(num, 0, sizeof num);
        for (int i = 1; i <= idx; i ++ ) cntA[tr[i].len] ++ ;
        for (int i = 1; i <= idx; i ++ ) cntA[i] += cntA[i - 1];
        for (int i = idx; i >= 1; i -- ) A[cntA[tr[i].len] -- ] = i;
        // 更行主串节点
        int temp = 1;
        for (int i = 0; i < n; i ++ ){
            num[temp = tr[temp].s[s[i] - 'a']] = 1;
        }
        /*拓扑更新*/
        for (int i = idx; i >= 1; i -- ){
            //basic
            int x = A[i];
            num[tr[x].fa] += num[x];
            //special
            ans[tr[x].len] = max(ans[tr[x].len], num[x]);
        }
        //special
        for (int i = tr[last].len; i > 1; i -- ){
            ans[i - 1] = max(ans[i - 1], ans[i]);
        }
    }
#ifdef RIGHT
    int get_right_between(int u, int l, int r) {
        return tree.query(tree.root[u], 1, n, l, r);
    }
    void dfs(int u, int father) {
        for (auto v : g[u]) {
            dfs(v, u);
            num[u] += num[v];
            tree.root[u] = tree.merge(tree.root[u], tree.root[v], 1, n);
        }
    }
    void build_sc_tree() {
        int temp = 1;
        for (int i = 0; i < n; i ++ ) {
            temp = tr[temp].s[s[i] - 'a'];
            end_pos[temp] = i + 1;
            num[temp] = 1;
        }
        for (int i = 2; i <= idx; i ++ ) {
            g[tr[i].fa].push_back(i);
        }
    }
    void extract_right() {
        build_sc_tree();
        tree.init();
        for (int u = 1; u <= idx; u ++ ) {
            if (end_pos[u]) {
                tree.root[u] = tree.modify(tree.root[u], 1, n, end_pos[u]);
            }
        }
        dfs(1, 0);
    }
#endif
}sam;

关于SAM的匹配问题

#ifdef MATCH
    int len[N * 2];
    int blen[N * 2];
    int get_longest_common_substr() {
        return *max_element(len, len + idx + 1);
    }
    i64 num_of_different_common_substr;
    void get_num_of_different_common_substr(int u) {
        for (auto v : g[u]) {
            get_num_of_different_common_substr(v);
            len[u] = max(len[u], len[v]);
        }
        int lenth = min(len[u], tr[u].len);
        num_of_different_common_substr += max(0, lenth - tr[tr[u].fa].len);
    }
    i64 get_num_of_different_common_substr() {
        num_of_different_common_substr = 0;
        get_num_of_different_common_substr(1);
        return num_of_different_common_substr;
    }
    void  match(vector<string> &S) {
        fill(len, len + idx + 1, 2e9);
        for (auto str : S) {
            fill(blen, blen + idx + 1, 0);
            int temp = 1, lenth = 0;
            for (int i = 0; i < str.size(); i ++ ) {
                char c = str[i];
                while (temp && !tr[temp].s[c - 'a']) {
                    temp = tr[temp].fa;
                    lenth = tr[temp].len;
                }
                if (!temp) {
                    lenth = 0;
                    temp = 1;
                } else {
                    lenth ++ ;
                    temp = tr[temp].s[c - 'a'];
                }
                blen[temp] = max(blen[temp], lenth);
            }
            for (int i = 0; i <= idx; i ++ ) {
                len[i] = min(len[i], blen[i]);
            }
        }
    }
#endif

广义SAM（对字典构造SAM）

constexpr int ALPHABET_SIZE = 26, N = 1000010;
struct Suffix_Automaton {
    struct Node {
        int s[ALPHABET_SIZE];
        int fa, len;
    }tr[N * 2];
    int last, idx;
    Suffix_Automaton() { clear(); };
    void clear() {
        last = idx = 1;
        tr[1].fa = tr[1].len = 0;
        memset(tr[1].s, 0, sizeof tr[1].s);
    }
    int extend(int pre, int c) {
        last = pre;
        int p = last;
        int np = last = ++ idx;
        tr[np].len = tr[p].len + 1; 
        memset(tr[np].s, 0, sizeof tr[np].s);
        while (p && !tr[p].s[c]) tr[p].s[c] = np, p = tr[p].fa;
        if (!p) {
            tr[np].fa = 1;
        } else {
            int q = tr[p].s[c];
            if (tr[q].len == tr[p].len + 1) {
                tr[np].fa = q;
            } else {
                int nq = ++ idx;
                tr[nq] = tr[q];
                tr[nq].len = tr[p].len + 1;
                tr[q].fa = tr[np].fa = nq;
                while (tr[p].s[c] == q) tr[p].s[c] = nq, p = tr[p].fa;
            }
        }
        return np;
    }
}sam;
struct Trie_tree {
    struct Node {
        int s[ALPHABET_SIZE];
    }tr[N];
    int root = 1, idx = 1;
    int sam_pos[N];
    void insert(string &s) {
        int p = root;
        for (auto c : s) {
            int u = c - 'a';
            if (!tr[p].s[u]) tr[p].s[u] = ++ idx;
            p = tr[p].s[u];
        }
    }
    void bfs() {
        queue<int> q;
        q.push(1);
        sam_pos[1] = 1;
        while (q.size()) {
            int u = q.front();
            q.pop();
            for (int i = 0; i < 26; i ++ ) {
                if (!tr[u].s[i]) continue;
                int v = tr[u].s[i];
                sam_pos[v] = sam.extend(sam_pos[u], i);
                q.push(v);
            }
        }
    }
}trie;