#ifndef HIDDEN_IN_VS // 折りたたみ用

// 警告の抑制
#define _CRT_SECURE_NO_WARNINGS

// ライブラリの読み込み
#include <bits/stdc++.h>
using namespace std;

// 型名の短縮
using ll = long long; // -2^63 ～ 2^63 = 9 * 10^18（int は -2^31 ～ 2^31 = 2 * 10^9）
using pii = pair<int, int>;	using pll = pair<ll, ll>;	using pil = pair<int, ll>;	using pli = pair<ll, int>;
using vi = vector<int>;		using vvi = vector<vi>;		using vvvi = vector<vvi>;	using vvvvi = vector<vvvi>;
using vl = vector<ll>;		using vvl = vector<vl>;		using vvvl = vector<vvl>;	using vvvvl = vector<vvvl>;
using vb = vector<bool>;	using vvb = vector<vb>;		using vvvb = vector<vvb>;
using vc = vector<char>;	using vvc = vector<vc>;		using vvvc = vector<vvc>;
using vd = vector<double>;	using vvd = vector<vd>;		using vvvd = vector<vvd>;
template <class T> using priority_queue_rev = priority_queue<T, vector<T>, greater<T>>;
using Graph = vvi;

// 定数の定義
const double PI = acos(-1);
const vi DX = { 1, 0, -1, 0 }; // 4 近傍（下，右，上，左）
const vi DY = { 0, 1, 0, -1 };
int INF = 1001001001; ll INFL = 4004004003104004004LL; // (int)INFL = 1010931620;
double EPS = 1e-15;

// 入出力高速化
struct fast_io { fast_io() { cin.tie(nullptr); ios::sync_with_stdio(false); cout << fixed << setprecision(18); } } fastIOtmp;

// 汎用マクロの定義
#define all(a) (a).begin(), (a).end()
#define sz(x) ((int)(x).size())
#define lbpos(a, x) (int)distance((a).begin(), std::lower_bound(all(a), x))
#define ubpos(a, x) (int)distance((a).begin(), std::upper_bound(all(a), x))
#define Yes(b) {cout << ((b) ? "Yes\n" : "No\n");}
#define rep(i, n) for(int i = 0, i##_len = int(n); i < i##_len; ++i) // 0 から n-1 まで昇順
#define repi(i, s, t) for(int i = int(s), i##_end = int(t); i <= i##_end; ++i) // s から t まで昇順
#define repir(i, s, t) for(int i = int(s), i##_end = int(t); i >= i##_end; --i) // s から t まで降順
#define repe(v, a) for(const auto& v : (a)) // a の全要素（変更不可能）
#define repea(v, a) for(auto& v : (a)) // a の全要素（変更可能）
#define repb(set, d) for(int set = 0; set < (1 << int(d)); ++set) // d ビット全探索（昇順）
#define repp(a) sort(all(a)); for(bool a##_perm = true; a##_perm; a##_perm = next_permutation(all(a))) // a の順列全て（昇順）
#define smod(n, m) ((((n) % (m)) + (m)) % (m)) // 非負mod
#define uniq(a) {sort(all(a)); (a).erase(unique(all(a)), (a).end());} // 重複除去
#define EXIT(a) {cout << (a) << endl; exit(0);} // 強制終了
#define inQ(x, y, u, l, d, r) ((u) <= (x) && (l) <= (y) && (x) < (d) && (y) < (r)) // 半開矩形内判定

// 汎用関数の定義
template <class T> inline ll pow(T n, int k) { ll v = 1; rep(i, k) v *= n; return v; }
template <class T> inline bool chmax(T& M, const T& x) { if (M < x) { M = x; return true; } return false; } // 最大値を更新（更新されたら true を返す）
template <class T> inline bool chmin(T& m, const T& x) { if (m > x) { m = x; return true; } return false; } // 最小値を更新（更新されたら true を返す）
template <class T> inline T get(T set, int i) { return (set >> i) & T(1); }

// 演算子オーバーロード
template <class T, class U> inline istream& operator>>(istream& is, pair<T, U>& p) { is >> p.first >> p.second; return is; }
template <class T> inline istream& operator>>(istream& is, vector<T>& v) { repea(x, v) is >> x; return is; }
template <class T> inline vector<T>& operator--(vector<T>& v) { repea(x, v) --x; return v; }
template <class T> inline vector<T>& operator++(vector<T>& v) { repea(x, v) ++x; return v; }

#endif // 折りたたみ用


#if __has_include(<atcoder/all>)
#include <atcoder/all>
using namespace atcoder;

#ifdef _MSC_VER
#include "localACL.hpp"
#endif

//using mint = modint1000000007;
using mint = modint998244353;
//using mint = modint; // mint::set_mod(m);

namespace atcoder {
	inline istream& operator>>(istream& is, mint& x) { ll x_; is >> x_; x = x_; return is; }
	inline ostream& operator<<(ostream& os, const mint& x) { os << x.val(); return os; }
}
using vm = vector<mint>; using vvm = vector<vm>; using vvvm = vector<vvm>; using vvvvm = vector<vvvm>;
#endif


#ifdef _MSC_VER // 手元環境（Visual Studio）
#include "local.hpp"
#else // 提出用（gcc）
inline int popcount(int n) { return __builtin_popcount(n); }
inline int popcount(ll n) { return __builtin_popcountll(n); }
inline int lsb(int n) { return n != 0 ? __builtin_ctz(n) : -1; }
inline int lsb(ll n) { return n != 0 ? __builtin_ctzll(n) : -1; }
inline int msb(int n) { return n != 0 ? (31 - __builtin_clz(n)) : -1; }
inline int msb(ll n) { return n != 0 ? (63 - __builtin_clzll(n)) : -1; }
#define gcd __gcd
#define dump(...)
#define dumpel(v)
#define dump_list(v)
#define dump_mat(v)
#define input_from_file(f)
#define output_to_file(f)
#define Assert(b) { if (!(b)) while (1) cout << "OLE"; }
#endif


// https://judge.yosupo.jp/submission/72030 から窃盗

// DSU_rollback
struct Data {
    int time, u, par;  // before `time`, `par` = par[u]
};

struct DSU {
    vector<int> par;
    vector<Data> change;

    DSU(int n) : par(n + 5, -1) {}

    // find root of x.
    // if par[x] < 0 then x is a root, and its tree has -par[x] nodes
    int getRoot(int x) {
        while (par[x] >= 0)
            x = par[x];
        return x;
    }

    // join components containing x and y.
    // t should be current time. We use it to update `change`.
    bool join(int x, int y, int t) {
        x = getRoot(x);
        y = getRoot(y);
        if (x == y) return false;

        //union by rank
        if (par[x] < par[y]) swap(x, y);
        //now x's tree has less nodes than y's tree
        change.push_back({ t, y, par[y] });
        par[y] += par[x];
        change.push_back({ t, x, par[x] });
        par[x] = y;
        return true;
    }

    // rollback all changes at time > t.
    void rollback(int t) {
        while (!change.empty() && change.back().time > t) {
            par[change.back().u] = change.back().par;
            change.pop_back();
        }
    }
};

// Directed MST
// Copied from chemthan
// Index from 0
using ll = long long;
struct Edge {
    int u, v;  // directed, u -> v
    ll cost;
};
struct HeapNode {  // lazy skew heap node
    Edge key;
    HeapNode* l, * r;
    ll delta;

    void prop() {
        key.cost += delta;
        if (l) l->delta += delta;
        if (r) r->delta += delta;
        delta = 0;
    }
    Edge top() {
        prop();
        return key;
    }
};
HeapNode* merge(HeapNode* a, HeapNode* b) {
    if (!a) return b;
    if (!b) return a;
    a->prop(); b->prop();
    if (a->key.cost > b->key.cost) swap(a, b);
    swap(a->l, (a->r = merge(b, a->r)));
    return a;
}
void pop(HeapNode*& a) {
    a->prop();
    a = merge(a->l, a->r);
}

// return {cost, parent[i]}
// Not found -> return {-1, {}}
pair<ll, vector<int>> directed_mst(int n, int root, vector<Edge>& edges) {
    DSU dsu(n);
    int dsu_time = 0;
    vector<HeapNode*> heap(n);
    for (const Edge& e : edges) {
        heap[e.v] = merge(heap[e.v], new HeapNode{ e });
    }

    ll res = 0;
    vector<int> seen(n, -1), path(n);
    seen[root] = root;
    vector<Edge> Q(n), in(n, { -1, -1 });
    deque<tuple<int, int, vector<Edge>>> cycs;
    for (int s = 0; s < n; ++s) {
        int u = s, qi = 0, w;
        while (seen[u] < 0) {
            if (!heap[u]) return { -1, {} };
            Edge e = heap[u]->top();
            heap[u]->delta -= e.cost;
            pop(heap[u]);
            Q[qi] = e;
            path[qi++] = u;
            seen[u] = s;
            res += e.cost;
            u = dsu.getRoot(e.u);

            if (seen[u] == s) {
                HeapNode* cyc = 0;
                int end = qi;
                int time = dsu_time;
                do {
                    cyc = merge(cyc, heap[w = path[--qi]]);
                } while (dsu.join(u, w, ++dsu_time));

                u = dsu.getRoot(u);
                heap[u] = cyc;
                seen[u] = -1;
                cycs.push_front({ u, time, {&Q[qi], &Q[end]} });
            }
        }
        for (int i = 0; i < qi; i++) in[dsu.getRoot(Q[i].v)] = Q[i];
    }

    for (auto& [u, t, comp] : cycs) {
        dsu.rollback(t);
        Edge inEdge = in[u];
        for (auto& e : comp) in[dsu.getRoot(e.v)] = e;
        in[dsu.getRoot(inEdge.v)] = inEdge;
    }

    vector<int> par(n);
    for (int i = 0; i < n; i++) par[i] = in[i].u;
    return { res, par };
}


int main() {
//	input_from_file("input.txt");
//	output_to_file("output.txt");

    int n, m;
    cin >> n >> m;

    vvl cost(n, vl(n, INF));
    rep(j, m) {
        int u, v; ll sc;
        cin >> u >> v >> sc;
       
        cost[u][v] = INF - sc;
    }

    vector<Edge> edges;
    rep(s, n) rep(t, n) if (s != t) edges.push_back({ s, t, cost[s][t] });

    ll cost_min = INFL; vi par_min; int root_min = -1;
    rep(root, n) {
        auto [total, par] = directed_mst(n, root, edges);
        if (chmin(cost_min, total)) {
            par_min = par;
            root_min = root;
        }
    }

    Graph g(n);
    rep(s, n) if (s != root_min) g[par_min[s]].push_back(s);

    cout << (ll)INF * (n - 1) - cost_min << endl;
    function<void(int)> dfs = [&](int s) {
        cout << s << endl;
        repe(t, g[s]) dfs(t);
    };
    dfs(root_min);
}