cp-library
競技プログラミング用 C++ ライブラリ
test/aho_corasick.test.cpp
test/aho_corasick.test.cpp · GitHub で見る
依存ライブラリ
コード
#define PROBLEM "https://judge.yosupo.jp/problem/aho_corasick"
#include "string/aho_corasick.hpp"
#include "components/fast_io.hpp"
#include <string>
#include <vector>
int main() {
FastInput input;
FastOutput output;
int n;
input.read(n);
std::vector<std::string> patterns(n);
int length = 0;
for (auto &s : patterns) input.read(s), length += static_cast<int>(s.size());
AhoCorasick<> ac;
ac.reserve(length, n);
for (const auto &s : patterns) ac.add(s);
ac.build();
output.println(ac.node_count());
for (int v = 1; v < ac.node_count(); ++v) {
output.write(ac.parent(v)), output.write(' '), output.println(ac.suffix_link(v));
}
for (int i = 0; i < n; ++i) {
if (i) output.write(' ');
output.write(ac.terminal(i));
}
output.write('\n');
}
#line 1 "test/aho_corasick.test.cpp"
#define PROBLEM "https://judge.yosupo.jp/problem/aho_corasick"
#line 1 "string/aho_corasick.hpp"
#include <array>
#include <bit>
#include <cassert>
#include <cstddef>
#include <cstdint>
#include <limits>
#include <string_view>
#include <vector>
#if defined(__x86_64__) && (defined(__GNUC__) || defined(__clang__))
#include <immintrin.h>
namespace aho_corasick_detail {
template<int Alphabet>
__attribute__((target("avx2"))) inline std::uint32_t complete(int *edges, const int *fallback) {
std::uint32_t mask = 0;
int i = 0;
for (; i + 8 <= Alphabet; i += 8) {
auto children = _mm256_loadu_si256(reinterpret_cast<const __m256i *>(edges + i));
auto inherited = _mm256_loadu_si256(reinterpret_cast<const __m256i *>(fallback + i));
auto missing = _mm256_cmpeq_epi32(children, _mm256_setzero_si256());
mask |= (std::uint32_t(~_mm256_movemask_ps(_mm256_castsi256_ps(missing))) & 255) << i;
_mm256_storeu_si256(reinterpret_cast<__m256i *>(edges + i), _mm256_blendv_epi8(children, inherited, missing));
}
for (; i < Alphabet; ++i) {
if (edges[i]) mask |= std::uint32_t(1) << i;
else edges[i] = fallback[i];
}
return mask;
}
}
#endif
template <int Alphabet = 26, int First = 'a'>
class AhoCorasick {
static_assert(0 < Alphabet && 0 <= First && First + Alphabet <= 256);
struct alignas(Alphabet >= 16 ? 64 : alignof(int)) Node {
std::array<int, Alphabet> next{};
int matches = 0;
int parent = 0, depth = 0, link = 0, output = -1, head = -1;
};
std::vector<Node> nodes_;
std::vector<int> terminal_, next_pattern_, order_;
bool built_ = false;
static unsigned symbol_(char c) { return static_cast<unsigned>(static_cast<unsigned char>(c) - First); }
void check_([[maybe_unused]] int v) const { assert(0 <= v && v < node_count()); }
void link_(int u, int f) {
nodes_[u].link = f;
nodes_[u].output = nodes_[f].head != -1 ? f : nodes_[f].output;
nodes_[u].matches += nodes_[f].matches;
order_.push_back(u);
}
public:
static constexpr int root = 0;
AhoCorasick() { nodes_.emplace_back(); }
void clear() {
nodes_.clear(), nodes_.emplace_back();
terminal_.clear(), next_pattern_.clear(), order_.clear();
built_ = false;
}
void reserve(int total_length, int patterns = 0) {
assert(0 <= total_length && total_length < std::numeric_limits<int>::max() && 0 <= patterns);
nodes_.reserve(std::size_t(total_length) + 1);
terminal_.reserve(patterns), next_pattern_.reserve(patterns);
}
int add(std::string_view pattern) {
assert(!built_ && size() < std::numeric_limits<int>::max());
int v = root;
for (char c : pattern) {
unsigned x = symbol_(c);
assert(x < Alphabet);
int u = nodes_[v].next[x];
if (!u) {
assert(node_count() < std::numeric_limits<int>::max());
u = node_count();
nodes_[v].next[x] = u;
nodes_.emplace_back();
nodes_[u].parent = v, nodes_[u].depth = nodes_[v].depth + 1;
}
v = u;
}
int id = size();
terminal_.push_back(v), next_pattern_.push_back(nodes_[v].head);
nodes_[v].head = id, ++nodes_[v].matches;
return id;
}
void build() {
if (built_) return;
order_.reserve(nodes_.size()), order_.push_back(root);
for (std::size_t i = 0; i < order_.size(); ++i) {
int v = order_[i];
auto &node = nodes_[v];
const auto &fallback = nodes_[node.link].next;
#if defined(__x86_64__) && (defined(__GNUC__) || defined(__clang__))
if constexpr (8 <= Alphabet && Alphabet <= 32) {
if (__builtin_cpu_supports("avx2")) {
auto mask = aho_corasick_detail::complete<Alphabet>(node.next.data(), fallback.data());
while (mask) {
int c = std::countr_zero(mask), u = node.next[c], f = v == root ? root : fallback[c];
mask &= mask - 1;
link_(u, f);
}
continue;
}
}
#endif
for (int c = 0; c < Alphabet; ++c) {
if (int u = node.next[c]) {
link_(u, v == root ? root : fallback[c]);
} else {
node.next[c] = fallback[c];
}
}
}
built_ = true;
}
int size() const { return static_cast<int>(terminal_.size()); }
bool empty() const { return terminal_.empty(); }
int node_count() const { return static_cast<int>(nodes_.size()); }
bool built() const { return built_; }
int terminal(int id) const {
assert(0 <= id && id < size());
return terminal_[id];
}
int length(int id) const { return nodes_[terminal(id)].depth; }
int parent(int v) const { check_(v); return nodes_[v].parent; }
int depth(int v) const { check_(v); return nodes_[v].depth; }
int child(int v, char c) const {
check_(v);
unsigned x = symbol_(c);
if (x >= Alphabet) return -1;
int u = nodes_[v].next[x];
return u && nodes_[u].parent == v ? u : -1;
}
int suffix_link(int v) const { check_(v); assert(built_); return nodes_[v].link; }
int output_link(int v) const { check_(v); assert(built_); return nodes_[v].output; }
int terminal_count(int v) const {
check_(v);
return built_ && v ? nodes_[v].matches - nodes_[nodes_[v].link].matches : nodes_[v].matches;
}
int match_count(int v) const { check_(v); assert(built_); return nodes_[v].matches; }
int transition(int v, char c) const {
check_(v), assert(built_);
unsigned x = symbol_(c);
return x < Alphabet ? nodes_[v].next[x] : root;
}
int walk(int v, std::string_view text) const {
check_(v), assert(built_);
for (char c : text) v = transition(v, c);
return v;
}
const std::vector<int> &bfs_order() const { assert(built_); return order_; }
template <class F>
void for_each_match(int v, F &&visit) const {
check_(v), assert(built_);
if (nodes_[v].head == -1) v = nodes_[v].output;
for (; v != -1; v = nodes_[v].output)
for (int id = nodes_[v].head; id != -1; id = next_pattern_[id]) visit(id);
}
template <class F>
void find_all(std::string_view text, F &&visit) const {
assert(built_);
int v = root;
for_each_match(v, [&](int id) { visit(std::size_t(0), id); });
for (std::size_t i = 0; i < text.size(); ++i) {
v = transition(v, text[i]);
for_each_match(v, [&](int id) { visit(i + 1, id); });
}
}
std::uint64_t count_matches(std::string_view text) const {
assert(built_);
int v = root;
std::uint64_t count = nodes_[root].matches;
for (char c : text) {
v = transition(v, c);
count += nodes_[v].matches;
}
return count;
}
template <class T>
void propagate(std::vector<T> &values) const {
assert(built_ && values.size() == nodes_.size());
for (std::size_t i = order_.size(); i-- > 1;) {
int v = order_[i];
values[nodes_[v].link] += values[v];
}
}
std::vector<std::uint64_t> occurrences(std::string_view text) const {
assert(built_);
std::vector<std::uint64_t> hits(nodes_.size());
int v = root;
++hits[root];
for (char c : text) v = transition(v, c), ++hits[v];
propagate(hits);
std::vector<std::uint64_t> count(size());
for (int id = 0; id < size(); ++id) count[id] = hits[terminal_[id]];
return count;
}
};
#line 1 "components/fast_io.hpp"
#include <algorithm>
#line 8 "components/fast_io.hpp"
#include <charconv>
#line 11 "components/fast_io.hpp"
#include <cstdio>
#include <cstring>
#include <memory>
#include <stdexcept>
#include <string>
#line 17 "components/fast_io.hpp"
#include <type_traits>
#if defined(__unix__) || defined(__APPLE__)
#include <sys/mman.h>
#include <sys/stat.h>
#include <unistd.h>
#endif
namespace fast_io_detail { struct BulkInput; struct BulkOutput; }
class FastInput {
friend struct fast_io_detail::BulkInput;
static constexpr std::size_t capacity = 1 << 16;
FILE *file = nullptr;
std::unique_ptr<char[]> buffer;
const char *current = "", *end = current;
std::string scratch;
void *mapping = nullptr;
std::size_t mapped_size = 0;
bool refill() {
if (!file) return false;
std::size_t n = std::fread(buffer.get(), 1, capacity, file);
current = buffer.get();
end = current + n;
return n;
}
static const char *scan(const char *p, const char *end) {
while (end - p >= 8) {
std::uint64_t x;
std::memcpy(&x, p, 8);
if ((x - 0x2121212121212121ULL) & ~x & 0x8080808080808080ULL) break;
p += 8;
}
while (p != end && static_cast<unsigned char>(*p) > ' ') ++p;
return p;
}
public:
explicit FastInput(FILE *input = stdin) {
#if defined(__unix__) || defined(__APPLE__)
struct stat status;
int fd = fileno(input);
auto offset = ftello(input);
if (fd >= 0 && offset >= 0 && !fstat(fd, &status) && S_ISREG(status.st_mode) && status.st_size > offset) {
mapped_size = status.st_size;
mapping = mmap(nullptr, mapped_size, PROT_READ, MAP_PRIVATE, fd, 0);
if (mapping != MAP_FAILED) {
current = static_cast<const char *>(mapping) + offset;
end = static_cast<const char *>(mapping) + mapped_size;
return;
}
mapping = nullptr;
}
#endif
file = input;
buffer = std::make_unique_for_overwrite<char[]>(capacity);
}
explicit FastInput(std::string_view input) : current(input.empty() ? "" : input.data()), end(current + input.size()) {}
FastInput(const FastInput &) = delete;
FastInput &operator=(const FastInput &) = delete;
~FastInput() {
#if defined(__unix__) || defined(__APPLE__)
if (mapping) munmap(mapping, mapped_size);
#endif
}
std::string_view token() {
while (true) {
while (current != end && static_cast<unsigned char>(*current) <= ' ') ++current;
if (current != end) break;
if (!refill()) return {};
}
const char *begin = current;
current = scan(current, end);
if (current != end || !file) return {begin, std::size_t(current - begin)};
scratch.assign(begin, end);
while (refill()) {
current = scan(current, end);
scratch.append(buffer.get(), std::size_t(current - buffer.get()));
if (current != end) break;
}
return scratch;
}
template <class T>
bool read(T &value) {
if constexpr (std::is_integral_v<T> && !std::is_same_v<T, bool>) {
while (true) {
while (current != end && static_cast<unsigned char>(*current) <= ' ') ++current;
if (current != end) break;
if (!refill()) return false;
}
if constexpr (std::is_unsigned_v<T> && sizeof(T) == 4 && std::endian::native == std::endian::little) {
if (end - current >= 11) {
std::uint64_t x;
std::memcpy(&x, current, 8);
if (!(((x + 0x4646464646464646ULL) | (x - 0x3030303030303030ULL)) & 0x8080808080808080ULL)) {
x &= 0x0f0f0f0f0f0f0f0fULL;
x = (x * 10 + (x >> 8)) & 0x00ff00ff00ff00ffULL;
x = (x * 100 + (x >> 16)) & 0x0000ffff0000ffffULL;
std::uint64_t v = std::uint32_t(x * 10000 + (x >> 32));
int n = 8;
for (; n < 10 && static_cast<unsigned char>(current[n]) - unsigned('0') < 10; ++n)
v = v * 10 + (current[n] - '0');
if (static_cast<unsigned char>(current[n]) <= ' ' && v <= 0xffffffffULL) {
current += n;
value = T(v);
return true;
}
}
}
}
T parsed{};
auto result = std::from_chars(current, end, parsed);
if (result.ec == std::errc{} && (result.ptr != end ? static_cast<unsigned char>(*result.ptr) <= ' ' : !file)) {
current = result.ptr;
value = parsed;
return true;
}
}
auto s = token();
if (s.empty()) return false;
if constexpr (std::is_same_v<T, bool>) {
assert(s == "0" || s == "1");
value = s[0] == '1';
} else if constexpr (std::is_integral_v<T>) {
auto result = std::from_chars(s.data(), s.data() + s.size(), value);
assert(result.ec == std::errc{} && result.ptr == s.data() + s.size());
(void)result;
} else value = T(s);
return true;
}
};
class FastOutput {
friend struct fast_io_detail::BulkOutput;
inline static constexpr auto quads = [] {
std::array<std::array<char, 4>, 10000> table{};
for (int i = 0; i < 10000; ++i)
table[i] = {char('0' + i / 1000), char('0' + i / 100 % 10), char('0' + i / 10 % 10), char('0' + i % 10)};
return table;
}();
FILE *file;
std::unique_ptr<char[]> buffer = std::make_unique_for_overwrite<char[]>(1 << 16);
std::size_t capacity = 1 << 16, length = 0;
char *reserve(std::size_t n) {
if (n > capacity - length) {
if (!flush()) throw std::runtime_error("output failed");
if (n > capacity) {
std::size_t next = std::max(n, 2 * capacity);
buffer = std::make_unique_for_overwrite<char[]>(next);
capacity = next;
}
}
return buffer.get() + length;
}
public:
explicit FastOutput(FILE *output = stdout) : file(output) {}
FastOutput(const FastOutput &) = delete;
FastOutput &operator=(const FastOutput &) = delete;
~FastOutput() { flush(); }
bool flush() {
std::size_t offset = 0;
while (offset < length) {
auto n = std::fwrite(buffer.get() + offset, 1, length - offset, file);
if (!n) {
std::memmove(buffer.get(), buffer.get() + offset, length - offset);
length -= offset;
return false;
}
offset += n;
}
length = 0;
return std::fflush(file) == 0;
}
void write(char c) { *reserve(1) = c; ++length; }
void write(__uint128_t value) {
char digits[40];
char *end = digits + sizeof(digits), *begin = end;
do { *--begin = char('0' + value % 10); value /= 10; } while (value);
write(std::string_view(begin, end - begin));
}
void write(__int128_t value) {
__uint128_t magnitude = __uint128_t(value);
if (value < 0) write('-'), magnitude = 0 - magnitude;
write(magnitude);
}
void write(std::string_view s) {
char *out = reserve(s.size());
if (!s.empty()) std::memcpy(out, s.data(), s.size());
length += s.size();
}
template <class T> requires std::is_integral_v<T>
void write(T value) {
char *out = reserve(32);
if constexpr (std::is_same_v<T, bool>) *out = '0' + value, ++length;
else if constexpr (std::is_unsigned_v<T> && sizeof(T) == 4) {
char *p = out;
if (value >= 100000000) {
auto high = value / 100000000, low = value % 100000000;
p = std::to_chars(p, p + 2, high).ptr;
std::memcpy(p, quads[low / 10000].data(), 4);
std::memcpy(p + 4, quads[low % 10000].data(), 4);
p += 8;
} else if (value >= 10000) {
p = std::to_chars(p, p + 4, value / 10000).ptr;
std::memcpy(p, quads[value % 10000].data(), 4);
p += 4;
} else p = std::to_chars(p, p + 4, value).ptr;
length += p - out;
} else length += std::to_chars(out, out + 32, value).ptr - out;
}
template <class T> requires requires(const T &value, char *out) { value.decimal_capacity(); value.to_chars(out); }
void write(const T &value) {
char *out = reserve(value.decimal_capacity());
length += value.to_chars(out) - out;
}
template <class T>
void println(const T &value) { write(value); write('\n'); }
};
#line 6 "test/aho_corasick.test.cpp"
int main() {
FastInput input;
FastOutput output;
int n;
input.read(n);
std::vector<std::string> patterns(n);
int length = 0;
for (auto &s : patterns) input.read(s), length += static_cast<int>(s.size());
AhoCorasick<> ac;
ac.reserve(length, n);
for (const auto &s : patterns) ac.add(s);
ac.build();
output.println(ac.node_count());
for (int v = 1; v < ac.node_count(); ++v) {
output.write(ac.parent(v)), output.write(' '), output.println(ac.suffix_link(v));
}
for (int i = 0; i < n; ++i) {
if (i) output.write(' ');
output.write(ac.terminal(i));
}
output.write('\n');
}