Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- #include <algorithm>
- #include <iostream>
- #include <set>
- #include <map>
- #include <string>
- #include <utility>
- #include <vector>
- #include <cmath>
- #include <numeric>
- using namespace std;
- const int MAX_RESULT_DOCUMENT_COUNT = 5;
- const double EPSILON = 1e-6;
- string ReadLine() {
- string s;
- getline(cin, s);
- return s;
- }
- int ReadLineWithNumber() {
- int result = 0;
- cin >> result;
- ReadLine();
- return result;
- }
- // type of doc
- enum class DocumentStatus {
- ACTUAL,
- IRRELEVANT,
- BANNED,
- REMOVED
- };
- vector<string> SplitIntoWords(const string& text) {
- vector<string> words;
- string word;
- for (const char c : text) {
- if (c == ' ') {
- if (!word.empty()) {
- words.push_back(word);
- word.clear();
- }
- }
- else {
- word += c;
- }
- }
- if (!word.empty()) {
- words.push_back(word);
- }
- return words;
- }
- struct Document {
- int id;
- double relevance;
- int rating;
- };
- bool FindWord(const string& word, const map<string, set<int>>& word_to_documents, const int& id) {
- return word_to_documents.count(word) && word_to_documents.at(word).count(id);
- }
- class SearchServer {
- public:
- void SetStopWords(const string& text) {
- for (const string& word : SplitIntoWords(text)) {
- stop_words_.insert(word);
- }
- }
- void AddDocument(int document_id, const string& document, DocumentStatus status, const vector<int>& ratings) {
- const vector<string> words = SplitIntoWordsNoStop(document);
- document_rating_[document_id] = ComputeAverageRating(ratings);
- document_status_[document_id] = status;
- //cout << static_cast<int>(status) << endl;
- double count_word = 0.0;
- for (const auto& w : words) {
- count_word = count(words.begin(), words.end(), w);
- word_to_documents_[w].insert(document_id);
- word_to_document_freqs_[w].insert({ document_id, count_word / words.size() });
- }
- ++documents_count_;
- }
- vector<Document> FindTopDocuments(const string& raw_query, DocumentStatus status = DocumentStatus::ACTUAL) const {
- const Query query_words = ParseQuery(raw_query);
- auto matched_documents = FindAllDocuments(query_words, status);
- sort(matched_documents.begin(), matched_documents.end(),
- [](const Document& lhs, const Document& rhs) {
- return lhs.relevance > rhs.relevance;
- });
- if (matched_documents.size() > MAX_RESULT_DOCUMENT_COUNT) {
- matched_documents.resize(MAX_RESULT_DOCUMENT_COUNT);
- }
- return matched_documents;
- }
- static int ComputeAverageRating(const vector<int>& ratings) {
- if (ratings.empty()) {
- return 0;
- }
- int document_rating = accumulate(ratings.begin(), ratings.end(), 0);
- return document_rating / static_cast<int>(ratings.size());
- }
- tuple<vector<string>, DocumentStatus> MatchDocument(const string& raw_query, int document_id) const {
- //исключаю пустой запрос и кривой id
- if (raw_query.empty() || document_id > documents_count_) {
- return { {}, document_status_.at(document_id) };
- }
- Query query;
- query = ParseQuery(raw_query);
- for (const auto& minus : query.minus_words) {
- if (FindWord(minus, word_to_documents_, document_id)) {
- return { {}, document_status_.at(document_id) };
- }
- }
- vector<string> words;
- for (const auto& plus : query.plus_words) {
- if (FindWord(plus, word_to_documents_, document_id)) {
- words.push_back(plus);
- }
- }
- sort(words.begin(), words.end());
- return { words, document_status_.at(document_id) };
- };
- int GetDocumentCount() {
- return documents_count_;
- }
- private:
- struct Status { // почему не хранить статус в документе?
- int id;
- double rating;
- DocumentStatus status;
- };
- struct Query {
- set<string> plus_words;
- set<string> minus_words;
- };
- map<string, set<int>> word_to_documents_;
- map<string, map<int, double>> word_to_document_freqs_;
- set<string> stop_words_;
- map<int, int> document_rating_;
- map<int, DocumentStatus> document_status_;
- int documents_count_ = 0;
- bool IsStopWord(const string& word) const {
- return stop_words_.count(word) > 0;
- }
- vector<string> SplitIntoWordsNoStop(const string& text) const {
- vector<string> words;
- for (const string& word : SplitIntoWords(text)) {
- if (!IsStopWord(word))
- {
- words.push_back(word);
- }
- }
- return words;
- }
- Query ParseQuery(const string& text) const {
- Query query;
- vector<string> raw_words = SplitIntoWordsNoStop(text);
- for (const string& word : raw_words) {
- if (word[0] == '-')
- {
- query.minus_words.insert(word.substr(1));
- }
- }
- for (const string& min : raw_words) {
- if (min[0] != '-' && !query.minus_words.count(min))
- {
- query.plus_words.insert(min);
- }
- }
- return query;
- }
- // IDF-TF
- vector<Document> FindAllDocuments(const Query& query_words, DocumentStatus status) const {
- vector<Document> matched_documents;
- map<int, double> document_to_relevance;
- double IDF = 0.0;
- //вынести в отдельную функцию учет plus слов
- for (const auto& plus : query_words.plus_words) {
- if (word_to_documents_.count(plus))
- {
- IDF = log(static_cast<double>(documents_count_) / word_to_documents_.at(plus).size());
- for (const auto& id : word_to_documents_.at(plus))
- {
- if (status == document_status_.at(id))
- {
- document_to_relevance[id] += IDF * word_to_document_freqs_.at(plus).at(id);
- }
- };
- }
- }
- //вынести в отдельную функцию учет minus слов
- for (const auto& minus : query_words.minus_words) {
- if (word_to_documents_.count(minus))
- {
- for (const auto& id : word_to_documents_.at(minus)) {
- document_to_relevance.erase(id);
- };
- }
- }
- for (const auto& [id, relevance] : document_to_relevance) {
- matched_documents.push_back({ id, relevance, document_rating_.at(id) });
- }
- return matched_documents;
- }
- };
- void PrintMatchDocumentResult(int document_id, const vector<string>& words, DocumentStatus status) {
- cout << "{ "s
- << "document_id = "s << document_id << ", "s
- << "status = "s << static_cast<int>(status) << ", "s
- << "words ="s;
- for (const string& word : words) {
- cout << ' ' << word;
- }
- cout << "}"s << endl;
- }
- int main() {
- using namespace std::literals;
- const std::vector<int> ratings1 = { 1, 2, 3, 4, 5 };
- const std::vector<int> ratings2 = { -1, -2, 30, -3, 44, 5 };
- const std::vector<int> ratings3 = { 12, -20, 80, 0, 8, 0, 0, 9, 67 };
- SearchServer search_server;
- search_server.SetStopWords("и в на"s);
- search_server.AddDocument(0, "белый кот и модный ошейник", DocumentStatus::ACTUAL, ratings1);
- search_server.AddDocument(1, "пушистый кот пушистый хвост", DocumentStatus::ACTUAL, ratings2);
- search_server.AddDocument(2, "ухоженный пёс выразительные глаза", DocumentStatus::ACTUAL,ratings3);
- search_server.AddDocument(3, "белый модный кот", DocumentStatus::IRRELEVANT, ratings1);
- search_server.AddDocument(4, "пушистый кот пёс", DocumentStatus::IRRELEVANT, ratings2);
- search_server.AddDocument(5, "ухоженный ошейник выразительные глаза",DocumentStatus::IRRELEVANT, ratings3);
- search_server.AddDocument(6, "кот и ошейник", DocumentStatus::BANNED, ratings1);
- search_server.AddDocument(7, "пёс и хвост", DocumentStatus::BANNED, ratings2);
- search_server.AddDocument(8, "модный пёс пушистый хвост", DocumentStatus::BANNED, ratings3);
- search_server.AddDocument(9, "кот пушистый ошейник", DocumentStatus::REMOVED, ratings1);
- search_server.AddDocument(10, "ухоженный кот и пёс", DocumentStatus::REMOVED, ratings2);
- search_server.AddDocument(11, "хвост и выразительные глаза", DocumentStatus::REMOVED, ratings3);
- const int document_count = search_server.GetDocumentCount();
- for (int document_id = 0; document_id < document_count; ++document_id) {
- const auto [words, status] = search_server.MatchDocument("пушистый ухоженный кот -ошейник", document_id);
- PrintMatchDocumentResult(document_id, words, status);
- }
- }
- /*
- * "пушистый кот -ошейник"
- * "пушистый -кот"
- * "пёс скворец -ухоженный"
- * "пушистый кот модный"
- * "-ухоженный"
- * "пёс скворец"
- * "шашлык автомобиль дом велосипед"
- * " "
- * ""
- * enum class DocumentStatus {
- ACTUAL,
- IRRELEVANT,
- BANNED,
- REMOVED
- };
- */
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement