Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- #include <algorithm>
- #include <cmath>
- #include <iostream>
- #include <map>
- #include <set>
- #include <stdexcept>
- #include <cstdlib>
- #include <string>
- #include <utility>
- #include <vector>
- using namespace std;
- const int MAX_RESULT_DOCUMENT_COUNT = 5;
- string ReadLine() {
- string s;
- getline(cin, s);
- return s;
- }
- int ReadLineWithNumber() {
- int result;
- cin >> result;
- ReadLine();
- return result;
- }
- 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 {
- Document() = default;
- Document(int id, double relevance, int rating)
- : id(id)
- , relevance(relevance)
- , rating(rating) {
- }
- int id = 0;
- double relevance = 0.0;
- int rating = 0;
- };
- template <typename StringContainer>
- set<string> MakeUniqueNonEmptyStrings(const StringContainer& strings) {
- set<string> non_empty_strings;
- for (const string& str : strings) {
- if (!str.empty()) {
- non_empty_strings.insert(str);
- }
- }
- return non_empty_strings;
- }
- enum class DocumentStatus {
- ACTUAL,
- IRRELEVANT,
- BANNED,
- REMOVED,
- };
- class SearchServer {
- public:
- template <typename StringContainer>
- explicit SearchServer(const StringContainer& stop_words)
- : stop_words_(MakeUniqueNonEmptyStrings(stop_words)) // Extract non-empty stop words
- {
- if (!all_of(stop_words_.begin(), stop_words_.end(), IsValidWord)) {
- throw invalid_argument("Some of stop words are invalid"s);
- }
- }
- explicit SearchServer(const string& stop_words_text)
- : SearchServer(
- SplitIntoWords(stop_words_text)) // Invoke delegating constructor from string container
- {
- }
- void AddDocument(int document_id, const string& document, DocumentStatus status,
- const vector<int>& ratings) {
- if ((document_id < 0) || (documents_.count(document_id) > 0)) {
- throw invalid_argument("Invalid document_id"s);
- }
- const auto words = SplitIntoWordsNoStop(document);
- const double inv_word_count = 1.0 / words.size();
- for (const string& word : words) {
- word_to_document_freqs_[word][document_id] += inv_word_count;
- }
- documents_.emplace(document_id, DocumentData{ComputeAverageRating(ratings), status});
- document_ids_.push_back(document_id);
- }
- template <typename DocumentPredicate>
- vector<Document> FindTopDocuments(const string& raw_query,
- DocumentPredicate document_predicate) const {
- const auto query = ParseQuery(raw_query);
- auto matched_documents = FindAllDocuments(query, document_predicate);
- sort(matched_documents.begin(), matched_documents.end(),
- [](const Document& lhs, const Document& rhs) {
- if (abs(lhs.relevance - rhs.relevance) < 1e-6) {
- return lhs.rating > rhs.rating;
- } else {
- return lhs.relevance > rhs.relevance;
- }
- });
- if (matched_documents.size() > MAX_RESULT_DOCUMENT_COUNT) {
- matched_documents.resize(MAX_RESULT_DOCUMENT_COUNT);
- }
- return matched_documents;
- }
- vector<Document> FindTopDocuments(const string& raw_query, DocumentStatus status) const {
- return FindTopDocuments(
- raw_query, [status](int document_id, DocumentStatus document_status, int rating) {
- return document_status == status;
- });
- }
- vector<Document> FindTopDocuments(const string& raw_query) const {
- return FindTopDocuments(raw_query, DocumentStatus::ACTUAL);
- }
- int GetDocumentCount() const {
- return documents_.size();
- }
- int GetDocumentId(int index) const {
- return document_ids_.at(index);
- }
- tuple<vector<string>, DocumentStatus> MatchDocument(const string& raw_query,
- int document_id) const {
- const auto query = ParseQuery(raw_query);
- vector<string> matched_words;
- for (const string& word : query.plus_words) {
- if (word_to_document_freqs_.count(word) == 0) {
- continue;
- }
- if (word_to_document_freqs_.at(word).count(document_id)) {
- matched_words.push_back(word);
- }
- }
- for (const string& word : query.minus_words) {
- if (word_to_document_freqs_.count(word) == 0) {
- continue;
- }
- if (word_to_document_freqs_.at(word).count(document_id)) {
- matched_words.clear();
- break;
- }
- }
- return {matched_words, documents_.at(document_id).status};
- }
- private:
- struct DocumentData {
- int rating;
- DocumentStatus status;
- };
- const set<string> stop_words_;
- map<string, map<int, double>> word_to_document_freqs_;
- map<int, DocumentData> documents_;
- vector<int> document_ids_;
- bool IsStopWord(const string& word) const {
- return stop_words_.count(word) > 0;
- }
- static bool IsValidWord(const string& word) {
- // A valid word must not contain special characters
- return none_of(word.begin(), word.end(), [](char c) {
- return c >= '\0' && c < ' ';
- });
- }
- vector<string> SplitIntoWordsNoStop(const string& text) const {
- vector<string> words;
- for (const string& word : SplitIntoWords(text)) {
- if (!IsValidWord(word)) {
- throw invalid_argument("Word "s + word + " is invalid"s);
- }
- if (!IsStopWord(word)) {
- words.push_back(word);
- }
- }
- return words;
- }
- static int ComputeAverageRating(const vector<int>& ratings) {
- if (ratings.empty()) {
- return 0;
- }
- int rating_sum = 0;
- for (const int rating : ratings) {
- rating_sum += rating;
- }
- return rating_sum / static_cast<int>(ratings.size());
- }
- struct QueryWord {
- string data;
- bool is_minus;
- bool is_stop;
- };
- QueryWord ParseQueryWord(const string& text) const {
- if (text.empty()) {
- throw invalid_argument("Query word is empty"s);
- }
- string word = text;
- bool is_minus = false;
- if (word[0] == '-') {
- is_minus = true;
- word = word.substr(1);
- }
- if (word.empty() || word[0] == '-' || !IsValidWord(word)) {
- throw invalid_argument("Query word "s + text + " is invalid");
- }
- return {word, is_minus, IsStopWord(word)};
- }
- struct Query {
- set<string> plus_words;
- set<string> minus_words;
- };
- Query ParseQuery(const string& text) const {
- Query result;
- for (const string& word : SplitIntoWords(text)) {
- const auto query_word = ParseQueryWord(word);
- if (!query_word.is_stop) {
- if (query_word.is_minus) {
- result.minus_words.insert(query_word.data);
- } else {
- result.plus_words.insert(query_word.data);
- }
- }
- }
- return result;
- }
- // Existence required
- double ComputeWordInverseDocumentFreq(const string& word) const {
- return log(GetDocumentCount() * 1.0 / word_to_document_freqs_.at(word).size());
- }
- template <typename DocumentPredicate>
- vector<Document> FindAllDocuments(const Query& query,
- DocumentPredicate document_predicate) const {
- map<int, double> document_to_relevance;
- for (const string& word : query.plus_words) {
- if (word_to_document_freqs_.count(word) == 0) {
- continue;
- }
- const double inverse_document_freq = ComputeWordInverseDocumentFreq(word);
- for (const auto [document_id, term_freq] : word_to_document_freqs_.at(word)) {
- const auto& document_data = documents_.at(document_id);
- if (document_predicate(document_id, document_data.status, document_data.rating)) {
- document_to_relevance[document_id] += term_freq * inverse_document_freq;
- }
- }
- }
- for (const string& word : query.minus_words) {
- if (word_to_document_freqs_.count(word) == 0) {
- continue;
- }
- for (const auto [document_id, _] : word_to_document_freqs_.at(word)) {
- document_to_relevance.erase(document_id);
- }
- }
- vector<Document> matched_documents;
- for (const auto [document_id, relevance] : document_to_relevance) {
- matched_documents.push_back(
- {document_id, relevance, documents_.at(document_id).rating});
- }
- return matched_documents;
- }
- };
- template <typename Iterator>
- class IteratorRange {
- public:
- IteratorRange (Iterator first, Iterator last) : first_(first), last_(last) {
- }
- Iterator first() const {
- return first_;
- }
- Iterator last() const {
- return last_;
- }
- int size() const {
- return distance(first_, last_);
- }
- private:
- Iterator first_;
- Iterator last_;
- };
- template <typename Iterator>
- ostream& operator<<(ostream& out, const IteratorRange<Iterator>& range) {
- auto current = range.first();
- while(current != range.last()) {
- out <<
- "{ "s <<
- "document_id = "s << current->id << ", "s <<
- "relevance = "s << current->relevance << ", "s <<
- "rating = "s << current->rating << " "s << "}"s;
- advance(current, 1);
- }
- return out;
- }
- /*
- ostream& operator<<(ostream& out, Document& doc) {
- out << "{ document_id = "s << doc.id << ", relevance = "s, doc.relevance << ", rating = "s << doc.rating << " }"s;
- return out;
- }
- template <typename Iterator>
- ostream& operator<<(ostream& out, const IteratorRange<Iterator>& range) {
- for (auto current = range.first(); current != range.last(); ++current) {
- out << current << endl << "Page break"s << endl;
- }
- return out;
- }
- */
- template <typename Iterator>
- class Paginator {
- public:
- Paginator(Iterator first, Iterator last, size_t page_size) {
- int d = distance(first, last);
- if (d <= page_size) {
- pages_.push_back(IteratorRange(first, last));
- return;
- }
- auto dvs = distance(first, last) % page_size;
- for (auto current = first; distance(current, last) > page_size; advance(current, page_size)) {
- pages_.push_back(IteratorRange(current, next(current, page_size)));
- }
- pages_.push_back(IteratorRange(prev(last, dvs), last));
- }
- auto begin () const {
- return pages_.begin();
- }
- auto end () const {
- return pages_.end();
- }
- private:
- vector<IteratorRange<Iterator>> pages_;
- };
- template <typename Container>
- auto Paginate(const Container& c, size_t page_size) {
- return Paginator(begin(c), end(c), page_size);
- }
- void outta (string s, SearchServer search_server, int page_size = 2) {
- const auto search_results = search_server.FindTopDocuments(s);
- const auto pages = Paginate(search_results, page_size);
- cout << "Size = "s << search_results.size() << endl;
- // Выводим найденные документы по страницам
- for (auto page = pages.begin(); page != pages.end(); ++page) {
- cout << *page << endl;
- cout << "Page break"s << endl;
- }
- cout << "<===>"s << endl;
- }
- int main() {
- SearchServer search_server("and with"s);
- search_server.AddDocument(1, "funny pet and nasty rat"s, DocumentStatus::ACTUAL, {7, 2, 7});
- search_server.AddDocument(2, "funny pet with curly hair"s, DocumentStatus::ACTUAL, {1, 2, 3});
- search_server.AddDocument(3, "big cat nasty hair"s, DocumentStatus::ACTUAL, {1, 2, 8});
- search_server.AddDocument(4, "big dog cat Vladislav"s, DocumentStatus::ACTUAL, {1, 3, 2});
- search_server.AddDocument(5, "big dog hamster Borya"s, DocumentStatus::ACTUAL, {1, 1, 1});
- int page_size = 2;
- vector<string> ss = {
- //"big pet"s, // 5 results
- //"rat"s, // 1 result
- "curly dog"s // 3 results
- //"shit"s // 0 results
- };
- //for (string s : ss) outta(s, search_server, page_size);
- //abort();
- string s = "big pet"s;
- //s = "curly dog"s;
- //s = "shit"s;
- const auto search_results = search_server.FindTopDocuments(s);
- const auto pages = Paginate(search_results, page_size);
- // Выводим найденные документы по страницам
- for (auto page = pages.begin(); page != pages.end(); ++page) {
- cout << *page << endl;
- cout << "Page break"s << endl;
- }
- }
Advertisement
Add Comment
Please, Sign In to add comment