SHOW:
|
|
- or go back to the newest paste.
| 1 | #include <algorithm> | |
| 2 | #include <cmath> | |
| 3 | #include <iostream> | |
| 4 | #include <map> | |
| 5 | #include <set> | |
| 6 | #include <string> | |
| 7 | #include <utility> | |
| 8 | #include <vector> | |
| 9 | #include <optional> | |
| 10 | ||
| 11 | using namespace std; | |
| 12 | ||
| 13 | constexpr double EPSILON = 1e-6; | |
| 14 | const int MAX_RESULT_DOCUMENT_COUNT = 5; | |
| 15 | ||
| 16 | string ReadLine() {
| |
| 17 | string s; | |
| 18 | getline(cin, s); | |
| 19 | return s; | |
| 20 | } | |
| 21 | ||
| 22 | int ReadLineWithNumber() {
| |
| 23 | int result; | |
| 24 | cin >> result; | |
| 25 | ReadLine(); | |
| 26 | return result; | |
| 27 | } | |
| 28 | ||
| 29 | vector<string> SplitIntoWords(const string& text) {
| |
| 30 | vector<string> words; | |
| 31 | string word; | |
| 32 | for (const char c : text) {
| |
| 33 | if (c == ' ') {
| |
| 34 | if (!word.empty()) {
| |
| 35 | words.push_back(word); | |
| 36 | word.clear(); | |
| 37 | } | |
| 38 | } else {
| |
| 39 | word += c; | |
| 40 | } | |
| 41 | } | |
| 42 | if (!word.empty()) {
| |
| 43 | words.push_back(word); | |
| 44 | } | |
| 45 | ||
| 46 | return words; | |
| 47 | } | |
| 48 | ||
| 49 | struct Document {
| |
| 50 | Document() = default; | |
| 51 | ||
| 52 | Document(int id, double relevance, int rating) | |
| 53 | : id(id) | |
| 54 | , relevance(relevance) | |
| 55 | , rating(rating) {
| |
| 56 | } | |
| 57 | ||
| 58 | int id = 0; | |
| 59 | double relevance = 0.0; | |
| 60 | int rating = 0; | |
| 61 | }; | |
| 62 | ||
| 63 | template <typename StringContainer> | |
| 64 | set<string> MakeUniqueNonEmptyStrings(const StringContainer& strings) {
| |
| 65 | set<string> non_empty_strings; | |
| 66 | for (const string& str : strings) {
| |
| 67 | if (!str.empty()) {
| |
| 68 | non_empty_strings.insert(str); | |
| 69 | } | |
| 70 | } | |
| 71 | return non_empty_strings; | |
| 72 | } | |
| 73 | ||
| 74 | enum class DocumentStatus {
| |
| 75 | ACTUAL, | |
| 76 | IRRELEVANT, | |
| 77 | BANNED, | |
| 78 | REMOVED, | |
| 79 | }; | |
| 80 | ||
| 81 | class SearchServer {
| |
| 82 | public: | |
| 83 | ||
| 84 | template <typename StringContainer> | |
| 85 | explicit SearchServer(const StringContainer& stop_words) : stop_words_(MakeUniqueNonEmptyStrings(stop_words)){
| |
| 86 | for (const auto& stop_word : stop_words_) {
| |
| 87 | if (IsValidWord(stop_word) == false) {
| |
| 88 | throw invalid_argument("Some of the stop words contain invalid characters"s);
| |
| 89 | } | |
| 90 | } | |
| 91 | } | |
| 92 | ||
| 93 | explicit SearchServer(const string& stop_words_text) : SearchServer(SplitIntoWords(stop_words_text)){
| |
| 94 | ||
| 95 | } | |
| 96 | ||
| 97 | bool AddDocument(int document_id, const string& document, DocumentStatus status, | |
| 98 | const vector<int>& ratings) {
| |
| 99 | if(document_id < 0){
| |
| 100 | throw invalid_argument("The document cannot have a negative ID"s);
| |
| 101 | } | |
| 102 | if(documents_.count(document_id)){
| |
| 103 | throw invalid_argument("ID belongs to an already added document"s);
| |
| 104 | } | |
| 105 | if(IsValidWord(document)==false){
| |
| 106 | throw invalid_argument("Document text contains invalid characters"s);
| |
| 107 | } | |
| 108 | const vector<string> words = SplitIntoWordsNoStop(document); | |
| 109 | const double inv_word_count = 1.0 / words.size(); | |
| 110 | for (const string& word : words) {
| |
| 111 | word_to_document_freqs_[word][document_id] += inv_word_count; | |
| 112 | } | |
| 113 | document_ids.push_back(document_id); | |
| 114 | documents_.emplace(document_id, DocumentData{ComputeAverageRating(ratings), status});
| |
| 115 | return true; | |
| 116 | } | |
| 117 | ||
| 118 | template <typename DocumentPredicate> | |
| 119 | vector<Document> FindTopDocuments(const string& raw_query, | |
| 120 | DocumentPredicate document_predicate) const {
| |
| 121 | const Query query = ParseQuery(raw_query); | |
| 122 | auto matched_documents = FindAllDocuments(query, document_predicate); | |
| 123 | sort(matched_documents.begin(), matched_documents.end(), | |
| 124 | [](const Document& lhs, const Document& rhs) {
| |
| 125 | if (abs(lhs.relevance - rhs.relevance) < EPSILON) {
| |
| 126 | return lhs.rating > rhs.rating; | |
| 127 | } else {
| |
| 128 | return lhs.relevance > rhs.relevance; | |
| 129 | }}); | |
| 130 | if (matched_documents.size() > MAX_RESULT_DOCUMENT_COUNT) {
| |
| 131 | matched_documents.resize(MAX_RESULT_DOCUMENT_COUNT); | |
| 132 | } | |
| 133 | return matched_documents; | |
| 134 | } | |
| 135 | ||
| 136 | vector<Document> FindTopDocuments(const string& raw_query, DocumentStatus status) const {
| |
| 137 | return FindTopDocuments(raw_query, [status](int document_id, DocumentStatus document_status, int rating) {
| |
| 138 | return document_status == status;}); | |
| 139 | } | |
| 140 | ||
| 141 | vector<Document> FindTopDocuments(const string& raw_query) const {
| |
| 142 | return FindTopDocuments(raw_query, DocumentStatus::ACTUAL); | |
| 143 | } | |
| 144 | ||
| 145 | int GetDocumentCount() const {
| |
| 146 | return documents_.size(); | |
| 147 | } | |
| 148 | ||
| 149 | int GetDocumentId(/*const*/ int index) const {
| |
| 150 | return document_ids.at(index); | |
| 151 | } | |
| 152 | ||
| 153 | tuple<vector<string>, DocumentStatus> MatchDocument(const string& raw_query, | |
| 154 | int document_id) const {
| |
| 155 | const Query query = ParseQuery(raw_query); | |
| 156 | vector<string> matched_words; | |
| 157 | for (const string& word : query.plus_words) {
| |
| 158 | if (word_to_document_freqs_.count(word) == 0) {
| |
| 159 | continue; | |
| 160 | } | |
| 161 | if (word_to_document_freqs_.at(word).count(document_id)) {
| |
| 162 | matched_words.push_back(word); | |
| 163 | } | |
| 164 | } | |
| 165 | for (const string& word : query.minus_words) {
| |
| 166 | if (word_to_document_freqs_.count(word) == 0) {
| |
| 167 | continue; | |
| 168 | } | |
| 169 | if (word_to_document_freqs_.at(word).count(document_id)) {
| |
| 170 | matched_words.clear(); | |
| 171 | break; | |
| 172 | } | |
| 173 | } | |
| 174 | auto result = tuple{matched_words, documents_.at(document_id).status};
| |
| 175 | return result; | |
| 176 | } | |
| 177 | ||
| 178 | private: | |
| 179 | struct DocumentData {
| |
| 180 | int rating; | |
| 181 | DocumentStatus status; | |
| 182 | }; | |
| 183 | ||
| 184 | vector<int> document_ids; | |
| 185 | set<string> stop_words_; | |
| 186 | map<string, map<int, double>> word_to_document_freqs_; | |
| 187 | map<int, DocumentData> documents_; | |
| 188 | ||
| 189 | bool IsStopWord(const string& word) const {
| |
| 190 | return stop_words_.count(word) > 0; | |
| 191 | } | |
| 192 | ||
| 193 | vector<string> SplitIntoWordsNoStop(const string& text) const {
| |
| 194 | vector<string> words; | |
| 195 | for (const string& word : SplitIntoWords(text)) {
| |
| 196 | if (!IsStopWord(word)) {
| |
| 197 | words.push_back(word); | |
| 198 | } | |
| 199 | } | |
| 200 | return words; | |
| 201 | } | |
| 202 | ||
| 203 | static int ComputeAverageRating(const vector<int>& ratings) {
| |
| 204 | if (ratings.empty()) {
| |
| 205 | return 0; | |
| 206 | } | |
| 207 | int rating_sum = 0; | |
| 208 | for (const int rating : ratings) {
| |
| 209 | rating_sum += rating; | |
| 210 | } | |
| 211 | return rating_sum / static_cast<int>(ratings.size()); | |
| 212 | } | |
| 213 | ||
| 214 | struct QueryWord {
| |
| 215 | string data; | |
| 216 | bool is_minus; | |
| 217 | bool is_stop; | |
| 218 | }; | |
| 219 | ||
| 220 | QueryWord ParseQueryWord(string text) const {
| |
| 221 | if(text.empty() == true){
| |
| 222 | throw invalid_argument("Search query is empty"s);
| |
| 223 | } | |
| 224 | if (text.back() == '-') {
| |
| 225 | throw invalid_argument("Missing text after minus sign in search query"s);
| |
| 226 | } | |
| 227 | ||
| 228 | bool is_minus = false; | |
| 229 | QueryWord result; | |
| 230 | ||
| 231 | if (text[0] == '-') {
| |
| 232 | is_minus = true; | |
| 233 | text = text.substr(1); | |
| 234 | } | |
| 235 | if(IsValidWord(text)==false){
| |
| 236 | throw invalid_argument("There are invalid characters in the search query words"s);
| |
| 237 | } | |
| 238 | if (text[0] == '-'){
| |
| 239 | throw invalid_argument("Presence of more than one minus before stop words"s);
| |
| 240 | } | |
| 241 | result = {text, is_minus, IsStopWord(text)};
| |
| 242 | return result; | |
| 243 | } | |
| 244 | ||
| 245 | struct Query {
| |
| 246 | set<string> plus_words; | |
| 247 | set<string> minus_words; | |
| 248 | }; | |
| 249 | ||
| 250 | Query ParseQuery(const string& text) const {
| |
| 251 | Query query; | |
| 252 | for (const string& word : SplitIntoWords(text)) {
| |
| 253 | const QueryWord query_word = ParseQueryWord(word); | |
| 254 | if (!query_word.is_stop) {
| |
| 255 | if (query_word.is_minus) {
| |
| 256 | query.minus_words.insert(query_word.data); | |
| 257 | } else {
| |
| 258 | query.plus_words.insert(query_word.data); | |
| 259 | } | |
| 260 | } | |
| 261 | } | |
| 262 | return query; | |
| 263 | } | |
| 264 | ||
| 265 | // Existence required | |
| 266 | double ComputeWordInverseDocumentFreq(const string& word) const {
| |
| 267 | return log(GetDocumentCount() * 1.0 / word_to_document_freqs_.at(word).size()); | |
| 268 | } | |
| 269 | ||
| 270 | template <typename DocumentPredicate> | |
| 271 | vector<Document> FindAllDocuments(const Query& query, | |
| 272 | DocumentPredicate document_predicate) const {
| |
| 273 | map<int, double> document_to_relevance; | |
| 274 | for (const string& word : query.plus_words) {
| |
| 275 | if (word_to_document_freqs_.count(word) == 0) {
| |
| 276 | continue; | |
| 277 | } | |
| 278 | const double inverse_document_freq = ComputeWordInverseDocumentFreq(word); | |
| 279 | for (const auto [document_id, term_freq] : word_to_document_freqs_.at(word)) {
| |
| 280 | const auto& document_data = documents_.at(document_id); | |
| 281 | if (document_predicate(document_id, document_data.status, document_data.rating)) {
| |
| 282 | document_to_relevance[document_id] += term_freq * inverse_document_freq; | |
| 283 | } | |
| 284 | } | |
| 285 | } | |
| 286 | ||
| 287 | for (const string& word : query.minus_words) {
| |
| 288 | if (word_to_document_freqs_.count(word) == 0) {
| |
| 289 | continue; | |
| 290 | } | |
| 291 | for (const auto [document_id, _] : word_to_document_freqs_.at(word)) {
| |
| 292 | document_to_relevance.erase(document_id); | |
| 293 | } | |
| 294 | } | |
| 295 | ||
| 296 | vector<Document> matched_documents; | |
| 297 | for (const auto [document_id, relevance] : document_to_relevance) {
| |
| 298 | matched_documents.push_back( | |
| 299 | {document_id, relevance, documents_.at(document_id).rating});
| |
| 300 | } | |
| 301 | return matched_documents; | |
| 302 | } | |
| 303 | ||
| 304 | static bool IsValidWord(const string& word) {
| |
| 305 | return none_of(word.begin(), word.end(), [](char c) {
| |
| 306 | return c >= '\0' && c <' ';}); | |
| 307 | } | |
| 308 | }; | |
| 309 | ||
| 310 | // ==================== для примера ========================= | |
| 311 | ||
| 312 | void PrintDocument(const Document& document) {
| |
| 313 | cout << "{ "s
| |
| 314 | << "document_id = "s << document.id << ", "s | |
| 315 | << "relevance = "s << document.relevance << ", "s | |
| 316 | << "rating = "s << document.rating << " }"s << endl; | |
| 317 | } | |
| 318 | int main() {
| |
| 319 | SearchServer search_server("and in on wi\x17th at a"s);
| |
| 320 | ||
| 321 | search_server.AddDocument(-1, "fluffy animals and feathered bird"s, DocumentStatus::ACTUAL, { -7, 1, 7, -1});
| |
| 322 | search_server.AddDocument(1, "fluffy animals and feathered bird"s, DocumentStatus::ACTUAL, { 3, 5 });
| |
| 323 | search_server.AddDocument(2, "a bird eats grain with a cheerful cat"s, DocumentStatus::ACTUAL, { 1, 2, 3 });
| |
| 324 | search_server.AddDocument(3, "a bird eat\x17s grain with a cheerful dog"s, DocumentStatus::ACTUAL, { -1, -2, -3 });
| |
| 325 | search_server.AddDocument(4, "bird and dog are good animals"s, DocumentStatus::ACTUAL, { 1, 1, 1 });
| |
| 326 | ||
| 327 | search_server.FindTopDocuments("fluffy -animals"s);
| |
| 328 | search_server.FindTopDocuments("--fluffy animals"s);
| |
| 329 | search_server.FindTopDocuments("animals -"s);
| |
| 330 | search_server.FindTopDocuments("-"s);
| |
| 331 | ||
| 332 | search_server.MatchDocument("good animals"s, 4);
| |
| 333 | search_server.MatchDocument("fluffy --animals"s, 1);
| |
| 334 | search_server.MatchDocument("-feathered bird"s, 1);
| |
| 335 | search_server.MatchDocument("bird dog -"s, 4);
| |
| 336 | search_server.MatchDocument("bird - dog"s, 14);
| |
| 337 | } | |
| 338 |