Add Homework 02 assignment

09-threads/lab/README.md

@@ -16,7 +16,7 @@ Sentiment analysis, или анализ на настроения, е техни
 "I love java" = 3 (moderately positive)
 "I hate bugs" = -2 (slightly negative)
 
-За да има смисъл от паралелна обработка обаче, може да си представим, че се анализират големи много файлове, с много текст в тях, например цели книги.
+За да има смисъл от паралелна обработка обаче, може да си представим, че се анализират много и големи файлове, с много текст в тях, например цели книги.
 
 Ще работим с няколко текстови потока:
  - поток със "стоп думи" - това са често срещани думи в даден език, които обикновено нямат особена семантична стойност или не допринасят за анализа на текста. Например "the", "is", "at", "which", "on" и т.н.

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/BookLoader.java

@@ -0,0 +1,28 @@
+package bg.sofia.uni.fmi.mjt.goodreads;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+import com.opencsv.CSVReader;
+import com.opencsv.exceptions.CsvException;
+
+import java.io.IOException;
+import java.io.Reader;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+public class BookLoader {
+
+    public static Set<Book> load(Reader reader) {
+
+        try (CSVReader csvReader = new CSVReader(reader)) {
+            return csvReader.readAll().stream()
+                    .skip(1)
+                    .map(Book::of)
+                    .collect(Collectors.toSet());
+
+        } catch (IOException | CsvException ex) {
+            throw new IllegalArgumentException("Could not load dataset", ex);
+        }
+
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/book/Book.java

@@ -0,0 +1,21 @@
+package bg.sofia.uni.fmi.mjt.goodreads.book;
+
+import java.util.ArrayList;
+import java.util.List;
+
+public record Book(
+        String ID,
+        String title,
+        String author,
+        String description,
+        List<String> genres,
+        double rating,
+        int ratingCount,
+        String URL
+) {
+
+    public static Book of(String[] tokens) {
+        throw new UnsupportedOperationException("Not yet implemented");
+
+    }
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/finder/BookFinder.java

@@ -0,0 +1,37 @@
+package bg.sofia.uni.fmi.mjt.goodreads.finder;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+import bg.sofia.uni.fmi.mjt.goodreads.sanitizer.StringTokenHandler;
+
+import java.util.List;
+import java.util.Set;
+
+public class BookFinder implements BookFinderAPI {
+
+    public BookFinder(Set<Book> books, StringTokenHandler tokenizer) {}
+
+    public Set<Book> allBooks() {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    @Override
+    public List<Book> searchByAuthor(String authorName) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    @Override
+    public Set<String> allGenres() {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    @Override
+    public List<Book> searchByGenres(Set<String> genres, MatchOption option) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    @Override
+    public List<Book> searchByKeywords(Set<String> keywords, MatchOption option) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/finder/BookFinderAPI.java

@@ -0,0 +1,57 @@
+package bg.sofia.uni.fmi.mjt.goodreads.finder;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+
+import java.util.List;
+import java.util.Set;
+
+public interface BookFinderAPI {
+
+    /**
+     * Retrieves all books
+     *
+     * @return a Set of all books.
+     */
+    Set<Book> allBooks();
+
+    /**
+     * Retrieves all genres
+     *
+     * @return a Set of all genres.
+     */
+    Set<String> allGenres();
+
+    /**
+     * Searches for books by the specified author name.
+     *
+     * @param authorName the name of the author to search for.
+     * @throws IllegalArgumentException if the author name if null or empty
+     * @return a List of books written by the specified author.
+     *         Returns an empty list if no books are found.
+     */
+    List<Book> searchByAuthor(String authorName);
+
+    /**
+     * Searches for books that belong to the specified genres.
+     * The search can be based on different match options (all or any genres).
+     *
+     * @param genres a Set of genres to search for.
+     * @throws IllegalArgumentException if {@param genres} is null
+     * @return a List of books that match the given genres according to the MatchOption
+     *         Returns an empty list if no books are found.
+     */
+    List<Book> searchByGenres(Set<String> genres, MatchOption option);
+
+    /**
+     * Searches for books that match the specified keywords.
+     * The search can be based on different match options (all or any keywords).
+     *
+     * @param keywords a {@code Set} of keywords to search for.
+     * @param option the {@code MatchOption} that defines the search criteria
+     *               (either {@link MatchOption#MATCH_ALL} or {@link MatchOption#MATCH_ANY}).
+     * @return a List of books that match the given keywords according to the MatchOption
+     *         Returns an empty list if no books are found.
+     */
+    List<Book> searchByKeywords(Set<String> keywords, MatchOption option);
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/finder/MatchOption.java

@@ -0,0 +1,6 @@
+package bg.sofia.uni.fmi.mjt.goodreads.finder;
+
+public enum MatchOption {
+    MATCH_ALL,
+    MATCH_ANY
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/recommender/BookRecommender.java

@@ -0,0 +1,21 @@
+package bg.sofia.uni.fmi.mjt.goodreads.recommender;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+
+import java.util.Comparator;
+import java.util.LinkedHashMap;
+import java.util.Map;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+public class BookRecommender implements BookRecommenderAPI {
+
+    public BookRecommender(Set<Book> initialBooks, SimilarityCalculator calculator) {}
+
+
+    @Override
+    public Map<Book, Double> recommendBooks(Book origin, int maxN) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/recommender/BookRecommenderAPI.java

@@ -0,0 +1,20 @@
+package bg.sofia.uni.fmi.mjt.goodreads.recommender;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+
+import java.util.Map;
+
+public interface BookRecommenderAPI {
+
+    /**
+     * Searches for books that are similar to the provided one.
+     *
+     * @param  originBook the book we should calculate similarity with.
+     * @param  maxN - the maximum number of entries returned
+     * @throws IllegalArgumentException if the originBook is null.
+     * @throws IllegalArgumentException if maxN is smaller or equal to 0.
+     * @return a Map<Book, Double> representing the top maxN closest books with their similarity to originBook ordered by their similarity score
+     */
+    Map<Book, Double> recommendBooks(Book originBook, int maxN);
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/recommender/similarityCalculator/composite/CompositeSimilarityCalculator.java

@@ -0,0 +1,18 @@
+package bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.composite;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+import bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.SimilarityCalculator;
+
+import java.util.Map;
+
+public class CompositeSimilarityCalculator implements SimilarityCalculator {
+
+    public CompositeSimilarityCalculator(Map<SimilarityCalculator, Double> similarityCalculatorMap) {}
+
+
+    @Override
+    public double calculateSimilarity(Book first, Book second) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/recommender/similarityCalculator/descriptions/TFIDF.java

@@ -0,0 +1,31 @@
+package bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.descriptions;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+import bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.SimilarityCalculator;
+import bg.sofia.uni.fmi.mjt.goodreads.sanitizer.StringTokenHandler;
+
+import java.util.*;
+import java.util.stream.Collectors;
+
+public class TFIDF implements SimilarityCalculator {
+
+    public TFIDF(Set<Book> books, TextTokenizer tokenizer) {}
+
+    @Override
+    public double descriptionSimilarity(Book first, Book second) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    public Map<String, Double> computeTFIDF(Book book) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    public Map<String, Double> computeTF(Book book) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    public Map<String, Double> computeIDF(Book book) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/recommender/similarityCalculator/genres/GenresOverlapSimilarityCalculator.java

@@ -0,0 +1,13 @@
+package bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.genres;
+
+import bg.sofia.uni.fmi.mjt.goodreads.book.Book;
+import bg.sofia.uni.fmi.mjt.goodreads.recommender.similarityCalculator.SimilarityCalculator;
+
+public class GenresOverlapSimilarityCalculator implements SimilarityCalculator {
+
+    @Override
+    public double genresSimilarity(Book first, Book second) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+}

homeworks/02-goodreads/resources/src/bg/sofia/uni/fmi/mjt/goodreads/tokenizer/TextTokenizer.java

@@ -0,0 +1,31 @@
+package bg.sofia.uni.fmi.mjt.goodreads.sanitizer;
+
+import java.io.BufferedReader;
+import java.io.IOException;
+import java.io.Reader;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Set;
+import java.util.stream.Collectors;
+
+public class TextTokenizer {
+
+    private final Set<String> stopwords;
+
+    public TextTokenizer(Reader stopwordsReader) {
+        try (var br = new BufferedReader(stopwordsReader)) {
+            stopwords = br.lines().collect(Collectors.toSet());
+        } catch (IOException ex) {
+            throw new IllegalArgumentException("Could not load dataset", ex);
+        }
+    }
+
+    public List<String> tokenize(String input) {
+        throw new UnsupportedOperationException("Not yet implemented");
+    }
+
+    public Set<String> stopwords() {
+        return stopwords;
+    }
+
+}