java/project/src/main/java/com/movieratings/crawler/strategy/BoxOfficeMojoCrawlerStrategy.java

package com.movieratings.crawler.strategy;

import com.movieratings.exception.CrawlerException;
import com.movieratings.model.Movie;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.io.IOException;
import java.util.ArrayList;
import java.util.List;

public class BoxOfficeMojoCrawlerStrategy extends AbstractCrawlerStrategy {
    private static final String URL = "https://www.boxofficemojo.com/chart/top_lifetime_gross/";

    @Override
    public String getSiteName() {
        return "Box Office Mojo";
    }

    @Override
    public List<Movie> crawl(int limit) {
        try {
            Document doc = connection(URL).get();
            List<Movie> movies = new ArrayList<>();
            for (Element row : doc.select("tr")) {
                if (movies.size() >= limit) {
                    break;
                }
                Elements cols = row.select("td");
                if (cols.size() < 4) {
                    continue;
                }
                Movie movie = parseRow(cols);
                if (movie.getTitle() != null && !movie.getTitle().isBlank()) {
                    movies.add(movie);
                }
            }
            return movies;
        } catch (IOException e) {
            throw new CrawlerException("Failed to crawl " + getSiteName(), e);
        }
    }

    private Movie parseRow(Elements cols) {
        int rank = parseCount(cols.get(0).text());
        Movie movie = new Movie();
        movie.setRank(rank);
        movie.setTitle(cols.get(1).text().trim());
        movie.setBoxOffice(parseMoney(cols.get(2).text()));
        movie.setReleaseYear(parseYear(cols.get(3).text()));
        movie.setRating(estimateRating(rank));
        movie.setDirector("Unknown");
        movie.setCountry("United States");
        movie.setReviewCount(0);
        movie.setPosterUrl("");
        movie.setQuote("Box Office Mojo lifetime gross chart entry");
        movie.setType("Movie");
        movie.setSourceSite(getSiteName());
        return movie;
    }

    private double parseMoney(String value) {
        if (value == null || value.isBlank()) {
            return 0.0;
        }
        String normalized = value.replace("$", "").replace(",", "").trim();
        try {
            return Double.parseDouble(normalized);
        } catch (NumberFormatException e) {
            return 0.0;
        }
    }

    private double estimateRating(int rank) {
        return Math.max(7.0, 8.8 - rank * 0.01);
    }
}