winscrape/agent-smith/src/UpdateScanner.java

import org.jsoup.Jsoup;

import java.io.IOException;
import java.util.*;

/**
 * downloads all files on each page, unpacks the cab file, and scans the driver files for desired imports.
 * inspection of the drivers must be done manually until further notice.
 */
public class UpdateScanner extends Thread {
    private static final String searchUrl = "https://www.catalog.update.microsoft.com/Search.aspx?q=";
    private String searchKey;
    private Map<String, String> dataMap = new HashMap<>();
    private static String[] imports;

    public UpdateScanner(String searchKey) {
        this.searchKey = searchKey;
    }

    public UpdateScanner(String searchKey, String[] imports) {
        this.searchKey = searchKey;
        this.imports = imports;
    }

    public UpdateScanner(String searchKey, Map<String, String> dataMap) {
        this.searchKey = searchKey;
        this.dataMap = dataMap;
    }

    /**
     * used to scan all pages with as many threads as possible.
     * Each page contains data needed to get to the next so as soon as that is obtained
     * we make a new thread to scan the next page.
     */
    public void run() {
        try {
            // get page
            var result = Jsoup
                    .connect(this.searchUrl.concat(this.searchKey.replace(" ", "+")))
                    .data(this.dataMap)
                    .post();

            // get the pages values
            var pageValues = new UpdatePageValues(result.getAllElements());
            System.out.printf("[+] scanning page %d of %d\n", pageValues.getPageIndex(), pageValues.getMaxPageIndex());

            // if there is a next page, fork a thread to process it whilst we process this page
            if (pageValues.hasNextPage())
                new UpdateScanner(this.searchKey, pageValues.getAspNetHiddenUrlEncoded()).start();

            // for each uuid on this page make a new thread to download and scan the file
            pageValues.getPageTableValues().getDriverUuids().forEach(uuid -> {
                new DriverDownloader(uuid, this.imports).start();
            });
        } catch (IOException e) { }
    }
}