_1242.java

packagecom.fishercoder.solutions.secondthousand;

importjava.util.*;
importjava.util.concurrent.*;

publicclass_1242 {
publicinterfaceHtmlParser {
publicList<String> getUrls(Stringurl);
 }

publicstaticclassSolution1 {
/*
 * credit: https://leetcode.com/problems/web-crawler-multithreaded/solutions/699006/java-blockingqueue-executorservice/
 */
publicList<String> crawl(StringstartUrl, HtmlParserhtmlParser) {
StringtargetHostName = getHostName(startUrl);
List<String> result = newArrayList<>();
BlockingQueue<String> queue = newLinkedBlockingQueue<>();
queue.offer(startUrl);
Set<String> visited = newHashSet<>();

Queue<Future> tasks = newLinkedList<>();
// create a thread pool to crawling the URLs
ExecutorServiceexecutorService =
Executors.newFixedThreadPool(
4,
r -> {
Threadt = newThread(r);
// LeetCode doesn't allow executor.shutdown(), so use daemon threads
// to let the program shutdown, otherwise TLE.
t.setDaemon(true);
returnt;
 });

while (true) {
Stringurl = queue.poll();
if (url != null) {
if (getHostName(url).equals(targetHostName) && visited.add(url)) {
result.add(url);
tasks.add(
executorService.submit(
 () -> {
List<String> urls = htmlParser.getUrls(url);
for (Stringu : urls) {
queue.offer(u);
 }
 }));
 }
 } else {
if (!tasks.isEmpty()) {
// wait for the next task to complete which might add new URLs into the
// queue
FuturenextTask = tasks.poll();
try {
nextTask.get();
 } catch (InterruptedException | ExecutionExceptione) {
 }
 } else {
// exit when all tasks are completed.
break;
 }
 }
 }

returnresult;
 }

privateStringgetHostName(Stringurl) {
url = url.substring("http://".length());
String[] parts = url.split("/");
returnparts[0];
 }
 }

publicstaticvoidmain(String... args) {
Solution1solution1 = newSolution1();
System.out.println(solution1.getHostName("http://news.yahoo.com"));
System.out.println(solution1.getHostName("http://news.yahoo.com/news"));
System.out.println(solution1.getHostName("http://news.yahoo.com/us"));
System.out.println(solution1.getHostName("http://news.yahoo.com"));
 }
}