Nutch 1.0 Fetcher 抓取模型解析
1. 介绍
2. 抓取流程分析
3. 结束
---------------
1. 介绍
Nutch是apache Lucene的一个子项目，它主要用来进行网页数据的收集和索引。它用结合apache的Hadoop和Lucene等子项目。Nutch的一般抓取流程如下：
1. 把初始网址inject到crawlDb中进行准备抓取
2. 用generate模块对crawlDb中的网址进行过滤
3. 用fetcher 模块对generate产生的网页进行抓取
4. 用parse后的网页进行解析
5. 用crawlDb的update工具对抓取网页中的外链接进行更新，使之成为下一轮抓取的种子
本文主要介绍一下Nutch中Fetch流程。流程一般介绍如下：
Nutch的抓取方法是通过基于多队列的生产者与消费者线程模型。
/**
* A queue-based fetcher.
*
* <p>
* This fetcher uses a well-known model of one producer (a QueueFeeder) and many
* consumers (FetcherThread-s).
*
* <p>
* QueueFeeder reads input fetchlists and populates a set of FetchItemQueue-s,
* which hold FetchItem-s that describe the items to be fetched. There are as
* many queues as there are unique hosts, but at any given time the total number
* of fetch items in all queues is less than a fixed number (currently set to a
* multiple of the number of threads).
*
* <p>
* As items are consumed from the queues, the QueueFeeder continues to add new
* input items, so that their total count stays fixed (FetcherThread-s may also
* add new items to the queues e.g. as a results of redirection) - until all
* input items are exhausted, at which point the number of items in the queues
* begins to decrease. When this number reaches 0 fetcher will finish.
*
* <p>
* This fetcher implementation handles per-host blocking itself, instead of
* delegating this work to protocol-specific plugins. Each per-host queue
* handles its own "politeness" settings, such as the maximum number of
* concurrent requests and crawl delay between consecutive requests - and also a
* list of requests in progress, and the time the last request was finished. As
* FetcherThread-s ask for new items to be fetched, queues may return eligible
* items or null if for "politeness" reasons this host's queue is not yet ready.
*
* <p>
* If there are still unfetched items in the queues, but none of the items are
* ready, FetcherThread-s will spin-wait until either some items become
* available, or a timeout is reached (at which point the Fetcher will abort,
* assuming the task is hung).
*
* @author Andrzej Bialecki
*/
2. 抓取流程解析
在Nutch包的org.apache.nutch中有一个Fetcher.java，它就是用来对generate产生的网页进行抓取的。
2.1 Main函数
其中有一个Mian方法，主要有三处参数，segment,threads和noParsing
view plaincopy to clipboardprint?
1. String usage = "Usage: Fetcher <segment> [-threads n] [-noParsing]";
2. if (args.length < 1) {
3. System.err.println(usage);
4. System.exit(-1);
5. }
6. Path segment = new Path(args[0]);
7. Configuration conf = NutchConfiguration.create();
8. int threads = conf.getInt("fetcher.threads.fetch", 10);
9. boolean parsing = true;
10. for (int i = 1; i < args.length; i++) { // parse command line
11. if (args[i].equals("-threads")) { // found -threads option
12. threads = Integer.parseInt(args[++i]);
13. } else if (args[i].equals("-noParsing"))
14. parsing = false;
15. }
16. conf.setInt("fetcher.threads.fetch", threads);
17. if (!parsing) {
18. conf.setBoolean("fetcher.parse", parsing);
19. }
20. Fetcher fetcher = new Fetcher(conf); // make a Fetcher
21. fetcher.fetch(segment, threads, parsing); // run the Fetcher
2.2 Fecther类中的fetch方法
这个方法主要用于对MapReduce的一些初始设置和启动Fecther的run方法，主要代码如下：
view plaincopy to clipboardprint?
1.
2. // set input path and input format class
3. FileInputFormat.addInputPath(job, new Path(segment,
4. CrawlDatum.GENERATE_DIR_NAME));
5. job.setInputFormat(InputFormat.class); // 设置读入类，进行相应的Split操作，这个类在Fetcher中定义
6. // set map runnable class
7. job.setMapRunnerClass(Fetcher.class); // 定义MapRunner类，这个类是Fetcher类，它继承自MapRunnable，这里只进行Map，没有Reduce过程
8. // set output path and output format class
9. FileOutputFormat.setOutputPath(job, segment);
10. job.setOutputFormat(FetcherOutputFormat.class); // 设置输出处理类，它继承自OutputFormat<Text,NutchWritable>
11. // set output key and value class 设置输出的Key和Value的格式，这两个类都可以被序列化到文件系统上。除非你自己定义OutputFormat
12. job.setOutputKeyClass(Text.class);
13. job.setOutputValueClass(NutchWritable.class);
14. JobClient.runJob(job); // 提交任务，运行 Map
2.3 Fetcher中的run方法
主要是用于启动生产者－消费者这个线程模型，这里的生产者是QueueFeeder，用于收集input得到的数据(网页地址元信息)放到多个队列中去，这里的队列ID是用queueID = proto + "://" + host;协议类型和host来组成唯一的队列ID。
view plaincopy to clipboardprint?
1. // crawl datum feed thread that used to feed the queue from
2. // RecordReader. 生产者进行网页抓取数据的生成
3. feeder = new QueueFeeder(input, fetchQueues, threadCount * 50); // threadCount * 50 为队列容量
4. // feeder.setPriority((Thread.MAX_PRIORITY + Thread.NORM_PRIORITY) / 2);
5. feeder.start();
6. // set non-blocking & no-robots mode for HTTP protocol plugins.
7. getConf().setBoolean(Protocol.CHECK_BLOCKING, false);
8. getConf().setBoolean(Protocol.CHECK_ROBOTS, false);
9. // 生成消费者线程，从公共队列中取出数据来进行抓取
10. for (int i = 0; i < threadCount; i++) { // spawn threads 启动抓了线程
11. new FetcherThread(getConf()).start();
12. }
13. // select a timeout that avoids a task timeout
14. long timeout = getConf().getInt("mapred.task.timeout", 10 * 60 * 1000) / 2;
15. do { // wait for threads to exit 等待线程结束
16. try {
17. Thread.sleep(1000);
18. } catch (InterruptedException e) {
19. }
20. reportStatus();
21. LOG.info("-activeThreads=" + activeThreads + ", spinWaiting="
22. + spinWaiting.get() + ", fetchQueues.totalSize="
23. + fetchQueues.getTotalSize());
24. if (!feeder.isAlive() && fetchQueues.getTotalSize() < 5) {
25. fetchQueues.dump();
26. }
27. // some requests seem to hang, despite all intentions
28. if ((System.currentTimeMillis() - lastRequestStart.get()) > timeout) {
29. if (LOG.isWarnEnabled()) {
30. LOG.warn("Aborting with " + activeThreads
31. + " hung threads.");
32. }
33. return;
34. }
35. } while (activeThreads.get() > 0);
36. LOG.info("-activeThreads=" + activeThreads);
2.4 QueueFeeder的run方法
主要代码如下：
view plaincopy to clipboardprint?
1. while (hasMore) {
2. int feed = size - queues.getTotalSize();
3. if (feed <= 0) {
4. // queues are full - spin-wait until they have some free
5. // space
6. try {
7. Thread.sleep(1000);
8. } catch (Exception e) {
9. }
10. ;
11. continue;
12. } else {
13. LOG.debug("-feeding " + feed + " input urls ...");
14. // add feed numbers of fetch items to queue until the feed
15. // number less that 0
16. while (feed > 0 && hasMore) {
17. try {
18. Text url = new Text();
19. CrawlDatum datum = new CrawlDatum();
20. hasMore = reader.next(url, datum); // 这里是用Map的Input中读出Key和Value,并且判断是否读取成功
21. if (hasMore) {
22. queues.addFetchItem(url, datum); // 放入队列
23. cnt++;
24. feed--;
25. }
26. } catch (IOException e) {
27. LOG.fatal(
28. "QueueFeeder error reading input, record "
29. + cnt, e);
30. return;
31. }
32. }
33. }
34. }
2.5 FetcherThread的run方法
从fetchQueues中读出一个Item，对其进行抓取，如果抓取成功，就从fetchQueue队列中删除，如果不成功，就做相应的处理，这里的判断条件都是根据抓取协议的返回状态来做判断的。具体的抓取协议类型都是从Nutch的插件库中读出来的。
还有一点要注意的是FetchItemQueue有两个队列，一个是queue用于存储Item项，另一个是inProgress列队，用于存储正在被抓取的项，当要从queue队列中得到一个抓取项时，它会从queue队列中把待抓取项移出后放入inProgress队列中，如果当inProgress列队中项大于最大线程数时，就停止返回数据项，这样可以防止很多数据项在等待抓取。在FetchItemQueue中还有一个要注意的是它有一个nextFetchTime，它是有来控制抓取间隔的。
2.5 FetcherThread中的output方法
把抓取的数据写到output中，这就是Map的输出，输出格式就是前面job中定义的job.setOutputKeyClass(Text.class);job.setOutputValueClass(NutchWritable.class);两个方法。
3. 结束
本文只是对Nutch流程的一个简单的介绍，其中一些细节还没有还得急展开。如FetcherOutputFormat等类的使用，这个会在下一次整理好后发出。也希望有兴趣的同学一起来讨论