学步园

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

String usage = "Usage: Fetcher <segment> [-threads n] [-noParsing]";<br /> if (args.length < 1) {<br /> System.err.println(usage);<br /> System.exit(-1);<br /> }<br /> Path segment = new Path(args[0]);<br /> Configuration conf = NutchConfiguration.create();<br /> int threads = conf.getInt("fetcher.threads.fetch", 10);<br /> boolean parsing = true;<br /> for (int i = 1; i < args.length; i++) { // parse command line<br /> if (args[i].equals("-threads")) { // found -threads option<br /> threads = Integer.parseInt(args[++i]);<br /> } else if (args[i].equals("-noParsing"))<br /> parsing = false;<br /> }<br /> conf.setInt("fetcher.threads.fetch", threads);<br /> if (!parsing) {<br /> conf.setBoolean("fetcher.parse", parsing);<br /> }<br /> Fetcher fetcher = new Fetcher(conf); // make a Fetcher<br /> 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：

 <br /> // set input path and input format class<br /> FileInputFormat.addInputPath(job, new Path(segment,<br /> CrawlDatum.GENERATE_DIR_NAME));<br /> 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<br /> // set map runnable class<br /> 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<br /> // set output path and output format class<br /> FileOutputFormat.setOutputPath(job, segment);<br /> job.setOutputFormat(FetcherOutputFormat.class); // 设置输出处理类，它继承自OutputFormat<Text,NutchWritable><br /> // 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<br /> job.setOutputKeyClass(Text.class);<br /> job.setOutputValueClass(NutchWritable.class);<br /> 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。

// crawl datum feed thread that used to feed the queue from<br /> // RecordReader. 生产者进行网页抓取数据的生成<br /> feeder = new QueueFeeder(input, fetchQueues, threadCount * 50); // threadCount *　50 为队列容量<br /> // feeder.setPriority((Thread.MAX_PRIORITY + Thread.NORM_PRIORITY) / 2);<br /> feeder.start();<br /> // set non-blocking & no-robots mode for HTTP protocol plugins.<br /> getConf().setBoolean(Protocol.CHECK_BLOCKING, false);<br /> getConf().setBoolean(Protocol.CHECK_ROBOTS, false);<br /> // 生成消费者线程，从公共队列中取出数据来进行抓取<br /> for (int i = 0; i < threadCount; i++) { // spawn threads 启动抓了线程<br /> new FetcherThread(getConf()).start();<br /> }<br /> // select a timeout that avoids a task timeout<br /> long timeout = getConf().getInt("mapred.task.timeout", 10 * 60 * 1000) / 2;<br /> do { // wait for threads to exit 等待线程结束<br /> try {<br /> Thread.sleep(1000);<br /> } catch (InterruptedException e) {<br /> }<br /> reportStatus();<br /> LOG.info("-activeThreads=" + activeThreads + ", spinWaiting="<br /> + spinWaiting.get() + ", fetchQueues.totalSize="<br /> + fetchQueues.getTotalSize());<br /> if (!feeder.isAlive() && fetchQueues.getTotalSize() < 5) {<br /> fetchQueues.dump();<br /> }<br /> // some requests seem to hang, despite all intentions<br /> if ((System.currentTimeMillis() - lastRequestStart.get()) > timeout) {<br /> if (LOG.isWarnEnabled()) {<br /> LOG.warn("Aborting with " + activeThreads<br /> + " hung threads.");<br /> }<br /> return;<br /> }<br /> } while (activeThreads.get() > 0);<br /> LOG.info("-activeThreads=" + activeThreads); 

2.4 QueueFeeder的run方法

主要代码如下：

while (hasMore) {<br /> int feed = size - queues.getTotalSize();<br /> if (feed <= 0) {<br /> // queues are full - spin-wait until they have some free<br /> // space<br /> try {<br /> Thread.sleep(1000);<br /> } catch (Exception e) {<br /> }<br /> ;<br /> continue;<br /> } else {<br /> LOG.debug("-feeding " + feed + " input urls ...");<br /> // add feed numbers of fetch items to queue until the feed<br /> // number less that 0<br /> while (feed > 0 && hasMore) {<br /> try {<br /> Text url = new Text();<br /> CrawlDatum datum = new CrawlDatum();<br /> 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<br /> if (hasMore) {<br /> queues.addFetchItem(url, datum); // 放入队列<br /> cnt++;<br /> feed--;<br /> }<br /> } catch (IOException e) {<br /> LOG.fatal(<br /> "QueueFeeder error reading input, record "<br /> + cnt, e);<br /> return;<br /> }<br /> }<br /> }<br /> } 

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。