import org.apache.lucene.benchmark.byTask.feeds.EnwikiContentSource; //导入依赖的package包/类
public static void main(String[] args) throws Exception {

    File wikipedia = null;
    File outputDir = new File("./enwiki");
    boolean keepImageOnlyDocs = true;
    for (int i = 0; i < args.length; i++) {
      String arg = args[i];
      if (arg.equals("--input") || arg.equals("-i")) {
        wikipedia = new File(args[i + 1]);
        i++;
      } else if (arg.equals("--output") || arg.equals("-o")) {
        outputDir = new File(args[i + 1]);
        i++;
      } else if (arg.equals("--discardImageOnlyDocs") || arg.equals("-d")) {
        keepImageOnlyDocs = false;
      }
    }
    
    Properties properties = new Properties();
    properties.setProperty("docs.file", wikipedia.getAbsolutePath());
    properties.setProperty("content.source.forever", "false");
    properties.setProperty("keep.image.only.docs", String.valueOf(keepImageOnlyDocs));
    Config config = new Config(properties);

    ContentSource source = new EnwikiContentSource();
    source.setConfig(config);
    
    DocMaker docMaker = new DocMaker();
    docMaker.setConfig(config, source);
    docMaker.resetInputs();
    if (wikipedia.exists()) {
      System.out.println("Extracting Wikipedia to: " + outputDir + " using EnwikiContentSource");
      outputDir.mkdirs();
      ExtractWikipedia extractor = new ExtractWikipedia(docMaker, outputDir);
      extractor.extract();
    } else {
      printUsage();
    }
  }