Method run

src/java/org/apache/nutch/parse/ParserChecker.java:77–204

(String[] args)

  }

  public int run(String[] args) throws Exception {
    boolean dumpText = false;
    boolean force = false;
    String contentType = null;
    String url = null;

    String usage = "Usage: ParserChecker [-dumpText] [-forceAs mimeType] url";

    if (args.length == 0) {
      LOG.error(usage);
      return (-1);
    }

    for (int i = 0; i < args.length; i++) {
      if (args[i].equals("-forceAs")) {
        force = true;
        contentType = args[++i];
      } else if (args[i].equals("-dumpText")) {
        dumpText = true;
      } else if (i != args.length - 1) {
        LOG.error(usage);
        System.exit(-1);
      } else {
        url = URLUtil.toASCII(args[i]);
      }
    }

    if (LOG.isInfoEnabled()) {
      LOG.info("fetching: " + url);
    }

    ProtocolFactory factory = new ProtocolFactory(conf);
    Protocol protocol = factory.getProtocol(url);
    WebPage page = WebPage.newBuilder().build();

    ProtocolOutput protocolOutput = protocol.getProtocolOutput(url, page);

    if (!protocolOutput.getStatus().isSuccess()) {
      LOG.error("Fetch failed with protocol status: "
          + ProtocolStatusUtils.getName(protocolOutput.getStatus().getCode())
          + ": " + ProtocolStatusUtils.getMessage(protocolOutput.getStatus()));
      return (-1);
    }
    Content content = protocolOutput.getContent();

    if (content == null) {
      LOG.error("No content for " + url);
      return (-1);
    }
    page.setBaseUrl(new org.apache.avro.util.Utf8(url));
    page.setContent(ByteBuffer.wrap(content.getContent()));

    if (force) {
      content.setContentType(contentType);
    } else {
      contentType = content.getContentType();
    }