Class Site

webmagic-core/src/main/java/us/codecraft/webmagic/Site.java:19–433

Object contains setting for crawler. @author code4crafter@gmail.com @see us.codecraft.webmagic.processor.PageProcessor @since 0.1.0

 * @since 0.1.0
 */
public class Site {

    private String domain;

    private String userAgent;

    private Map<String, String> defaultCookies = new LinkedHashMap<String, String>();

    private Map<String, Map<String, String>> cookies = new HashMap<String, Map<String, String>>();

    private String charset;

    private String defaultCharset;

    private int sleepTime = 5000;

    private int retryTimes = 0;

    private int cycleRetryTimes = 0;

    private int retrySleepTime = 1000;

    private int timeOut = 5000;

    private static final Set<Integer> DEFAULT_STATUS_CODE_SET = new HashSet<Integer>();

    private Set<Integer> acceptStatCode = DEFAULT_STATUS_CODE_SET;

    private Map<String, String> headers = new HashMap<String, String>();

    private boolean useGzip = true;

    private boolean disableCookieManagement = false;

    static {
        DEFAULT_STATUS_CODE_SET.add(HttpConstant.StatusCode.CODE_200);
    }

    /**
     * new a Site
     *
     * @return new site
     */
    public static Site me() {
        return new Site();
    }

    /**
     * Add a cookie with domain {@link #getDomain()}
     *
     * @param name name
     * @param value value
     * @return this
     */
    public Site addCookie(String name, String value) {
        defaultCookies.put(name, value);
        return this;
    }