使用httpclient返回 header too long

最近些的爬虫请求某代理网站时运行几天就会持续报错400,header too long，google无答案，于是看了下源码，主要原因为cookie的累积导致（可以理解为你的浏览器很久没有清理缓存），以下为排查过程，解决方案见文章最后。

httclient请求调用链路：

org.apache.http.impl.client.InternalHttpClient#doExecute

org.apache.http.impl.client.InternalHttpClient#setupContext

if (context.getAttribute(HttpClientContext.COOKIE_STORE) == null) {
    context.setAttribute(HttpClientContext.COOKIE_STORE, this.cookieStore);
}

 
如果没有显示设置cooki_store，取this(httpclient)的成员变量cookieStore，而通常我们httpClient只有1个实例，那么cookieStore也等同于是单例的。

org.apache.http.impl.execchain.RedirectExec#execute:108

org.apache.http.impl.execchain.RetryExec#execute:86

org.apache.http.impl.execchain.ProtocolExec#execute

org.apache.http.impl.execchain.MainClientExec#execute

org.apache.http.protocol.HttpRequestExecutor#execute

conn.sendRequestHeader(request);

 org.apache.http.protocol.HttpRequestExecutor#doSendRequest

org.apache.http.impl.conn.CPoolProxy#sendRequestHeader

org.apache.http.impl.io.AbstractMessageWriter#write

Cookie的发送：

for (final HeaderIterator it = message.headerIterator(); it.hasNext(); ) {
final Header header = it.nextHeader();
    this.sessionBuffer.writeLine
        (lineFormatter.formatHeader(this.lineBuf, header));
}

 

Cookie的保存：

org.apache.http.impl.execchain.ProtocolExec#execute:200

org.apache.http.HttpResponseInterceptor#process

org.apache.http.client.protocol.ResponseProcessCookies#process

org.apache.http.client.protocol.ResponseProcessCookies#processCookies:114

cookieStore.addCookie(cookie);

 

/**
 * Adds an {@link Cookie HTTP cookie}, replacing any existing equivalent cookies.
 * If the given cookie has already expired it will not be added, but existing
 * values will still be removed.
 *
 * @param cookie the {@link Cookie cookie} to be added
 *
 * @see #addCookies(Cookie[])
 *
 */
public synchronized void addCookie(final Cookie cookie) {
if (cookie != null) {
// first remove any old cookie that is equivalent
cookies.remove(cookie);
        if (!cookie.isExpired(new Date())) {
cookies.add(cookie);
}
    }
}

 

注意到这里添加一个cookie时会先移除，然后判断cookie是否已经失效，没有失效才会add，这样看是不会出问题的，那问题到底出在哪里？而通过调试发现我们的第三方网站的sessionID的cookie的name居然是会变的！导致老的cookie无法删除，越积越多。

解决方案①：禁用cookie

CloseableHttpClient httpClient = HttpClientBuilder.create().setConnectionManager(connManager)
                              .setRetryHandler(retryHandler).setDefaultRequestConfig(config).disableCookieManagement().build();

 disableCookieManagement()方法会停止发送和接收cookie。

/**
 * Disables state (cookie) management.
 * <p/>
* Please note this value can be overridden by the {@link #setHttpProcessor(
 * org.apache.http.protocol.HttpProcessor)} method.
 */
public final HttpClientBuilder disableCookieManagement() {
this.cookieManagementDisabled = true;
    return this;
}

 

启用后org.apache.http.protocol.ImmutableHttpProcessor#responseInterceptors response拦截器里不再包含ResponseProcessCookies这一拦截器，不再执行存储cookie操作，观察后续请求，header里也不再包含cookie字段。

拦截器注册代码：org.apache.http.impl.client.HttpClientBuilder#build:839

if (!cookieManagementDisabled) {
    b.add(new RequestAddCookies());
}
if (!cookieManagementDisabled) {
    b.add(new ResponseProcessCookies());
}

 

解放方案②：设置单独的context

HttpClientContext context = HttpClientContext.create();
context.setCookieStore(new BasicCookieStore());
CloseableHttpResponse response = httpClient.execute(httpGet, context);

 

设置后因为context的cookieStore不为null，将不再默认取httpclient的成员变量cookiestore。

以上两种方案，可根据自身情况进行选择。

参考文档：http://hc.apache.org/httpcomponents-client-4.3.x/tutorial/html/statemgmt.html#d5e565