优雅地记录http请求和响应的数据
经常会遇到需要处理http请求以及响应body的场景。而这里比较大的一个问题是servlet的requestBody或responseBody流一旦被读取了。就无法二次读取了。针对这个问题,spring本身提供了解决方案,即ContentCachingRequestWrapper/ContentCachingResponseWrapper。
我们编写一个过滤器:
public abstract class HttpBodyRecorderFilter extends OncePerRequestFilter {private static final int DEFAULT_MAX_PAYLOAD_LENGTH = 1024 * 512;private int maxPayloadLength = DEFAULT_MAX_PAYLOAD_LENGTH;protected void doFilterInternal(HttpServletRequest request, HttpServletResponse response, FilterChain filterChain)throws ServletException, IOException {boolean isFirstRequest = !isAsyncDispatch(request);HttpServletRequest requestToUse = request;if (isFirstRequest&& !(request instanceof ContentCachingRequestWrapper)&& (request.getMethod().equals(HttpMethod.PUT.name())|| request.getMethod().equals(HttpMethod.POST.name()))) {requestToUse = new ContentCachingRequestWrapper(request);}HttpServletResponse responseToUse = response;if (!(response instanceof ContentCachingResponseWrapper)&& (request.getMethod().equals(HttpMethod.PUT.name())|| request.getMethod().equals(HttpMethod.POST.name()))) {responseToUse = new ContentCachingResponseWrapper(response);}boolean hasException = false;try {filterChain.doFilter(requestToUse, responseToUse);} catch (final Exception e) {hasException = true;throw e;} finally {int code = hasException ? 500 : response.getStatus();if (!isAsyncStarted(requestToUse)&& (this.codeMatched(code, AdvancedHunterConfigManager.recordCode()))) {recordBody(createRequest(requestToUse), createResponse(responseToUse));} else {writeResponseBack(responseToUse);}}}protected String createRequest(HttpServletRequest request) {String payload = "";ContentCachingRequestWrapper wrapper =WebUtils.getNativeRequest(request, ContentCachingRequestWrapper.class);if (wrapper != null) {byte[] buf = wrapper.getContentAsByteArray();payload = genPayload(payload, buf, wrapper.getCharacterEncoding());}return payload;}protected String createResponse(HttpServletResponse resp) {String response = "";ContentCachingResponseWrapper wrapper = WebUtils.getNativeResponse(resp, ContentCachingResponseWrapper.class);if (wrapper != null) {byte[] buf = wrapper.getContentAsByteArray();try {wrapper.copyBodyToResponse();} catch (IOException e) {e.printStackTrace();}response = genPayload(response, buf, wrapper.getCharacterEncoding());}return response;}protected void writeResponseBack(HttpServletResponse resp) {ContentCachingResponseWrapper wrapper = WebUtils.getNativeResponse(resp, ContentCachingResponseWrapper.class);if (wrapper != null) {try {wrapper.copyBodyToResponse();} catch (IOException e) {LOG.error("Fail to write response body back", e);}}}private String genPayload(String payload, byte[] buf, String characterEncoding) {if (buf.length > 0 && buf.length < getMaxPayloadLength()) {try {payload = new String(buf, 0, buf.length, characterEncoding);} catch (UnsupportedEncodingException ex) {payload = "[unknown]";}}return payload;}public int getMaxPayloadLength() {return maxPayloadLength;}private boolean codeMatched(int responseStatus, String statusCode) {if (statusCode.matches("^[0-9,]*$")) {String[] filteredCode = statusCode.split(",");return Stream.of(filteredCode).map(Integer::parseInt).collect(Collectors.toList()).contains(responseStatus);} else {return false;}}protected abstract void recordBody(String payload, String response);protected abstract String recordCode();}
这样自定义一个filter继承HttpBodyRecorderFilter,重写recordBody方法就能自定义自己的处理逻辑了。另外,recordCode方法可用于定义在请求响应码为多少的时候才会去记录body,例如可以定义为只有遇到400或500时才记录body,用于错误侦测。
过滤器的匹配规则比较简单,如果想要像springmvc那样进行匹配,我们可以使用AntPathMatcher。
class PatternMappingFilterProxy implements Filter {private final Filter delegate;private final ListpathUrlPatterns = new ArrayList(); private PathMatcher pathMatcher;public PatternMappingFilterProxy(Filter delegate, String... urlPatterns) {Assert.notNull(delegate, "A delegate Filter is required");this.delegate = delegate;int length = urlPatterns.length;pathMatcher = new AntPathMatcher();for (int index = 0; index < length; ++index) {String urlPattern = urlPatterns[index];this.pathUrlPatterns.add(urlPattern);}}public void doFilter(ServletRequest request, ServletResponse response, FilterChain filterChain)throws IOException, ServletException {HttpServletRequest httpRequest = (HttpServletRequest) request;String path = httpRequest.getRequestURI();if (this.matches(path)) {this.delegate.doFilter(request, response, filterChain);} else {filterChain.doFilter(request, response);}}private boolean matches(String requestPath) {for (String pattern : pathUrlPatterns) {if (pathMatcher.match(pattern, requestPath)) {return true;}}return false;}public void init(FilterConfig filterConfig) throws ServletException {this.delegate.init(filterConfig);}public void destroy() {this.delegate.destroy();}public ListgetPathUrlPatterns() {return pathUrlPatterns;}public void setPathUrlPatterns(ListurlPatterns) {pathUrlPatterns.clear();pathUrlPatterns.addAll(urlPatterns);}}
这样子,PatternMappingFilterProxy装饰了真正的HttpBodyRecorderFilter,支持传入urlPatterns,从而实现像springmvc那样的ant style的匹配。例如对于以下接口:
public Object test( final Integer index) {//do something}
可以设置urlPattern为/test/{id:[0-9]+}。

腾讯、阿里、滴滴后台面试题汇总总结 — (含答案)
面试:史上最全多线程面试题 !
最新阿里内推Java后端面试题
JVM难学?那是因为你没认真看完这篇文章

关注作者微信公众号 —《JAVA烂猪皮》
了解更多java后端架构知识以及最新面试宝典


看完本文记得给作者点赞+在看哦~~~大家的支持,是作者源源不断出文的动力
作者:fredalxin
地址:https://fredal.xin/http-body-recorder
评论
