Capture and log the response body [duplicate]

This question already has answers here: How to read and copy the HTTP servlet response output stream content for logging (6 answers) 开发者_如何学Python Closed 6 years ago.

I have a servlet that handle certain HTTP requests and responses. I want to log the response body before sending back to the client. Is there any way that I can capture the response body before it is send as a HttpServletResponse object from the servlet?

---

If I understand you correctly, you want to log the response body? That's a pretty expensive task, but if that's the business requirement...

As @duffymo pointed, a Filter is a suitable place for this. You can capture the response body by replacing the passed-in ServletResponse with a HttpServletResponseWrapper implementation which replaces the HttpServletResponse#getWriter() with an own implementation which copies the response body into some buffer. After continuing the filter chain with the replaced response, just log the copy.

Here's a kickoff example how the doFilter() method can look like:

public void doFilter(ServletRequest request, final ServletResponse response, FilterChain chain) throws IOException, ServletException {
    final CopyPrintWriter writer = new CopyPrintWriter(response.getWriter());
    chain.doFilter(request, new HttpServletResponseWrapper((HttpServletResponse) response) {
        @Override public PrintWriter getWriter() {
            return writer;
        }
    });
    logger.log(writer.getCopy());
}

Here's how the CopyPrintWriter can look like:

public class CopyPrintWriter extends PrintWriter {

    private StringBuilder copy = new StringBuilder();

    public CopyPrintWriter(Writer writer) {
        super(writer);
    }

    @Override
    public void write(int c) {
        copy.append((char) c); // It is actually a char, not an int.
        super.write(c);
    }

    @Override
    public void write(char[] chars, int offset, int length) {
        copy.append(chars, offset, length);
        super.write(chars, offset, length);
    }

    @Override
    public void write(String string, int offset, int length) {
        copy.append(string, offset, length);
        super.write(string, offset, length);
    }

    public String getCopy() {
        return copy.toString();
    }

}

Map this filter on an url-pattern for which you'd like to log responses for. Keep in mind that binary/static content like images, CSS, JS files and so on won't be logged this way. You'd like to exclude them by using a specific enough url-pattern, e.g. *.jsp or just on the servlet-name of the servlet in question. If you want to log binary/static content anyway (for which I don't see any benefit), then you need to replace the HttpServletResponse#getOutputStream() the same way as well.

---

An alternative to BalusC answer Using the TeeOutputStream to write into two outputstreams at time.

public void doFilterInternal(HttpServletRequest req, HttpServletResponse res, FilterChain chain) throws IOException, ServletException {
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    final PrintStream ps = new PrintStream(baos);

    chain.doFilter(req,new HttpServletResponseWrapper(res) {
         @Override
         public ServletOutputStream getOutputStream() throws IOException {
            return new DelegatingServletOutputStream(new TeeOutputStream(super.getOutputStream(), ps)
            );
         }
         @Override
         public  PrintWriter getWriter() throws IOException {
            return new PrintWriter(new DelegatingServletOutputStream (new TeeOutputStream(super.getOutputStream(), ps))
            );
         }
      });

    //Get Response body calling baos.toString();
}

---

Maybe a servlet filter can help you. Think of it as aspect-oriented programming for HTTP.