背景

在使用Android retrofit+rxjava時，想獲知網路請求的一些引數，方便除錯，比如：請求地址、請求響應時間、請求響應訊息體等內容，雖然部分可以通過每個介面進行獲知，但是這樣極其不方便，那麼有沒有可以統一設定的方法呢？請接下去看。

日誌攔截器

retrofit是使用okhttp3，做為網路請求，okhttp3有個Interceptor介面，可以對請求和響應進行攔截。通過這個機制，我們可以設計自己的一套日誌攔截器，列印一些必要的資訊。

• 自定義攔截器

  首先要自己定義一個攔截器類，並實現Interceptor介面，在裡面列印一些資訊，如下LogInterceptor.java類。

public class LogInterceptor implements Interceptor{
    public static final String TAG = "LogInterceptor.java";
    @Override
    public Response intercept(Chain chain) throws IOException {
        Request request = chain.request();
        //the request url
        String url = request.url().toString();
        //the request method
 

        String method = request.method();
        long t1 = System.nanoTime();
        HLog.d(TAG,String.format(Locale.getDefault(),"Sending %s request [url = %s]",method,url));
        //the request body
        RequestBody requestBody = request.body();
        if(requestBody!= null) {
            StringBuilder sb = new
 
 StringBuilder("Request Body [");
            okio.Buffer buffer = new okio.Buffer();
            requestBody.writeTo(buffer);
            Charset charset = Charset.forName("UTF-8");
            MediaType contentType = requestBody.contentType();
            if (contentType != null) {
                charset = contentType.charset(charset);
            }
            if(isPlaintext(buffer)){
                sb.append(buffer.readString(charset));
                sb.append(" (Content-Type = ").append(contentType.toString()).append(",")
                        .append(requestBody.contentLength()).append("-byte body)");
            }else {
                sb.append(" (Content-Type = ").append(contentType.toString())
                        .append(",binary ").append(requestBody.contentLength()).append("-byte body omitted)");
            }
            sb.append("]");
            HLog.d(TAG, String.format(Locale.getDefault(), "%s %s", method, sb.toString()));
        }
        Response response = chain.proceed(request);
        long t2 = System.nanoTime();
        //the response time
        HLog.d(TAG,String.format(Locale.getDefault(),"Received response for [url = %s] in %.1fms",url, (t2-t1)/1e6d));

        //the response state
        HLog.d(TAG,String.format(Locale.CHINA,"Received response is %s ,message[%s],code[%d]",response.isSuccessful()?"success":"fail",response.message(),response.code()));

        //the response data
        ResponseBody body = response.body();

        BufferedSource source = body.source();
        source.request(Long.MAX_VALUE); // Buffer the entire body.
        Buffer buffer = source.buffer();
        Charset charset = Charset.defaultCharset();
        MediaType contentType = body.contentType();
        if (contentType != null) {
            charset = contentType.charset(charset);
        }
        String bodyString = buffer.clone().readString(charset);
        HLog.d(TAG,String.format("Received response json string [%s]",bodyString));

        return response;
    }

    static boolean isPlaintext(Buffer buffer){
        try {
            Buffer prefix = new Buffer();
            long byteCount = buffer.size() < 64 ? buffer.size() : 64;
            buffer.copyTo(prefix, 0, byteCount);
            for (int i = 0; i < 16; i++) {
                if (prefix.exhausted()) {
                    break;
                }
                int codePoint = prefix.readUtf8CodePoint();
                if (Character.isISOControl(codePoint) && !Character.isWhitespace(codePoint)) {
                    return false;
                }
            }
            return true;
        } catch (EOFException e) {
            return false; // Truncated UTF-8 sequence.
        }
    }

}

在上面的實現中，列印了請求的url、請求方法（get\post\put\delete…）、響應時間、響應狀態、響應內容等。這裡需要主要的是，在列印響應內容體的時候，或許你會遇到下面這個異常：

 java.lang.IllegalStateException: closed

這是因為你列印的時候直接呼叫了ResponseBody.string()，我們看下原始碼的實現

public final String string() throws IOException {
    return new String(bytes(), charset().name());
}

它呼叫了bytes(),我在接著看下

public final byte[] bytes() throws IOException {
    long contentLength = contentLength();
    if (contentLength > Integer.MAX_VALUE) {
      throw new IOException("Cannot buffer entire body for content length: " + contentLength);
    }

    BufferedSource source = source();
    byte[] bytes;
    try {
      bytes = source.readByteArray();
    } finally {
      Util.closeQuietly(source);
    }
    if (contentLength != -1 && contentLength != bytes.length) {
      throw new IOException("Content-Length and stream length disagree");
    }
    return bytes;
  }

看到finally部分，它已經將 BufferedSource 關閉掉了，這樣到我們自己介面真正要處理的時候，它就會報錯了，已經關閉的緩衝區就不能操作了。
那麼上面LogInterceptor實際上是clone了一份BufferedSource 進行操作，這樣不影響原來的BufferedSource。
也可以使用下面第二種方法，原理相同：

ResponseBody originalBody = response.body();
ResponseBody body = response.peekBody(originalBody.contentLength());
bodyString = body.string();

• 給okhttp設定攔截器

OkHttpClient.Builder builder = new OkHttpClient().newBuilder();
builder.addNetworkInterceptor(new NetworkInterceptor());
if(BuildConfig.DEBUG){
    //日誌攔截器
    builder.addNetworkInterceptor(new LogInterceptor());
}
OkHttpClient client = builder.build();
retrofit = new Retrofit.Builder()
        .baseUrl(APIService.BASE_URL)
        .addConverterFactory(GsonConverterFactory.create())
        .addCallAdapterFactory(RxJavaCallAdapterFactory.create())
        .client(client)
        .build();

相關Exception

 java.lang.IllegalStateException: closed