
org.tinymediamanager.scraper.http.TmmHttpLoggingInterceptor Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of api-scraper Show documentation
Show all versions of api-scraper Show documentation
API for tinyMediaManager scrapers
The newest version!
/*
* Copyright 2012 - 2019 Manuel Laggner
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.tinymediamanager.scraper.http;
import java.io.EOFException;
import java.io.IOException;
import java.nio.charset.Charset;
import java.util.concurrent.TimeUnit;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import okhttp3.Connection;
import okhttp3.Headers;
import okhttp3.Interceptor;
import okhttp3.MediaType;
import okhttp3.Request;
import okhttp3.RequestBody;
import okhttp3.Response;
import okhttp3.ResponseBody;
import okhttp3.internal.http.HttpHeaders;
import okio.Buffer;
import okio.BufferedSource;
import okio.GzipSource;
/**
* This class is used to provide a logging interceptor for tinyMediaManager which is able to log request headers/responses and the body of text
* responses at trace logging level
*
* @author Manuel Laggner
*/
public class TmmHttpLoggingInterceptor implements Interceptor {
private static final Logger LOGGER = LoggerFactory.getLogger(TmmHttpLoggingInterceptor.class);
private static final Charset UTF8 = Charset.forName("UTF-8");
@Override
public Response intercept(Chain chain) throws IOException {
Request request = chain.request();
try {
RequestBody requestBody = request.body();
boolean hasRequestBody = requestBody != null;
Connection connection = chain.connection();
String requestStartMessage = "--> " + request.method() + ' ' + request.url() + (connection != null ? " " + connection.protocol() : "");
LOGGER.trace(requestStartMessage);
if (hasRequestBody) {
// Request body headers are only present when installed as a network interceptor. Force
// them to be included (when available) so there values are known.
if (requestBody.contentType() != null) {
LOGGER.trace("Content-Type: {}", requestBody.contentType());
}
if (requestBody.contentLength() != -1) {
LOGGER.trace("Content-Length: {}", requestBody.contentLength());
}
}
Headers headersRequest = request.headers();
for (int i = 0, count = headersRequest.size(); i < count; i++) {
String name = headersRequest.name(i);
// Skip headers from the request body as they are explicitly logged above.
if (!"Content-Type".equalsIgnoreCase(name) && !"Content-Length".equalsIgnoreCase(name)) {
LOGGER.trace("{} : {}", headersRequest.name(i), headersRequest.value(i));
}
}
if (!hasRequestBody || bodyHasUnknownEncoding(request.headers())) {
LOGGER.trace("--> END {}", request.method());
}
else {
Buffer buffer = new Buffer();
requestBody.writeTo(buffer);
Charset charset = UTF8;
MediaType contentType = requestBody.contentType();
if (contentType != null) {
charset = contentType.charset(UTF8);
}
LOGGER.trace("");
if (isPlaintext(buffer)) {
String content = buffer.readString(charset);
// only log the first 10k characters
if (content.length() > 10000) {
LOGGER.trace("{}...", content.substring(0, 10000)); // NOSONAR
}
else {
LOGGER.trace(content);
}
LOGGER.trace("--> END {} ({}-byte body)", request.method(), requestBody.contentLength());
}
else {
LOGGER.trace("--> END {}", request.method());
}
}
}
catch (Exception e) {
LOGGER.error("Problem in HTTP logging detected: {}", e.getMessage());
}
long startNs = System.nanoTime();
Response response;
try {
response = chain.proceed(request);
}
catch (Exception e) {
LOGGER.trace("<-- HTTP FAILED: {}", e.getMessage());
throw e;
}
long tookMs = TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - startNs);
try {
ResponseBody responseBody = response.body();
long contentLength = responseBody.contentLength();
String bodySize = contentLength != -1 ? contentLength + "-byte" : "unknown-length";
LOGGER.debug("<-- " + response.code() + (response.message().isEmpty() ? "" : ' ' + response.message()) + ' ' + response.request().url() + " ("
+ tookMs + "ms" + ", " + bodySize + " body" + ')');
Headers headersResponse = response.headers();
for (int i = 0, count = headersResponse.size(); i < count; i++) {
LOGGER.trace("{} : {}", headersResponse.name(i), headersResponse.value(i));
}
if (!HttpHeaders.hasBody(response) || bodyHasUnknownEncoding(response.headers())) {
LOGGER.trace("<-- END HTTP");
}
else if (isTextResponse(response)) {
BufferedSource source = responseBody.source();
source.request(Long.MAX_VALUE); // Buffer the entire body.
Buffer buffer = source.buffer();
Long gzippedLength = null;
if ("gzip".equalsIgnoreCase(headersResponse.get("Content-Encoding"))) {
gzippedLength = buffer.size();
GzipSource gzippedResponseBody = null;
try {
gzippedResponseBody = new GzipSource(buffer.clone());
buffer = new Buffer();
buffer.writeAll(gzippedResponseBody);
}
finally {
if (gzippedResponseBody != null) {
gzippedResponseBody.close();
}
}
}
Charset charset = UTF8;
MediaType contentType = responseBody.contentType();
if (contentType != null) {
charset = contentType.charset(UTF8);
}
if (!isPlaintext(buffer)) {
LOGGER.trace("");
LOGGER.trace("<-- END HTTP (binary {}-byte body omitted)", buffer.size());
return response;
}
if (contentLength != 0) {
LOGGER.trace("");
String content = buffer.clone().readString(charset);
// only log the first 10k characters
if (content.length() > 10000) {
LOGGER.trace("{}...", content.substring(0, 10000)); // NOSONAR
}
else {
LOGGER.trace(content);
}
}
if (gzippedLength != null) {
LOGGER.trace("<-- END HTTP ({}-byte, {}-gzipped-byte body)", buffer.size(), gzippedLength);
}
else {
LOGGER.trace("<-- END HTTP ({}-byte body)", buffer.size());
}
}
}
catch (Exception e) {
LOGGER.error("Problem in HTTP logging detected: {}", e.getMessage());
}
return response;
}
private static boolean bodyHasUnknownEncoding(Headers headers) {
String contentEncoding = headers.get("Content-Encoding");
return contentEncoding != null && !contentEncoding.equalsIgnoreCase("identity") && !contentEncoding.equalsIgnoreCase("gzip");
}
private static boolean isTextResponse(Response response) {
MediaType type = response.body().contentType();
if ("text".equalsIgnoreCase(type.type()) || "xml".equalsIgnoreCase(type.subtype()) || "json".equalsIgnoreCase(type.subtype())) {
return true;
}
return false;
}
/**
* Returns true if the body in question probably contains human readable text. Uses a small sample of code points to detect unicode control
* characters commonly used in binary file signatures.
*/
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.
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy