
org.opensearch.rest.RestRequest Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of opensearch Show documentation
Show all versions of opensearch Show documentation
OpenSearch subproject :server
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/
/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch licenses this file to you 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.
*/
/*
* Modifications Copyright OpenSearch Contributors. See
* GitHub history for details.
*/
package org.opensearch.rest;
import org.apache.lucene.util.SetOnce;
import org.opensearch.OpenSearchParseException;
import org.opensearch.common.Booleans;
import org.opensearch.common.CheckedConsumer;
import org.opensearch.common.Nullable;
import org.opensearch.common.Strings;
import org.opensearch.common.bytes.BytesArray;
import org.opensearch.common.bytes.BytesReference;
import org.opensearch.common.collect.Tuple;
import org.opensearch.common.unit.ByteSizeValue;
import org.opensearch.common.unit.TimeValue;
import org.opensearch.common.xcontent.LoggingDeprecationHandler;
import org.opensearch.common.xcontent.NamedXContentRegistry;
import org.opensearch.common.xcontent.ToXContent;
import org.opensearch.common.xcontent.XContentParser;
import org.opensearch.common.xcontent.XContentType;
import org.opensearch.http.HttpChannel;
import org.opensearch.http.HttpRequest;
import java.io.IOException;
import java.io.InputStream;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.atomic.AtomicLong;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import static org.opensearch.common.unit.ByteSizeValue.parseBytesSizeValue;
import static org.opensearch.common.unit.TimeValue.parseTimeValue;
public class RestRequest implements ToXContent.Params {
// tchar pattern as defined by RFC7230 section 3.2.6
private static final Pattern TCHAR_PATTERN = Pattern.compile("[a-zA-z0-9!#$%&'*+\\-.\\^_`|~]+");
private static final AtomicLong requestIdGenerator = new AtomicLong();
private final NamedXContentRegistry xContentRegistry;
private final Map params;
private final Map> headers;
private final String rawPath;
private final Set consumedParams = new HashSet<>();
private final SetOnce xContentType = new SetOnce<>();
private final HttpChannel httpChannel;
private HttpRequest httpRequest;
private boolean contentConsumed = false;
private final long requestId;
public boolean isContentConsumed() {
return contentConsumed;
}
protected RestRequest(
NamedXContentRegistry xContentRegistry,
Map params,
String path,
Map> headers,
HttpRequest httpRequest,
HttpChannel httpChannel
) {
this(xContentRegistry, params, path, headers, httpRequest, httpChannel, requestIdGenerator.incrementAndGet());
}
private RestRequest(
NamedXContentRegistry xContentRegistry,
Map params,
String path,
Map> headers,
HttpRequest httpRequest,
HttpChannel httpChannel,
long requestId
) {
final XContentType xContentType;
try {
xContentType = parseContentType(headers.get("Content-Type"));
} catch (final IllegalArgumentException e) {
throw new ContentTypeHeaderException(e);
}
if (xContentType != null) {
this.xContentType.set(xContentType);
}
this.xContentRegistry = xContentRegistry;
this.httpRequest = httpRequest;
this.httpChannel = httpChannel;
this.params = params;
this.rawPath = path;
this.headers = Collections.unmodifiableMap(headers);
this.requestId = requestId;
}
protected RestRequest(RestRequest restRequest) {
this(
restRequest.getXContentRegistry(),
restRequest.params(),
restRequest.path(),
restRequest.getHeaders(),
restRequest.getHttpRequest(),
restRequest.getHttpChannel(),
restRequest.getRequestId()
);
}
/**
* Invoke {@link HttpRequest#releaseAndCopy()} on the http request in this instance and replace a pooled http request
* with an unpooled copy. This is supposed to be used before passing requests to {@link RestHandler} instances that can not safely
* handle http requests that use pooled buffers as determined by {@link RestHandler#allowsUnsafeBuffers()}.
*/
void ensureSafeBuffers() {
httpRequest = httpRequest.releaseAndCopy();
}
/**
* Creates a new REST request. This method will throw {@link BadParameterException} if the path cannot be
* decoded
*
* @param xContentRegistry the content registry
* @param httpRequest the http request
* @param httpChannel the http channel
* @throws BadParameterException if the parameters can not be decoded
* @throws ContentTypeHeaderException if the Content-Type header can not be parsed
*/
public static RestRequest request(NamedXContentRegistry xContentRegistry, HttpRequest httpRequest, HttpChannel httpChannel) {
Map params = params(httpRequest.uri());
String path = path(httpRequest.uri());
return new RestRequest(
xContentRegistry,
params,
path,
httpRequest.getHeaders(),
httpRequest,
httpChannel,
requestIdGenerator.incrementAndGet()
);
}
private static Map params(final String uri) {
final Map params = new HashMap<>();
int index = uri.indexOf('?');
if (index >= 0) {
try {
RestUtils.decodeQueryString(uri, index + 1, params);
} catch (final IllegalArgumentException e) {
throw new BadParameterException(e);
}
}
return params;
}
private static String path(final String uri) {
final int index = uri.indexOf('?');
if (index >= 0) {
return uri.substring(0, index);
} else {
return uri;
}
}
/**
* Creates a new REST request. The path is not decoded so this constructor will not throw a
* {@link BadParameterException}.
*
* @param xContentRegistry the content registry
* @param httpRequest the http request
* @param httpChannel the http channel
* @throws ContentTypeHeaderException if the Content-Type header can not be parsed
*/
public static RestRequest requestWithoutParameters(
NamedXContentRegistry xContentRegistry,
HttpRequest httpRequest,
HttpChannel httpChannel
) {
Map params = Collections.emptyMap();
return new RestRequest(
xContentRegistry,
params,
httpRequest.uri(),
httpRequest.getHeaders(),
httpRequest,
httpChannel,
requestIdGenerator.incrementAndGet()
);
}
public enum Method {
GET,
POST,
PUT,
DELETE,
OPTIONS,
HEAD,
PATCH,
TRACE,
CONNECT
}
/**
* Returns the HTTP method used in the REST request.
*
* @return the {@link Method} used in the REST request
* @throws IllegalArgumentException if the HTTP method is invalid
*/
public Method method() {
return httpRequest.method();
}
/**
* The uri of the rest request, with the query string.
*/
public String uri() {
return httpRequest.uri();
}
/**
* The non decoded, raw path provided.
*/
public String rawPath() {
return rawPath;
}
/**
* The path part of the URI (without the query string), decoded.
*/
public final String path() {
return RestUtils.decodeComponent(rawPath());
}
public boolean hasContent() {
return content(false).length() > 0;
}
public BytesReference content() {
return content(true);
}
protected BytesReference content(final boolean contentConsumed) {
this.contentConsumed = this.contentConsumed | contentConsumed;
return httpRequest.content();
}
/**
* @return content of the request body or throw an exception if the body or content type is missing
*/
public final BytesReference requiredContent() {
if (hasContent() == false) {
throw new OpenSearchParseException("request body is required");
} else if (xContentType.get() == null) {
throw new IllegalStateException("unknown content type");
}
return content();
}
/**
* Get the value of the header or {@code null} if not found. This method only retrieves the first header value if multiple values are
* sent. Use of {@link #getAllHeaderValues(String)} should be preferred
*/
public final String header(String name) {
List values = headers.get(name);
if (values != null && values.isEmpty() == false) {
return values.get(0);
}
return null;
}
/**
* Get all values for the header or {@code null} if the header was not found
*/
public final List getAllHeaderValues(String name) {
List values = headers.get(name);
if (values != null) {
return Collections.unmodifiableList(values);
}
return null;
}
/**
* Get all of the headers and values associated with the headers. Modifications of this map are not supported.
*/
public final Map> getHeaders() {
return headers;
}
public final long getRequestId() {
return requestId;
}
/**
* The {@link XContentType} that was parsed from the {@code Content-Type} header. This value will be {@code null} in the case of
* a request without a valid {@code Content-Type} header, a request without content ({@link #hasContent()}, or a plain text request
*/
@Nullable
public final XContentType getXContentType() {
return xContentType.get();
}
public HttpChannel getHttpChannel() {
return httpChannel;
}
public HttpRequest getHttpRequest() {
return httpRequest;
}
public final boolean hasParam(String key) {
return params.containsKey(key);
}
@Override
public final String param(String key) {
consumedParams.add(key);
return params.get(key);
}
@Override
public final String param(String key, String defaultValue) {
consumedParams.add(key);
String value = params.get(key);
if (value == null) {
return defaultValue;
}
return value;
}
public Map params() {
return params;
}
/**
* Returns a list of parameters that have been consumed. This method returns a copy, callers
* are free to modify the returned list.
*
* @return the list of currently consumed parameters.
*/
List consumedParams() {
return new ArrayList<>(consumedParams);
}
/**
* Returns a list of parameters that have not yet been consumed. This method returns a copy,
* callers are free to modify the returned list.
*
* @return the list of currently unconsumed parameters.
*/
List unconsumedParams() {
return params.keySet().stream().filter(p -> !consumedParams.contains(p)).collect(Collectors.toList());
}
public float paramAsFloat(String key, float defaultValue) {
String sValue = param(key);
if (sValue == null) {
return defaultValue;
}
try {
return Float.parseFloat(sValue);
} catch (NumberFormatException e) {
throw new IllegalArgumentException("Failed to parse float parameter [" + key + "] with value [" + sValue + "]", e);
}
}
public int paramAsInt(String key, int defaultValue) {
String sValue = param(key);
if (sValue == null) {
return defaultValue;
}
try {
return Integer.parseInt(sValue);
} catch (NumberFormatException e) {
throw new IllegalArgumentException("Failed to parse int parameter [" + key + "] with value [" + sValue + "]", e);
}
}
public long paramAsLong(String key, long defaultValue) {
String sValue = param(key);
if (sValue == null) {
return defaultValue;
}
try {
return Long.parseLong(sValue);
} catch (NumberFormatException e) {
throw new IllegalArgumentException("Failed to parse long parameter [" + key + "] with value [" + sValue + "]", e);
}
}
@Override
public boolean paramAsBoolean(String key, boolean defaultValue) {
String rawParam = param(key);
// Treat empty string as true because that allows the presence of the url parameter to mean "turn this on"
if (rawParam != null && rawParam.length() == 0) {
return true;
} else {
return Booleans.parseBoolean(rawParam, defaultValue);
}
}
@Override
public Boolean paramAsBoolean(String key, Boolean defaultValue) {
return Booleans.parseBoolean(param(key), defaultValue);
}
public TimeValue paramAsTime(String key, TimeValue defaultValue) {
return parseTimeValue(param(key), defaultValue, key);
}
public ByteSizeValue paramAsSize(String key, ByteSizeValue defaultValue) {
return parseBytesSizeValue(param(key), defaultValue, key);
}
public String[] paramAsStringArray(String key, String[] defaultValue) {
String value = param(key);
if (value == null) {
return defaultValue;
}
return Strings.splitStringByCommaToArray(value);
}
public String[] paramAsStringArrayOrEmptyIfAll(String key) {
String[] params = paramAsStringArray(key, Strings.EMPTY_ARRAY);
if (Strings.isAllOrWildcard(params)) {
return Strings.EMPTY_ARRAY;
}
return params;
}
/**
* Get the {@link NamedXContentRegistry} that should be used to create parsers from this request.
*/
public NamedXContentRegistry getXContentRegistry() {
return xContentRegistry;
}
/**
* A parser for the contents of this request if there is a body, otherwise throws an {@link OpenSearchParseException}. Use
* {@link #applyContentParser(CheckedConsumer)} if you want to gracefully handle when the request doesn't have any contents. Use
* {@link #contentOrSourceParamParser()} for requests that support specifying the request body in the {@code source} param.
*/
public final XContentParser contentParser() throws IOException {
BytesReference content = requiredContent(); // will throw exception if body or content type missing
return xContentType.get().xContent().createParser(xContentRegistry, LoggingDeprecationHandler.INSTANCE, content.streamInput());
}
/**
* If there is any content then call {@code applyParser} with the parser, otherwise do nothing.
*/
public final void applyContentParser(CheckedConsumer applyParser) throws IOException {
if (hasContent()) {
try (XContentParser parser = contentParser()) {
applyParser.accept(parser);
}
}
}
/**
* Does this request have content or a {@code source} parameter? Use this instead of {@link #hasContent()} if this
* {@linkplain RestHandler} treats the {@code source} parameter like the body content.
*/
public final boolean hasContentOrSourceParam() {
return hasContent() || hasParam("source");
}
/**
* A parser for the contents of this request if it has contents, otherwise a parser for the {@code source} parameter if there is one,
* otherwise throws an {@link OpenSearchParseException}. Use {@link #withContentOrSourceParamParserOrNull(CheckedConsumer)} instead
* if you need to handle the absence request content gracefully.
*/
public final XContentParser contentOrSourceParamParser() throws IOException {
Tuple tuple = contentOrSourceParam();
return tuple.v1().xContent().createParser(xContentRegistry, LoggingDeprecationHandler.INSTANCE, tuple.v2().streamInput());
}
/**
* Call a consumer with the parser for the contents of this request if it has contents, otherwise with a parser for the {@code source}
* parameter if there is one, otherwise with {@code null}. Use {@link #contentOrSourceParamParser()} if you should throw an exception
* back to the user when there isn't request content.
*/
public final void withContentOrSourceParamParserOrNull(CheckedConsumer withParser) throws IOException {
if (hasContentOrSourceParam()) {
Tuple tuple = contentOrSourceParam();
BytesReference content = tuple.v2();
XContentType xContentType = tuple.v1();
try (
InputStream stream = content.streamInput();
XContentParser parser = xContentType.xContent().createParser(xContentRegistry, LoggingDeprecationHandler.INSTANCE, stream)
) {
withParser.accept(parser);
}
} else {
withParser.accept(null);
}
}
/**
* Get the content of the request or the contents of the {@code source} param or throw an exception if both are missing.
* Prefer {@link #contentOrSourceParamParser()} or {@link #withContentOrSourceParamParserOrNull(CheckedConsumer)} if you need a parser.
*/
public final Tuple contentOrSourceParam() {
if (hasContentOrSourceParam() == false) {
throw new OpenSearchParseException("request body or source parameter is required");
} else if (hasContent()) {
return new Tuple<>(xContentType.get(), requiredContent());
}
String source = param("source");
String typeParam = param("source_content_type");
if (source == null || typeParam == null) {
throw new IllegalStateException("source and source_content_type parameters are required");
}
BytesArray bytes = new BytesArray(source);
final XContentType xContentType = parseContentType(Collections.singletonList(typeParam));
if (xContentType == null) {
throw new IllegalStateException("Unknown value for source_content_type [" + typeParam + "]");
}
return new Tuple<>(xContentType, bytes);
}
/**
* Parses the given content type string for the media type. This method currently ignores parameters.
*/
// TODO stop ignoring parameters such as charset...
public static XContentType parseContentType(List header) {
if (header == null || header.isEmpty()) {
return null;
} else if (header.size() > 1) {
throw new IllegalArgumentException("only one Content-Type header should be provided");
}
String rawContentType = header.get(0);
final String[] elements = rawContentType.split("[ \t]*;");
if (elements.length > 0) {
final String[] splitMediaType = elements[0].split("/");
if (splitMediaType.length == 2
&& TCHAR_PATTERN.matcher(splitMediaType[0]).matches()
&& TCHAR_PATTERN.matcher(splitMediaType[1].trim()).matches()) {
return XContentType.fromMediaType(elements[0]);
} else {
throw new IllegalArgumentException("invalid Content-Type header [" + rawContentType + "]");
}
}
throw new IllegalArgumentException("empty Content-Type header");
}
public static class ContentTypeHeaderException extends RuntimeException {
ContentTypeHeaderException(final IllegalArgumentException cause) {
super(cause);
}
}
public static class BadParameterException extends RuntimeException {
BadParameterException(final IllegalArgumentException cause) {
super(cause);
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy