All Downloads are FREE. Search and download functionalities are using the official Maven repository.

org.elasticsearch.search.highlight.HighlighterParseElement Maven / Gradle / Ivy

There is a newer version: 8.15.1
Show newest version
/*
 * Licensed to ElasticSearch and Shay Banon 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.
 */

package org.elasticsearch.search.highlight;

import com.google.common.collect.Lists;
import org.elasticsearch.common.lucene.search.vectorhighlight.SimpleBoundaryScanner2;
import org.elasticsearch.common.xcontent.XContentParser;
import org.elasticsearch.search.SearchParseElement;
import org.elasticsearch.search.SearchParseException;
import org.elasticsearch.search.internal.SearchContext;

import java.util.List;

import static com.google.common.collect.Lists.newArrayList;

/**
 * 
 * highlight : {
 *  tags_schema : "styled",
 *  pre_tags : ["tag1", "tag2"],
 *  post_tags : ["tag1", "tag2"],
 *  order : "score",
 *  highlight_filter : true,
 *  fields : {
 *      field1 : {  },
 *      field2 : { fragment_size : 100, number_of_fragments : 2 },
 *      field3 : { number_of_fragments : 5, order : "simple", tags_schema : "styled" },
 *      field4 : { number_of_fragments: 0, pre_tags : ["openingTagA", "openingTagB"], post_tags : ["closingTag"] }
 *  }
 * }
 * 
*/ public class HighlighterParseElement implements SearchParseElement { private static final String[] DEFAULT_PRE_TAGS = new String[]{""}; private static final String[] DEFAULT_POST_TAGS = new String[]{""}; private static final String[] STYLED_PRE_TAG = { "", "", "", "", "", "", "", "", "", "" }; private static final String[] STYLED_POST_TAGS = {""}; @Override public void parse(XContentParser parser, SearchContext context) throws Exception { XContentParser.Token token; String topLevelFieldName = null; List fields = newArrayList(); String[] globalPreTags = DEFAULT_PRE_TAGS; String[] globalPostTags = DEFAULT_POST_TAGS; boolean globalScoreOrdered = false; boolean globalHighlightFilter = false; boolean globalRequireFieldMatch = false; int globalFragmentSize = 100; int globalNumOfFragments = 5; String globalEncoder = "default"; int globalBoundaryMaxScan = SimpleBoundaryScanner2.DEFAULT_MAX_SCAN; char[] globalBoundaryChars = SimpleBoundaryScanner2.DEFAULT_BOUNDARY_CHARS; while ((token = parser.nextToken()) != XContentParser.Token.END_OBJECT) { if (token == XContentParser.Token.FIELD_NAME) { topLevelFieldName = parser.currentName(); } else if (token == XContentParser.Token.START_ARRAY) { if ("pre_tags".equals(topLevelFieldName) || "preTags".equals(topLevelFieldName)) { List preTagsList = Lists.newArrayList(); while ((token = parser.nextToken()) != XContentParser.Token.END_ARRAY) { preTagsList.add(parser.text()); } globalPreTags = preTagsList.toArray(new String[preTagsList.size()]); } else if ("post_tags".equals(topLevelFieldName) || "postTags".equals(topLevelFieldName)) { List postTagsList = Lists.newArrayList(); while ((token = parser.nextToken()) != XContentParser.Token.END_ARRAY) { postTagsList.add(parser.text()); } globalPostTags = postTagsList.toArray(new String[postTagsList.size()]); } } else if (token.isValue()) { if ("order".equals(topLevelFieldName)) { globalScoreOrdered = "score".equals(parser.text()); } else if ("tags_schema".equals(topLevelFieldName) || "tagsSchema".equals(topLevelFieldName)) { String schema = parser.text(); if ("styled".equals(schema)) { globalPreTags = STYLED_PRE_TAG; globalPostTags = STYLED_POST_TAGS; } } else if ("highlight_filter".equals(topLevelFieldName) || "highlightFilter".equals(topLevelFieldName)) { globalHighlightFilter = parser.booleanValue(); } else if ("fragment_size".equals(topLevelFieldName) || "fragmentSize".equals(topLevelFieldName)) { globalFragmentSize = parser.intValue(); } else if ("number_of_fragments".equals(topLevelFieldName) || "numberOfFragments".equals(topLevelFieldName)) { globalNumOfFragments = parser.intValue(); } else if ("encoder".equals(topLevelFieldName)) { globalEncoder = parser.text(); } else if ("require_field_match".equals(topLevelFieldName) || "requireFieldMatch".equals(topLevelFieldName)) { globalRequireFieldMatch = parser.booleanValue(); } else if ("boundary_max_scan".equals(topLevelFieldName) || "boundaryMaxScan".equals(topLevelFieldName)) { globalBoundaryMaxScan = parser.intValue(); } else if ("boundary_chars".equals(topLevelFieldName) || "boundaryChars".equals(topLevelFieldName)) { globalBoundaryChars = parser.text().toCharArray(); } } else if (token == XContentParser.Token.START_OBJECT) { if ("fields".equals(topLevelFieldName)) { String highlightFieldName = null; while ((token = parser.nextToken()) != XContentParser.Token.END_OBJECT) { if (token == XContentParser.Token.FIELD_NAME) { highlightFieldName = parser.currentName(); } else if (token == XContentParser.Token.START_OBJECT) { SearchContextHighlight.Field field = new SearchContextHighlight.Field(highlightFieldName); String fieldName = null; while ((token = parser.nextToken()) != XContentParser.Token.END_OBJECT) { if (token == XContentParser.Token.FIELD_NAME) { fieldName = parser.currentName(); } else if (token == XContentParser.Token.START_ARRAY) { if ("pre_tags".equals(fieldName) || "preTags".equals(fieldName)) { List preTagsList = Lists.newArrayList(); while ((token = parser.nextToken()) != XContentParser.Token.END_ARRAY) { preTagsList.add(parser.text()); } field.preTags(preTagsList.toArray(new String[preTagsList.size()])); } else if ("post_tags".equals(fieldName) || "postTags".equals(fieldName)) { List postTagsList = Lists.newArrayList(); while ((token = parser.nextToken()) != XContentParser.Token.END_ARRAY) { postTagsList.add(parser.text()); } field.postTags(postTagsList.toArray(new String[postTagsList.size()])); } } else if (token.isValue()) { if ("fragment_size".equals(fieldName) || "fragmentSize".equals(fieldName)) { field.fragmentCharSize(parser.intValue()); } else if ("number_of_fragments".equals(fieldName) || "numberOfFragments".equals(fieldName)) { field.numberOfFragments(parser.intValue()); } else if ("fragment_offset".equals(fieldName) || "fragmentOffset".equals(fieldName)) { field.fragmentOffset(parser.intValue()); } else if ("highlight_filter".equals(fieldName) || "highlightFilter".equals(fieldName)) { field.highlightFilter(parser.booleanValue()); } else if ("order".equals(fieldName)) { field.scoreOrdered("score".equals(parser.text())); } else if ("require_field_match".equals(fieldName) || "requireFieldMatch".equals(fieldName)) { field.requireFieldMatch(parser.booleanValue()); } else if ("boundary_max_scan".equals(topLevelFieldName) || "boundaryMaxScan".equals(topLevelFieldName)) { field.boundaryMaxScan(parser.intValue()); } else if ("boundary_chars".equals(topLevelFieldName) || "boundaryChars".equals(topLevelFieldName)) { field.boundaryChars(parser.text().toCharArray()); } } } fields.add(field); } } } } } if (globalPreTags != null && globalPostTags == null) { throw new SearchParseException(context, "Highlighter global preTags are set, but global postTags are not set"); } // now, go over and fill all fields with default values from the global state for (SearchContextHighlight.Field field : fields) { if (field.preTags() == null) { field.preTags(globalPreTags); } if (field.postTags() == null) { field.postTags(globalPostTags); } if (field.highlightFilter() == null) { field.highlightFilter(globalHighlightFilter); } if (field.scoreOrdered() == null) { field.scoreOrdered(globalScoreOrdered); } if (field.fragmentCharSize() == -1) { field.fragmentCharSize(globalFragmentSize); } if (field.numberOfFragments() == -1) { field.numberOfFragments(globalNumOfFragments); } if (field.encoder() == null) { field.encoder(globalEncoder); } if (field.requireFieldMatch() == null) { field.requireFieldMatch(globalRequireFieldMatch); } if (field.boundaryMaxScan() == -1) { field.boundaryMaxScan(globalBoundaryMaxScan); } if (field.boundaryChars() == null) { field.boundaryChars(globalBoundaryChars); } } context.highlight(new SearchContextHighlight(fields)); } }