From 355f80adc92cc69ed35cbc4936f0f8334ee55307 Mon Sep 17 00:00:00 2001 From: Simon Willnauer Date: Fri, 26 Apr 2013 19:16:44 +0200 Subject: [PATCH] Added temporary fix for LUCENE-4899 where FastVectorHighlihgter failed with StringIndexOutOfBoundsException if a single highlight phrase or term was greater than the fragCharSize producing negative string offsets The fixed BaseFragListBuilder was added as XSimpleFragListBuilder which triggers an assert once Elasticsearch upgrades to Lucene 4.3 --- .../XSimpleFragListBuilder.java | 168 ++++++++++++++++++ .../search/highlight/HighlightPhase.java | 6 +- .../highlight/HighlighterSearchTests.java | 70 ++++++++ 3 files changed, 239 insertions(+), 5 deletions(-) create mode 100644 src/main/java/org/apache/lucene/search/vectorhighlight/XSimpleFragListBuilder.java diff --git a/src/main/java/org/apache/lucene/search/vectorhighlight/XSimpleFragListBuilder.java b/src/main/java/org/apache/lucene/search/vectorhighlight/XSimpleFragListBuilder.java new file mode 100644 index 0000000000000..86494a676f0ba --- /dev/null +++ b/src/main/java/org/apache/lucene/search/vectorhighlight/XSimpleFragListBuilder.java @@ -0,0 +1,168 @@ +package org.apache.lucene.search.vectorhighlight; + +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF 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. + */ + +import java.util.ArrayList; +import java.util.Iterator; +import java.util.List; + +import org.apache.lucene.search.vectorhighlight.FieldPhraseList.WeightedPhraseInfo; +import org.apache.lucene.util.Version; +import org.elasticsearch.common.lucene.Lucene; + +/** + * A non-abstract copy of the abstract {@link BaseFragListBuilder}. that works + * in the same way as {@link SimpleFragListBuilder} but isn't prone to the + * problem of negative offsets. This is fixed in Lucene 4.3 and this class + * should be removed once Elasticsearch upgraded to Lucene 4.3 + *

+ * LUCENE-4899: FastVectorHighlihgter failed with + * {@link StringIndexOutOfBoundsException} if a single highlight phrase or term was + * greater than the fragCharSize producing negative string offsets + *

+ */ +public final class XSimpleFragListBuilder implements FragListBuilder { + + public static final int MARGIN_DEFAULT = 6; + public static final int MIN_FRAG_CHAR_SIZE_FACTOR = 3; + + final int margin; + final int minFragCharSize; + static { + assert Version.LUCENE_42 == Lucene.VERSION: "Elasticsearch has upgraded to Lucene Version: [" + Lucene.VERSION + "] this should can be removed"; + } + + public XSimpleFragListBuilder(int margin) { + if (margin < 0) + throw new IllegalArgumentException("margin(" + margin + ") is too small. It must be 0 or higher."); + + this.margin = margin; + this.minFragCharSize = Math.max(1, margin * MIN_FRAG_CHAR_SIZE_FACTOR); + } + + public XSimpleFragListBuilder() { + this(MARGIN_DEFAULT); + } + + @Override + public FieldFragList createFieldFragList(FieldPhraseList fieldPhraseList, int fragCharSize) { + return createFieldFragList(fieldPhraseList, new SimpleFieldFragList(fragCharSize), fragCharSize); + } + + protected FieldFragList createFieldFragList(FieldPhraseList fieldPhraseList, FieldFragList fieldFragList, int fragCharSize) { + if (fragCharSize < minFragCharSize) + throw new IllegalArgumentException("fragCharSize(" + fragCharSize + ") is too small. It must be " + minFragCharSize + " or higher."); + + List wpil = new ArrayList(); + IteratorQueue queue = new IteratorQueue(fieldPhraseList.getPhraseList().iterator()); + WeightedPhraseInfo phraseInfo = null; + int startOffset = 0; + while ((phraseInfo = queue.top()) != null) { + // if the phrase violates the border of previous fragment, discard + // it and try next phrase + if (phraseInfo.getStartOffset() < startOffset) { + queue.removeTop(); + continue; + } + + wpil.clear(); + final int currentPhraseStartOffset = phraseInfo.getStartOffset(); + int currentPhraseEndOffset = phraseInfo.getEndOffset(); + int spanStart = Math.max(currentPhraseStartOffset - margin, startOffset); + int spanEnd = Math.max(currentPhraseEndOffset, spanStart + fragCharSize); + if (acceptPhrase(queue.removeTop(), currentPhraseEndOffset - currentPhraseStartOffset, fragCharSize)) { + wpil.add(phraseInfo); + } + while ((phraseInfo = queue.top()) != null) { // pull until we crossed the current spanEnd + if (phraseInfo.getEndOffset() <= spanEnd) { + currentPhraseEndOffset = phraseInfo.getEndOffset(); + if (acceptPhrase(queue.removeTop(), currentPhraseEndOffset - currentPhraseStartOffset, fragCharSize)) { + wpil.add(phraseInfo); + } + } else { + break; + } + } + if (wpil.isEmpty()) { + continue; + } + + final int matchLen = currentPhraseEndOffset - currentPhraseStartOffset; + // now recalculate the start and end position to "center" the result + final int newMargin = Math.max(0, (fragCharSize - matchLen) / 2); + // matchLen can be > fragCharSize prevent IAOOB here + spanStart = currentPhraseStartOffset - newMargin; + if (spanStart < startOffset) { + spanStart = startOffset; + } + // whatever is bigger here we grow this out + spanEnd = spanStart + Math.max(matchLen, fragCharSize); + startOffset = spanEnd; + fieldFragList.add(spanStart, spanEnd, wpil); + } + return fieldFragList; + } + + /** + * A predicate to decide if the given {@link WeightedPhraseInfo} should be + * accepted as a highlighted phrase or if it should be discarded. + *

+ * The default implementation discards phrases that are composed of more + * than one term and where the matchLength exceeds the fragment character + * size. + * + * @param info + * the phrase info to accept + * @param matchLength + * the match length of the current phrase + * @param fragCharSize + * the configured fragment character size + * @return true if this phrase info should be accepted as a + * highligh phrase + */ + protected boolean acceptPhrase(WeightedPhraseInfo info, int matchLength, int fragCharSize) { + return info.getTermsOffsets().size() <= 1 || matchLength <= fragCharSize; + } + + private static final class IteratorQueue { + private final Iterator iter; + private T top; + + public IteratorQueue(Iterator iter) { + this.iter = iter; + T removeTop = removeTop(); + assert removeTop == null; + } + + public T top() { + return top; + } + + public T removeTop() { + T currentTop = top; + if (iter.hasNext()) { + top = iter.next(); + } else { + top = null; + } + return currentTop; + } + + } + +} diff --git a/src/main/java/org/elasticsearch/search/highlight/HighlightPhase.java b/src/main/java/org/elasticsearch/search/highlight/HighlightPhase.java index 6bef775164108..28d3e64ad4676 100644 --- a/src/main/java/org/elasticsearch/search/highlight/HighlightPhase.java +++ b/src/main/java/org/elasticsearch/search/highlight/HighlightPhase.java @@ -267,11 +267,7 @@ public int compare(TextFragment o1, TextFragment o2) { fragmentsBuilder = new SourceSimpleFragmentsBuilder(mapper, context, field.preTags(), field.postTags(), boundaryScanner); } } else { - if (field.fragmentOffset() == -1) - fragListBuilder = new SimpleFragListBuilder(); - else - fragListBuilder = new SimpleFragListBuilder(field.fragmentOffset()); - + fragListBuilder = field.fragmentOffset() == -1 ? new XSimpleFragListBuilder() : new XSimpleFragListBuilder(field.fragmentOffset()); if (field.scoreOrdered()) { if (mapper.fieldType().stored()) { fragmentsBuilder = new XScoreOrderFragmentsBuilder(field.preTags(), field.postTags(), boundaryScanner); diff --git a/src/test/java/org/elasticsearch/test/integration/search/highlight/HighlighterSearchTests.java b/src/test/java/org/elasticsearch/test/integration/search/highlight/HighlighterSearchTests.java index e1a8ddd0e56fe..65ffabec45aad 100644 --- a/src/test/java/org/elasticsearch/test/integration/search/highlight/HighlighterSearchTests.java +++ b/src/test/java/org/elasticsearch/test/integration/search/highlight/HighlighterSearchTests.java @@ -19,6 +19,12 @@ package org.elasticsearch.test.integration.search.highlight; +import org.apache.lucene.index.Term; +import org.apache.lucene.search.BooleanQuery; +import org.apache.lucene.search.PhraseQuery; +import org.apache.lucene.search.TermQuery; +import org.apache.lucene.search.BooleanClause.Occur; +import org.apache.lucene.search.vectorhighlight.FieldQuery; import org.elasticsearch.ElasticSearchException; import org.elasticsearch.action.search.SearchPhaseExecutionException; import org.elasticsearch.action.search.SearchResponse; @@ -31,6 +37,7 @@ import org.elasticsearch.common.xcontent.XContentFactory; import org.elasticsearch.index.query.FilterBuilders; import org.elasticsearch.index.query.MatchQueryBuilder; +import org.elasticsearch.index.query.MatchQueryBuilder.Type; import org.elasticsearch.index.query.QueryBuilders; import org.elasticsearch.indices.IndexMissingException; import org.elasticsearch.rest.RestStatus; @@ -81,6 +88,69 @@ protected Client getClient() { return client("server1"); } + + @Test + public void testEnsureNoNegativeOffsets() throws Exception { + try { + client.admin().indices().prepareDelete("test").execute().actionGet(); + } catch (Exception e) { + // ignore + } + + client.admin().indices().prepareCreate("test").setSettings(ImmutableSettings.settingsBuilder().put("index.number_of_shards", 2)) + .addMapping("type1", jsonBuilder().startObject().startObject("type1").startObject("properties") + // we don't store title, now lets see if it works... + .startObject("no_long_term").field("type", "string").field("store", "no").field("term_vector", "with_positions_offsets").endObject() + .startObject("long_term").field("type", "string").field("store", "no").field("term_vector", "with_positions_offsets").endObject() + .endObject().endObject().endObject()) + .execute().actionGet(); + + + client.prepareIndex("test", "type1", "1") + .setSource(XContentFactory.jsonBuilder().startObject() + .startArray("no_long_term") + .value("This is a test where foo is highlighed and should be highlighted") + .endArray() + .startArray("long_term") + .value("This is a test thisisaverylongwordandmakessurethisfails where foo is highlighed and should be highlighted") + .endArray() + .endObject()) + .setRefresh(true).execute().actionGet(); + + + SearchResponse search = client.prepareSearch() + .setQuery(matchQuery("long_term", "thisisaverylongwordandmakessurethisfails foo highlighed")) + .addHighlightedField("long_term", 18, 1) + .execute().actionGet(); + + assertThat(search.getHits().totalHits(), equalTo(1l)); + assertThat(search.getHits().hits().length, equalTo(1)); + + assertThat(search.getHits().hits()[0].highlightFields().get("long_term").fragments().length, equalTo(1)); + assertThat(search.getHits().hits()[0].highlightFields().get("long_term").fragments()[0].string(), equalTo("thisisaverylongwordandmakessurethisfails")); + + + search = client.prepareSearch() + .setQuery(matchQuery("no_long_term", "test foo highlighed").type(Type.PHRASE).slop(3)) + .addHighlightedField("no_long_term", 18, 1).setHighlighterPostTags("").setHighlighterPreTags("") + .execute().actionGet(); + assertThat(search.getHits().totalHits(), equalTo(1l)); + assertThat(search.getHits().hits().length, equalTo(1)); + assertThat(search.getHits().hits()[0].highlightFields().size(), equalTo(0)); + + + search = client.prepareSearch() + .setQuery(matchQuery("no_long_term", "test foo highlighed").type(Type.PHRASE).slop(3)) + .addHighlightedField("no_long_term", 30, 1).setHighlighterPostTags("").setHighlighterPreTags("") + .execute().actionGet(); + + assertThat(search.getHits().totalHits(), equalTo(1l)); + assertThat(search.getHits().hits().length, equalTo(1)); + + assertThat(search.getHits().hits()[0].highlightFields().get("no_long_term").fragments().length, equalTo(1)); + assertThat(search.getHits().hits()[0].highlightFields().get("no_long_term").fragments()[0].string(), equalTo("a test where foo is highlighed and")); + } + @Test public void testSourceLookupHighlightingUsingPlainHighlighter() throws Exception { try {