Skip to content

BlendedTermQuery should ignore fields that don't exist in the index #41125

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Apr 16, 2019
Merged

BlendedTermQuery should ignore fields that don't exist in the index #41125

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8ee6ad5
BlendedTermQuery should ignore fields that don't exists in the index
jimczi Apr 11, 2019
2a3f674
Merge branch 'master' into bug/cross_fields_no_term
jimczi Apr 15, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 5 additions & 25 deletions server/src/main/java/org/apache/lucene/queries/BlendedTermQuery.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,23 +113,17 @@ protected void blend(final TermStates[] contexts, int maxDoc, IndexReader reader
// TODO: Maybe it could also make sense to assume independent distributions of documents and eg. have:
// df = df1 + df2 - (df1 * df2 / maxDoc)?
max = Math.max(df, max);
if (minSumTTF != -1 && ctx.totalTermFreq() != -1) {
if (ctx.totalTermFreq() > 0) {
// we need to find out the minimum sumTTF to adjust the statistics
// otherwise the statistics don't match
minSumTTF = Math.min(minSumTTF, reader.getSumTotalTermFreq(terms[i].field()));
} else {
minSumTTF = -1;
}

}
if (minSumTTF != -1 && maxDoc > minSumTTF) {
maxDoc = (int)minSumTTF;
}

if (max == 0) {
return; // we are done that term doesn't exist at all
}
long sumTTF = minSumTTF == -1 ? -1 : 0;
long sumTTF = 0;
final int[] tieBreak = new int[contexts.length];
for (int i = 0; i < tieBreak.length; ++i) {
tieBreak[i] = i;
Expand Down Expand Up @@ -165,29 +159,20 @@ protected int compare(int i, int j) {
}
contexts[i] = ctx = adjustDF(reader.getContext(), ctx, Math.min(maxDoc, actualDf));
prev = current;
if (sumTTF >= 0 && ctx.totalTermFreq() >= 0) {
sumTTF += ctx.totalTermFreq();
} else {
sumTTF = -1; // omit once TF is omitted anywhere!
}
sumTTF += ctx.totalTermFreq();
}
sumTTF = Math.min(sumTTF, minSumTTF);
for (int i = 0; i < contexts.length; i++) {
int df = contexts[i].docFreq();
if (df == 0) {
continue;
}
// the blended sumTTF can't be greater than the sumTTTF on the field
final long fixedTTF = sumTTF == -1 ? -1 : sumTTF;
contexts[i] = adjustTTF(reader.getContext(), contexts[i], fixedTTF);
contexts[i] = adjustTTF(reader.getContext(), contexts[i], sumTTF);
}
}

private TermStates adjustTTF(IndexReaderContext readerContext, TermStates termContext, long sumTTF) throws IOException {
assert termContext.wasBuiltFor(readerContext);
if (sumTTF == -1 && termContext.totalTermFreq() == -1) {
return termContext;
}
TermStates newTermContext = new TermStates(readerContext);
List<LeafReaderContext> leaves = readerContext.leaves();
final int len;
Expand All @@ -213,12 +198,7 @@ private TermStates adjustTTF(IndexReaderContext readerContext, TermStates termCo
private static TermStates adjustDF(IndexReaderContext readerContext, TermStates ctx, int newDocFreq) throws IOException {
assert ctx.wasBuiltFor(readerContext);
// Use a value of ttf that is consistent with the doc freq (ie. gte)
long newTTF;
if (ctx.totalTermFreq() < 0) {
newTTF = -1;
} else {
newTTF = Math.max(ctx.totalTermFreq(), newDocFreq);
}
long newTTF = Math.max(ctx.totalTermFreq(), newDocFreq);
List<LeafReaderContext> leaves = readerContext.leaves();
final int len;
if (leaves == null) {
Expand Down
59 changes: 59 additions & 0 deletions server/src/test/java/org/apache/lucene/queries/BlendedTermQueryTests.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,10 +28,12 @@
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.MultiReader;
import org.apache.lucene.index.Term;
import org.apache.lucene.index.TermStates;
import org.apache.lucene.search.BooleanClause;
import org.apache.lucene.search.BooleanQuery;
import org.apache.lucene.search.DisjunctionMaxQuery;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.QueryUtils;
import org.apache.lucene.search.ScoreDoc;
import org.apache.lucene.search.ScoreMode;
Expand All @@ -52,6 +54,8 @@

import static org.hamcrest.Matchers.containsInAnyOrder;
import static org.hamcrest.Matchers.equalTo;
import static org.hamcrest.Matchers.greaterThan;
import static org.hamcrest.Matchers.instanceOf;

public class BlendedTermQueryTests extends ESTestCase {
public void testDismaxQuery() throws IOException {
Expand Down Expand Up @@ -114,6 +118,61 @@ public void testDismaxQuery() throws IOException {
assertEquals(Integer.toString(1), reader.document(scoreDocs[0].doc).getField("id").stringValue());

}
{
// test with an unknown field
String[] fields = new String[] {"username", "song", "unknown_field"};
Query query = BlendedTermQuery.dismaxBlendedQuery(toTerms(fields, "foo"), 1.0f);
Query rewrite = searcher.rewrite(query);
assertThat(rewrite, instanceOf(BooleanQuery.class));
for (BooleanClause clause : (BooleanQuery) rewrite) {
assertThat(clause.getQuery(), instanceOf(TermQuery.class));
TermQuery termQuery = (TermQuery) clause.getQuery();
TermStates termStates = termQuery.getTermStates();
if (termQuery.getTerm().field().equals("unknown_field")) {
assertThat(termStates.docFreq(), equalTo(0));
assertThat(termStates.totalTermFreq(), equalTo(0L));
} else {
assertThat(termStates.docFreq(), greaterThan(0));
assertThat(termStates.totalTermFreq(), greaterThan(0L));
}
}
assertThat(searcher.search(query, 10).totalHits.value, equalTo((long) iters + username.length));
}
{
// test with an unknown field and an unknown term
String[] fields = new String[] {"username", "song", "unknown_field"};
Query query = BlendedTermQuery.dismaxBlendedQuery(toTerms(fields, "unknown_term"), 1.0f);
Query rewrite = searcher.rewrite(query);
assertThat(rewrite, instanceOf(BooleanQuery.class));
for (BooleanClause clause : (BooleanQuery) rewrite) {
assertThat(clause.getQuery(), instanceOf(TermQuery.class));
TermQuery termQuery = (TermQuery) clause.getQuery();
TermStates termStates = termQuery.getTermStates();
assertThat(termStates.docFreq(), equalTo(0));
assertThat(termStates.totalTermFreq(), equalTo(0L));
}
assertThat(searcher.search(query, 10).totalHits.value, equalTo(0L));
}
{
// test with an unknown field and a term that is present in only one field
String[] fields = new String[] {"username", "song", "id", "unknown_field"};
Query query = BlendedTermQuery.dismaxBlendedQuery(toTerms(fields, "fan"), 1.0f);
Query rewrite = searcher.rewrite(query);
assertThat(rewrite, instanceOf(BooleanQuery.class));
for (BooleanClause clause : (BooleanQuery) rewrite) {
assertThat(clause.getQuery(), instanceOf(TermQuery.class));
TermQuery termQuery = (TermQuery) clause.getQuery();
TermStates termStates = termQuery.getTermStates();
if (termQuery.getTerm().field().equals("username")) {
assertThat(termStates.docFreq(), equalTo(1));
assertThat(termStates.totalTermFreq(), equalTo(1L));
} else {
assertThat(termStates.docFreq(), equalTo(0));
assertThat(termStates.totalTermFreq(), equalTo(0L));
}
}
assertThat(searcher.search(query, 10).totalHits.value, equalTo(1L));
}
reader.close();
w.close();
dir.close();
Expand Down