Add zeroTermsQuery for matchPhraseQuery (#648)

Add zeroTermsQuery for matchPhraseQuery
Yelp · Apr 16, 2024 · ed29d9f · ed29d9f
1 parent 732d655
commit ed29d9f
Show file tree

Hide file tree

Showing 4 changed files with 72 additions and 1 deletion.
diff --git a/clientlib/src/main/proto/yelp/nrtsearch/search.proto b/clientlib/src/main/proto/yelp/nrtsearch/search.proto
@@ -190,6 +190,15 @@ message MatchPhraseQuery {
     string query = 2; // The text to query with.
     int32 slop = 3; // Edit distance between respective positions of tokens generated by analyzing this query and the positions of terms in a document.
     Analyzer analyzer = 4; // Analyzer used to analyze the query. If not provided, the default search analyzer for the field would be used instead.
+    ZeroTerms zeroTermsQuery = 5; // Indicates whether none or all documents are returned if the analyzer removes all tokens. Valid values are NONE_ZERO_TERMS and ALL_ZERO_TERMS.
+
+    // Zero Terms options when analyzer removes all tokens.
+    enum ZeroTerms {
+        // No documents are returned if the analyzer removes all tokens.
+        NONE_ZERO_TERMS = 0;
+        // All documents are returned if the analyzer removes all tokens.
+        ALL_ZERO_TERMS = 1;
+    }
 }
 
 // A query that matches documents containing terms in the same order as those in the analyzed query string. The final analyzed token is treated as a prefix

diff --git a/docs/queries/match_phrase.rst b/docs/queries/match_phrase.rst
@@ -12,4 +12,13 @@ Proto definition:
        string query = 2; // The text to query with.
        int32 slop = 3; // Edit distance between respective positions of tokens generated by analyzing this query and the positions of terms in a document.
        Analyzer analyzer = 4; // Analyzer used to analyze the query. If not provided, the default search analyzer for the field would be used instead.
+       ZeroTerms zeroTermsQuery = 5; // Indicates whether none or all documents are returned if the analyzer removes all tokens. Valid values are NONE_ZERO_TERMS and ALL_ZERO_TERMS.
+
+       // Zero Terms options when analyzer removes all tokens.
+       enum ZeroTerms {
+           // No documents are returned if the analyzer removes all tokens.
+           NONE_ZERO_TERMS = 0;
+           // All documents are returned if the analyzer removes all tokens.
+           ALL_ZERO_TERMS = 1;
+       }
    }
diff --git a/src/main/java/com/yelp/nrtsearch/server/luceneserver/QueryNodeMapper.java b/src/main/java/com/yelp/nrtsearch/server/luceneserver/QueryNodeMapper.java
@@ -393,7 +393,18 @@ private Query getMatchPhraseQuery(MatchPhraseQuery matchPhraseQuery, IndexState
 
     // This can happen if there are no tokens found after analyzing the query text
     if (phraseQuery == null) {
-      return new MatchNoDocsQuery();
+      MatchPhraseQuery.ZeroTerms zeroTermsQuery = matchPhraseQuery.getZeroTermsQuery();
+      switch (zeroTermsQuery) {
+        case NONE_ZERO_TERMS -> {
+          return new MatchNoDocsQuery();
+        }
+        case ALL_ZERO_TERMS -> {
+          return new MatchAllDocsQuery();
+        }
+        default -> throw new IllegalArgumentException(
+            zeroTermsQuery
+                + " not valid. ZeroTermsQuery should be NONE_ZERO_TERMS or ALL_ZERO_TERMS");
+      }
     }
     return phraseQuery;
   }

diff --git a/src/test/java/com/yelp/nrtsearch/server/grpc/QueryTest.java b/src/test/java/com/yelp/nrtsearch/server/grpc/QueryTest.java
@@ -725,6 +725,48 @@ public void testSearchMatchPhraseQueryEmptyAfterAnalysis() {
     testQuery(query, responseTester);
   }
 
+  @Test
+  public void testSearchMatchPhraseQueryZeroTermsIsNone() {
+    Query query =
+        Query.newBuilder()
+            .setMatchPhraseQuery(
+                MatchPhraseQuery.newBuilder()
+                    .setField("vendor_name")
+                    .setQuery("/?/ ?//?")
+                    .setSlop(1)
+                    .setZeroTermsQuery(MatchPhraseQuery.ZeroTerms.NONE_ZERO_TERMS))
+            .build();
+
+    Consumer<SearchResponse> responseTester =
+        searchResponse -> {
+          assertEquals(0, searchResponse.getTotalHits().getValue());
+          assertEquals(0, searchResponse.getHitsList().size());
+        };
+
+    testQuery(query, responseTester);
+  }
+
+  @Test
+  public void testSearchMatchPhraseQueryZeroTermsIsAll() {
+    Query query =
+        Query.newBuilder()
+            .setMatchPhraseQuery(
+                MatchPhraseQuery.newBuilder()
+                    .setField("vendor_name")
+                    .setQuery("/?/ ?//?")
+                    .setSlop(1)
+                    .setZeroTermsQuery(MatchPhraseQuery.ZeroTerms.ALL_ZERO_TERMS))
+            .build();
+
+    Consumer<SearchResponse> responseTester =
+        searchResponse -> {
+          assertEquals(2, searchResponse.getTotalHits().getValue());
+          assertEquals(2, searchResponse.getHitsList().size());
+        };
+
+    testQuery(query, responseTester);
+  }
+
   @Test
   public void testSearchMatchPhraseQueryCustomAnalyzer() {
     Query query =