Change how `max_matches` affects `target_field` option. (#47982)

Prior to this change the `target_field` would always be a json array
field in the document being ingested. This to take into account that
multiple enrich documents could be inserted into the `target_field`.

However the default `max_matches` is `1`. Meaning that by default
only a single enrich document would be added to `target_field` json
array field.

This commit changes this; if `max_matches` is set to `1` then the single
document would be added as a json object to the `target_field` and
if it is configured to a higher value then the enrich documents will be
added as a json array (even if a single enrich document happens to be
enriched).

docs/reference/ingest/apis/enrich/put-enrich-policy.asciidoc

@@ -323,15 +323,13 @@ The API returns the following response:
   "_seq_no": 55,
   "_primary_term": 1,
   "_source": {
-    "geo_data": [
-        {
-            "location": {
-                "type": "envelope",
-                "coordinates": [[13.0, 53.0], [14.0, 52.0]]
-            },
-            "postal_code": "96598"
-        }
-    ],
+    "geo_data": {
+      "location": {
+        "type": "envelope",
+        "coordinates": [[13.0, 53.0], [14.0, 52.0]]
+      },
+      "postal_code": "96598"
+    },
     "first_name": "Mardy",
     "last_name": "Brown",
     "geo_location": "POINT (13.5 52.5)"

docs/reference/ingest/enrich.asciidoc

@@ -167,7 +167,8 @@ PUT /_ingest/pipeline/user_lookup
       "enrich" : {
         "policy_name": "users-policy",
         "field" : "email",
-        "target_field": "user"
+        "target_field": "user",
+        "max_matches": "1"
       }
     }
   ]
@@ -175,34 +176,38 @@ PUT /_ingest/pipeline/user_lookup
 ----
 // TEST[continued]
 
+Because the enrich policy type is `match`,
+the enrich processor matches incoming documents
+to documents in the enrich index
+based on match field values.
+The enrich processor then appends the enrich field data 
+from matching documents in the enrich index
+to the target field of incoming documents.
+
+Because the `max_matches` option for the enrich processor is `1`,
+the enrich processor appends the data from only the best matching document
+to each incoming document's target field as an object.
+
+If the `max_matches` option were greater than `1`,
+the processor could append data from up to the `max_matches` number of documents
+to the target field as an array.
+
+If the incoming document matches no documents in the enrich index,
+the processor appends no data.
+
 You also can add other <<ingest-processors,processors>>
 to your ingest pipeline.
 You can use these processors to change or drop incoming documents
 based on your criteria.
-
 See <<ingest-processors>> for a list of built-in processors.
 
+
 [float]
 [[ingest-enrich-docs]]
 ==== Ingest and enrich documents
 
 Index incoming documents using your ingest pipeline.
 
-Because the enrich policy type is `match`,
-the enrich processor matches incoming documents
-to documents in the enrich index
-based on match field values.
-The processor then appends the enrich field data 
-from any matching document in the enrich index
-to target field of the incoming document.
-
-The enrich processor appends all data to the target field as an array.
-If the incoming document matches more than one document in the enrich index,
-the processor appends data from those documents to the array.
-
-If the incoming document matches no documents in the enrich index,
-the processor appends no data.
-
 The following <<docs-index_,index API>> request uses the ingest pipeline
 to index a document
 containing the `email` field
@@ -240,16 +245,14 @@ The API returns the following response:
   "_seq_no": 55,
   "_primary_term": 1,
   "_source": {
-    "user": [
-      {
-          "email": "mardy.brown@asciidocsmith.com",
-          "first_name": "Mardy",
-          "last_name": "Brown",
-          "zip": 70116,
-          "city": "New Orleans",
-          "state": "LA"
-      }
-    ],
+    "user": {
+      "email": "mardy.brown@asciidocsmith.com",
+      "first_name": "Mardy",
+      "last_name": "Brown",
+      "zip": 70116,
+      "city": "New Orleans",
+      "state": "LA"
+    },
     "email": "mardy.brown@asciidocsmith.com"
   }
 }

docs/reference/ingest/processors/enrich.asciidoc

@@ -16,18 +16,9 @@ check out the <<ingest-enriching-data,tutorial>> to get familiar with enrich pol
 | `field`            | yes       | -                    | The field in the input document that matches the policies match_field used to retrieve the enrichment data.
 | `target_field`     | yes       | -                    | The field that will be used for the enrichment data.
 | `ignore_missing`   | no        | false                | If `true` and `field` does not exist, the processor quietly exits without modifying the document
-| `override`         | no        | true                 | If processor will update fields with pre-existing non-null-valued field. When set to `false`, such fields will                                                             not be touched.
-| `max_matches`      | no        | 1                    | The maximum number of matched documents to include under the configured target field. In order to avoid                                                                    documents getting too large, the maximum allowed value is 128.
-| `shape_relation`   | no        | `INTERSECTS`         a| Spatial relation operator
-used to match the <<geo-shape,geo_shape>> of incoming documents
-to documents in the enrich index.
-+
-This option is only used for `geo_match` enrich policy types.
-+
-The <<spatial-strategy, geo_shape strategy>> mapping parameter determines
-which spatial relation operators are availlble.
-See <<_spatial_relations>>
-for operators and more information.
+| `override`         | no        | true                 | If processor will update fields with pre-existing non-null-valued field. When set to `false`, such fields will not be touched.
+| `max_matches`      | no        | 1                    | The maximum number of matched documents to include under the configured target field. The `target_field` will be turned into a json array if `max_matches` is higher than 1, otherwise `target_field` will become a json object. In order to avoid documents getting too large, the maximum allowed value is 128.
+| `shape_relation`   | no        | `INTERSECTS`         | A spatial relation operator used to match the <<geo-shape,geo_shape>> of incoming documents to documents in the enrich index. This option is only used for `geo_match` enrich policy types. The <<spatial-strategy, geo_shape strategy>> mapping parameter determines which spatial relation operators are available. See <<_spatial_relations>> for operators and more information.
 
 include::common-options.asciidoc[]
 |======

x-pack/plugin/enrich/qa/common/src/main/java/org/elasticsearch/test/enrich/CommonEnrichRestTestCase.java

@@ -80,13 +80,12 @@ public abstract class CommonEnrichRestTestCase extends ESRestTestCase {
         // Check if document has been enriched
         Request getRequest = new Request("GET", "/my-index/_doc/1");
         Map<String, Object> response = toMap(client().performRequest(getRequest));
-        List<?> entries = (List<?>) ((Map<?, ?>) response.get("_source")).get("entry");
-        Map<?, ?> _source = (Map<?, ?>) entries.get(0);
-        assertThat(_source.size(), equalTo(4));
-        assertThat(_source.get("host"), equalTo("elastic.co"));
-        assertThat(_source.get("tld"), equalTo("co"));
-        assertThat(_source.get("globalRank"), equalTo(25));
-        assertThat(_source.get("tldRank"), equalTo(7));
+        Map<?, ?> entry = (Map<?, ?>) ((Map<?, ?>) response.get("_source")).get("entry");
+        assertThat(entry.size(), equalTo(4));
+        assertThat(entry.get("host"), equalTo("elastic.co"));
+        assertThat(entry.get("tld"), equalTo("co"));
+        assertThat(entry.get("globalRank"), equalTo(25));
+        assertThat(entry.get("tldRank"), equalTo(7));
 
         if (deletePipeilne) {
             // delete the pipeline so the policies can be deleted

x-pack/plugin/enrich/src/main/java/org/elasticsearch/xpack/enrich/AbstractEnrichProcessor.java

@@ -96,12 +96,17 @@ public abstract class AbstractEnrichProcessor extends AbstractProcessor {
                 }
 
                 if (overrideEnabled || ingestDocument.hasField(targetField) == false) {
-                    List<Map<String, Object>> enrichDocuments = new ArrayList<>(searchHits.length);
-                    for (SearchHit searchHit : searchHits) {
-                        Map<String, Object> enrichDocument = searchHit.getSourceAsMap();
-                        enrichDocuments.add(enrichDocument);
+                    if (maxMatches == 1) {
+                        Map<String, Object> firstDocument = searchHits[0].getSourceAsMap();
+                        ingestDocument.setFieldValue(targetField, firstDocument);
+                    } else {
+                        List<Map<String, Object>> enrichDocuments = new ArrayList<>(searchHits.length);
+                        for (SearchHit searchHit : searchHits) {
+                            Map<String, Object> enrichDocument = searchHit.getSourceAsMap();
+                            enrichDocuments.add(enrichDocument);
+                        }
+                        ingestDocument.setFieldValue(targetField, enrichDocuments);
                     }
-                    ingestDocument.setFieldValue(targetField, enrichDocuments);
                 }
                 handler.accept(ingestDocument, null);
             });

x-pack/plugin/enrich/src/test/java/org/elasticsearch/xpack/enrich/BasicEnrichTests.java

@@ -157,9 +157,11 @@ public class BasicEnrichTests extends ESSingleNodeTestCase {
 
         GetResponse getResponse = client().get(new GetRequest("my-index", "_id")).actionGet();
         Map<String, Object> source = getResponse.getSourceAsMap();
-        List<?> entries = (List<?>) source.get("enriched");
+        Map<?, ?> entries = (Map) source.get("enriched");
         assertThat(entries, notNullValue());
-        assertThat(entries.size(), equalTo(1));
+        assertThat(entries.size(), equalTo(2));
+        assertThat(entries.containsKey(matchField), is(true));
+        assertThat(entries.get(enrichField), equalTo("94040"));
 
         EnrichStatsAction.Response statsResponse =
             client().execute(EnrichStatsAction.INSTANCE, new EnrichStatsAction.Request()).actionGet();
@@ -208,7 +210,7 @@ public class BasicEnrichTests extends ESSingleNodeTestCase {
             GetResponse getResponse = client().get(new GetRequest("my-index", Integer.toString(i))).actionGet();
             Map<String, Object> source = getResponse.getSourceAsMap();
             assertThat(source.size(), equalTo(2));
-            assertThat(source.get("target"), equalTo(Arrays.asList(mapOf("key", "key", "value", "val" + i))));
+            assertThat(source.get("target"), equalTo(mapOf("key", "key", "value", "val" + i)));
         }
     }
 

x-pack/plugin/enrich/src/test/java/org/elasticsearch/xpack/enrich/EnrichMultiNodeIT.java

@@ -154,8 +154,7 @@ public class EnrichMultiNodeIT extends ESIntegTestCase {
         for (int i = 0; i < numDocs; i++) {
             GetResponse getResponse = client().get(new GetRequest("my-index", Integer.toString(i))).actionGet();
             Map<String, Object> source = getResponse.getSourceAsMap();
-            List<?> entries = (List<?>) source.get("user");
-            Map<?, ?> userEntry = (Map<?, ?>) entries.get(0);
+            Map<?, ?> userEntry = (Map<?, ?>) source.get("user");
             assertThat(userEntry.size(), equalTo(DECORATE_FIELDS.length + 1));
             assertThat(keys.contains(userEntry.get(MATCH_FIELD)), is(true));
             for (String field : DECORATE_FIELDS) {

x-pack/plugin/enrich/src/test/java/org/elasticsearch/xpack/enrich/GeoMatchProcessorTests.java

@@ -95,8 +95,13 @@ public class GeoMatchProcessorTests extends ESTestCase {
         assertThat(shapeQueryBuilder.shape(), equalTo(expectedGeometry));
 
         // Check result
-        List<?> entries = ingestDocument.getFieldValue("entry", List.class);
-        Map<?, ?> entry = (Map<?, ?>) entries.get(0);
+        Map<?, ?> entry;
+        if (maxMatches == 1) {
+            entry = ingestDocument.getFieldValue("entry", Map.class);
+        } else {
+            List<?> entries = ingestDocument.getFieldValue("entry", List.class);
+            entry = (Map<?, ?>) entries.get(0);
+        }
         assertThat(entry.size(), equalTo(2));
         assertThat(entry.get("zipcode"), equalTo(94040));
 

x-pack/plugin/enrich/src/test/java/org/elasticsearch/xpack/enrich/MatchProcessorTests.java

@@ -209,10 +209,9 @@ public class MatchProcessorTests extends ESTestCase {
     }
 
     public void testNumericValue() {
-        int maxMatches = randomIntBetween(1, 8);
         MockSearchFunction mockSearch = mockedSearchFunction(mapOf(2, mapOf("globalRank", 451, "tldRank", 23, "tld", "co")));
         MatchProcessor processor =
-            new MatchProcessor("_tag", mockSearch, "_name", "domain", "entry", false, true, "domain", maxMatches);
+            new MatchProcessor("_tag", mockSearch, "_name", "domain", "entry", false, true, "domain", 1);
         IngestDocument ingestDocument =
             new IngestDocument("_index", "_type", "_id", "_routing", 1L, VersionType.INTERNAL, mapOf("domain", 2));
 
@@ -230,8 +229,7 @@ public class MatchProcessorTests extends ESTestCase {
         assertThat(termQueryBuilder.value(), equalTo(2));
 
         // Check result
-        List<?> entries = ingestDocument.getFieldValue("entry", List.class);
-        Map<?, ?> entry = (Map<?, ?>) entries.get(0);
+        Map<?, ?> entry = ingestDocument.getFieldValue("entry", Map.class);
         assertThat(entry.size(), equalTo(3));
         assertThat(entry.get("globalRank"), equalTo(451));
         assertThat(entry.get("tldRank"), equalTo(23));
@@ -239,11 +237,10 @@ public class MatchProcessorTests extends ESTestCase {
     }
 
     public void testArray() {
-        int maxMatches = randomIntBetween(1, 8);
         MockSearchFunction mockSearch =
             mockedSearchFunction(mapOf(Arrays.asList("1", "2"), mapOf("globalRank", 451, "tldRank", 23, "tld", "co")));
         MatchProcessor processor =
-            new MatchProcessor("_tag", mockSearch, "_name", "domain", "entry", false, true, "domain", maxMatches);
+            new MatchProcessor("_tag", mockSearch, "_name", "domain", "entry", false, true, "domain", 1);
         IngestDocument ingestDocument =
             new IngestDocument("_index", "_type", "_id", "_routing", 1L, VersionType.INTERNAL, mapOf("domain", Arrays.asList("1", "2")));
 
@@ -263,8 +260,7 @@ public class MatchProcessorTests extends ESTestCase {
         assertThat(termQueryBuilder.values().get(1), equalTo("2"));
 
         // Check result
-        List<?> entries = ingestDocument.getFieldValue("entry", List.class);
-        Map<?, ?> entry = (Map<?, ?>) entries.get(0);
+        Map<?, ?> entry = ingestDocument.getFieldValue("entry", Map.class);
         assertThat(entry.size(), equalTo(3));
         assertThat(entry.get("globalRank"), equalTo(451));
         assertThat(entry.get("tldRank"), equalTo(23));