`value_type` is useful regardless of scripting. (#22160)

Today we only expose `value_type` in scriptable aggregations, however it is
also useful with unmapped fields. I suspect we never noticed because
`value_type` was not documented (fixed) and most aggregations are scriptable.

Closes #20163

core/src/main/java/org/elasticsearch/search/aggregations/support/MissingValues.java

@@ -80,7 +80,7 @@ public enum MissingValues {
     }
 
     public static ValuesSource.Numeric replaceMissing(final ValuesSource.Numeric valuesSource, final Number missing) {
-        final boolean missingIsFloat = missing.longValue() != (long) missing.doubleValue();
+        final boolean missingIsFloat = missing.doubleValue() % 1 != 0;
         final boolean isFloatingPoint = valuesSource.isFloatingPoint() || missingIsFloat;
         return new ValuesSource.Numeric() {
 

core/src/main/java/org/elasticsearch/search/aggregations/support/ValuesSourceParserHelper.java

@@ -67,6 +67,17 @@ public final class ValuesSourceParserHelper {
         objectParser.declareField(ValuesSourceAggregationBuilder::missing, XContentParser::objectText,
                 new ParseField("missing"), ObjectParser.ValueType.VALUE);
 
+        objectParser.declareField(ValuesSourceAggregationBuilder::valueType, p -> {
+            ValueType valueType = ValueType.resolveForScript(p.text());
+            if (targetValueType != null && valueType.isNotA(targetValueType)) {
+                throw new ParsingException(p.getTokenLocation(),
+                        "Aggregation [" + objectParser.getName() + "] was configured with an incompatible value type ["
+                                + valueType + "]. It can only work on value of type ["
+                                + targetValueType + "]");
+            }
+            return valueType;
+        }, new ParseField("value_type", "valueType"), ObjectParser.ValueType.STRING);
+
         if (formattable) {
             objectParser.declareField(ValuesSourceAggregationBuilder::format, XContentParser::text,
                     new ParseField("format"), ObjectParser.ValueType.STRING);
@@ -75,17 +86,6 @@ public final class ValuesSourceParserHelper {
         if (scriptable) {
             objectParser.declareField(ValuesSourceAggregationBuilder::script, org.elasticsearch.script.Script::parse,
                     Script.SCRIPT_PARSE_FIELD, ObjectParser.ValueType.OBJECT_OR_STRING);
-
-            objectParser.declareField(ValuesSourceAggregationBuilder::valueType, p -> {
-                ValueType valueType = ValueType.resolveForScript(p.text());
-                if (targetValueType != null && valueType.isNotA(targetValueType)) {
-                    throw new ParsingException(p.getTokenLocation(),
-                            "Aggregation [" + objectParser.getName() + "] was configured with an incompatible value type ["
-                                    + valueType + "]. It can only work on value of type ["
-                                    + targetValueType + "]");
-                }
-                return valueType;
-            }, new ParseField("value_type", "valueType"), ObjectParser.ValueType.STRING);
         }
 
         if (timezoneAware) {

core/src/test/java/org/elasticsearch/search/aggregations/support/MissingValuesTests.java

@@ -293,4 +293,9 @@ public class MissingValuesTests extends ESTestCase {
             }
         }
     }
+
+    public void testFloatingPointDetection() {
+        assertFalse(MissingValues.replaceMissing(ValuesSource.Numeric.EMPTY, 3).isFloatingPoint());
+        assertTrue(MissingValues.replaceMissing(ValuesSource.Numeric.EMPTY, 3.5).isFloatingPoint());
+    }
 }

docs/reference/aggregations.asciidoc

@@ -92,11 +92,15 @@ enables defining all the "dynamic" expressions in the script as parameters, whic
 between calls (this will ensure the use of the cached compiled scripts in Elasticsearch).
 ===============================
 
-Scripts can generate a single value or multiple values per document. When generating multiple values, one can use the
-`script_values_sorted` settings to indicate whether these values are sorted or not. Internally, Elasticsearch can
-perform optimizations when dealing with sorted values (for example, with the `min` aggregations, knowing the values are
-sorted, Elasticsearch will skip the iterations over all the values and rely on the first value in the list to be the
-minimum value among all other values associated with the same document).
+Elasticsearch uses the type of the field in the mapping in order to figure out
+how to run the aggregation and format the response. However there are two cases
+in which Elasticsearch cannot figure out this information: unmapped fields (for
+instance in the case of a search request across multiple indices, and only some
+of them have a mapping for the field) and pure scripts. For those cases, it is
+possible to give Elasticsearch a hint using the `value_type` option, which
+accepts the following values: `string`, `long` (works for all integer types),
+`double` (works for all decimal types like `float` or `scaled_float`), `date`,
+`ip` and `boolean`.
 
 --
 

rest-api-spec/src/main/resources/rest-api-spec/test/search.aggregation/20_terms.yaml

@@ -465,3 +465,132 @@ setup:
   - match: { aggregations.str_terms.buckets.0.key: 1234 }
 
   - match: { aggregations.str_terms.buckets.0.doc_count: 2 }
+
+---
+"Unmapped strings":
+
+  - do:
+      index:
+        index: test_1
+        type: test
+        id: 1
+        body: {}
+
+  - do:
+      indices.refresh: {}
+
+  - do:
+      search:
+        body: { "size" : 0, "aggs" : { "string_terms" : { "terms" : { "field" : "unmapped_string", "value_type" : "string", "missing": "abc" } } } }
+
+  - match: { hits.total: 1 }
+
+  - length: { aggregations.string_terms.buckets: 1 }
+
+  - match: { aggregations.string_terms.buckets.0.key: "abc" }
+
+  - match: { aggregations.string_terms.buckets.0.doc_count: 1 }
+
+---
+"Unmapped booleans":
+
+  - do:
+      index:
+        index: test_1
+        type: test
+        id: 1
+        body: {}
+
+  - do:
+      indices.refresh: {}
+
+  - do:
+      search:
+        body: { "size" : 0, "aggs" : { "boolean_terms" : { "terms" : { "field" : "unmapped_boolean", "value_type" : "boolean", "missing": true } } } }
+
+  - match: { hits.total: 1 }
+
+  - length: { aggregations.boolean_terms.buckets: 1 }
+
+  - match: { aggregations.boolean_terms.buckets.0.key: 1 }
+
+  - match: { aggregations.boolean_terms.buckets.0.key_as_string: "true" }
+
+  - match: { aggregations.boolean_terms.buckets.0.doc_count: 1 }
+
+---
+"Unmapped dates":
+
+  - do:
+      index:
+        index: test_1
+        type: test
+        id: 1
+        body: {}
+
+  - do:
+      indices.refresh: {}
+
+  - do:
+      search:
+        body: { "size" : 0, "aggs" : { "date_terms" : { "terms" : { "field" : "unmapped_date", "value_type" : "date", "missing": "2016-05-11" } } } }
+
+  - match: { hits.total: 1 }
+
+  - length: { aggregations.date_terms.buckets: 1 }
+
+  - match: { aggregations.date_terms.buckets.0.key: 1462924800000 }
+
+  - match: { aggregations.date_terms.buckets.0.key_as_string: "2016-05-11T00:00:00.000Z" }
+
+  - match: { aggregations.date_terms.buckets.0.doc_count: 1 }
+
+---
+"Unmapped longs":
+
+  - do:
+      index:
+        index: test_1
+        type: test
+        id: 1
+        body: {}
+
+  - do:
+      indices.refresh: {}
+
+  - do:
+      search:
+        body: { "size" : 0, "aggs" : { "long_terms" : { "terms" : { "field" : "unmapped_long", "value_type" : "long", "missing": 3 } } } }
+
+  - match: { hits.total: 1 }
+
+  - length: { aggregations.long_terms.buckets: 1 }
+
+  - match: { aggregations.long_terms.buckets.0.key: 3 }
+
+  - match: { aggregations.long_terms.buckets.0.doc_count: 1 }
+
+---
+"Unmapped doubles":
+
+  - do:
+      index:
+        index: test_1
+        type: test
+        id: 1 
+        body: {}
+
+  - do:
+      indices.refresh: {}
+
+  - do:
+      search:
+        body: { "size" : 0, "aggs" : { "double_terms" : { "terms" : { "field" : "unmapped_double", "value_type" : "double", "missing": 3.5 } } } }
+
+  - match: { hits.total: 1 }
+
+  - length: { aggregations.double_terms.buckets: 1 }
+
+  - match: { aggregations.double_terms.buckets.0.key: 3.5 }
+
+  - match: { aggregations.double_terms.buckets.0.doc_count: 1 }