use standard config in ingest processor intead of always return list (opensearch-project#2985) (opensearch-project#3031)

Signed-off-by: Mingshi Liu <mingshl@amazon.com>
(cherry picked from commit f4b4724)

Co-authored-by: Mingshi Liu <mingshl@amazon.com>

Author: opensearch-trigger-bot[bot]

common/build.gradle

@@ -35,6 +35,7 @@ dependencies {
         exclude group: 'com.google.j2objc', module: 'j2objc-annotations'
         exclude group: 'com.google.guava', module: 'listenablefuture'
     }
+    compileOnly 'com.jayway.jsonpath:json-path:2.9.0'
 }
 
 lombok {

common/src/main/java/org/opensearch/ml/common/utils/StringUtils.java

@@ -28,6 +28,7 @@
 import com.google.gson.JsonElement;
 import com.google.gson.JsonParser;
 import com.google.gson.JsonSyntaxException;
+import com.jayway.jsonpath.JsonPath;
 
 import lombok.extern.log4j.Log4j2;
 
@@ -293,4 +294,29 @@ public static String getJsonPath(String jsonPathWithSource) {
         // Extract the substring from the startIndex to the end of the input string
         return (startIndex != -1) ? jsonPathWithSource.substring(startIndex) : jsonPathWithSource;
     }
+
+    /**
+     * Checks if the given input string matches the JSONPath format.
+     *
+     * <p>The JSONPath format is a way to navigate and extract data from JSON documents.
+     * It uses a syntax similar to XPath for XML documents. This method attempts to compile
+     * the input string as a JSONPath expression using the {@link com.jayway.jsonpath.JsonPath}
+     * library. If the compilation succeeds, it means the input string is a valid JSONPath
+     * expression.
+     *
+     * @param input the input string to be checked for JSONPath format validity
+     * @return true if the input string is a valid JSONPath expression, false otherwise
+     */
+    public static boolean isValidJSONPath(String input) {
+        if (input == null || input.isBlank()) {
+            return false;
+        }
+        try {
+            JsonPath.compile(input); // This will throw an exception if the path is invalid
+            return true;
+        } catch (Exception e) {
+            return false;
+        }
+    }
+
 }

plugin/src/main/java/org/opensearch/ml/processor/MLInferenceIngestProcessor.java

@@ -7,7 +7,6 @@
 import static org.opensearch.ml.processor.InferenceProcessorAttributes.*;
 
 import java.io.IOException;
-import java.util.ArrayList;
 import java.util.Collection;
 import java.util.HashMap;
 import java.util.HashSet;
@@ -37,9 +36,7 @@
 import org.opensearch.script.ScriptService;
 import org.opensearch.script.TemplateScript;
 
-import com.jayway.jsonpath.Configuration;
 import com.jayway.jsonpath.JsonPath;
-import com.jayway.jsonpath.Option;
 
 /**
  * MLInferenceIngestProcessor requires a modelId string to call model inferences
@@ -75,11 +72,6 @@ public class MLInferenceIngestProcessor extends AbstractProcessor implements Mod
     public static final String DEFAULT_MODEl_INPUT = "{ \"parameters\": ${ml_inference.parameters} }";
     private final NamedXContentRegistry xContentRegistry;
 
-    private Configuration suppressExceptionConfiguration = Configuration
-        .builder()
-        .options(Option.SUPPRESS_EXCEPTIONS, Option.DEFAULT_PATH_LEAF_TO_NULL, Option.ALWAYS_RETURN_LIST)
-        .build();
-
     protected MLInferenceIngestProcessor(
         String modelId,
         List<Map<String, String>> inputMaps,
@@ -320,24 +312,29 @@ private void getMappedModelInputFromDocuments(
             Object documentFieldValue = ingestDocument.getFieldValue(originalFieldPath, Object.class);
             String documentFieldValueAsString = toString(documentFieldValue);
             updateModelParameters(modelInputFieldName, documentFieldValueAsString, modelParameters);
+            return;
         }
-        // else when cannot find field path in document, try check for nested array using json path
-        else {
-            if (documentFieldName.contains(DOT_SYMBOL)) {
-
-                Map<String, Object> sourceObject = ingestDocument.getSourceAndMetadata();
-                ArrayList<Object> fieldValueList = JsonPath
-                    .using(suppressExceptionConfiguration)
-                    .parse(sourceObject)
-                    .read(documentFieldName);
-                if (!fieldValueList.isEmpty()) {
-                    updateModelParameters(modelInputFieldName, toString(fieldValueList), modelParameters);
-                } else if (!ignoreMissing) {
-                    throw new IllegalArgumentException("cannot find field name defined from input map: " + documentFieldName);
+        // If the standard dot path fails, try to check for a nested array using JSON path
+        if (StringUtils.isValidJSONPath(documentFieldName)) {
+            Map<String, Object> sourceObject = ingestDocument.getSourceAndMetadata();
+            Object fieldValue = JsonPath.using(suppressExceptionConfiguration).parse(sourceObject).read(documentFieldName);
+
+            if (fieldValue != null) {
+                if (fieldValue instanceof List) {
+                    List<?> fieldValueList = (List<?>) fieldValue;
+                    if (!fieldValueList.isEmpty()) {
+                        updateModelParameters(modelInputFieldName, toString(fieldValueList), modelParameters);
+                    } else if (!ignoreMissing) {
+                        throw new IllegalArgumentException("Cannot find field name defined from input map: " + documentFieldName);
+                    }
+                } else {
+                    updateModelParameters(modelInputFieldName, toString(fieldValue), modelParameters);
                 }
             } else if (!ignoreMissing) {
-                throw new IllegalArgumentException("cannot find field name defined from input map: " + documentFieldName);
+                throw new IllegalArgumentException("Cannot find field name defined from input map: " + documentFieldName);
             }
+        } else {
+            throw new IllegalArgumentException("Cannot find field name defined from input map: " + documentFieldName);
         }
     }