Skip to content
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

Adds preset contentRegistry for IngestProcessors #3281

Merged
merged 4 commits into from
Dec 31, 2024
Merged

Adds preset contentRegistry for IngestProcessors #3281

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
67c562a
add preset xContentRegistry to ingestProcessors for custom parametize…
brianf-aws Dec 16, 2024
30316f9
Adds UT for proving models depend on xContentRegistry for prediction
brianf-aws Dec 17, 2024
f9caf1b
apply spotless
brianf-aws Dec 17, 2024
d44d07b
Adds IT for Asymmetric Embedding scenario with MLInferenceIngestProce…
brianf-aws Dec 23, 2024
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
4 changes: 3 additions & 1 deletion plugin/src/main/java/org/opensearch/ml/plugin/MachineLearningPlugin.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1062,10 +1062,12 @@ public void loadExtensions(ExtensionLoader loader) {
@Override
public Map<String, org.opensearch.ingest.Processor.Factory> getProcessors(org.opensearch.ingest.Processor.Parameters parameters) {
Map<String, org.opensearch.ingest.Processor.Factory> processors = new HashMap<>();
NamedXContentRegistry contentRegistry = new NamedXContentRegistry(getNamedXContent());

processors
.put(
MLInferenceIngestProcessor.TYPE,
new MLInferenceIngestProcessor.Factory(parameters.scriptService, parameters.client, xContentRegistry)
new MLInferenceIngestProcessor.Factory(parameters.scriptService, parameters.client, contentRegistry)
);
return Collections.unmodifiableMap(processors);
}
Expand Down
55 changes: 55 additions & 0 deletions plugin/src/test/java/org/opensearch/ml/processor/MLInferenceIngestProcessorTests.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@
import org.opensearch.core.action.ActionListener;
import org.opensearch.core.xcontent.NamedXContentRegistry;
import org.opensearch.ingest.IngestDocument;
import org.opensearch.ml.common.FunctionName;
import org.opensearch.ml.common.dataset.remote.RemoteInferenceInputDataSet;
import org.opensearch.ml.common.input.MLInput;
import org.opensearch.ml.common.output.model.MLResultDataType;
Expand Down Expand Up @@ -138,6 +139,60 @@ public void testExecute_Exception() throws Exception {

}

/**
* Models that use the parameters field need to have a valid NamedXContentRegistry object to create valid MLInputs. For example
* <pre>
* PUT /_plugins/_ml/_predict/text_embedding/model_id
* {
* "parameters": {
* "content_type" : "query"
* },
* "text_docs" : ["what day is it today?"],
* "target_response" : ["sentence_embedding"]
* }
* </pre>
* These types of models like Local Asymmetric embedding models use the parameters field.
* And as such we need to test that having the contentRegistry throws an exception as it can not
* properly create a valid MLInput to perform prediction
*
* @implNote If you check the stack trace of the test you will see it tells you that it's a direct consequence of xContentRegistry being null
*/
public void testExecute_xContentRegistryNullWithLocalModel_throwsException() throws Exception {
brianf-aws marked this conversation as resolved.
Show resolved Hide resolved
// Set the registry to null and reset after exiting the test
xContentRegistry = null;

String localModelInput =
"{ \"text_docs\": [\"What day is it today?\"],\"target_response\": [\"sentence_embedding\"], \"parameters\": { \"contentType\" : \"query\"} }";

MLInferenceIngestProcessor processor = createMLInferenceProcessor(
"local_model_id",
null,
null,
null,
false,
FunctionName.TEXT_EMBEDDING.toString(),
false,
false,
false,
localModelInput
);
try {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more. 

// Act & Assert: Verify NullPointerException and its message
    NullPointerException exception = assertThrows(
        NullPointerException.class,
        () -> processor.execute(ingestDocument, handler),
        "Expected NullPointerException due to null xContentRegistry"
    );

    assertTrue(exception.getMessage().contains("Cannot invoke"),
               "Exception message should indicate a failure due to null mlInput");

What do you think about this?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I like this, but the problem is that the exception is passed by the handler its not done by the method itself. So this wouldn't be possible thats the reason why this class and more specifically this method has a catch to make sure that an exception is not possible. i.e. the handler passes an exception only.

String npeMessage =
"Cannot invoke \"org.opensearch.ml.common.input.MLInput.setAlgorithm(org.opensearch.ml.common.FunctionName)\" because \"mlInput\" is null";

processor.execute(ingestDocument, handler);
verify(handler)
.accept(
isNull(),
argThat(exception -> exception instanceof NullPointerException && exception.getMessage().equals(npeMessage))
);
} catch (Exception e) {
assertEquals("this catch block should not get executed.", e.getMessage());
}
// reset to mocked object
xContentRegistry = mock(NamedXContentRegistry.class);
}

/**
* test nested object document with array of Map<String,String>
*/
Expand Down
123 changes: 122 additions & 1 deletion plugin/src/test/java/org/opensearch/ml/rest/RestMLInferenceIngestProcessorIT.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
package org.opensearch.ml.rest;

import static org.opensearch.ml.common.MLTask.MODEL_ID_FIELD;
import static org.opensearch.ml.utils.TestData.SENTENCE_TRANSFORMER_MODEL_HASH_VALUE;
import static org.opensearch.ml.utils.TestData.SENTENCE_TRANSFORMER_MODEL_URL;
import static org.opensearch.ml.utils.TestHelper.makeRequest;

Expand All @@ -28,6 +29,7 @@
import org.opensearch.ml.utils.TestHelper;

import com.google.common.collect.ImmutableList;
import com.jayway.jsonpath.DocumentContext;
import com.jayway.jsonpath.JsonPath;

public class RestMLInferenceIngestProcessorIT extends MLCommonsRestTestCase {
Expand Down Expand Up @@ -434,6 +436,110 @@ public void testMLInferenceProcessorLocalModelObjectField() throws Exception {
Assert.assertEquals(0.49191704, (Double) embedding2.get(0), 0.005);
}

public void testMLInferenceIngestProcessor_simulatesIngestPipelineSuccessfully_withAsymmetricEmbeddingModelUsingPassageContentType()
throws Exception {
String taskId = registerModel(TestHelper.toJsonString(registerAsymmetricEmbeddingModelInput()));
waitForTask(taskId, MLTaskState.COMPLETED);
getTask(client(), taskId, response -> {
assertNotNull(response.get(MODEL_ID_FIELD));
this.localModelId = (String) response.get(MODEL_ID_FIELD);
try {
String deployTaskID = deployModel(this.localModelId);
waitForTask(deployTaskID, MLTaskState.COMPLETED);

getModel(client(), this.localModelId, model -> { assertEquals("DEPLOYED", model.get("model_state")); });
} catch (IOException | InterruptedException e) {
throw new RuntimeException(e);
}
});

String asymmetricPipelineName = "asymmetric_embedding_pipeline";
String createPipelineRequestBody = "{\n"
+ " \"description\": \"ingest PASSAGE text and generate a embedding using an asymmetric model\",\n"
+ " \"processors\": [\n"
+ " {\n"
+ " \"ml_inference\": {\n"
+ "\n"
+ " \"model_input\": \"{\\\"text_docs\\\":[\\\"${input_map.text_docs}\\\"],\\\"target_response\\\":[\\\"sentence_embedding\\\"],\\\"parameters\\\":{\\\"content_type\\\":\\\"passage\\\"}}\",\n"
+ " \"function_name\": \"text_embedding\",\n"
+ " \"model_id\": \""
+ this.localModelId
+ "\",\n"
+ " \"input_map\": [\n"
+ " {\n"
+ " \"text_docs\": \"description\"\n"
+ " }\n"
+ " ],\n"
+ " \"output_map\": [\n"
+ " {\n"
+ "\n"
+ " "
+ " \"fact_embedding\": \"$.inference_results[0].output[0].data\"\n"
+ " }\n"
+ " ]\n"
+ " }\n"
+ " }\n"
+ " ]\n"
+ "}";

createPipelineProcessor(createPipelineRequestBody, asymmetricPipelineName);
String sampleDocuments = "{\n"
+ " \"docs\": [\n"
+ " {\n"
+ " \"_index\": \"my-index\",\n"
+ " \"_id\": \"1\",\n"
+ " \"_source\": {\n"
+ " \"title\": \"Central Park\",\n"
+ " \"description\": \"A large public park in the heart of New York City, offering a wide range of recreational activities.\"\n"
+ " }\n"
+ " },\n"
+ " {\n"
+ " \"_index\": \"my-index\",\n"
+ " \"_id\": \"2\",\n"
+ " \"_source\": {\n"
+ " \"title\": \"Empire State Building\",\n"
+ " \"description\": \"An iconic skyscraper in New York City offering breathtaking views from its observation deck.\"\n"
+ " }\n"
+ " }\n"
+ " ]\n"
+ "}\n";

Map simulateResponseDocuments = simulateIngestPipeline(asymmetricPipelineName, sampleDocuments);

DocumentContext documents = JsonPath.parse(simulateResponseDocuments);

List centralParkFactEmbedding = documents.read("docs.[0].*._source.fact_embedding.*");
assertEquals(768, centralParkFactEmbedding.size());
Assert.assertEquals(0.5137818, (Double) centralParkFactEmbedding.get(0), 0.005);

List empireStateBuildingFactEmbedding = documents.read("docs.[1].*._source.fact_embedding.*");
assertEquals(768, empireStateBuildingFactEmbedding.size());
Assert.assertEquals(0.4493039, (Double) empireStateBuildingFactEmbedding.get(0), 0.005);
}

private MLRegisterModelInput registerAsymmetricEmbeddingModelInput() {
MLModelConfig modelConfig = TextEmbeddingModelConfig
.builder()
.modelType("bert")
.frameworkType(TextEmbeddingModelConfig.FrameworkType.SENTENCE_TRANSFORMERS)
.embeddingDimension(768)
.queryPrefix("query >>")
.passagePrefix("passage >> ")
.build();

return MLRegisterModelInput
.builder()
.modelName("test_model_name")
.version("1.0.0")
.functionName(FunctionName.TEXT_EMBEDDING)
.modelFormat(MLModelFormat.TORCH_SCRIPT)
.modelConfig(modelConfig)
.url(SENTENCE_TRANSFORMER_MODEL_URL)
.deployModel(false)
.hashValue(SENTENCE_TRANSFORMER_MODEL_HASH_VALUE)
.build();
}

// TODO: add tests for other local model types such as sparse/cross encoders
public void testMLInferenceProcessorLocalModelNestedField() throws Exception {

Expand Down Expand Up @@ -560,6 +666,21 @@ protected void createPipelineProcessor(String requestBody, final String pipeline

}

protected Map simulateIngestPipeline(String pipelineName, String sampleDocuments) throws IOException {
Response ingestionResponse = TestHelper
.makeRequest(
client(),
"POST",
"/_ingest/pipeline/" + pipelineName + "/_simulate",
null,
sampleDocuments,
ImmutableList.of(new BasicHeader(HttpHeaders.USER_AGENT, ""))
);
assertEquals(200, ingestionResponse.getStatusLine().getStatusCode());

return parseResponseToMap(ingestionResponse);
}

protected void createIndex(String indexName, String requestBody) throws Exception {
Response response = makeRequest(
client(),
Expand Down Expand Up @@ -602,7 +723,7 @@ protected MLRegisterModelInput registerModelInput() throws IOException, Interrup
.modelConfig(modelConfig)
.url(SENTENCE_TRANSFORMER_MODEL_URL)
.deployModel(false)
.hashValue("e13b74006290a9d0f58c1376f9629d4ebc05a0f9385f40db837452b167ae9021")
.hashValue(SENTENCE_TRANSFORMER_MODEL_HASH_VALUE)
.build();
}

Expand Down
1 change: 1 addition & 0 deletions plugin/src/test/java/org/opensearch/ml/utils/TestData.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ public class TestData {
"https://github.com/opensearch-project/ml-commons/blob/2.x/ml-algorithms/src/test/resources/org/opensearch/ml/engine/algorithms/text_embedding/all-MiniLM-L6-v2_torchscript_huggingface.zip?raw=true";
public static final String SENTENCE_TRANSFORMER_MODEL_URL =
"https://github.com/opensearch-project/ml-commons/blob/2.x/ml-algorithms/src/test/resources/org/opensearch/ml/engine/algorithms/text_embedding/traced_small_model.zip?raw=true";
public static final String SENTENCE_TRANSFORMER_MODEL_HASH_VALUE = "e13b74006290a9d0f58c1376f9629d4ebc05a0f9385f40db837452b167ae9021";
public static final String TIME_FIELD = "timestamp";
public static final String HUGGINGFACE_TRANSFORMER_MODEL_HASH_VALUE =
"e13b74006290a9d0f58c1376f9629d4ebc05a0f9385f40db837452b167ae9021";
Expand Down
Loading