Merge branch 'nosqlbench:main' into main

adapter-dynamodb/pom.xml

@@ -43,7 +43,7 @@
         <dependency>
             <groupId>com.amazonaws</groupId>
             <artifactId>aws-java-sdk-dynamodb</artifactId>
-            <version>1.12.464</version>
+            <version>1.12.466</version>
         </dependency>
 
     </dependencies>

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/PineconeDriverAdapter.java

@@ -17,6 +17,7 @@
 package io.nosqlbench.adapter.pinecone;
 
 import io.nosqlbench.adapter.pinecone.ops.PineconeOp;
+import io.nosqlbench.api.config.standard.NBConfigModel;
 import io.nosqlbench.api.config.standard.NBConfiguration;
 import io.nosqlbench.engine.api.activityimpl.OpMapper;
 import io.nosqlbench.engine.api.activityimpl.uniform.BaseDriverAdapter;
@@ -41,4 +42,9 @@ public class PineconeDriverAdapter extends BaseDriverAdapter<PineconeOp, Pinecon
         return (s) -> new PineconeSpace(s,cfg);
     }
 
+    @Override
+    public NBConfigModel getConfigModel() {
+        return super.getConfigModel().add(PineconeSpace.getConfigModel());
+    }
+
 }

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/opdispensers/PineconeDeleteOpDispenser.java

@@ -16,6 +16,7 @@
 
 package io.nosqlbench.adapter.pinecone.opdispensers;
 
+import com.google.protobuf.ListValue;
 import com.google.protobuf.Struct;
 import com.google.protobuf.Value;
 import io.nosqlbench.adapter.pinecone.PineconeDriverAdapter;
@@ -27,7 +28,10 @@ import io.pinecone.proto.DeleteRequest;
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 
-import java.util.*;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Map;
+import java.util.Optional;
 import java.util.function.LongFunction;
 
 
@@ -61,7 +65,6 @@ public class PineconeDeleteOpDispenser extends PineconeOpDispenser {
     /**
      * @param op The ParsedOp used to build the Request
      * @return A function that will take a long (the current cycle) and return a Pinecone DeleteRequest
-     *
      * The pattern used here is to accommodate the way Request types are constructed for Pinecone.
      * Requests use a Builder pattern, so at time of instantiation the methods should be chained together.
      * For each method in the chain a function is created here and added to the chain of functions
@@ -95,16 +98,10 @@ public class PineconeDeleteOpDispenser extends PineconeOpDispenser {
             rFunc = l -> finalFunc.apply(l).setDeleteAll(af.apply(l));
         }
 
-        Optional<LongFunction<String>> filterFunction = op.getAsOptionalFunction("filter", String.class);
+        Optional<LongFunction<Map>> filterFunction = op.getAsOptionalFunction("filter", Map.class);
         if (filterFunction.isPresent()) {
             LongFunction<DeleteRequest.Builder> finalFunc = rFunc;
-            LongFunction<Struct> builtFilter = l -> {
-                String[] filterFields = filterFunction.get().apply(l).split(" ");
-                return Struct.newBuilder().putFields(filterFields[0],
-                    Value.newBuilder().setStructValue(Struct.newBuilder().putFields(filterFields[1],
-                            Value.newBuilder().setNumberValue(Integer.parseInt(filterFields[2])).build()))
-                            .build()).build();
-            };
+            LongFunction<Struct> builtFilter = buildFilterStruct(filterFunction.get());
             rFunc = l -> finalFunc.apply(l).setFilter(builtFilter.apply(l));
         }
 
@@ -112,5 +109,4 @@ public class PineconeDeleteOpDispenser extends PineconeOpDispenser {
         return l -> finalRFunc.apply(l).build();
     }
 
-
 }

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/opdispensers/PineconeDescribeIndexStatsOpDispenser.java

@@ -27,6 +27,7 @@ import io.pinecone.proto.DescribeIndexStatsRequest;
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 
+import java.util.Map;
 import java.util.Optional;
 import java.util.function.LongFunction;
 
@@ -63,18 +64,14 @@ public class PineconeDescribeIndexStatsOpDispenser extends PineconeOpDispenser {
      */
     private LongFunction<DescribeIndexStatsRequest> createDescribeIndexStatsRequestFunction(ParsedOp op) {
         LongFunction<DescribeIndexStatsRequest.Builder> rFunc = l -> DescribeIndexStatsRequest.newBuilder();
-        Optional<LongFunction<String>> filterFunction = op.getAsOptionalFunction("filter", String.class);
+
+        Optional<LongFunction<Map>> filterFunction = op.getAsOptionalFunction("filter", Map.class);
         if (filterFunction.isPresent()) {
             LongFunction<DescribeIndexStatsRequest.Builder> finalFunc = rFunc;
-            LongFunction<Struct> builtFilter = l -> {
-                String[] filterFields = filterFunction.get().apply(l).split(" ");
-                return Struct.newBuilder().putFields(filterFields[0],
-                        Value.newBuilder().setStructValue(Struct.newBuilder().putFields(filterFields[1],
-                                        Value.newBuilder().setNumberValue(Integer.parseInt(filterFields[2])).build()))
-                                .build()).build();
-            };
+            LongFunction<Struct> builtFilter = buildFilterStruct(filterFunction.get());
             rFunc = l -> finalFunc.apply(l).setFilter(builtFilter.apply(l));
         }
+
         LongFunction<DescribeIndexStatsRequest.Builder> finalRFunc = rFunc;
         return l -> finalRFunc.apply(l).build();
     }

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/opdispensers/PineconeOpDispenser.java

@@ -16,12 +16,17 @@
 
 package io.nosqlbench.adapter.pinecone.opdispensers;
 
+import com.google.protobuf.ListValue;
+import com.google.protobuf.Struct;
+import com.google.protobuf.Value;
 import io.nosqlbench.adapter.pinecone.PineconeDriverAdapter;
 import io.nosqlbench.adapter.pinecone.PineconeSpace;
 import io.nosqlbench.adapter.pinecone.ops.PineconeOp;
 import io.nosqlbench.engine.api.activityimpl.BaseOpDispenser;
 import io.nosqlbench.engine.api.templating.ParsedOp;
 
+import java.util.List;
+import java.util.Map;
 import java.util.function.LongFunction;
 
 public abstract class PineconeOpDispenser extends BaseOpDispenser<PineconeOp, PineconeSpace> {
@@ -37,4 +42,37 @@ public abstract class PineconeOpDispenser extends BaseOpDispenser<PineconeOp, Pi
         this.targetFunction = targetFunction;
     }
 
+    protected LongFunction<Struct> buildFilterStruct(LongFunction<Map> filterFunction) {
+        return  l -> {
+            Map<String,Object> filterFields = filterFunction.apply(l);
+            Value comparatorVal;
+            Object comparator = filterFields.get("comparator");
+            if (comparator instanceof String) {
+                comparatorVal = Value.newBuilder().setStringValue((String) comparator).build();
+            } else if (comparator instanceof Number) {
+                comparatorVal = Value.newBuilder().setNumberValue((Double) comparator).build();
+            } else if (comparator instanceof List) {
+                comparatorVal = Value.newBuilder().setListValue(generateListValue((List) comparator)).build();
+            } else {
+                throw new RuntimeException("Invalid type for filter comparator specified");
+            }
+            return Struct.newBuilder().putFields((String) filterFields.get("filterfield"),
+                Value.newBuilder().setStructValue(
+                        Struct.newBuilder().putFields((String) filterFields.get("operator"),
+                            comparatorVal))
+                    .build()).build();
+        };
+    }
+
+    protected ListValue generateListValue(List comparator) {
+        ListValue.Builder listValueBuilder = ListValue.newBuilder();
+        for (Object entry : comparator) {
+            Value value = Value.newBuilder().setStringValue(String.valueOf(entry)).build();
+            listValueBuilder.addValues(value);
+        }
+        return listValueBuilder.build();
+    }
+
+
+
 }

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/opdispensers/PineconeQueryOpDispenser.java

@@ -17,7 +17,6 @@
 package io.nosqlbench.adapter.pinecone.opdispensers;
 
 import com.google.protobuf.Struct;
-import com.google.protobuf.Value;
 import io.nosqlbench.adapter.pinecone.PineconeDriverAdapter;
 import io.nosqlbench.adapter.pinecone.PineconeSpace;
 import io.nosqlbench.adapter.pinecone.ops.PineconeOp;
@@ -58,14 +57,14 @@ public class PineconeQueryOpDispenser extends PineconeOpDispenser {
     /**
      * @param op The ParsedOp used to build the Request
      * @return A function that will take a long (the current cycle) and return a Pinecone QueryRequest Builder
-     *
+     * <p>
      * The pattern used here is to accommodate the way Request types are constructed for Pinecone.
      * Requests use a Builder pattern, so at time of instantiation the methods should be chained together.
      * For each method in the chain a function is created here and added to the chain of functions
      * called at time of instantiation.
-     *
+     * <p>
      * The QueryVector objects used by the QueryRequest as sufficiently sophisticated in their own building process
-     * that it has been broken out into a separate method. At runtime they are built separately and then added
+     * that it has been broken out into a separate method. At runtime, they are built separately and then added
      * to the build chain by the builder returned by this method.
      */
     private LongFunction<QueryRequest.Builder> createQueryRequestFunc(ParsedOp op) {
@@ -114,16 +113,10 @@ public class PineconeQueryOpDispenser extends PineconeOpDispenser {
             rFunc = l -> finalFunc.apply(l).addAllVector(alf.apply(l));
         }
 
-        Optional<LongFunction<String>> filterFunction = op.getAsOptionalFunction("filter", String.class);
+        Optional<LongFunction<Map>> filterFunction = op.getAsOptionalFunction("filter", Map.class);
         if (filterFunction.isPresent()) {
             LongFunction<QueryRequest.Builder> finalFunc = rFunc;
-            LongFunction<Struct> builtFilter = l -> {
-                String[] filterFields = filterFunction.get().apply(l).split(" ");
-                return Struct.newBuilder().putFields(filterFields[0],
-                        Value.newBuilder().setStructValue(Struct.newBuilder().putFields(filterFields[1],
-                                        Value.newBuilder().setNumberValue(Integer.parseInt(filterFields[2])).build()))
-                                .build()).build();
-            };
+            LongFunction<Struct> builtFilter = buildFilterStruct(filterFunction.get());
             rFunc = l -> finalFunc.apply(l).setFilter(builtFilter.apply(l));
         }
 
@@ -133,7 +126,7 @@ public class PineconeQueryOpDispenser extends PineconeOpDispenser {
     /**
      * @param op the ParsedOp from which the Query Vector objects will be built
      * @return an Iterable Collection of QueryVector objects to be added to a Pinecone QueryRequest
-     *
+     * <p>
      * This method interrogates the subsection of the ParsedOp defined for QueryVector parameters and constructs
      * a list of QueryVectors based on the included values, or returns null if this section is not populated. The
      * base function returns either the List of vectors or null, while the interior function builds the vectors
@@ -158,11 +151,10 @@ public class PineconeQueryOpDispenser extends PineconeOpDispenser {
                     qvb.setTopK((Integer) vector.get("top_k"));
                 }
                 if (vector.containsKey("filter")) {
-                    String[] rawVals = ((String)vector.get("filter")).split(" ");
-                    qvb.setFilter(Struct.newBuilder().putFields(rawVals[0],
-                        Value.newBuilder().setStructValue(Struct.newBuilder().putFields(rawVals[1],
-                                Value.newBuilder().setNumberValue(Integer.parseInt(rawVals[2])).build()))
-                            .build()).build());
+                    LongFunction<Struct> builtFilter = buildFilterStruct(l2 -> {
+                        return (Map) vector.get("filter");
+                    });
+                    qvb.setFilter(builtFilter.apply(l));
                 }
                 if (vector.containsKey("sparse_values")) {
                     Map<String,String> sparse_values = (Map<String, String>) vector.get("sparse_values");

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/opdispensers/PineconeUpsertOpDispenser.java

@@ -72,6 +72,8 @@ public class PineconeUpsertOpDispenser extends PineconeOpDispenser {
             List<Map<String, Object>> vectors = listLongFunction.apply(l);
             for (Map<String, Object> vector : vectors) {
                 Vector.Builder vb = Vector.newBuilder();
+                // No need to check for key, it is invalid if id is not there, let it throw an exception
+                vb.setId(vector.get("id").toString());
                 String[] rawValues = ((String) vector.get("values")).split(",");
                 ArrayList<Float> floatValues = new ArrayList<>();
                 for (String val : rawValues) {

adapter-pinecone/src/main/java/io/nosqlbench/adapter/pinecone/ops/PineconeQueryOp.java

@@ -46,6 +46,9 @@ public class PineconeQueryOp extends PineconeOp {
     public void run() {
         QueryResponse response = connection.getBlockingStub().query(request);
         if (logger.isDebugEnabled()) {
+            for (ScoredVector scored : response.getMatchesList()) {
+                logger.debug(scored.getId() + ": " + scored.getScore());
+            }
             for (SingleQueryResults results : response.getResultsList()) {
                 for (ScoredVector scored : results.getMatchesList()) {
                     logger.debug(scored.getId() + ": " + scored.getScore());

adapter-pinecone/src/main/resources/activities/pinecone-delete.yaml

@@ -0,0 +1,16 @@
+scenarios:
+  default:
+    delete: run driver=pinecone tags==blocks:delete cycles=1
+
+blocks:
+  delete:
+    ops:
+      op1:
+        delete: "test-index"
+        namespace: "example_namespace"
+        #ids: "item_0,item_6"
+        deleteall: false
+        filter:
+          filterfield: "category"
+          operator: "$eq"
+          comparator: "sports"

adapter-pinecone/src/main/resources/activities/pinecone-example.yaml

@@ -1,19 +0,0 @@
-bindings:
-  vector: Need function to generate a list of float values here
-  comparator: Uniform(1,1000) -> int
-
-blocks:
-  query:
-    ops:
-      query-op1:
-        query: "test-index"
-        vector: {vector}
-        namespace: "test-namespace"
-        top_k: 10
-        filters:
-          - filter_field: "value"
-            operator: "$lt"
-            comparator: {comparator}
-        include_values: true
-        include_metadata: true
-

adapter-pinecone/src/main/resources/activities/pinecone-fetch.yaml

@@ -0,0 +1,11 @@
+scenarios:
+  default:
+    fetch: run driver=pinecone tags==blocks:fetch cycles=1
+
+blocks:
+  fetch:
+    ops:
+      op1:
+        fetch: "test-index"
+        namespace: "example_namespace"
+        ids: "item_2,item_3,item_4"

adapter-pinecone/src/main/resources/activities/pinecone-query.yaml

@@ -0,0 +1,18 @@
+scenarios:
+  default:
+    query: run driver=pinecone tags==blocks:query cycles=10
+
+blocks:
+  query:
+    ops:
+      op1:
+        query: "test-index"
+        vector: "0.8602578079921012,0.12103044768221516,0.7737329191858439,0.4521093269320254,0.29351661477669416,0.4261807015226558,0.14131665592103335,0.882370813029422,0.4412833140430886,0.9916525700115515"
+        namespace: "example_namespace"
+        topk: 10
+        include_values: true
+        include_metadata: true
+        filter:
+          filterfield: "category"
+          operator: "$eq"
+          comparator: "movies"

adapter-pinecone/src/main/resources/activities/pinecone-stats.yaml

@@ -0,0 +1,10 @@
+scenarios:
+  default:
+    fetch: run driver=pinecone tags==blocks:describeindexstats cycles=1
+
+blocks:
+  describeindexstats:
+    ops:
+      op1:
+        describeindexstats: "test-index"
+        #filter: ""

adapter-pinecone/src/main/resources/activities/pinecone-upsert.yaml

@@ -0,0 +1,22 @@
+scenarios:
+  default:
+    query: run driver=pinecone tags==blocks:upsert cycles=1000
+
+blocks:
+  upsert:
+    ops:
+      op1:
+        upsert: "test-index"
+        namespace: "example_namespace"
+        upsert_vectors:
+          - id: "item_0"
+            values: "0.8238042071878214,0.6579519242642494,0.2772098082162267,0.11434681368630942,0.7496610470760962,0.08914691629812001,0.13249049306247204,0.8875869003282576,0.21615924382746318,0.8454796243176312"
+            metadata:
+              category: "sports"
+              colors: "blue,red,green"
+              games_played: 100
+          - id: "item_1"
+            values: "0.45058789301987323,0.41195231042505165,0.07228720002814844,0.24340801453997024,0.4010484623888162,0.9761375066056239,0.6844972055528982,0.4376943130598503,0.5966792562087805,0.7210604043175083"
+            metadata:
+              category: "music"
+              colors: "purple,green,yellow"

adapter-pinecone/src/main/resources/pinecone.md

@@ -43,7 +43,10 @@ ops:
     # The number of results to return for each query.
     top_k: int_query_topk
     # You can use vector metadata to limit your search. See https://www.pinecone.io/docs/metadata-filtering/
-    filter: <field operator compval>
+    filter:
+      filterfield: metadata_field
+      operator: [$lt, $eq, $gt, ...]
+      comparator: value
     # Indicates whether vector values are included in the response.
     include_values: boolean
     # Indicates whether metadata is included in the response as well as the ids.
@@ -53,7 +56,10 @@ ops:
         values: csv_separated_floats
         top_k: int_val
         namespace: string_val
-        filter: <field operator compval>
+        filter:
+          filterfield: metadata_field
+          operator: [$lt, $eq, $gt, ...]
+          comparator: value
         sparse_values:
           indices: list_of_ints
           values: list_of_floats
@@ -61,7 +67,10 @@ ops:
         values: csv_separated_floats
         top_k: int_val
         namespace: string_val
-        filter: <field operator compval>
+        filter:
+          filterfield: metadata_field
+          operator: [$lt, $eq, $gt, ...]
+          comparator: value
         sparse_values:
           indices: list_of_ints
           values: list_of_floats
@@ -76,13 +85,19 @@ ops:
     namespace: delete_namespace
     ids: csv_list_of_vectors_to_delete
     deleteall: [true,false]
-    filter: <field operator compval>
+    filter:
+      filterfield: metadata_field
+      operator: [$lt, $eq, $gt, ...]
+      comparator: value
 
 # A describe index stats op. Specify metadata filters to narrow the range of indices described.
   describe-index-stats-example:
     type: describe-index-stats
     index: describe_index
-    filter: <field operator compval>
+    filter:
+      filterfield: metadata_field
+      operator: [$lt, $eq, $gt, ...]
+      comparator: value
 
 # A pinecone fetch op
   fetch-example:

adapter-pinecone/src/test/java/io/nosqlbench/adapter/pinecone/PineconeOpMapperTest.java

@@ -64,31 +64,40 @@ public class PineconeOpMapperTest {
     @Test
     public void testQueryOpDispenserSimple() {
         ParsedOp pop = parsedOpFor("""
-                ops:
-                  op1:
-                     type: "query"
-                     index: "test-index"
-                     vector: "1.0,2.0,3.0"
+            ops:
+              op1:
+                 type: "query"
+                 index: "test-index"
+                 vector: "1.0,2.0,3.0"
+                 namespace: "test-namespace"
+                 top_k: 10
+                 filter:
+                   filterfield: "field"
+                   operator: "$gt"
+                   comparator: 2.0
+                 include_values: true
+                 include_metadata: true
+                 query_vectors:
+                   - id: 1
+                     values: "1.0,2.0,3.0"
+                     top_k: 8
                      namespace: "test-namespace"
-                     top_k: 10
-                     filter: "value $lt 2"
-                     include_values: true
-                     include_metadata: true
-                     query_vectors:
-                       - id: 1
-                         values: "1.0,2.0,3.0"
-                         top_k: 8
-                         namespace: "test-namespace"
-                         filter: "value $lt 2"
-                         sparse_values:
-                           indices: "1,2,3"
-                           values: "1.0,2.0,3.0"
-                       - id: 2
-                         values: "4.0,5.0,6.0"
-                         top_k: 11
-                         namespace: "test-namespace"
-                         filter: "value $gt 10"
-                """);
+                     filter:
+                       filterfield: "field"
+                       operator: "$lt"
+                       comparator: 9.0
+                     sparse_values:
+                       indices: "1,2,3"
+                       values: "1.0,2.0,3.0"
+                   - id: 2
+                     values: "4.0,5.0,6.0"
+                     top_k: 11
+                     namespace: "test-namespace"
+                     filter:
+                       filterfield: "field"
+                       operator: "$eq"
+                       comparator: "val"
+            """);
         OpDispenser<? extends PineconeOp> dispenser = mapper.apply(pop);
         assert(dispenser instanceof PineconeQueryOpDispenser);
         PineconeOp op = dispenser.apply(0);
@@ -105,7 +114,10 @@ public class PineconeOpMapperTest {
                  ids: "1.0,2.0,3.0"
                  namespace: "test-namespace"
                  deleteall: true
-                 filter: "value $gt 10"
+                 filter:
+                   filterfield: "key"
+                   operator: "$eq"
+                   comparator: "val"
             """);
         OpDispenser<? extends PineconeOp> dispenser = mapper.apply(pop);
         assert(dispenser instanceof PineconeDeleteOpDispenser);
@@ -120,7 +132,13 @@ public class PineconeOpMapperTest {
               op1:
                  type: "describeindexstats"
                  index: "test-index"
-                 filter: "value $gt 10"
+                 filter:
+                   filterfield: "color"
+                   operator: "$eq"
+                   comparator:
+                     - "green"
+                     - "yellow"
+                     - "red"
             """);
         OpDispenser<? extends PineconeOp> dispenser = mapper.apply(pop);
         assert(dispenser instanceof PineconeDescribeIndexStatsOpDispenser);

mvn-defaults/pom.xml

@@ -300,7 +300,7 @@
             <dependency>
                 <groupId>com.amazonaws</groupId>
                 <artifactId>aws-java-sdk-s3</artifactId>
-                <version>1.12.465</version>
+                <version>1.12.466</version>
             </dependency>
 
             <dependency>