merge dev into vec-330

.github/workflows/extensive_vector_search_tests.yml

@@ -4,11 +4,6 @@ on:
   push:
     branches:
       - dev
-      - main
-  # TODO remove this after testing it in the PR
-  pull_request:
-    branches:
-      - dev
 
 jobs:
   test-exhaustive-vector-search:
@@ -74,7 +69,7 @@ jobs:
 
         sleep 5
 
-        python -m pytest standard -s --host 0.0.0.0 --port 5000 --extensive_vector_search --cov=aerospike_vector_search -vs
+        python -m pytest standard -s --host 0.0.0.0 --port 5000 --extensive_vector_search -vs
 
         mv .coverage coverage_data
       working-directory: tests

tests/standard/aio/test_extensive_vector_search.py

@@ -1,7 +1,7 @@
 import numpy as np
 import asyncio
 import pytest
-import random
+import time
 from aerospike_vector_search import types, AVSServerError
 
 import grpc
@@ -184,6 +184,11 @@ async def test_vector_search(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo1")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -221,6 +226,11 @@ async def test_vector_search_with_set_same_as_index(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo2")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -262,6 +272,11 @@ async def test_vector_search_with_set_different_than_name(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo3")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -303,6 +318,11 @@ async def test_vector_search_with_index_storage_different_than_name(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo4")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -344,6 +364,11 @@ async def test_vector_search_with_index_storage_different_location(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo5")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -385,6 +410,11 @@ async def test_vector_search_with_separate_namespace(
         session_vector_client.wait_for_index_completion(namespace="test", name="demo6")
     )
     await asyncio.gather(*tasks)
+
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     await grade_results(
         base_numpy,
         truth_numpy,
@@ -395,39 +425,6 @@ async def test_vector_search_with_separate_namespace(
     )
 
 
-async def test_vector_is_indexed(
-    session_vector_client, session_admin_client, with_latency
-):
-
-    result = await session_vector_client.is_indexed(
-        namespace="test",
-        key=str(random.randrange(10_000)),
-        index_name="demo2",
-        set_name="demo2",
-    )
-    assert result is True
-
-
-async def test_vector_is_indexed_timeout(
-    session_vector_client, session_admin_client, with_latency
-):
-    if not with_latency:
-        pytest.skip("Server latency too low to test timeout")
-    for i in range(10):
-        try:
-            result = await session_vector_client.is_indexed(
-                namespace="test",
-                key=str(random.randrange(10_000)),
-                index_name="demo2",
-                timeout=0.0001,
-            )
-        except AVSServerError as se:
-            if se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED:
-                assert se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED
-                return
-    assert "In several attempts, the timeout did not happen" == "TEST FAIL"
-
-
 async def test_vector_vector_search_timeout(
     session_vector_client, session_admin_client, with_latency
 ):

tests/standard/aio/test_vector_client_is_indexed.py

@@ -0,0 +1,85 @@
+import pytest
+import random
+from aerospike_vector_search import AVSServerError
+from aerospike_vector_search.aio import Client, AdminClient
+
+import asyncio
+import grpc
+
+# Define module-level constants for common arguments
+NAMESPACE = "test"
+SET_NAME = "isidxset"
+INDEX_NAME = "isidx"
+DIMENSIONS = 3
+VECTOR_FIELD = "vector"
+
+@pytest.fixture(scope="module", autouse=True)
+async def setup_teardown(session_vector_client: Client, session_admin_client: AdminClient):
+    # Setup: Create index and upsert records
+    await session_admin_client.index_create(
+        namespace=NAMESPACE,
+        sets=SET_NAME,
+        name=INDEX_NAME,
+        dimensions=DIMENSIONS,
+        vector_field=VECTOR_FIELD,
+    )
+    tasks = []
+    for i in range(10):
+        tasks.append(session_vector_client.upsert(
+            namespace=NAMESPACE,
+            set_name=SET_NAME,
+            key=str(i),
+            record_data={VECTOR_FIELD: [float(i)] * DIMENSIONS},
+        ))
+    tasks.append(session_vector_client.wait_for_index_completion(
+        namespace=NAMESPACE,
+        name=INDEX_NAME,
+    ))
+    await asyncio.gather(*tasks)
+    yield
+    # Teardown: remove records
+    tasks = []
+    for i in range(10):
+        tasks.append(session_vector_client.delete(
+            namespace=NAMESPACE,
+            set_name=SET_NAME,
+            key=str(i)
+        ))
+    await asyncio.gather(*tasks)
+
+    # Teardown: Drop index
+    await session_admin_client.index_drop(
+        namespace=NAMESPACE,
+        name=INDEX_NAME
+    )
+
+async def test_vector_is_indexed(
+    session_vector_client,
+):
+    result = await session_vector_client.is_indexed(
+        namespace=NAMESPACE,
+        key="0",
+        index_name=INDEX_NAME,
+        set_name=SET_NAME,
+    )
+    assert result is True
+
+
+async def test_vector_is_indexed_timeout(
+    session_vector_client, with_latency
+):
+    if not with_latency:
+        pytest.skip("Server latency too low to test timeout")
+    for _ in range(10):
+        try:
+            await session_vector_client.is_indexed(
+                namespace=NAMESPACE,
+                key="0",
+                index_name=INDEX_NAME,
+                timeout=0.0001,
+            )
+        except AVSServerError as se:
+            if se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED:
+                assert se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED
+                return
+    assert "In several attempts, the timeout did not happen" == "TEST FAIL"

tests/standard/sync/test_extensive_vector_search.py

@@ -1,6 +1,6 @@
 import numpy as np
 import pytest
-import random
+import time
 from aerospike_vector_search import types
 from aerospike_vector_search import AVSServerError
 import grpc
@@ -176,6 +176,10 @@ def test_vector_search(
 
     session_vector_client.wait_for_index_completion(namespace="test", name="demo1")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     grade_results(
         base_numpy,
         truth_numpy,
@@ -211,6 +215,10 @@ def test_vector_search_with_set_same_as_index(
 
     session_vector_client.wait_for_index_completion(namespace="test", name="demo2")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     grade_results(
         base_numpy,
         truth_numpy,
@@ -245,6 +253,10 @@ def test_vector_search_with_set_different_than_name(
     for j, vector in enumerate(base_numpy):
         put_vector(session_vector_client, vector, j, "example1")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     session_vector_client.wait_for_index_completion(namespace="test", name="demo3")
 
     grade_results(
@@ -283,6 +295,10 @@ def test_vector_search_with_index_storage_different_than_name(
 
     session_vector_client.wait_for_index_completion(namespace="test", name="demo4")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     grade_results(
         base_numpy,
         truth_numpy,
@@ -319,6 +335,10 @@ def test_vector_search_with_index_storage_different_location(
 
     session_vector_client.wait_for_index_completion(namespace="test", name="demo5")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     grade_results(
         base_numpy,
         truth_numpy,
@@ -355,6 +375,10 @@ def test_vector_search_with_separate_namespace(
 
     session_vector_client.wait_for_index_completion(namespace="test", name="demo6")
 
+    # Wait for index completion isn't perfect
+    # give the index some extra time since accuracy is the point of this test
+    time.sleep(5)
+
     grade_results(
         base_numpy,
         truth_numpy,
@@ -365,36 +389,6 @@ def test_vector_search_with_separate_namespace(
     )
 
 
-def test_vector_is_indexed(session_vector_client, session_admin_client):
-
-    result = session_vector_client.is_indexed(
-        namespace="test",
-        key=str(random.randrange(10_000)),
-        index_name="demo2",
-        set_name="demo2",
-    )
-
-    assert result is True
-
-
-def test_vector_is_indexed_timeout(
-    session_vector_client, session_admin_client, with_latency
-):
-    if not with_latency:
-        pytest.skip("Server latency too low to test timeout")
-
-    for i in range(10):
-        try:
-            result = session_vector_client.is_indexed(
-                namespace="test", key=500, index_name="demo2", timeout=0.0001
-            )
-        except AVSServerError as se:
-            if se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED:
-                assert se.rpc_error.code() == grpc.StatusCode.DEADLINE_EXCEEDED
-                return
-    assert "In several attempts, the timeout did not happen" == "TEST FAIL"
-
-
 def test_vector_vector_search_timeout(
     session_vector_client, session_admin_client, with_latency
 ):