[Fix] Fetch sparse-only vector over grpc (#467)

## Problem

An error is raised when fetching sparse vector data over grpc.

## Solution

- Add test case for fetch
- Modify response parsing so that error does not occur

## Type of Change

- [x] Bug fix (non-breaking change which fixes an issue)

Author: jhamon

.github/actions/test-data-asyncio/action.yaml

@@ -38,7 +38,7 @@ runs:
     - name: Run data plane tests
       id: data-plane-asyncio-tests
       shell: bash
-      run: poetry run pytest tests/integration/data_asyncio -s -vv
+      run: poetry run pytest tests/integration/data_asyncio --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
       env:
         PINECONE_API_KEY: ${{ inputs.PINECONE_API_KEY }}
         USE_GRPC: ${{ inputs.use_grpc }}

.github/actions/test-data-plane/action.yaml

@@ -52,7 +52,7 @@ runs:
     - name: Run data plane tests
       id: data-plane-tests
       shell: bash
-      run: poetry run pytest tests/integration/data
+      run: poetry run pytest tests/integration/data --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
       env:
         PINECONE_API_KEY: ${{ inputs.PINECONE_API_KEY }}
         USE_GRPC: ${{ inputs.use_grpc }}

.github/workflows/testing-integration-asyncio.yaml

@@ -26,7 +26,7 @@ jobs:
       - name: Run data plane tests
         id: data-plane-asyncio-tests
         shell: bash
-        run: poetry run pytest tests/integration/data_asyncio -s -vv
+        run: poetry run pytest tests/integration/data_asyncio --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
         env:
             PINECONE_API_KEY: ${{ secrets.PINECONE_API_KEY }}
 
@@ -51,6 +51,6 @@ jobs:
           include_asyncio: true
           include_dev: true
       - name: 'db_control asyncio'
-        run: poetry run pytest tests/integration/control_asyncio -s -vv
+        run: poetry run pytest tests/integration/control_asyncio --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
         env:
           PINECONE_API_KEY: '${{ secrets.PINECONE_API_KEY }}'

.github/workflows/testing-integration.yaml

@@ -21,7 +21,7 @@ jobs:
         with:
           include_asyncio: true
       - name: 'Run integration tests'
-        run: poetry run pytest tests/integration/inference -s -vv
+        run: poetry run pytest tests/integration/inference --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
         env:
           PINECONE_DEBUG_CURL: 'true'
           PINECONE_API_KEY: '${{ secrets.PINECONE_API_KEY }}'
@@ -113,7 +113,7 @@ jobs:
       - name: Setup Poetry
         uses: ./.github/actions/setup-poetry
       - name: 'Run integration tests (REST)'
-        run: poetry run pytest tests/integration/control/serverless -s -vv
+        run: poetry run pytest tests/integration/control/serverless --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
         env:
           PINECONE_DEBUG_CURL: 'true'
           PINECONE_API_KEY: '${{ secrets.PINECONE_API_KEY }}'

.github/workflows/testing-unit.yaml

@@ -58,7 +58,7 @@ jobs:
           include_asyncio: true
 
       - name: Run unit tests (REST)
-        run: poetry run pytest --cov=pinecone --timeout=120 tests/unit
+        run: poetry run pytest --cov=pinecone --timeout=120 tests/unit --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG
       - name: Run unit tests (GRPC)
         if: ${{ matrix.use_grpc == true }}
-        run: poetry run pytest --cov=pinecone/grpc --timeout=120 tests/unit_grpc
+        run: poetry run pytest --cov=pinecone/grpc --timeout=120 tests/unit_grpc --retries 5 --retry-delay 35 -s -vv --log-cli-level=DEBUG

pinecone/grpc/utils.py

@@ -48,8 +48,8 @@ def parse_fetch_response(response: Message):
     for id, vec in vectors.items():
         vd[id] = _Vector(
             id=vec["id"],
-            values=vec["values"],
-            sparse_values=parse_sparse_values(vec.get("sparseValues")),
+            values=vec.get("values", None),
+            sparse_values=parse_sparse_values(vec.get("sparseValues", None)),
             metadata=vec.get("metadata", None),
             _check_type=False,
         )

pinecone/utils/tqdm.py

@@ -1,6 +1,12 @@
+import warnings
+
+__all__ = ["tqdm"]
+
 try:
-    # Use the notebook-friendly auto selection if tqdm is installed.
-    from tqdm.auto import tqdm
+    # Suppress the specific tqdm warning about IProgress
+    with warnings.catch_warnings():
+        warnings.filterwarnings("ignore", category=UserWarning, module="tqdm")
+        from tqdm.auto import tqdm
 except ImportError:
     # Fallback: define a dummy tqdm that supports the same interface.
     class tqdm:  # type: ignore

poetry.lock

@@ -85,6 +85,7 @@ pytest = "8.2.0"
 pytest-asyncio = "^0.25.2"
 pytest-cov = "2.10.1"
 pytest-mock = "3.6.1"
+pytest-retry = "^1.7.0"
 pytest-timeout = "2.2.0"
 pytest-benchmark = [
   { version = '5.0.0', python = ">=3.9,<4.0" }

pyproject.toml

@@ -143,7 +143,6 @@ def test_fetch_unspecified_namespace(self, idx):
         assert results.vectors["1"].values is not None
         assert results.vectors["4"].metadata is not None
 
-    @pytest.mark.skip(reason="Backend implementation not ready")
     def test_fetch_sparse_index(self, sparse_idx):
         sparse_idx.upsert(
             vectors=[
@@ -165,9 +164,11 @@ def test_fetch_sparse_index(self, sparse_idx):
         assert fetch_results.namespace == ""
         assert len(fetch_results.vectors) == 10
         for i in range(10):
+            logger.debug(fetch_results.vectors[str(i)])
             assert fetch_results.vectors[str(i)].id == str(i)
-            assert fetch_results.vectors[str(i)].values is not None
-            assert len(fetch_results.vectors[str(i)].values) == 2
+            assert fetch_results.vectors[str(i)].sparse_values is not None
+            assert len(fetch_results.vectors[str(i)].sparse_values.indices) == 2
+            assert len(fetch_results.vectors[str(i)].sparse_values.values) == 2
             assert fetch_results.vectors[str(i)].metadata is not None
             assert fetch_results.vectors[str(i)].metadata["genre"] == "action"
             assert fetch_results.vectors[str(i)].metadata["runtime"] == 120