new: add type annotations, add comment for hdf5 dataset

Author: joein

dataset_reader/h5_reader.py

@@ -27,6 +27,14 @@ def read_data(self) -> Iterator[Record]:
 if __name__ == '__main__':
     import os
     from benchmark.settings import DATASET_DIR
+
+    # h5py file 4 keys:
+    # `train` - float vectors (num vectors 1183514)
+    # `test` - float vectors (num vectors 10000)
+    # `neighbors` - int - indices of nearest neighbors for test (num items 10k, each item
+    # contains info about 100 nearest neighbors)
+    # `distances` - float - distances for nearest neighbors for test vectors
+
     test_path = os.path.join(DATASET_DIR, 'glove-100-angular', 'glove-100-angular.hdf5')
     record = next(H5Reader(test_path).read_data())
-    print(record)
+    print(record)

engine/base_client/search.py

@@ -9,12 +9,12 @@ class BaseSearcher:
     MP_CONTEXT = None
 
     @classmethod
-    def init_client(cls, host, connection_params, search_params):
+    def init_client(cls, host: str, connection_params: dict, search_params: dict):
         cls.search_params = search_params
         raise NotImplementedError()
 
     @classmethod
-    def search_one(cls, vector, meta_conditions) -> List[Tuple[int, float]]:
+    def search_one(cls, vector: List[float], meta_conditions) -> List[Tuple[int, float]]:
         raise NotImplementedError()
 
     @classmethod

engine/base_client/upload.py

@@ -16,7 +16,14 @@ def init_client(cls, host, connection_params: dict):
         raise NotImplementedError()
 
     @classmethod
-    def upload(cls, url, records: Iterable[Record], batch_size, parallel, connection_params):
+    def upload(
+        cls,
+        url: str,
+        records: Iterable[Record],
+        batch_size: int,
+        parallel: int,
+        connection_params: dict,
+    ) -> List[float]:
         latencies = []
 
         if parallel == 1:
@@ -38,16 +45,15 @@ def upload(cls, url, records: Iterable[Record], batch_size, parallel, connection
         return latencies
 
     @classmethod
-    def _upload_batch(cls, ids, vectors, metadata) -> float:
+    def _upload_batch(
+        cls, ids: List[int], vectors: List[list], metadata: List[Optional[dict]]
+    ) -> float:
         start = time.perf_counter()
         cls.upload_batch(ids, vectors, metadata)
         return time.perf_counter() - start
 
     @classmethod
     def upload_batch(
-            cls,
-            ids: List[int],
-            vectors: List[list],
-            metadata: List[Optional[dict]]
+        cls, ids: List[int], vectors: List[list], metadata: List[Optional[dict]]
     ):
         raise NotImplementedError()