Refactor README and Vicinity class to support any serializable item type

README.md

@@ -35,8 +35,7 @@ Vicinity is a light-weight, low-dependency vector store. It provides a simple an
 
 There are many nearest neighbors packages and methods out there. However, we found it difficult to compare them. Every package has its own interface, quirks, and limitations, and learning a new package can be time-consuming. In addition to that, how do you effectively evaluate different packages? How do you know which one is the best for your use case?
 
-
- This is where Vicinity comes in. Instead of learning a new interface for each new package or backend, Vicinity provides a unified interface for all backends. This allows you to easily experiment with different indexing methods and distance metrics and choose the best one for your use case. Vicinity also provides a simple way to evaluate the performance of different backends, allowing you to measure the queries per second and recall.
+This is where Vicinity comes in. Instead of learning a new interface for each new package or backend, Vicinity provides a unified interface for all backends. This allows you to easily experiment with different indexing methods and distance metrics and choose the best one for your use case. Vicinity also provides a simple way to evaluate the performance of different backends, allowing you to measure the queries per second and recall.
 
 ## Quickstart
 
@@ -49,13 +48,13 @@ Optionally, [install any of the supported backends](#installation), or simply in
 pip install vicinity[all]
 ```
 
-
 The following code snippet demonstrates how to use Vicinity for nearest neighbor search:
+
 ```python
 import numpy as np
 from vicinity import Vicinity, Backend, Metric
 
-# Create some dummy data
+# Create some dummy data as strings or other serializable objects
 items = ["triforce", "master sword", "hylian shield", "boomerang", "hookshot"]
 vectors = np.random.rand(len(items), 128)
 
@@ -82,12 +81,14 @@ results = vicinity.query(query_vectors, k=3)
 ```
 
 Saving and loading a vector store:
+
 ```python
 vicinity.save('my_vector_store')
 vicinity = Vicinity.load('my_vector_store')
 ```
 
 Evaluating a backend:
+
 ```python
 # Use the first 1000 vectors as query vectors
 query_vectors = vectors[:1000]
@@ -100,6 +101,7 @@ qps, recall = vicinity.evaluate(
 ```
 
 ## Main Features
+
 Vicinity provides the following features:
 - Lightweight: Minimal dependencies and fast performance.
 - Flexible Backend Support: Use different backends for vector storage and search.
@@ -108,6 +110,7 @@ Vicinity provides the following features:
 - Easy to Use: Simple and intuitive API.
 
 ## Supported Backends
+
 The following backends are supported:
 - `BASIC`: A simple (exact matching) flat index for vector storage and search.
 - [HNSW](https://github.com/nmslib/hnswlib): Hierarchical Navigable Small World Graph (HNSW) for ANN search using hnswlib.
@@ -126,8 +129,6 @@ The following backends are supported:
   - `ivfpqr`: Inverted file search with product quantizer and refinement.
 - [VOYAGER](https://github.com/spotify/voyager): Voyager is a library for performing fast approximate nearest-neighbor searches on an in-memory collection of vectors.
 
-
-
 NOTE: the ANN backends do not support dynamic deletion. To delete items, you need to recreate the index. Insertion is supported in the following backends: `FAISS`, `HNSW`, and `Usearch`. The `BASIC` backend supports both insertion and deletion.
 
 ### Backend Parameters
@@ -159,7 +160,9 @@ NOTE: the ANN backends do not support dynamic deletion. To delete items, you nee
 |                 | `m`                 | The number of connections between nodes in the tree’s internal data structure.                                                              | `16`                |
 
 ## Installation
+
 The following installation options are available:
+
 ```bash
 # Install the base package
 pip install vicinity

tests/conftest.py

@@ -24,7 +24,7 @@
 @pytest.fixture(scope="session")
 def items() -> list[str]:
     """Fixture providing a list of item names."""
-    return [f"item{i}" for i in range(1, 10001)]
+    return [f"item{i}" if i % 2 == 0 else {"name": f"item{i}", "id": i} for i in range(1, 10001)]
 
 
 @pytest.fixture(scope="session")

tests/test_vicinity.py

@@ -183,18 +183,17 @@ def test_index_vector_store(vicinity_with_basic_backend_and_store: Vicinity, vec
         vicinity_with_basic_backend_and_store.get_vector_by_index([-1])
 
 
-def test_vicinity_insert_duplicate(vicinity_instance: Vicinity, query_vector: np.ndarray) -> None:
+def test_vicinity_insert_duplicate(items: list[str], vicinity_instance: Vicinity, query_vector: np.ndarray) -> None:
     """
     Test that Vicinity.insert raises ValueError when inserting duplicate items.
 
     :param vicinity_instance: A Vicinity instance.
     :raises ValueError: If inserting items that already exist.
     """
-    new_items = ["item1"]
     new_vector = query_vector
 
     with pytest.raises(ValueError):
-        vicinity_instance.insert(new_items, new_vector[None, :])
+        vicinity_instance.insert(items[0], new_vector[None, :])
 
 
 def test_vicinity_delete_nonexistent(vicinity_instance: Vicinity) -> None:
@@ -281,22 +280,23 @@ def test_vicinity_delete_and_query(vicinity_instance: Vicinity, items: list[str]
         return
 
     # Delete some items from the Vicinity instance
-    items_to_delete = ["item2", "item4", "item6"]
+    non_existing_items_indices = [0, 1, 2]
+    items_to_delete = [items[i] for i in non_existing_items_indices]
     vicinity_instance.delete(items_to_delete)
 
     # Ensure the items are no longer in the items list
     for item in items_to_delete:
         assert item not in vicinity_instance.items
 
     # Query using a vector of an item that wasn't deleted
-    item3_index = items.index("item3")
-    item3_vector = vectors[item3_index]
+    existsing_item_index = 3
+    item3_vector = vectors[existsing_item_index]
 
     results = vicinity_instance.query(item3_vector, k=10)
     returned_items = [item for item, _ in results[0]]
 
     # Check that the queried item is in the results
-    assert "item3" in returned_items
+    assert items[existsing_item_index] in returned_items
 
 
 def test_vicinity_evaluate(vicinity_instance: Vicinity, vectors: np.ndarray) -> None:

vicinity/vicinity.py

@@ -29,7 +29,7 @@ class Vicinity:
 
     def __init__(
         self,
-        items: Sequence[str],
+        items: Sequence[Any],
         backend: AbstractBackend,
         metadata: Union[dict[str, Any], None] = None,
         vector_store: BasicVectorStore | None = None,
@@ -49,7 +49,7 @@ def __init__(
             raise ValueError(
                 "Your vector space and list of items are not the same length: " f"{len(backend)} != {len(items)}"
             )
-        self.items: list[str] = list(items)
+        self.items: list[Any] = list(items)
         self.backend: AbstractBackend = backend
         self.metadata = metadata or {}
         self.vector_store = vector_store
@@ -74,7 +74,7 @@ def __len__(self) -> int:
     def from_vectors_and_items(
         cls: type[Vicinity],
         vectors: npt.NDArray,
-        items: Sequence[str],
+        items: Sequence[Any],
         backend_type: Backend | str = Backend.BASIC,
         store_vectors: bool = False,
         **kwargs: Any,
@@ -211,7 +211,7 @@ def load(cls, filename: PathLike) -> Vicinity:
 
         with open(folder_path / "data.json", "rb") as file_handle:
             data: dict[str, Any] = orjson.loads(file_handle.read())
-        items: Sequence[str] = data["items"]
+        items: Sequence[Any] = data["items"]
 
         metadata: dict[str, Any] = data["metadata"]
         backend_type = Backend(data["backend_type"])
@@ -227,7 +227,7 @@ def load(cls, filename: PathLike) -> Vicinity:
 
         return instance
 
-    def insert(self, tokens: Sequence[str], vectors: npt.NDArray) -> None:
+    def insert(self, tokens: Sequence[Any], vectors: npt.NDArray) -> None:
         """
         Insert new items into the vector space.
 
@@ -241,16 +241,16 @@ def insert(self, tokens: Sequence[str], vectors: npt.NDArray) -> None:
         if vectors.shape[1] != self.dim:
             raise ValueError("The inserted vectors must have the same dimension as the backend.")
 
-        item_set = set(self.items)
         for token in tokens:
-            if token in item_set:
-                raise ValueError(f"Token {token} is already in the vector space.")
+            for item in self.items:
+                if item == token:
+                    raise ValueError(f"Token {token} is already in the vector space.")
             self.items.append(token)
         self.backend.insert(vectors)
         if self.vector_store is not None:
             self.vector_store.insert(vectors)
 
-    def delete(self, tokens: Sequence[str]) -> None:
+    def delete(self, tokens: Sequence[Any]) -> None:
         """
         Delete tokens from the vector space.
 
@@ -260,10 +260,15 @@ def delete(self, tokens: Sequence[str]) -> None:
         :param tokens: A list of tokens to remove from the vector space.
         :raises ValueError: If any passed tokens are not in the vector space.
         """
-        try:
-            curr_indices = [self.items.index(token) for token in tokens]
-        except ValueError as exc:
-            raise ValueError(f"Token {exc} was not in the vector space.") from exc
+        seen_tokens = []
+        curr_indices = []
+        for idx, elem in enumerate(self.items):
+            for token in tokens:
+                if elem == token:
+                    curr_indices.append(idx)
+                    seen_tokens.append(token)
+        if len(seen_tokens) < len(tokens):
+            raise ValueError("Not all tokens were in the vector space.")
 
         self.backend.delete(curr_indices)
         if self.vector_store is not None: