Fix

Author: Nikos Papailiou

.github/workflows/quarto-render.yml

@@ -36,7 +36,7 @@ jobs:
       - name: "Quarto render"
         shell: bash
         run: |
-          pip install quartodoc PyYAML click
+          pip install quartodoc PyYAML click griffe==0.32.3
           # create a symlink to the tiledbvcf python package, so it doesn't have to be installed
           #ln -s apis/python/src/tiledb/vector_search
           quartodoc build

_quarto.yml

@@ -39,7 +39,7 @@ quartodoc:
           - index.Index
           - flat_index.FlatIndex
           - ivf_flat_index.IVFFlatIndex
-          - ingestion.ingest
+          - ingestion
 
 website:
   favicon: "documentation/assets/tiledb.ico"

apis/python/src/tiledb/vector_search/flat_index.py

@@ -13,8 +13,8 @@ class FlatIndex(Index):
     Parameters
     ----------
     uri: str
-        URI of datataset
-    config: None
+        URI of the index
+    config: Optional[Mapping[str, Any]]
         config dictionary, defaults to None
     """
 
@@ -52,7 +52,7 @@ def __init__(
 
     def query_internal(
         self,
-        targets: np.ndarray,
+        queries: np.ndarray,
         k: int = 10,
         nthreads: int = 8,
     ):
@@ -61,22 +61,20 @@ def query_internal(
 
         Parameters
         ----------
-        targets: numpy.ndarray
-            ND Array of query targets
+        queries: numpy.ndarray
+            ND Array of queries
         k: int
-            Number of top results to return per target
-        nqueries: int
-            Number of queries
+            Number of top results to return per query
         nthreads: int
             Number of threads to use for query
         """
         # TODO:
-        # - typecheck targets
+        # - typecheck queries
         # - add all the options and query strategies
 
-        assert targets.dtype == np.float32
+        assert queries.dtype == np.float32
 
-        targets_m = array_to_matrix(np.transpose(targets))
-        r = query_vq_heap(self._db, targets_m, self._ids, k, nthreads)
+        queries_m = array_to_matrix(np.transpose(queries))
+        r = query_vq_heap(self._db, queries_m, self._ids, k, nthreads)
 
         return np.transpose(np.array(r))

apis/python/src/tiledb/vector_search/index.py

@@ -8,9 +8,19 @@
 
 
 class Index:
+    """
+    Open a Vector index
+
+    Parameters
+    ----------
+    uri: str
+        URI of the index
+    config: Optional[Mapping[str, Any]]
+        config dictionary, defaults to None
+    """
     def __init__(
         self,
-        uri,
+        uri: str,
         config: Optional[Mapping[str, Any]] = None,
     ):
         # If the user passes a tiledb python Config object convert to a dictionary
@@ -26,14 +36,14 @@ def __init__(
         self.index_version = self.group.meta.get("index_version", "")
 
 
-    def query(self, targets: np.ndarray, k, **kwargs):
+    def query(self, queries: np.ndarray, k, **kwargs):
         # TODO merge results based on scores and use higher k to improve retrieval
         updated_ids = set(self.read_updated_ids())
-        internal_results = self.query_internal(targets, k, **kwargs)
+        internal_results = self.query_internal(queries, k, **kwargs)
         if self.update_arrays_uri is None:
             return internal_results
-        addition_results = self.query_additions(targets, k)
-        merged_results = np.zeros((targets.shape[0], k), dtype=np.uint64)
+        addition_results = self.query_additions(queries, k)
+        merged_results = np.zeros((queries.shape[0], k), dtype=np.uint64)
         query_id = 0
         for query in internal_results:
             res_id = 0
@@ -48,17 +58,17 @@ def query(self, targets: np.ndarray, k, **kwargs):
             query_id += 1
         return merged_results
 
-    def query_internal(self, targets: np.ndarray, k, **kwargs):
+    def query_internal(self, queries: np.ndarray, k, **kwargs):
         raise NotImplementedError
 
-    def query_additions(self, targets: np.ndarray, k):
-        assert targets.dtype == np.float32
+    def query_additions(self, queries: np.ndarray, k):
+        assert queries.dtype == np.float32
 
         additions_vectors, additions_external_ids = self.read_additions()
-        targets_m = array_to_matrix(np.transpose(targets))
+        queries_m = array_to_matrix(np.transpose(queries))
         r = query_vq_heap_pyarray(
             array_to_matrix(np.transpose(additions_vectors).astype(self.dtype)),
-            targets_m,
+            queries_m,
             StdVector_u64(additions_external_ids),
             k,
             8)

apis/python/src/tiledb/vector_search/ingestion.py

@@ -2,9 +2,6 @@
 from functools import partial
 
 from tiledb.cloud.dag import Mode
-from tiledb.vector_search.index import Index
-from tiledb.vector_search.flat_index import FlatIndex
-from tiledb.vector_search.ivf_flat_index import IVFFlatIndex
 from tiledb.vector_search._tiledbvspy import *
 import numpy as np
 
@@ -31,7 +28,7 @@ def ingest(
     verbose: bool = False,
     trace_id: Optional[str] = None,
     mode: Mode = Mode.LOCAL,
-) -> Index:
+):
     """
     Ingest vectors into TileDB.
 
@@ -100,6 +97,9 @@ def ingest(
     from tiledb.cloud.utilities import get_logger
     from tiledb.cloud.utilities import set_aws_context
     from tiledb.vector_search.storage_formats import storage_formats, STORAGE_VERSION
+    from tiledb.vector_search.index import Index
+    from tiledb.vector_search.flat_index import FlatIndex
+    from tiledb.vector_search.ivf_flat_index import IVFFlatIndex
 
     # use index_group_uri for internal clarity
     index_group_uri = index_uri

apis/python/src/tiledb/vector_search/ivf_flat_index.py

@@ -22,16 +22,16 @@ class IVFFlatIndex(Index):
     Parameters
     ----------
     uri: str
-        URI of datataset
-    config: None
+        URI of the index
+    config: Optional[Mapping[str, Any]]
         config dictionary, defaults to None
     memory_budget: int
         Main memory budget. If not provided, no memory budget is applied.
     """
 
     def __init__(
         self,
-        uri,
+        uri: str,
         config: Optional[Mapping[str, Any]] = None,
         memory_budget: int = -1,
     ):
@@ -100,7 +100,7 @@ def query_internal(
         queries: numpy.ndarray
             ND Array of queries
         k: int
-            Number of top results to return per target
+            Number of top results to return per query
         nprobe: int
             number of probes
         nthreads: int
@@ -193,13 +193,11 @@ def taskgraph_query(
         queries: numpy.ndarray
             ND Array of queries
         k: int
-            Number of top results to return per target
+            Number of top results to return per query
         nprobe: int
             number of probes
         nthreads: int
             Number of threads to use for query
-        use_nuv_implementation: bool
-            wether to use the nuv query implementation. Default: False
         mode: Mode
             If provided the query will be executed using TileDB cloud taskgraphs.
             For distributed execution you can use REALTIME or BATCH mode
@@ -209,6 +207,8 @@ def taskgraph_query(
         num_workers: int
             Only relevant for taskgraph based execution.
             If provided, this is the number of workers to use for the query execution.
+        config: None
+            config dictionary, defaults to None
         """
         from tiledb.cloud import dag
         from tiledb.cloud.dag import Mode