Merge pull request #2 from turbopuffer/fix_query_bug

Fix assert being hit when querying without vector values

RELEASE.md

@@ -0,0 +1,9 @@
+# Release Build Instructions
+
+To build and publish the PyPI package:
+
+1. `python -m pip install --upgrade build`
+1. `python -m pip install --upgrade twine`
+1. `rm -rf dist/`
+2. `python -m build`
+3. `python -m twine upload dist/*`

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "turbopuffer"
-version = "0.1.1"
+version = "0.1.2"
 description = "Python Client for accessing the turbopuffer API"
 authors = ["turbopuffer Inc. <info@turbopuffer.com>"]
 maintainers = ["Jacob Wirth"]

tests/test_vectors.py

@@ -64,7 +64,7 @@ def test_delete_vectors():
         ns.upsert([tpuf.VectorRow(id=2)])
         assert False, "Upserting to delete should not be allowed"
     except ValueError as err:
-        assert err.args == ('VectorRow.vector cannot be None, use Namespace.delete([ids...]) instead.',)
+        assert err.args == ('upsert() call would result in a vector deletion, use Namespace.delete([ids...]) instead.',)
 
     try:
         ns.upsert([tpuf.VectorRow(id=2, dist=5)])
@@ -77,7 +77,7 @@ def test_delete_vectors():
         ns.upsert([{'id': 6}])
         assert False, "Upserting to delete should not be allowed"
     except ValueError as err:
-        assert err.args == ('VectorRow.vector cannot be None, use Namespace.delete([ids...]) instead.',)
+        assert err.args == ('upsert() call would result in a vector deletion, use Namespace.delete([ids...]) instead.',)
 
     # Test delete single row
     ns.delete(2)
@@ -155,11 +155,15 @@ def check_result(row, expected):
         assert row.id == expected.id
         assert row.attributes == expected.attributes
         if isinstance(expected.vector, list):
+            assert isinstance(row.vector, list)
             assert abs(row.vector[0] - expected.vector[0]) < 0.000001
             assert abs(row.vector[1] - expected.vector[1]) < 0.000001
         else:
             assert row.vector == expected.vector
-        assert abs(row.dist - expected.dist) < 0.000001
+        if isinstance(expected.dist, float):
+            assert abs(row.dist - expected.dist) < 0.000001
+        else:
+            assert row.dist == expected.dist
 
     expected = [
         tpuf.VectorRow(id=7, vector=[0.7, 0.7], attributes={'hello': 'world'}, dist=0.01),
@@ -209,6 +213,22 @@ def check_result(row, expected):
         check_result(row, expected[i])
         i += 1
 
+    # Test query with no vectors
+    expected = [
+        tpuf.VectorRow(id=10),
+        tpuf.VectorRow(id=11),
+        tpuf.VectorRow(id=12),
+    ]
+    vector_set = ns.query(
+        top_k=3,
+        include_vectors=False,
+        filters={
+            'id': [['In', [10, 11, 12]]]
+        },
+    )
+    for i in range(len(vector_set)): # Use VectorResult in index mode
+        check_result(vector_set[i], expected[i])
+
 def test_list_vectors():
     ns = tpuf.Namespace('client_test')
 

turbopuffer/namespace.py

@@ -14,6 +14,13 @@ class Namespace:
     backend: Backend
 
     def __init__(self, name: str, api_key: Optional[str] = None):
+        """
+        Creates a new turbopuffer.Namespace object for querying the turbopuffer API.
+
+        This function does not make any API calls on its own.
+
+        Specifying an api_key here will override the global configuration for API calls to this namespace.
+        """
         self.name = name
         self.backend = Backend(api_key)
 

turbopuffer/vectors.py

@@ -37,9 +37,7 @@ class _(JSONSerializable.Meta):
     def __post_init__(self):
         if not isinstance(self.id, int):
             raise ValueError('VectorRow.id must be an int, got:', type(self.id))
-        if self.vector is None and self.dist is None:
-            raise ValueError('VectorRow.vector cannot be None, use Namespace.delete([ids...]) instead.')
-        if not isinstance(self.vector, list) and self.dist is None:
+        if self.vector is not None and not isinstance(self.vector, list):
             raise ValueError('VectorRow.vector must be a list, got:', type(self.vector))
         if self.attributes is not None and not isinstance(self.attributes, dict):
             raise ValueError('VectorRow.attributes must be a dict, got:', type(self.attributes))

turbopuffer/version.py

@@ -1 +1 @@
-VERSION = '0.1.1'
+VERSION = '0.1.2'