nucliadb 6.2.1.post2798__py3-none-any.whl → 6.2.1.post2803__py3-none-any.whl

nucliadb/ingest/orm/knowledgebox.py

@@ -146,6 +146,27 @@ class KnowledgeBox:
                 kb_shards.actual = -1
 
                 vs_external_indexes = []
+
+                # HACK! Currently, we share responsibility of deciding where to
+                # store extracted vectors with processing. Depending on whether
+                # it sends the vectorset id or not (in the extracted vectors
+                # wrapper) nucliadb will store the extracted vectors in a different place
+                #
+                # Right now, processing behaviour is not setting vectorset ids
+                # if there's only one semantic model configured. This is done
+                # for Bw/c with KBs previous to vectorsets.
+                #
+                # We now hardcode this assumption here, so we can annotate each
+                # vectorset with a mark and don't depend on processing
+                # information to decide the storage key. Once this is done we'll
+                # be able to force processing to always send vectorset ids and
+                # remove that bw/c behavior
+                #
+                if len(semantic_models) == 1:
+                    storage_key_kind = knowledgebox_pb2.VectorSetConfig.StorageKeyKind.LEGACY
+                else:
+                    storage_key_kind = knowledgebox_pb2.VectorSetConfig.StorageKeyKind.VECTORSET_PREFIX
+
                 for vectorset_id, semantic_model in semantic_models.items():  # type: ignore
                     # if this KB uses a matryoshka model, we can choose a different
                     # dimension
@@ -172,6 +193,7 @@ class KnowledgeBox:
                             vector_dimension=dimension,
                         ),
                         matryoshka_dimensions=semantic_model.matryoshka_dimensions,
+                        storage_key_kind=storage_key_kind,
                     )
                     await datamanagers.vectorsets.set(txn, kbid=kbid, config=vectorset_config)
 

{nucliadb-6.2.1.post2798.dist-info → nucliadb-6.2.1.post2803.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb
-Version: 6.2.1.post2798
+Version: 6.2.1.post2803
 Home-page: https://docs.nuclia.dev/docs/management/nucliadb/intro
 Author: NucliaDB Community
 Author-email: nucliadb@nuclia.com
@@ -22,10 +22,10 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: >=3.9, <4
 Description-Content-Type: text/markdown
-Requires-Dist: nucliadb-telemetry[all]>=6.2.1.post2798
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post2798
-Requires-Dist: nucliadb-protos>=6.2.1.post2798
-Requires-Dist: nucliadb-models>=6.2.1.post2798
+Requires-Dist: nucliadb-telemetry[all]>=6.2.1.post2803
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post2803
+Requires-Dist: nucliadb-protos>=6.2.1.post2803
+Requires-Dist: nucliadb-models>=6.2.1.post2803
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nucliadb-node-binding>=2.26.0
 Requires-Dist: nuclia-models>=0.24.2

{nucliadb-6.2.1.post2798.dist-info → nucliadb-6.2.1.post2803.dist-info}/RECORD

@@ -138,7 +138,7 @@ nucliadb/ingest/orm/brain.py,sha256=Hzq-3aarKaUCiUoa8H83unRUfduRE9TsQH1dEq0mvZY,
 nucliadb/ingest/orm/broker_message.py,sha256=JYYUJIZEL_EqovQuw6u-FmEkjyoYlxIXJq9hFekOiks,6441
 nucliadb/ingest/orm/entities.py,sha256=2PslT1FZ6yCvJtjR0UpKTSzxJrtS-C_gZx4ZTWHunTc,15759
 nucliadb/ingest/orm/exceptions.py,sha256=k4Esv4NtL4TrGTcsQpwrSfDhPQpiYcRbB1SpYmBX5MY,1432
-nucliadb/ingest/orm/knowledgebox.py,sha256=dBetjoJBYT6JuGmMHiqjcfJeD8qJrK3MQt9X03IrHRA,23228
+nucliadb/ingest/orm/knowledgebox.py,sha256=UpWJrVaVfCtk8R4qfSR6h6vzwOKXa8Teuwkna5QSljE,24508
 nucliadb/ingest/orm/metrics.py,sha256=OkwMSPKLZcKba0ZTwtTiIxwBgaLMX5ydhGieKvi2y7E,1096
 nucliadb/ingest/orm/resource.py,sha256=FIC7Tzef_0hoN92tFwN8PUi8IHlUkyifPJxsA1_XumM,55727
 nucliadb/ingest/orm/utils.py,sha256=vCe_9UxHu26JDFGLwQ0wH-XyzJIpQCTK-Ow9dtZR5Vg,2716
@@ -335,9 +335,9 @@ nucliadb/writer/tus/local.py,sha256=7jYa_w9b-N90jWgN2sQKkNcomqn6JMVBOVeDOVYJHto,
 nucliadb/writer/tus/s3.py,sha256=vF0NkFTXiXhXq3bCVXXVV-ED38ECVoUeeYViP8uMqcU,8357
 nucliadb/writer/tus/storage.py,sha256=ToqwjoYnjI4oIcwzkhha_MPxi-k4Jk3Lt55zRwaC1SM,2903
 nucliadb/writer/tus/utils.py,sha256=MSdVbRsRSZVdkaum69_0wku7X3p5wlZf4nr6E0GMKbw,2556
-nucliadb-6.2.1.post2798.dist-info/METADATA,sha256=4dRh7zEqoaQHFsY1knkDYmSJ7ujs66iLDvrcK4GEvn4,4689
-nucliadb-6.2.1.post2798.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-nucliadb-6.2.1.post2798.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.2.1.post2798.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.2.1.post2798.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-nucliadb-6.2.1.post2798.dist-info/RECORD,,
+nucliadb-6.2.1.post2803.dist-info/METADATA,sha256=NZ6FUiG4RGSNyA919ez19edmW1geeUPg9AIFVSRrt4k,4689
+nucliadb-6.2.1.post2803.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+nucliadb-6.2.1.post2803.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.2.1.post2803.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.2.1.post2803.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+nucliadb-6.2.1.post2803.dist-info/RECORD,,