nucliadb 6.3.4.post3796__py3-none-any.whl → 6.3.4.post3812__py3-none-any.whl

nucliadb/ingest/orm/knowledgebox.py

@@ -502,11 +502,12 @@ class KnowledgeBox:
         shard_manager = get_shard_manager()
         await shard_manager.create_vectorset(self.kbid, config)
 
-    async def delete_vectorset(self, vectorset_id: str):
-        vectorset_count = await datamanagers.vectorsets.count(self.txn, kbid=self.kbid)
-        if vectorset_count == 1:
-            raise VectorSetConflict("Deletion of your last vectorset is not allowed")
+    async def vectorset_marked_for_deletion(self, vectorset_id: str) -> bool:
+        key = KB_VECTORSET_TO_DELETE.format(kbid=self.kbid, vectorset=vectorset_id)
+        value = await self.txn.get(key)
+        return value is not None
 
+    async def delete_vectorset(self, vectorset_id: str):
         deleted = await datamanagers.vectorsets.delete(
             self.txn, kbid=self.kbid, vectorset_id=vectorset_id
         )
@@ -514,6 +515,10 @@ class KnowledgeBox:
             # already deleted
             return
 
+        vectorset_count = await datamanagers.vectorsets.count(self.txn, kbid=self.kbid)
+        if vectorset_count == 0:
+            raise VectorSetConflict("Deletion of your last vectorset is not allowed")
+
         # mark vectorset for async deletion
         deletion_mark_key = KB_VECTORSET_TO_DELETE.format(kbid=self.kbid, vectorset=vectorset_id)
         payload = VectorSetPurge(storage_key_kind=deleted.storage_key_kind)

nucliadb/search/predict.py

@@ -19,6 +19,7 @@
 #
 import base64
 import json
+import logging
 import os
 import random
 from enum import Enum
@@ -216,13 +217,16 @@ class PredictEngine:
         else:
             return {"X-STF-KBID": kbid}
 
-    async def check_response(self, resp: aiohttp.ClientResponse, expected_status: int = 200) -> None:
+    async def check_response(
+        self, kbid: str, resp: aiohttp.ClientResponse, expected_status: int = 200
+    ) -> None:
         if resp.status == expected_status:
             return
 
         if resp.status == 402:
             data = await resp.json()
             raise LimitsExceededError(402, data["detail"])
+
         try:
             data = await resp.json()
             try:
@@ -234,10 +238,22 @@ class PredictEngine:
             aiohttp.client_exceptions.ContentTypeError,
         ):
             detail = await resp.text()
-        if str(resp.status).startswith("5"):
-            logger.error(f"Predict API error at {resp.url}: {detail}")
-        else:
-            logger.info(f"Predict API error at {resp.url}: {detail}")
+
+        is_5xx_error = resp.status > 499
+        # NOTE: 512 is a special status code sent by learning predict api indicating that the error
+        # is related to an external generative model, so we don't want to log it as an error
+        is_external_generative_error = resp.status == 512
+        log_level = logging.ERROR if is_5xx_error and not is_external_generative_error else logging.INFO
+        logger.log(
+            log_level,
+            "Predict API error",
+            extra=dict(
+                kbid=kbid,
+                url=resp.url,
+                status_code=resp.status,
+                detail=detail,
+            ),
+        )
         raise ProxiedPredictAPIError(status=resp.status, detail=detail)
 
     @backoff.on_exception(
@@ -265,7 +281,7 @@ class PredictEngine:
             json=item.model_dump(),
             headers=self.get_predict_headers(kbid),
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         return await _parse_rephrase_response(resp)
 
     @predict_observer.wrap({"type": "chat_ndjson"})
@@ -294,7 +310,7 @@ class PredictEngine:
             headers=headers,
             timeout=None,
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         ident = resp.headers.get(NUCLIA_LEARNING_ID_HEADER)
         model = resp.headers.get(NUCLIA_LEARNING_MODEL_HEADER)
         return ident, model, get_chat_ndjson_generator(resp)
@@ -348,7 +364,7 @@ class PredictEngine:
             params=params,
             headers=self.get_predict_headers(kbid),
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         data = await resp.json()
         return QueryInfo(**data)
 
@@ -368,7 +384,7 @@ class PredictEngine:
             params={"text": sentence},
             headers=self.get_predict_headers(kbid),
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         data = await resp.json()
         return convert_relations(data)
 
@@ -387,7 +403,7 @@ class PredictEngine:
             headers=self.get_predict_headers(kbid),
             timeout=None,
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         data = await resp.json()
         return SummarizedResponse.model_validate(data)
 
@@ -405,7 +421,7 @@ class PredictEngine:
             json=item.model_dump(),
             headers=self.get_predict_headers(kbid),
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         data = await resp.json()
         return RerankResponse.model_validate(data)
 
@@ -423,7 +439,7 @@ class PredictEngine:
             json=item.model_dump(),
             headers=self.get_predict_headers(kbid),
         )
-        await self.check_response(resp, expected_status=200)
+        await self.check_response(kbid, resp, expected_status=200)
         data = await resp.json()
         return RunAgentsResponse.model_validate(data)
 

nucliadb/writer/api/v1/knowledgebox.py

@@ -66,7 +66,7 @@ async def create_kb_endpoint(request: Request, item: KnowledgeBoxConfig) -> Know
     except KnowledgeBoxConflict:
         raise HTTPException(status_code=419, detail="Knowledge box already exists")
     except ExternalIndexCreationError as exc:
-        raise HTTPException(status_code=502, detail=str(exc))
+        raise HTTPException(status_code=512, detail=str(exc))
     except Exception:
         logger.exception("Could not create KB")
         raise HTTPException(status_code=500, detail="Error creating knowledge box")

nucliadb/writer/api/v1/vectorsets.py

@@ -58,16 +58,23 @@ async def add_vectorset(request: Request, kbid: str, vectorset_id: str) -> Creat
             detail=err.content,
         )
 
-    except VectorSetConflict:
+    except VectorSetConflict as err:
         raise HTTPException(
             status_code=409,
-            detail="A vectorset with this embedding model already exists in your KB",
+            detail=str(err),
         )
 
     return CreatedVectorSet(id=vectorset_id)
 
 
 async def _add_vectorset(kbid: str, vectorset_id: str) -> None:
+    storage = await get_storage()
+
+    async with datamanagers.with_ro_transaction() as txn:
+        kbobj = KnowledgeBox(txn, storage, kbid)
+        if await kbobj.vectorset_marked_for_deletion(vectorset_id):
+            raise VectorSetConflict("Vectorset is already being deleted. Please try again later.")
+
     # First off, add the vectorset to the learning configuration if it's not already there
     lconfig = await learning_proxy.get_configuration(kbid)
     assert lconfig is not None
@@ -79,7 +86,6 @@ async def _add_vectorset(kbid: str, vectorset_id: str) -> None:
         assert lconfig is not None
 
     # Then, add the vectorset to the index if it's not already there
-    storage = await get_storage()
     vectorset_config = get_vectorset_config(lconfig, vectorset_id)
     async with datamanagers.with_rw_transaction() as txn:
         kbobj = KnowledgeBox(txn, storage, kbid)

nucliadb/writer/tus/gcs.py

@@ -275,7 +275,7 @@ class GCloudFileStorageManager(FileStorageManager):
                     data = {"text": text}
                 if resp.status not in (200, 204, 404):
                     if resp.status == 404:
-                        logger.error(
+                        logger.debug(
                             f"Attempt to delete not found gcloud: {data}, status: {resp.status}",
                             exc_info=True,
                         )

{nucliadb-6.3.4.post3796.dist-info → nucliadb-6.3.4.post3812.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.3.4.post3796
+Version: 6.3.4.post3812
 Summary: NucliaDB
 Author-email: Nuclia <nucliadb@nuclia.com>
 License: AGPL
@@ -20,11 +20,11 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: <4,>=3.9
 Description-Content-Type: text/markdown
-Requires-Dist: nucliadb-telemetry[all]>=6.3.4.post3796
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.4.post3796
-Requires-Dist: nucliadb-protos>=6.3.4.post3796
-Requires-Dist: nucliadb-models>=6.3.4.post3796
-Requires-Dist: nidx-protos>=6.3.4.post3796
+Requires-Dist: nucliadb-telemetry[all]>=6.3.4.post3812
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.4.post3812
+Requires-Dist: nucliadb-protos>=6.3.4.post3812
+Requires-Dist: nucliadb-models>=6.3.4.post3812
+Requires-Dist: nidx-protos>=6.3.4.post3812
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn

{nucliadb-6.3.4.post3796.dist-info → nucliadb-6.3.4.post3812.dist-info}/RECORD

@@ -143,7 +143,7 @@ nucliadb/ingest/orm/brain.py,sha256=A8H1J7Bo95sNzDgYr0_UNoemQhWOFEFz9UlYfs6ug-8,
 nucliadb/ingest/orm/broker_message.py,sha256=XWaiZgDOz94NPOPT-hqbRr5ZkpVimUw6PjUJNftfoVw,7514
 nucliadb/ingest/orm/entities.py,sha256=3_n6lKhBy2GsdmNmkh0_mvxP8md20OZsbtTNEmfJ8Hg,14888
 nucliadb/ingest/orm/exceptions.py,sha256=k4Esv4NtL4TrGTcsQpwrSfDhPQpiYcRbB1SpYmBX5MY,1432
-nucliadb/ingest/orm/knowledgebox.py,sha256=IGOPvBR1qXqDxE5DeiOdYCLdPgjzOVVpsASJ2zYvWwQ,23651
+nucliadb/ingest/orm/knowledgebox.py,sha256=Bfb4-MIQWlaJrQAUDbgs_iIsXCYjS7s5YiiGl_Jb4jo,23887
 nucliadb/ingest/orm/metrics.py,sha256=OkwMSPKLZcKba0ZTwtTiIxwBgaLMX5ydhGieKvi2y7E,1096
 nucliadb/ingest/orm/resource.py,sha256=oFD7APhmG1A72h7DTKumZWQRpIDM0o_FytP1P-CcNq0,45918
 nucliadb/ingest/orm/utils.py,sha256=vCe_9UxHu26JDFGLwQ0wH-XyzJIpQCTK-Ow9dtZR5Vg,2716
@@ -193,7 +193,7 @@ nucliadb/search/__init__.py,sha256=tnypbqcH4nBHbGpkINudhKgdLKpwXQCvDtPchUlsyY4,1
 nucliadb/search/app.py,sha256=-WEX1AZRA8R_9aeOo9ovOTwjXW_7VfwWN7N2ccSoqXg,3387
 nucliadb/search/lifecycle.py,sha256=V_Pj5PRP0yyDY8d5LytO4X8p9HhN7UomqRG6Ri0UaFA,2206
 nucliadb/search/openapi.py,sha256=t3Wo_4baTrfPftg2BHsyLWNZ1MYn7ZRdW7ht-wFOgRs,1016
-nucliadb/search/predict.py,sha256=z2-RkhMkH-5T6PtFkfESxNof07XiS5FxicLHPRyCUXc,22284
+nucliadb/search/predict.py,sha256=VJr5Itx8FE7CZIGYcP-fRgd2YGxAnP9Qj9NxiwWiwcc,22819
 nucliadb/search/predict_models.py,sha256=ZAe0dneUsPmV9uBar57cCFADCGOrYDsJHuqKlA5zWag,5937
 nucliadb/search/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nucliadb/search/run.py,sha256=aFb-CXRi_C8YMpP_ivNj8KW1BYhADj88y8K9Lr_nUPI,1402
@@ -331,7 +331,7 @@ nucliadb/writer/api/utils.py,sha256=wIQHlU8RQiIGVLI72suvyVIKlCU44Unh0Ae0IiN6Qwo,
 nucliadb/writer/api/v1/__init__.py,sha256=akI9A_jloNLb0dU4T5zjfdyvmSAiDeIdjAlzNx74FlU,1128
 nucliadb/writer/api/v1/export_import.py,sha256=elf-EQY5DD3mhw8kWb9tQpDcbrF9sY6VFYqxQOjuVP0,8201
 nucliadb/writer/api/v1/field.py,sha256=FySCMpcruSAKGeepeAlOihjwxyUPcDO73Uilq5VDWRk,18514
-nucliadb/writer/api/v1/knowledgebox.py,sha256=MLeIuym4jPrJgfy1NTcN9CpUGwuBiqDHMcx0hY9DR7g,9530
+nucliadb/writer/api/v1/knowledgebox.py,sha256=PHEYDFa-sN5JrI8-EiVVg5FDOsRuCLT43kyAB4xt-xA,9530
 nucliadb/writer/api/v1/learning_config.py,sha256=CKBjqcbewkfPwGUPLDWzZSpro6XkmCaVppe5Qtpu5Go,3117
 nucliadb/writer/api/v1/resource.py,sha256=jV9HM-ID1PPYypfy4Sl4_9aSPF87v7gSJZUSzHjHcQ4,19740
 nucliadb/writer/api/v1/router.py,sha256=RjuoWLpZer6Kl2BW_wznpNo6XL3BOpdTGqXZCn3QrrQ,1034
@@ -339,7 +339,7 @@ nucliadb/writer/api/v1/services.py,sha256=3AUjk-SmvqJx76v7y89DZx6oyasojPliGYeniR
 nucliadb/writer/api/v1/slug.py,sha256=xlVBDBpRi9bNulpBHZwhyftVvulfE0zFm1XZIWl-AKY,2389
 nucliadb/writer/api/v1/transaction.py,sha256=d2Vbgnkk_-FLGSTt3vfldwiJIUf0XoyD0wP1jQNz_DY,2430
 nucliadb/writer/api/v1/upload.py,sha256=hLMHXSaqEOE-vjKjhIupgdx8klJc3mVQp_oMwx5N-7o,33800
-nucliadb/writer/api/v1/vectorsets.py,sha256=mESaXkkI9f-jWWMW61ZZgv7E5YWXKemyc6vwT0lFXns,6747
+nucliadb/writer/api/v1/vectorsets.py,sha256=F3iMViL5G95_Tns4aO2SOA0DwAzxK2_P8MXxtd_XLRE,6973
 nucliadb/writer/resource/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/writer/resource/audit.py,sha256=FvxMZPzrNHtd31HgpZEvxzwAkbxJTZRhPLqRYYJi3tA,1426
 nucliadb/writer/resource/basic.py,sha256=_zdAr110C7rtEzOKoBRMzPjAnQ0pAtRfGjB8qCzodvI,11767
@@ -349,13 +349,13 @@ nucliadb/writer/tus/__init__.py,sha256=huWpKnDnjsrKlBBJk30ta5vamlA-4x0TbPs_2Up8h
 nucliadb/writer/tus/azure.py,sha256=XhWAlWTM0vmXcXtuEPYjjeEhuZjiZXZu8q9WsJ7omFE,4107
 nucliadb/writer/tus/dm.py,sha256=bVoXqt_dpNvTjpffPYhj1JfqK6gfLoPr0hdkknUCZ9E,5488
 nucliadb/writer/tus/exceptions.py,sha256=WfZSSjsHfoy63wUFlH3QoHx7FMoCNA1oKJmWpZZDnCo,2156
-nucliadb/writer/tus/gcs.py,sha256=yM9GSO0mV7c6ZFqK9LpjiBNu96uf1_rofAj9R0NC7xg,14079
+nucliadb/writer/tus/gcs.py,sha256=OnE-YUnp7eyfWFlnh-vlGoxEPS8cUBSSmSm6iJ1Kva0,14079
 nucliadb/writer/tus/local.py,sha256=7jYa_w9b-N90jWgN2sQKkNcomqn6JMVBOVeDOVYJHto,5193
 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-6.3.4.post3796.dist-info/METADATA,sha256=CAxQ5xeSrVmN4NWzZwAdBUv2q5O5cP1INsDQ14RJYdk,4291
-nucliadb-6.3.4.post3796.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-nucliadb-6.3.4.post3796.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.3.4.post3796.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.3.4.post3796.dist-info/RECORD,,
+nucliadb-6.3.4.post3812.dist-info/METADATA,sha256=yjF1rSCHEhQiCv7vEdwAPyTtjrWrsKJgae-L39Y_zE8,4291
+nucliadb-6.3.4.post3812.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+nucliadb-6.3.4.post3812.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.3.4.post3812.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.3.4.post3812.dist-info/RECORD,,