nucliadb 6.2.1.post3067__py3-none-any.whl → 6.2.1.post3072__py3-none-any.whl

nucliadb/common/cluster/manager.py

@@ -104,6 +104,14 @@ class KBShardManager:
             errors.capture_exception(exc)
             raise NodeError("Node unavailable for operation") from exc
 
+        for result in results:
+            if isinstance(result, Exception):
+                errors.capture_exception(result)
+                raise NodeError(
+                    f"Error while applying {aw.__name__} for all shards. Other similar errors may have been shadowed.\n"
+                    f"{type(result).__name__}: {result}"
+                ) from result
+
         return results
 
     # TODO: move to data manager

nucliadb/ingest/orm/entities.py

@@ -26,7 +26,6 @@ from nucliadb.common.cluster.base import AbstractIndexNode
 from nucliadb.common.cluster.exceptions import (
     AlreadyExists,
     EntitiesGroupNotFound,
-    NodeError,
 )
 from nucliadb.common.cluster.utils import get_shard_manager
 from nucliadb.common.datamanagers.entities import (
@@ -54,7 +53,6 @@ from nucliadb_protos.nodereader_pb2 import (
 )
 from nucliadb_protos.utils_pb2 import RelationNode
 from nucliadb_protos.writer_pb2 import GetEntitiesResponse
-from nucliadb_telemetry import errors
 
 from .exceptions import EntityManagementException
 
@@ -226,10 +224,6 @@ class EntitiesManager:
             settings.relation_search_timeout,
             use_read_replica_nodes=self.use_read_replica_nodes,
         )
-        for result in results:
-            if isinstance(result, Exception):
-                errors.capture_exception(result)
-                raise NodeError("Error while querying relation index")
 
         entities = {}
         for result in results:
@@ -305,6 +299,7 @@ class EntitiesManager:
         shard_manager = get_shard_manager()
 
         async def query_indexed_entities_group_names(node: AbstractIndexNode, shard_id: str) -> set[str]:
+            """Search all relation types"""
             request = SearchRequest(
                 shard=shard_id,
                 result_per_page=0,
@@ -316,10 +311,11 @@ class EntitiesManager:
             response: SearchResponse = await query_shard(node, shard_id, request)
             try:
                 facetresults = response.document.facets["/e"].facetresults
-                return {facet.tag.split("/")[-1] for facet in facetresults}
             except KeyError:
                 # No entities found
                 return set()
+            else:
+                return {facet.tag.split("/")[-1] for facet in facetresults}
 
         results = await shard_manager.apply_for_all_shards(
             self.kbid,
@@ -327,10 +323,6 @@ class EntitiesManager:
             settings.relation_types_timeout,
             use_read_replica_nodes=self.use_read_replica_nodes,
         )
-        for result in results:
-            if isinstance(result, Exception):
-                errors.capture_exception(result)
-                raise NodeError("Error while looking for relations types")
 
         if not results:
             return set()

nucliadb/search/search/chat/query.py

@@ -50,7 +50,12 @@ from nucliadb_models.search import (
     parse_rephrase_prompt,
 )
 from nucliadb_protos import audit_pb2
-from nucliadb_protos.nodereader_pb2 import RelationSearchResponse, SearchRequest, SearchResponse
+from nucliadb_protos.nodereader_pb2 import (
+    EntitiesSubgraphRequest,
+    RelationSearchResponse,
+    SearchRequest,
+    SearchResponse,
+)
 from nucliadb_protos.utils_pb2 import RelationNode
 from nucliadb_telemetry.errors import capture_exception
 from nucliadb_utils.utilities import get_audit
@@ -245,10 +250,16 @@ async def get_relations_results_from_entities(
     timeout: Optional[float] = None,
     only_with_metadata: bool = False,
     only_agentic_relations: bool = False,
+    deleted_entities: set[str] = set(),
 ) -> Relations:
     request = SearchRequest()
     request.relation_subgraph.entry_points.extend(entities)
     request.relation_subgraph.depth = 1
+
+    deleted = EntitiesSubgraphRequest.DeletedEntities()
+    deleted.node_values.extend(deleted_entities)
+    request.relation_subgraph.deleted_entities.append(deleted)
+
     results: list[SearchResponse]
     (
         results,

nucliadb/search/search/graph_strategy.py

@@ -337,7 +337,14 @@ async def get_graph_results(
                     or graph_strategy.query_entity_detection == QueryEntityDetection.PREDICT
                 ):
                     try:
-                        entities_to_explore = await predict.detect_entities(kbid, query)
+                        # Purposely ignore the entity subtype. This is done so we find all entities that match
+                        # the entity by name. e.g: in a query like "2000", predict might detect the number as
+                        # a year entity or as a currency entity. We want graph results for both, so we ignore the
+                        # subtype just in this case.
+                        entities_to_explore = [
+                            RelationNode(ntype=r.ntype, value=r.value, subtype="")
+                            for r in await predict.detect_entities(kbid, query)
+                        ]
                     except Exception as e:
                         capture_exception(e)
                         logger.exception("Error in detecting entities for graph strategy")
@@ -365,19 +372,14 @@ async def get_graph_results(
                     timeout=5.0,
                     only_with_metadata=True,
                     only_agentic_relations=graph_strategy.agentic_graph_only,
+                    deleted_entities=explored_entities,
                 )
             except Exception as e:
                 capture_exception(e)
                 logger.exception("Error in getting query relations for graph strategy")
                 new_relations = Relations(entities={})
 
-            # Removing the relations connected to the entities that were already explored
-            # XXX: This could be optimized by implementing a filter in the index
-            # so we don't have to remove them after
-            new_subgraphs = {
-                entity: filter_subgraph(subgraph, explored_entities)
-                for entity, subgraph in new_relations.entities.items()
-            }
+            new_subgraphs = new_relations.entities
 
             explored_entities.update(new_subgraphs.keys())
 
@@ -842,15 +844,6 @@ async def build_graph_response(
     )
 
 
-def filter_subgraph(subgraph: EntitySubgraph, entities_to_remove: Collection[str]) -> EntitySubgraph:
-    """
-    Removes the relationships with entities in `entities_to_remove` from the subgraph.
-    """
-    return EntitySubgraph(
-        related_to=[rel for rel in subgraph.related_to if rel.entity not in entities_to_remove]
-    )
-
-
 def relations_match_to_text_block_match(
     paragraph_match: RelationsParagraphMatch,
 ) -> TextBlockMatch:

{nucliadb-6.2.1.post3067.dist-info → nucliadb-6.2.1.post3072.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb
-Version: 6.2.1.post3067
+Version: 6.2.1.post3072
 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.post3067
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post3067
-Requires-Dist: nucliadb-protos>=6.2.1.post3067
-Requires-Dist: nucliadb-models>=6.2.1.post3067
+Requires-Dist: nucliadb-telemetry[all]>=6.2.1.post3072
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post3072
+Requires-Dist: nucliadb-protos>=6.2.1.post3072
+Requires-Dist: nucliadb-models>=6.2.1.post3072
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn

{nucliadb-6.2.1.post3067.dist-info → nucliadb-6.2.1.post3072.dist-info}/RECORD

@@ -47,7 +47,7 @@ nucliadb/common/cluster/base.py,sha256=kklDqyvsubNX0W494ttl9f3E58lGaX6AXqAd8XX8Z
 nucliadb/common/cluster/exceptions.py,sha256=V3c_fgH00GyJ-a5CaGLhwTuhwhUNR9YAGvS5jaRuc_Y,1495
 nucliadb/common/cluster/grpc_node_dummy.py,sha256=L85wBnfab7Rev0CfsfUjPxQC6DiHPsETKrZAOLx9XHg,3510
 nucliadb/common/cluster/index_node.py,sha256=g38H1kiAliF3Y6et_CWYInpn_xPxf7THAFJ7RtgLNZo,3246
-nucliadb/common/cluster/manager.py,sha256=3UnYwVb-ZykYfLndxM7TLw7-2T_vxqoFXMu0Pzxh5-A,15327
+nucliadb/common/cluster/manager.py,sha256=cj8yNIspsmdyGUq3vbyW--r-LcxAUqA4S9tocz6MsCM,15695
 nucliadb/common/cluster/rebalance.py,sha256=jSEYsPgs_Dobv3FOaKl5arBko4s8JlWkahm8LOzgNnE,9135
 nucliadb/common/cluster/rollover.py,sha256=dx6AF9ywKP10iBNlcoJgRV40921fOPpVWaCUU54hztE,25823
 nucliadb/common/cluster/settings.py,sha256=TMoym-cZsQ2soWfLAce0moSa2XncttQyhahL43LrWTo,3384
@@ -128,7 +128,7 @@ nucliadb/ingest/fields/text.py,sha256=tFvSQJAe0W7ePpp2_WDfLiE2yglR1OTU0Zht9acvOF
 nucliadb/ingest/orm/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/ingest/orm/brain.py,sha256=UND5EsNUdd7XdjScYqRqg4r_xCx3l-My8alGw5M9CWg,28398
 nucliadb/ingest/orm/broker_message.py,sha256=ZEMueoGuuRKO4tHgzc0P0AM1Ls1TTYey_4UvRQf0BpY,6915
-nucliadb/ingest/orm/entities.py,sha256=5d6Gfo-Yz-rns_mNJeRqiGaPeWpUMgSKZnmWIGMLCKo,15537
+nucliadb/ingest/orm/entities.py,sha256=sBhg8eahsWVwO34KoAJV2YRix4Uw5GINx3srJWxRC9k,15148
 nucliadb/ingest/orm/exceptions.py,sha256=k4Esv4NtL4TrGTcsQpwrSfDhPQpiYcRbB1SpYmBX5MY,1432
 nucliadb/ingest/orm/knowledgebox.py,sha256=IGOPvBR1qXqDxE5DeiOdYCLdPgjzOVVpsASJ2zYvWwQ,23651
 nucliadb/ingest/orm/metrics.py,sha256=OkwMSPKLZcKba0ZTwtTiIxwBgaLMX5ydhGieKvi2y7E,1096
@@ -210,7 +210,7 @@ nucliadb/search/search/fetch.py,sha256=XJHIFnZmXM_8Kb37lb4lg1GYG7cZ1plT-qAIb_Qzi
 nucliadb/search/search/filters.py,sha256=1MkHlJjAQqoRCj7e5cEzK2HvBxGLE17I_omsjiklbtw,6476
 nucliadb/search/search/find.py,sha256=yQbttt85wQFc4NEaj2RNGgozP7IQx_bjAOhHke3fXY0,9890
 nucliadb/search/search/find_merge.py,sha256=_R_YpHAZv5BHh3XABQ8MRd1Ci0seclGYf26yJHJ7H0I,17178
-nucliadb/search/search/graph_strategy.py,sha256=cFcu6nrOqHfobsbUu1pOwfBndrN4ppEvv3-4jV4_4bg,31977
+nucliadb/search/search/graph_strategy.py,sha256=Egcq_zn895gTUYmyQTsXj8YaUMa3HBKhcSa1GBvgzAM,31877
 nucliadb/search/search/hydrator.py,sha256=-R37gCrGxkyaiHQalnTWHNG_FCx11Zucd7qA1vQCxuw,6985
 nucliadb/search/search/merge.py,sha256=i_PTBFRqC5iTTziOMEltxLIlmokIou5hjjgR4BnoLBE,22635
 nucliadb/search/search/metrics.py,sha256=81X-tahGW4n2CLvUzCPdNxNClmZqUWZjcVOGCUHoiUM,2872
@@ -228,7 +228,7 @@ nucliadb/search/search/chat/ask.py,sha256=K85Size6WAb-q4sCn0u1drrPnqIvqCy6YbfCxQ
 nucliadb/search/search/chat/exceptions.py,sha256=Siy4GXW2L7oPhIR86H3WHBhE9lkV4A4YaAszuGGUf54,1356
 nucliadb/search/search/chat/images.py,sha256=PA8VWxT5_HUGfW1ULhKTK46UBsVyINtWWqEM1ulzX1E,3095
 nucliadb/search/search/chat/prompt.py,sha256=r2JTiRWH3YHPdeRAG5w6gD0g0fWVxdTjYIR86qAVa7k,47106
-nucliadb/search/search/chat/query.py,sha256=y7W5VuKl1XiZuNsxZIcxxHcFXSG6It2W5CoftZ-ekAc,15428
+nucliadb/search/search/chat/query.py,sha256=rBssR6MPSx8h2DASRMTLODaz9oGE5tNVVVeDncSrEp4,15684
 nucliadb/search/search/query_parser/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/search/search/query_parser/exceptions.py,sha256=tuzl7ZyvVsRz6u0_3zMe60vx39nd3pi641prs-5nC0E,872
 nucliadb/search/search/query_parser/models.py,sha256=-VlCDXUCgOroAZw1Leqhj2VMgRv_CD2w40PXXOBLaUM,2332
@@ -329,9 +329,9 @@ 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.post3067.dist-info/METADATA,sha256=V33nguAqDHoZDoT_3Vfc2w3HTnobJN1C_uB-tlKWGMU,4603
-nucliadb-6.2.1.post3067.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-nucliadb-6.2.1.post3067.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.2.1.post3067.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.2.1.post3067.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-nucliadb-6.2.1.post3067.dist-info/RECORD,,
+nucliadb-6.2.1.post3072.dist-info/METADATA,sha256=Spf7lqAgt29pV4iYGRWhP-Gn1yZLgJh1uzhjqOoWSoU,4603
+nucliadb-6.2.1.post3072.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+nucliadb-6.2.1.post3072.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.2.1.post3072.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.2.1.post3072.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+nucliadb-6.2.1.post3072.dist-info/RECORD,,