nucliadb 6.3.4.post3812__py3-none-any.whl → 6.3.4.post3821__py3-none-any.whl

nucliadb/common/datamanagers/resources.py

@@ -103,7 +103,8 @@ async def modify_slug(txn: Transaction, *, kbid: str, rid: str, new_slug: str) -
 async def get_resource_shard_id(
     txn: Transaction, *, kbid: str, rid: str, for_update: bool = False
 ) -> Optional[str]:
-    shard = await txn.get(KB_RESOURCE_SHARD.format(kbid=kbid, uuid=rid, for_update=for_update))
+    key = KB_RESOURCE_SHARD.format(kbid=kbid, uuid=rid)
+    shard = await txn.get(key, for_update=for_update)
     if shard is not None:
         return shard.decode()
     else:

nucliadb/ingest/orm/entities.py

@@ -36,7 +36,7 @@ from nucliadb.common.datamanagers.entities import (
 from nucliadb.common.maindb.driver import Transaction
 from nucliadb.ingest.orm.knowledgebox import KnowledgeBox
 from nucliadb.ingest.settings import settings
-from nucliadb.search.search.shards import query_shard
+from nucliadb.search.search.shards import graph_search_shard, query_shard
 from nucliadb_protos.knowledgebox_pb2 import (
     DeletedEntitiesGroups,
     EntitiesGroup,
@@ -45,9 +45,8 @@ from nucliadb_protos.knowledgebox_pb2 import (
 )
 from nucliadb_protos.nodereader_pb2 import (
     Faceted,
-    RelationNodeFilter,
-    RelationPrefixSearchRequest,
-    RelationSearchResponse,
+    GraphSearchRequest,
+    GraphSearchResponse,
     SearchRequest,
     SearchResponse,
 )
@@ -203,18 +202,16 @@ class EntitiesManager:
     async def get_indexed_entities_group(self, group: str) -> Optional[EntitiesGroup]:
         shard_manager = get_shard_manager()
 
-        async def do_entities_search(node: AbstractIndexNode, shard_id: str) -> RelationSearchResponse:
-            request = SearchRequest(
-                shard=shard_id,
-                relation_prefix=RelationPrefixSearchRequest(
-                    prefix="",
-                    node_filters=[
-                        RelationNodeFilter(node_type=RelationNode.NodeType.ENTITY, node_subtype=group)
-                    ],
-                ),
-            )
-            response = await query_shard(node, shard_id, request)
-            return response.relation
+        async def do_entities_search(node: AbstractIndexNode, shard_id: str) -> GraphSearchResponse:
+            request = GraphSearchRequest()
+            # XXX: this is a wild guess. Are those enough or too many?
+            request.top_k = 500
+            request.kind = GraphSearchRequest.QueryKind.NODES
+            request.query.path.path.source.node_type = RelationNode.NodeType.ENTITY
+            request.query.path.path.source.node_subtype = group
+            request.query.path.path.undirected = True
+            response = await graph_search_shard(node, shard_id, request)
+            return response
 
         results = await shard_manager.apply_for_all_shards(
             self.kbid,
@@ -224,7 +221,7 @@ class EntitiesManager:
 
         entities = {}
         for result in results:
-            entities.update({node.value: Entity(value=node.value) for node in result.prefix.nodes})
+            entities.update({node.value: Entity(value=node.value) for node in result.nodes})
 
         if not entities:
             return None

nucliadb/search/search/graph_strategy.py

@@ -44,7 +44,6 @@ from nucliadb.search.search.find_merge import (
     hydrate_and_rerank,
 )
 from nucliadb.search.search.hydrator import ResourceHydrationOptions, TextBlockHydrationOptions
-from nucliadb.search.search.merge import merge_relation_prefix_results
 from nucliadb.search.search.metrics import RAGMetrics
 from nucliadb.search.search.rerankers import Reranker, RerankingOptions
 from nucliadb.search.utilities import get_predict
@@ -65,6 +64,7 @@ from nucliadb_models.search import (
     NucliaDBClientType,
     QueryEntityDetection,
     RelatedEntities,
+    RelatedEntity,
     RelationDirection,
     RelationRanking,
     Relations,
@@ -443,26 +443,35 @@ async def fuzzy_search_entities(
 ) -> Optional[RelatedEntities]:
     """Fuzzy find entities in KB given a query using the same methodology as /suggest, but split by words."""
 
-    request = nodereader_pb2.SearchRequest()
-    request.relation_prefix.query = query
+    # Build an OR for each word in the query matching with fuzzy any word in any
+    # node in any position. I.e., for the query "Rose Hamiltn", it'll match
+    # "Rosa Parks" and "Margaret Hamilton"
+    request = nodereader_pb2.GraphSearchRequest()
+    # XXX Are those enough results? Too many?
+    request.top_k = 50
+    request.kind = nodereader_pb2.GraphSearchRequest.QueryKind.NODES
+    for word in query.split():
+        subquery = nodereader_pb2.GraphQuery.PathQuery()
+        subquery.path.source.value = word
+        subquery.path.source.fuzzy.kind = nodereader_pb2.GraphQuery.Node.MatchLocation.WORDS
+        subquery.path.source.fuzzy.distance = 1
+        subquery.path.undirected = True
+        request.query.path.bool_or.operands.append(subquery)
 
-    results: list[nodereader_pb2.SearchResponse]
     try:
-        (
-            results,
-            _,
-            _,
-        ) = await node_query(
-            kbid,
-            Method.SEARCH,
-            request,
-        )
-        return merge_relation_prefix_results(results)
-    except Exception as e:
-        capture_exception(e)
+        results, _, _ = await node_query(kbid, Method.GRAPH, request)
+    except Exception as exc:
+        capture_exception(exc)
         logger.exception("Error in finding entities in query for graph strategy")
         return None
 
+    # merge shard results while deduplicating repeated entities across shards
+    unique_entities: set[RelatedEntity] = set()
+    for response in results:
+        unique_entities.update((RelatedEntity(family=e.subtype, value=e.value) for e in response.nodes))
+
+    return RelatedEntities(entities=list(unique_entities), total=len(unique_entities))
+
 
 async def rank_relations_reranker(
     relations: Relations,

nucliadb/search/search/merge.py

@@ -632,19 +632,6 @@ async def merge_suggest_entities_results(
     return RelatedEntities(entities=list(unique_entities), total=len(unique_entities))
 
 
-def merge_relation_prefix_results(
-    responses: list[SearchResponse],
-) -> RelatedEntities:
-    unique_entities: Set[RelatedEntity] = set()
-    for response in responses:
-        response_entities = (
-            RelatedEntity(family=e.subtype, value=e.value) for e in response.relation.prefix.nodes
-        )
-        unique_entities.update(response_entities)
-
-    return RelatedEntities(entities=list(unique_entities), total=len(unique_entities))
-
-
 async def merge_suggest_results(
     suggest_responses: list[SuggestResponse],
     kbid: str,

nucliadb/search/search/query_parser/parsers/graph.py

@@ -221,11 +221,12 @@ def _parse_relation_query(
 def _set_node_to_pb(node: graph_requests.GraphNode, pb: nodereader_pb2.GraphQuery.Node):
     if node.value is not None:
         pb.value = node.value
-        if node.match == graph_requests.NodeMatchKind.EXACT:
-            pb.match_kind = nodereader_pb2.GraphQuery.Node.MatchKind.DEPRECATED_EXACT
+        if node.match == graph_requests.NodeMatchKindName.EXACT:
+            pb.exact.kind = nodereader_pb2.GraphQuery.Node.MatchLocation.FULL
 
-        elif node.match == graph_requests.NodeMatchKind.FUZZY:
-            pb.match_kind = nodereader_pb2.GraphQuery.Node.MatchKind.DEPRECATED_FUZZY
+        elif node.match == graph_requests.NodeMatchKindName.FUZZY:
+            pb.fuzzy.kind = nodereader_pb2.GraphQuery.Node.MatchLocation.PREFIX
+            pb.fuzzy.distance = 1
 
         else:  # pragma: nocover
             # This is a trick so mypy generates an error if this branch can be reached,

{nucliadb-6.3.4.post3812.dist-info → nucliadb-6.3.4.post3821.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.3.4.post3812
+Version: 6.3.4.post3821
 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.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-telemetry[all]>=6.3.4.post3821
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.4.post3821
+Requires-Dist: nucliadb-protos>=6.3.4.post3821
+Requires-Dist: nucliadb-models>=6.3.4.post3821
+Requires-Dist: nidx-protos>=6.3.4.post3821
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn

{nucliadb-6.3.4.post3812.dist-info → nucliadb-6.3.4.post3821.dist-info}/RECORD

@@ -78,7 +78,7 @@ nucliadb/common/datamanagers/fields.py,sha256=9KqBzTssAT68FR5hd17Xu_CSwAYdKFuYic
 nucliadb/common/datamanagers/kb.py,sha256=P7EhF4tApIUG2jw_HH1oMufTKG9__kuOLKnrCNGbDM4,6156
 nucliadb/common/datamanagers/labels.py,sha256=Zm0GQpSPoGXEEysUY7VsDIcyKSIIQsMVphj23IyM9_c,4502
 nucliadb/common/datamanagers/processing.py,sha256=ByxdZzdbAfJGqC6__mY-zryjk040TyQfcUq3rxujeoY,1587
-nucliadb/common/datamanagers/resources.py,sha256=cuwcVL-GEjS1VHigJtz5SG_dRhl09UADQ9MSYOOqs70,10779
+nucliadb/common/datamanagers/resources.py,sha256=VwFdCyHSnzMU3ASYRhC-wuCjCQEjOKEF7tIob4lTcPg,10793
 nucliadb/common/datamanagers/rollover.py,sha256=BM1hJ2cEU91xekM5PtmnA0SN3i3w0WmodiyTpO8YZZs,7865
 nucliadb/common/datamanagers/search_configurations.py,sha256=O-8eW43CE46GcxO6TB5hpi27NBguv4BL4SI1vLlN8os,2463
 nucliadb/common/datamanagers/synonyms.py,sha256=zk3GEH38KF5vV_VcuL6DCg-2JwgXJfQl7Io6VPqv2cw,1566
@@ -141,7 +141,7 @@ 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=A8H1J7Bo95sNzDgYr0_UNoemQhWOFEFz9UlYfs6ug-8,29407
 nucliadb/ingest/orm/broker_message.py,sha256=XWaiZgDOz94NPOPT-hqbRr5ZkpVimUw6PjUJNftfoVw,7514
-nucliadb/ingest/orm/entities.py,sha256=3_n6lKhBy2GsdmNmkh0_mvxP8md20OZsbtTNEmfJ8Hg,14888
+nucliadb/ingest/orm/entities.py,sha256=a-aYuKBUQhxDKFtXOzTAkLlY_t2JiTfaptw2vt3AQDQ,14915
 nucliadb/ingest/orm/exceptions.py,sha256=k4Esv4NtL4TrGTcsQpwrSfDhPQpiYcRbB1SpYmBX5MY,1432
 nucliadb/ingest/orm/knowledgebox.py,sha256=Bfb4-MIQWlaJrQAUDbgs_iIsXCYjS7s5YiiGl_Jb4jo,23887
 nucliadb/ingest/orm/metrics.py,sha256=OkwMSPKLZcKba0ZTwtTiIxwBgaLMX5ydhGieKvi2y7E,1096
@@ -229,10 +229,10 @@ nucliadb/search/search/filters.py,sha256=1MkHlJjAQqoRCj7e5cEzK2HvBxGLE17I_omsjik
 nucliadb/search/search/find.py,sha256=pjc-i55j_p_F1H0La4qc9slZSl_gICoFzl3fy6FLYTM,10314
 nucliadb/search/search/find_merge.py,sha256=3FnzKFEnVemg6FO_6zveulbAU7klvsiPEBvLrpBBMg8,17450
 nucliadb/search/search/graph_merge.py,sha256=OiUNiXOWwrUVKqStuRcoUJwvDbDYamqIgiAy_FwPdMI,3405
-nucliadb/search/search/graph_strategy.py,sha256=gisL2GpbSIa_SucyOwEt7TWdqURyAQqxvD_-PkXQct8,32339
+nucliadb/search/search/graph_strategy.py,sha256=SPJdDHQcTFsNb1IEWdWzklC5j1Vv9igibo0dYQAgcy0,33113
 nucliadb/search/search/hydrator.py,sha256=-R37gCrGxkyaiHQalnTWHNG_FCx11Zucd7qA1vQCxuw,6985
 nucliadb/search/search/ingestion_agents.py,sha256=NeJr4EEX-bvFFMGvXOOwLv8uU7NuQ-ntJnnrhnKfMzY,3174
-nucliadb/search/search/merge.py,sha256=bPO51Kc3Ec69UuCbF5ulokd02gWZhp7zlJSuMjGDo9Y,23141
+nucliadb/search/search/merge.py,sha256=fh5WnA_xnXp-Iiq5Cud9hIGole7_0OW2b3Oymk32D6Y,22689
 nucliadb/search/search/metrics.py,sha256=GGGtXHLhK79_ESV277xkBVjcaMURXHCxYG0EdGamUd8,2886
 nucliadb/search/search/paragraphs.py,sha256=pNAEiYqJGGUVcEf7xf-PFMVqz0PX4Qb-WNG-_zPGN2o,7799
 nucliadb/search/search/pgcatalog.py,sha256=s_J98fsX_RuFXwpejpkGqG-tD9ELuzz4YQ6U3ew5h2g,9313
@@ -258,7 +258,7 @@ nucliadb/search/search/query_parser/old_filters.py,sha256=-zbfN-RsXoj_DRjh3Lfp-w
 nucliadb/search/search/query_parser/parsers/__init__.py,sha256=ySCNSdbesLXGZyR88919njulA6UE10_3PhqMG_Yj1o4,1034
 nucliadb/search/search/query_parser/parsers/catalog.py,sha256=XdBiTweGTQkj8m_V_i2xbwp7P5pPO8K1Tud692XKhMw,7149
 nucliadb/search/search/query_parser/parsers/find.py,sha256=q3wH_i0DGceeKckYEH3c5MqM5EvRiMCL7r-6nCAId9Q,4666
-nucliadb/search/search/query_parser/parsers/graph.py,sha256=43S8iSg4j9I_XD8zpW1VggDspQD-NMyS26B5Mw6Dktw,9358
+nucliadb/search/search/query_parser/parsers/graph.py,sha256=QJs-pybNXPsMSEkIHctb0Q0xQG-aArks8BtUxbJL5rU,9386
 nucliadb/standalone/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/standalone/api_router.py,sha256=hgq9FXpihzgjHkwcVGfGCSwyXy67fqXTfLFHuINzIi0,5567
 nucliadb/standalone/app.py,sha256=mAApNK_iVsQgJyd-mtwCeZq5csSimwnXmlQGH9a70pE,5586
@@ -354,8 +354,8 @@ 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-6.3.4.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,,
+nucliadb-6.3.4.post3821.dist-info/METADATA,sha256=I_IqkABVw9DZGcLJyK5KVtV7Tr4d8x1Ut3GX-Oncydo,4291
+nucliadb-6.3.4.post3821.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+nucliadb-6.3.4.post3821.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.3.4.post3821.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.3.4.post3821.dist-info/RECORD,,