nucliadb 6.2.1.post2835__py3-none-any.whl → 6.2.1.post2842__py3-none-any.whl

nucliadb/search/search/hydrator.py

@@ -66,6 +66,9 @@ class TextBlockHydrationOptions(BaseModel):
     # list of exact matches to highlight
     ematches: Optional[list[str]] = None
 
+    # If true, only hydrate the text block if its text is not already populated
+    only_hydrate_empty: bool = False
+
 
 @hydrator_observer.wrap({"type": "resource_text"})
 async def hydrate_resource_text(
@@ -161,6 +164,8 @@ async def hydrate_text_block(
     `text_block` object.
 
     """
+    if options.only_hydrate_empty and text_block.text:
+        return text_block
     async with AsyncExitStack() as stack:
         if concurrency_control is not None:
             await stack.enter_async_context(concurrency_control)
@@ -188,4 +193,5 @@ def text_block_to_find_paragraph(text_block: TextBlockMatch) -> FindParagraph:
         reference=text_block.representation_file,
         page_with_visual=text_block.page_with_visual,
         position=text_block.position,
+        relevant_relations=text_block.relevant_relations,
     )

nucliadb/search/search/merge.py

@@ -23,6 +23,7 @@ import math
 from typing import Any, Optional, Set, Union
 
 from nucliadb.common.ids import FieldId, ParagraphId
+from nucliadb.common.models_utils import from_proto
 from nucliadb.common.models_utils.from_proto import RelationTypePbMap
 from nucliadb.search.search import cache
 from nucliadb.search.search.cut import cut_page
@@ -442,15 +443,38 @@ async def merge_paragraph_results(
 async def merge_relations_results(
     relations_responses: list[RelationSearchResponse],
     query: EntitiesSubgraphRequest,
+    only_with_metadata: bool = False,
+    only_agentic: bool = False,
 ) -> Relations:
     loop = asyncio.get_event_loop()
-    return await loop.run_in_executor(None, _merge_relations_results, relations_responses, query)
+    return await loop.run_in_executor(
+        None,
+        _merge_relations_results,
+        relations_responses,
+        query,
+        only_with_metadata,
+        only_agentic,
+    )
 
 
 def _merge_relations_results(
     relations_responses: list[RelationSearchResponse],
     query: EntitiesSubgraphRequest,
+    only_with_metadata: bool,
+    only_agentic: bool,
 ) -> Relations:
+    """
+    Merge relation search responses into a single Relations object while applying filters.
+
+    Args:
+        relations_responses: List of relation search responses
+        query: EntitiesSubgraphRequest object
+        only_with_metadata: If True, only include relations with metadata. This metadata includes paragraph_id and entity positions among other things.
+        only_agentic: If True, only include relations extracted by a Graph Extraction Agent.
+
+    Returns:
+        Relations
+    """
     relations = Relations(entities={})
 
     for entry_point in query.entry_points:
@@ -462,27 +486,37 @@ def _merge_relations_results(
             destination = relation.to
             relation_type = RelationTypePbMap[relation.relation]
             relation_label = relation.relation_label
-
-            if origin.value in relations.entities:
-                relations.entities[origin.value].related_to.append(
-                    DirectionalRelation(
-                        entity=destination.value,
-                        entity_type=relation_node_type_to_entity_type(destination.ntype),
-                        relation=relation_type,
-                        relation_label=relation_label,
-                        direction=RelationDirection.OUT,
+            metadata = relation.metadata if relation.HasField("metadata") else None
+            # If only_with_metadata is True, we check that metadata for the relation is not None
+            # If only_agentic is True, we check that metadata for the relation is not None and that it has a data_augmentation_task_id
+            # TODO: This is suboptimal, we should be able to filter this in the query to the index,
+            if (not only_with_metadata or metadata) and (
+                not only_agentic or (metadata and metadata.data_augmentation_task_id)
+            ):
+                if origin.value in relations.entities:
+                    relations.entities[origin.value].related_to.append(
+                        DirectionalRelation(
+                            entity=destination.value,
+                            entity_type=relation_node_type_to_entity_type(destination.ntype),
+                            entity_subtype=destination.subtype,
+                            relation=relation_type,
+                            relation_label=relation_label,
+                            direction=RelationDirection.OUT,
+                            metadata=from_proto.relation_metadata(metadata) if metadata else None,
+                        )
                     )
-                )
-            elif destination.value in relations.entities:
-                relations.entities[destination.value].related_to.append(
-                    DirectionalRelation(
-                        entity=origin.value,
-                        entity_type=relation_node_type_to_entity_type(origin.ntype),
-                        relation=relation_type,
-                        relation_label=relation_label,
-                        direction=RelationDirection.IN,
+                elif destination.value in relations.entities:
+                    relations.entities[destination.value].related_to.append(
+                        DirectionalRelation(
+                            entity=origin.value,
+                            entity_type=relation_node_type_to_entity_type(origin.ntype),
+                            entity_subtype=origin.subtype,
+                            relation=relation_type,
+                            relation_label=relation_label,
+                            direction=RelationDirection.IN,
+                            metadata=from_proto.relation_metadata(metadata) if metadata else None,
+                        )
                     )
-                )
 
     return relations
 
@@ -584,8 +618,6 @@ async def merge_suggest_entities_results(
 async def merge_suggest_results(
     suggest_responses: list[SuggestResponse],
     kbid: str,
-    show: list[ResourceProperties],
-    field_type_filter: list[FieldTypeName],
     highlight: bool = False,
 ) -> KnowledgeboxSuggestResults:
     api_results = KnowledgeboxSuggestResults()

{nucliadb-6.2.1.post2835.dist-info → nucliadb-6.2.1.post2842.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb
-Version: 6.2.1.post2835
+Version: 6.2.1.post2842
 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.post2835
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post2835
-Requires-Dist: nucliadb-protos>=6.2.1.post2835
-Requires-Dist: nucliadb-models>=6.2.1.post2835
+Requires-Dist: nucliadb-telemetry[all]>=6.2.1.post2842
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.2.1.post2842
+Requires-Dist: nucliadb-protos>=6.2.1.post2842
+Requires-Dist: nucliadb-models>=6.2.1.post2842
 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.post2835.dist-info → nucliadb-6.2.1.post2842.dist-info}/RECORD

@@ -38,7 +38,7 @@ nucliadb/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nucliadb/common/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/common/constants.py,sha256=QpigxJh_CtD85Evy0PtV5cVq6x0U_f9xfIcXz1ymkUg,869
 nucliadb/common/counters.py,sha256=yhJEmmrglTSrDmB8OjaFLkZ__TwhTxayyQrtacnB55I,957
-nucliadb/common/ids.py,sha256=P92d9aNJCd1VpGgbqnab6g35iQalpUIW_eaaxEin2mE,7439
+nucliadb/common/ids.py,sha256=HMb213Kz9HaY4IsBwaQJFhUErntKWV-29s0UHaGcf1E,8004
 nucliadb/common/locking.py,sha256=RL0CabZVPzxHZyUjYeUyLvsJTm7W3J9o4fEgsY_ufNc,5896
 nucliadb/common/nidx.py,sha256=D74oNdniWjbc6gBBDwZP74NH-egTIORHhbfzgIto8DE,8667
 nucliadb/common/cluster/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
@@ -80,7 +80,7 @@ nucliadb/common/datamanagers/synonyms.py,sha256=zk3GEH38KF5vV_VcuL6DCg-2JwgXJfQl
 nucliadb/common/datamanagers/utils.py,sha256=McHlXvE4P3x-bBY3pr0n8djbTDQvI1G5WusJrnRdhLA,1827
 nucliadb/common/datamanagers/vectorsets.py,sha256=XgHNQRw13GpWWymE6qu_ymdzuwL6hDiBKq50fN_sEMM,4007
 nucliadb/common/external_index_providers/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
-nucliadb/common/external_index_providers/base.py,sha256=jijdPxWNLUUO3xZISpO7vvlmm8rOxNY56z8s6QVys6o,8809
+nucliadb/common/external_index_providers/base.py,sha256=yfPkCigT4unXFvAyzy1tXSy2UgWC481GcZAS9bdE4NI,8871
 nucliadb/common/external_index_providers/exceptions.py,sha256=nDhhOIkb66hjCrBk4Spvl2vN1SuW5gbwrMCDmrdjHHE,1209
 nucliadb/common/external_index_providers/manager.py,sha256=aFSrrKKYG1ydpTSyq4zYD0LOxFS7P-CO6rcKC0hiF4I,4267
 nucliadb/common/external_index_providers/pinecone.py,sha256=afglJq6FfifFNTONGrFBCe5yuEL2h3pDFwZkKiA4_6o,39802
@@ -203,7 +203,7 @@ nucliadb/search/api/v1/knowledgebox.py,sha256=uPaMF5c1168Owd6fy_x7IUlcTBzDp2Qm6i
 nucliadb/search/api/v1/predict_proxy.py,sha256=QrGzo0hKjtmyGZ6pjlJHYAh4hxwVUIOTcVcerRCw7eE,3047
 nucliadb/search/api/v1/router.py,sha256=mtT07rBZcVfpa49doaw9b1tj3sdi3qLH0gn9Io6NYM0,988
 nucliadb/search/api/v1/search.py,sha256=vCj5V9kozoti0JrgU_XJhTcBucWzI4SY1B0yCSj9EQw,13638
-nucliadb/search/api/v1/suggest.py,sha256=SXxRVKT5hDSHNKlBYo8XozHHq9bGyvJOlo286lEruLE,5979
+nucliadb/search/api/v1/suggest.py,sha256=S0YUTAWukzZSYZJzN3T5MUgPM3599HQvG76GOCBuAbQ,5907
 nucliadb/search/api/v1/summarize.py,sha256=VAHJvE6V3xUgEBfqNKhgoxmDqCvh30RnrEIBVhMcNLU,2499
 nucliadb/search/api/v1/utils.py,sha256=5Ve-frn7LAE2jqAgB85F8RSeqxDlyA08--gS-AdOLS4,1434
 nucliadb/search/api/v1/resource/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
@@ -219,8 +219,9 @@ 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/hydrator.py,sha256=7Zi44uf2m9b2X_b1aOV2lrWu1Vmbo9lXYgPVUGK0RGI,6728
-nucliadb/search/search/merge.py,sha256=mswe5E5aUMNiMruxpw4iuZTdSMxCVcRtjiXMSYgggzU,20519
+nucliadb/search/search/graph_strategy.py,sha256=Yw6gPiIcKSj6HtbBcIT_5Poypc7AHkqKhcf8RRzJMnI,33513
+nucliadb/search/search/hydrator.py,sha256=-R37gCrGxkyaiHQalnTWHNG_FCx11Zucd7qA1vQCxuw,6985
+nucliadb/search/search/merge.py,sha256=g0PxUejWtYIYWG-VroArMCgwB6AOp3lZMkzoHAPYBKE,22183
 nucliadb/search/search/metrics.py,sha256=81X-tahGW4n2CLvUzCPdNxNClmZqUWZjcVOGCUHoiUM,2872
 nucliadb/search/search/paragraphs.py,sha256=pNAEiYqJGGUVcEf7xf-PFMVqz0PX4Qb-WNG-_zPGN2o,7799
 nucliadb/search/search/pgcatalog.py,sha256=IaNK4dAxdXs38PoIkTdgqMDuZDjeiOtcXn3LeaT-OMw,8855
@@ -232,11 +233,11 @@ nucliadb/search/search/shards.py,sha256=mM2aCHWhl_gwkCENXDShPukS-_qnB5tFS3UAJuzM
 nucliadb/search/search/summarize.py,sha256=ksmYPubEQvAQgfPdZHfzB_rR19B2ci4IYZ6jLdHxZo8,4996
 nucliadb/search/search/utils.py,sha256=iF2tbBA56gRMJH1TlE2hMrqeXqjoeOPt4KgRdp2m9Ek,3313
 nucliadb/search/search/chat/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
-nucliadb/search/search/chat/ask.py,sha256=_emhBlraFGsScWOmEbc8MbYi-aES9XZ1NnhSXR_79qQ,34138
+nucliadb/search/search/chat/ask.py,sha256=4PHueXt4dMF6NHuBYII--reCslbAVCzAE4LgWFAiKdY,35143
 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=KKXdw49rQJgWFrd7aYQ2P373L1soY1nbC4xOQqJnYdo,46867
-nucliadb/search/search/chat/query.py,sha256=4av8tdgIt8WQXNzWt_cQ5cQ12I6X_jr0x3ndPjiINjc,14463
+nucliadb/search/search/chat/prompt.py,sha256=HHzBOMUDta7We_zQqpI-eBLNCFdB0BfYDhijvTmw1k0,46952
+nucliadb/search/search/chat/query.py,sha256=4cmTxnqnvQGPDKTdaNJL4Au8aop1reHLXzkGNhf4NWg,15345
 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
@@ -336,9 +337,9 @@ 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.post2835.dist-info/METADATA,sha256=zFMnPRZlaUaL-SDj_s_Ypi9XpuWMO0gFSUKHYn7R34Q,4689
-nucliadb-6.2.1.post2835.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-nucliadb-6.2.1.post2835.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.2.1.post2835.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.2.1.post2835.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-nucliadb-6.2.1.post2835.dist-info/RECORD,,
+nucliadb-6.2.1.post2842.dist-info/METADATA,sha256=CmYjCtJ5NGjNjopTPvSGiJNDrhzypADVZGncMsY834k,4689
+nucliadb-6.2.1.post2842.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+nucliadb-6.2.1.post2842.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.2.1.post2842.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.2.1.post2842.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+nucliadb-6.2.1.post2842.dist-info/RECORD,,