nucliadb 6.3.4.post3698__py3-none-any.whl → 6.3.4.post3705__py3-none-any.whl

nucliadb/ingest/orm/resource.py

@@ -233,7 +233,7 @@ class Resource:
         # Some basic fields are computed off field metadata.
         # This means we need to recompute upon field deletions.
         if deleted_fields is not None and len(deleted_fields) > 0:
-            remove_field_classifications(self.basic, deleted_fields=deleted_fields)
+            delete_basic_computedmetadata_classifications(self.basic, deleted_fields=deleted_fields)
 
         await datamanagers.resources.set_basic(
             self.txn, kbid=self.kb.kbid, rid=self.uuid, basic=self.basic
@@ -848,7 +848,7 @@ class Resource:
 
         maybe_update_basic_thumbnail(self.basic, field_metadata.metadata.metadata.thumbnail)
 
-        add_field_classifications(self.basic, field_metadata)
+        update_basic_computedmetadata_classifications(self.basic, field_metadata)
         self.modified = True
 
     async def _apply_extracted_vectors(
@@ -1000,37 +1000,57 @@ async def get_file_page_positions(field: File) -> FilePagePositions:
     return positions
 
 
-def remove_field_classifications(basic: PBBasic, deleted_fields: list[FieldID]):
+def delete_basic_computedmetadata_classifications(basic: PBBasic, deleted_fields: list[FieldID]) -> bool:
     """
-    Clean classifications of fields that have been deleted
+    We keep a copy of field classifications computed by the processing engine at the basic object
+    so that users can easily access them without having to load the field metadata from the storage.
+
+    This funcion removes the field classifications for the fields that have been deleted.
+    Returns whether the basic was modified.
     """
-    field_classifications = [
+    if len(deleted_fields) == 0:
+        # Nothing to delete
+        return False
+    new_field_classifications = [
         fc for fc in basic.computedmetadata.field_classifications if fc.field not in deleted_fields
     ]
+    if len(new_field_classifications) == len(basic.computedmetadata.field_classifications):
+        # No changes
+        return False
+
     basic.computedmetadata.ClearField("field_classifications")
-    basic.computedmetadata.field_classifications.extend(field_classifications)
+    basic.computedmetadata.field_classifications.extend(new_field_classifications)
+    return True
 
 
-def add_field_classifications(basic: PBBasic, fcmw: FieldComputedMetadataWrapper) -> bool:
+def update_basic_computedmetadata_classifications(
+    basic: PBBasic, fcmw: FieldComputedMetadataWrapper
+) -> bool:
     """
-    Returns whether some new field classifications were added
+    We keep a copy of field classifications computed by the processing engine at the basic object
+    so that users can easily access them without having to load the field metadata from the storage.
+
+    This function updates the basic object with the new field computed metadata.
+    Returns whether the basic was modified.
     """
-    if len(fcmw.metadata.metadata.classifications) == 0 and all(
-        len(split.classifications) == 0 for split in fcmw.metadata.split_metadata.values()
-    ):
-        return False
+    some_deleted = delete_basic_computedmetadata_classifications(basic, [fcmw.field])
 
-    remove_field_classifications(basic, [fcmw.field])
     fcfs = FieldClassifications()
     fcfs.field.CopyFrom(fcmw.field)
-    fcfs.classifications.extend(fcmw.metadata.metadata.classifications)
+
+    some_added = False
+    if len(fcmw.metadata.metadata.classifications) > 0:
+        some_added = True
+        fcfs.classifications.extend(fcmw.metadata.metadata.classifications)
 
     for split_id, split in fcmw.metadata.split_metadata.items():
         if split_id not in fcmw.metadata.deleted_splits:
-            fcfs.classifications.extend(split.classifications)
-
-    basic.computedmetadata.field_classifications.append(fcfs)
-    return True
+            if len(split.classifications) > 0:
+                some_added = True
+                fcfs.classifications.extend(split.classifications)
+    if some_added:
+        basic.computedmetadata.field_classifications.append(fcfs)
+    return some_added or some_deleted
 
 
 def maybe_update_basic_summary(basic: PBBasic, summary_text: str) -> bool:

{nucliadb-6.3.4.post3698.dist-info → nucliadb-6.3.4.post3705.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nucliadb
-Version: 6.3.4.post3698
+Version: 6.3.4.post3705
 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.post3698
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.4.post3698
-Requires-Dist: nucliadb-protos>=6.3.4.post3698
-Requires-Dist: nucliadb-models>=6.3.4.post3698
-Requires-Dist: nidx-protos>=6.3.4.post3698
+Requires-Dist: nucliadb-telemetry[all]>=6.3.4.post3705
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.3.4.post3705
+Requires-Dist: nucliadb-protos>=6.3.4.post3705
+Requires-Dist: nucliadb-models>=6.3.4.post3705
+Requires-Dist: nidx-protos>=6.3.4.post3705
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn

{nucliadb-6.3.4.post3698.dist-info → nucliadb-6.3.4.post3705.dist-info}/RECORD

@@ -144,7 +144,7 @@ nucliadb/ingest/orm/entities.py,sha256=3_n6lKhBy2GsdmNmkh0_mvxP8md20OZsbtTNEmfJ8
 nucliadb/ingest/orm/exceptions.py,sha256=k4Esv4NtL4TrGTcsQpwrSfDhPQpiYcRbB1SpYmBX5MY,1432
 nucliadb/ingest/orm/knowledgebox.py,sha256=IGOPvBR1qXqDxE5DeiOdYCLdPgjzOVVpsASJ2zYvWwQ,23651
 nucliadb/ingest/orm/metrics.py,sha256=OkwMSPKLZcKba0ZTwtTiIxwBgaLMX5ydhGieKvi2y7E,1096
-nucliadb/ingest/orm/resource.py,sha256=tukAbD32HrRbVa_Ppo9H2YaYQyq659BzOQ45643sp7E,44964
+nucliadb/ingest/orm/resource.py,sha256=oFD7APhmG1A72h7DTKumZWQRpIDM0o_FytP1P-CcNq0,45918
 nucliadb/ingest/orm/utils.py,sha256=vCe_9UxHu26JDFGLwQ0wH-XyzJIpQCTK-Ow9dtZR5Vg,2716
 nucliadb/ingest/orm/processor/__init__.py,sha256=Aqd9wCNTvggkMkCY3WvoI8spdr94Jnqk-0iq9XpLs18,922
 nucliadb/ingest/orm/processor/auditing.py,sha256=TeYhXGJRyQ7ROytbb2u8R0fIh_FYi3HgTu3S1ribY3U,4623
@@ -353,8 +353,8 @@ 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.post3698.dist-info/METADATA,sha256=eId6hciRE0AtrTvOycJk5qjNygXc6kuyn-qX5nSO5eU,4291
-nucliadb-6.3.4.post3698.dist-info/WHEEL,sha256=beeZ86-EfXScwlR_HKu4SllMC9wUEj_8Z_4FJ3egI2w,91
-nucliadb-6.3.4.post3698.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.3.4.post3698.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.3.4.post3698.dist-info/RECORD,,
+nucliadb-6.3.4.post3705.dist-info/METADATA,sha256=BUOECm6cOmDYEdufJ2_6TfQX3ib7DgUIltnFx2y6X9k,4291
+nucliadb-6.3.4.post3705.dist-info/WHEEL,sha256=beeZ86-EfXScwlR_HKu4SllMC9wUEj_8Z_4FJ3egI2w,91
+nucliadb-6.3.4.post3705.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.3.4.post3705.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.3.4.post3705.dist-info/RECORD,,