nucliadb 6.9.0.post5078__py3-none-any.whl → 6.9.0.post5086__py3-none-any.whl

nucliadb/ingest/fields/conversation.py

@@ -58,7 +58,19 @@ class Conversation(Field[PBConversation]):
         self._splits_metadata: Optional[SplitsMetadata] = None
         self.metadata = None
 
+    async def delete_value(self):
+        await self.resource.txn.delete_by_prefix(
+            CONVERSATION_METADATA.format(kbid=self.kbid, uuid=self.uuid, type=self.type, field=self.id)
+        )
+        self._split_metadata = None
+        self.metadata = None
+        self.value.clear()
+
     async def set_value(self, payload: PBConversation):
+        if payload.replace_field:
+            # As we need to overwrite the value of the conversation, first delete any previous data.
+            await self.delete_value()
+
         metadata = await self.get_metadata()
         metadata.extract_strategy = payload.extract_strategy
         metadata.split_strategy = payload.split_strategy

nucliadb/writer/api/v1/field.py

@@ -249,9 +249,10 @@ async def parse_conversation_field_adapter(
     writer: BrokerMessage,
     toprocess: PushPayload,
     resource_classifications: ResourceClassifications,
+    replace_field: bool = False,
 ):
     return await parse_conversation_field(
-        field_id, field_payload, writer, toprocess, kbid, rid, resource_classifications
+        field_id, field_payload, writer, toprocess, kbid, rid, resource_classifications, replace_field
     )
 
 
@@ -380,7 +381,9 @@ async def add_resource_field_conversation_rslug_prefix(
     field_id: FieldIdString,
     field_payload: models.InputConversationField,
 ) -> ResourceFieldAdded:
-    return await add_field_to_resource_by_slug(request, kbid, rslug, field_id, field_payload)
+    return await add_field_to_resource_by_slug(
+        request, kbid, rslug, field_id, field_payload, replace_field=True
+    )
 
 
 @api.put(
@@ -399,7 +402,7 @@ async def add_resource_field_conversation_rid_prefix(
     field_id: FieldIdString,
     field_payload: models.InputConversationField,
 ) -> ResourceFieldAdded:
-    return await add_field_to_resource(request, kbid, rid, field_id, field_payload)
+    return await add_field_to_resource(request, kbid, rid, field_id, field_payload, replace_field=True)
 
 
 @api.put(
@@ -466,7 +469,9 @@ async def append_messages_to_conversation_field_rslug_prefix(
         field = models.InputConversationField(messages=messages)
     except pydantic.ValidationError as e:
         raise HTTPException(status_code=422, detail=str(e))
-    return await add_field_to_resource_by_slug(request, kbid, rslug, field_id, field)
+    return await add_field_to_resource_by_slug(
+        request, kbid, rslug, field_id, field, replace_field=False
+    )
 
 
 @api.put(
@@ -489,7 +494,7 @@ async def append_messages_to_conversation_field_rid_prefix(
         field = models.InputConversationField(messages=messages)
     except pydantic.ValidationError as e:
         raise HTTPException(status_code=422, detail=str(e))
-    return await add_field_to_resource(request, kbid, rid, field_id, field)
+    return await add_field_to_resource(request, kbid, rid, field_id, field, replace_field=False)
 
 
 @api.delete(

nucliadb/writer/resource/field.py

@@ -228,6 +228,7 @@ async def parse_fields(
             kbid,
             uuid,
             resource_classifications,
+            replace_field=True,
         )
 
 
@@ -431,19 +432,15 @@ async def parse_conversation_field(
     kbid: str,
     uuid: str,
     resource_classifications: ResourceClassifications,
+    replace_field: bool,
 ) -> None:
-    # Make sure that the max number of messages is not exceeded
-    current_message_count = await get_current_conversation_message_count(kbid, uuid, key)
-    if len(conversation_field.messages) + current_message_count > MAX_CONVERSATION_MESSAGES:
-        raise HTTPException(
-            status_code=422,
-            detail=f"Conversation fields cannot have more than {MAX_CONVERSATION_MESSAGES} messages.",
-        )
-
+    if not replace_field:
+        # Appending messages to conversation
+        await _conversation_append_checks(kbid, uuid, key, conversation_field)
     classif_labels = resource_classifications.for_field(key, resources_pb2.FieldType.CONVERSATION)
     storage = await get_storage(service_name=SERVICE_NAME)
     processing = get_processing()
-    field_value = resources_pb2.Conversation()
+    field_value = resources_pb2.Conversation(replace_field=replace_field)
     convs = processing_models.PushConversation()
     for message in conversation_field.messages:
         cm = resources_pb2.Message()
@@ -554,13 +551,31 @@ async def get_stored_resource_classifications(
     return rc
 
 
-async def get_current_conversation_message_count(kbid: str, rid: str, field_id: str) -> int:
+async def _conversation_append_checks(
+    kbid: str, rid: str, field_id: str, input: models.InputConversationField
+):
     async with datamanagers.with_ro_transaction() as txn:
         resource_obj = await datamanagers.resources.get_resource(txn, kbid=kbid, rid=rid)
         if resource_obj is None:
-            return 0
-        field_obj: Conversation = await resource_obj.get_field(
+            return
+        conv: Conversation = await resource_obj.get_field(
             field_id, resources_pb2.FieldType.CONVERSATION, load=False
         )
-        metadata = await field_obj.get_metadata()
-        return metadata.total
+
+        # Make sure that the max number of messages is not exceeded
+        current_message_count = (await conv.get_metadata()).total
+        if len(input.messages) + current_message_count > MAX_CONVERSATION_MESSAGES:
+            raise HTTPException(
+                status_code=422,
+                detail=f"Conversation fields cannot have more than {MAX_CONVERSATION_MESSAGES} messages.",
+            )
+
+        # Make sure input messages use unique idents
+        existing_message_ids = set((await conv.get_splits_metadata()).metadata.keys())
+        input_message_ids = {message.ident for message in input.messages}
+        intersection = input_message_ids.intersection(existing_message_ids)
+        if intersection != set():
+            raise HTTPException(
+                status_code=422,
+                detail=f"Message identifiers must be unique field={field_id}: {list(intersection)[:50]}",
+            )

{nucliadb-6.9.0.post5078.dist-info → nucliadb-6.9.0.post5086.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.9.0.post5078
+Version: 6.9.0.post5086
 Summary: NucliaDB
 Author-email: Nuclia <nucliadb@nuclia.com>
 License-Expression: AGPL-3.0-or-later
@@ -19,11 +19,11 @@ 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]>=6.9.0.post5078
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.9.0.post5078
-Requires-Dist: nucliadb-protos>=6.9.0.post5078
-Requires-Dist: nucliadb-models>=6.9.0.post5078
-Requires-Dist: nidx-protos>=6.9.0.post5078
+Requires-Dist: nucliadb-telemetry[all]>=6.9.0.post5086
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.9.0.post5086
+Requires-Dist: nucliadb-protos>=6.9.0.post5086
+Requires-Dist: nucliadb-models>=6.9.0.post5086
+Requires-Dist: nidx-protos>=6.9.0.post5086
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.50.0
 Requires-Dist: uvicorn[standard]

{nucliadb-6.9.0.post5078.dist-info → nucliadb-6.9.0.post5086.dist-info}/RECORD

@@ -155,7 +155,7 @@ nucliadb/ingest/consumer/shard_creator.py,sha256=UKIk0yaS_jC_nGQqymn9NGJWzwZEqhI
 nucliadb/ingest/consumer/utils.py,sha256=jpX8D4lKzuPCpArQLZeX_Zczq3pfen_zAf8sPJfOEZU,2642
 nucliadb/ingest/fields/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/ingest/fields/base.py,sha256=D8NzawonF7hivDW9zvQBbV938TKA6e2OCGqV4kS96RU,24405
-nucliadb/ingest/fields/conversation.py,sha256=Jb6nDvlGQiX60VM6NzX1-okrNexlXuUSZ-Ge_yIJYVw,9098
+nucliadb/ingest/fields/conversation.py,sha256=KkOvNM1rZFQRg2RsfGd3Jrz3lpx0HpGpN1cmlpz_mZw,9563
 nucliadb/ingest/fields/exceptions.py,sha256=sZBk21BSrXFdOdo1qUdCAyD-9YMYakSLdn4_WdIPCIQ,1217
 nucliadb/ingest/fields/file.py,sha256=1v4jLg3balUua2VmSV8hHkAwPFShTUCOzufZvIUQcQw,4740
 nucliadb/ingest/fields/generic.py,sha256=elgtqv15aJUq3zY7X_g0bli_2BpcwPArVvzhe54Y4Ig,1547
@@ -361,7 +361,7 @@ nucliadb/writer/api/constants.py,sha256=SCdqGDbEmpdczQdTfbTlpHzVjbLqccPtMQ25MPIF
 nucliadb/writer/api/utils.py,sha256=wIQHlU8RQiIGVLI72suvyVIKlCU44Unh0Ae0IiN6Qwo,1313
 nucliadb/writer/api/v1/__init__.py,sha256=akI9A_jloNLb0dU4T5zjfdyvmSAiDeIdjAlzNx74FlU,1128
 nucliadb/writer/api/v1/export_import.py,sha256=v0sU55TtRSqDzwkDgcwv2uSaqKCuQTtGcMpYoHQYBQA,8192
-nucliadb/writer/api/v1/field.py,sha256=qcuniSwR9tR9vn5abpK3rB_olpuUTEj_0LcL_1eAiLw,18972
+nucliadb/writer/api/v1/field.py,sha256=nO3IEV6v5hokdIo5HoaecdwDqvr1PzCJlh5DafzcNTw,19130
 nucliadb/writer/api/v1/knowledgebox.py,sha256=kioqjD3yN-y1cDTgmXAAOwivXHX9NXxwblcSzGqJup0,9533
 nucliadb/writer/api/v1/learning_config.py,sha256=DTLEzKJ3dHvi8pbZscjElUqCH_ZvLc6WZgvalFqHo10,4450
 nucliadb/writer/api/v1/resource.py,sha256=IfcT6HXnR5sC5wSnQSuKmFzEWcLTh7OzZEAV4hYmXnA,20442
@@ -374,7 +374,7 @@ nucliadb/writer/api/v1/vectorsets.py,sha256=F3iMViL5G95_Tns4aO2SOA0DwAzxK2_P8MXx
 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=44GK8M9EEVoAUfGiabdLrrpENqeFwNn7qwxF2AHhQGg,10504
-nucliadb/writer/resource/field.py,sha256=eM2KFxhcG3u6-ldniZDSYqGzhJ5bpWgIQBGXXFwskqw,22195
+nucliadb/writer/resource/field.py,sha256=kJFxOgmizGbEuTRPb5o0cNqonZ8sa9ehVlSfRk-ektY,22866
 nucliadb/writer/resource/origin.py,sha256=pvhUDdU0mlWPUcpoQi4LDUJaRtfjzVVrA8XcGVI_N8k,2021
 nucliadb/writer/tus/__init__.py,sha256=Kera0BtxoDX0ngPftXiMjNgjrhtQ3l2XFc5nJqSBOJY,5498
 nucliadb/writer/tus/azure.py,sha256=yxoRi4PhGDikTqVK3PiuVyguy8H9DOS66JpZCY4hpUY,4177
@@ -385,8 +385,8 @@ nucliadb/writer/tus/local.py,sha256=7jYa_w9b-N90jWgN2sQKkNcomqn6JMVBOVeDOVYJHto,
 nucliadb/writer/tus/s3.py,sha256=vu1BGg4VqJ_x2P1u2BxqPKlSfw5orT_a3R-Ln5oPUpU,8483
 nucliadb/writer/tus/storage.py,sha256=ToqwjoYnjI4oIcwzkhha_MPxi-k4Jk3Lt55zRwaC1SM,2903
 nucliadb/writer/tus/utils.py,sha256=MSdVbRsRSZVdkaum69_0wku7X3p5wlZf4nr6E0GMKbw,2556
-nucliadb-6.9.0.post5078.dist-info/METADATA,sha256=x9K2fNkUdKoA0BNuiVFTg6IkhDHeiAvUZntKj6P4g94,4158
-nucliadb-6.9.0.post5078.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nucliadb-6.9.0.post5078.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.9.0.post5078.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.9.0.post5078.dist-info/RECORD,,
+nucliadb-6.9.0.post5086.dist-info/METADATA,sha256=kaY1HSGwZVkCNZzUpp-_XgKI32QB48_6dRb-kK9GUSA,4158
+nucliadb-6.9.0.post5086.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nucliadb-6.9.0.post5086.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.9.0.post5086.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.9.0.post5086.dist-info/RECORD,,