nucliadb 6.1.0.post2533__py3-none-any.whl → 6.1.0.post2536__py3-none-any.whl

nucliadb/search/search/chat/prompt.py

@@ -680,6 +680,7 @@ async def conversation_prompt_context(
             if field_type == "c" and paragraph.score_type in (
                 SCORE_TYPE.VECTOR,
                 SCORE_TYPE.BOTH,
+                SCORE_TYPE.BM25,
             ):
                 field_unique_id = "-".join([rid, field_type, field_id])
                 if field_unique_id in analyzed_fields:
@@ -695,20 +696,31 @@ async def conversation_prompt_context(
 
                 attachments: List[resources_pb2.FieldRef] = []
                 if conversational_strategy.full:
+                    extracted_text = await field_obj.get_extracted_text()
                     for current_page in range(1, cmetadata.pages + 1):
                         conv = await field_obj.db_get_value(current_page)
+
                         for message in conv.messages:
-                            text = message.content.text.strip()
-                            pid = f"{rid}/{field_type}/{field_id}/{message.ident}/0-{len(message.content.text) + 1}"
+                            ident = message.ident
+                            if extracted_text is not None:
+                                text = extracted_text.split_text.get(ident, message.content.text.strip())
+                            else:
+                                text = message.content.text.strip()
+                            pid = f"{rid}/{field_type}/{field_id}/{ident}/0-{len(text) + 1}"
                             context[pid] = text
                             attachments.extend(message.content.attachments_fields)
                 else:
                     # Add first message
+                    extracted_text = await field_obj.get_extracted_text()
                     first_page = await field_obj.db_get_value()
                     if len(first_page.messages) > 0:
                         message = first_page.messages[0]
-                        text = message.content.text.strip()
-                        pid = f"{rid}/{field_type}/{field_id}/{message.ident}/0-{len(message.content.text) + 1}"
+                        ident = message.ident
+                        if extracted_text is not None:
+                            text = extracted_text.split_text.get(ident, message.content.text.strip())
+                        else:
+                            text = message.content.text.strip()
+                        pid = f"{rid}/{field_type}/{field_id}/{ident}/0-{len(text) + 1}"
                         context[pid] = text
                         attachments.extend(message.content.attachments_fields)
 
@@ -746,7 +758,7 @@ async def conversation_prompt_context(
                         extracted_text = await field.get_extracted_text()
                         if extracted_text is not None:
                             pid = f"{rid}/{field_type}/{attachment.field_id}/0-{len(extracted_text.text) + 1}"
-                            context[pid] = extracted_text.text
+                            context[pid] = f"Attachment {attachment.field_id}: {extracted_text.text}\n\n"
 
                 if conversational_strategy.attachments_images and visual_llm:
                     for attachment in attachments:

{nucliadb-6.1.0.post2533.dist-info → nucliadb-6.1.0.post2536.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nucliadb
-Version: 6.1.0.post2533
+Version: 6.1.0.post2536
 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]>=6.1.0.post2533
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.1.0.post2533
-Requires-Dist: nucliadb-protos>=6.1.0.post2533
-Requires-Dist: nucliadb-models>=6.1.0.post2533
+Requires-Dist: nucliadb-telemetry[all]>=6.1.0.post2536
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.1.0.post2536
+Requires-Dist: nucliadb-protos>=6.1.0.post2536
+Requires-Dist: nucliadb-models>=6.1.0.post2536
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nucliadb-node-binding>=2.26.0
 Requires-Dist: uvicorn

{nucliadb-6.1.0.post2533.dist-info → nucliadb-6.1.0.post2536.dist-info}/RECORD

@@ -230,7 +230,7 @@ nucliadb/search/search/chat/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn
 nucliadb/search/search/chat/ask.py,sha256=2ZIz3TbCG18X0eJZSfLc8nHyKToyrlv_SPFzOOcc5wU,33865
 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=mreHIjhIWCIGDlJ2CA7kv7KLgh8d09meoMX3-bvWje8,46152
+nucliadb/search/search/chat/prompt.py,sha256=TIzjI_882hJ--KLKCY8rJomtJ_CMJ-MHYtHqivgG8Lk,46819
 nucliadb/search/search/chat/query.py,sha256=gKtlj2ms81m417Id29-DtHFxE3M4TtJvYNB03gAgpYo,14402
 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
@@ -331,9 +331,9 @@ 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-6.1.0.post2533.dist-info/METADATA,sha256=92uKuzSMwFjVzxVwrt2ogNuoKMquNvZY64BtiIz1rEY,4390
-nucliadb-6.1.0.post2533.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-nucliadb-6.1.0.post2533.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.1.0.post2533.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.1.0.post2533.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-nucliadb-6.1.0.post2533.dist-info/RECORD,,
+nucliadb-6.1.0.post2536.dist-info/METADATA,sha256=UQH-519ROPaHACjS8TGo9X-Ex9RD_kvJqcp6CFwbcIM,4390
+nucliadb-6.1.0.post2536.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+nucliadb-6.1.0.post2536.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.1.0.post2536.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.1.0.post2536.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+nucliadb-6.1.0.post2536.dist-info/RECORD,,