nucliadb 6.6.1.post4590__py3-none-any.whl → 6.6.1.post4596__py3-none-any.whl

nucliadb/ingest/orm/processor/pgcatalog.py

@@ -67,8 +67,10 @@ async def pgcatalog_update(txn: Transaction, kbid: str, resource: Resource, inde
     async with _pg_transaction(txn).connection.cursor() as cur:
         # Labels from the resource and classification labels from each field
         labels = [label for label in index_message.labels]
-        for field in index_message.texts.values():
-            labels += [label for label in field.labels if label.startswith("/l/")]
+        for field in (await resource.get_fields()).values():
+            meta = await field.get_field_metadata()
+            if meta:
+                labels += [f"/l/{c.labelset}/{c.label}" for c in meta.metadata.classifications]
 
         await cur.execute(
             """

nucliadb/search/search/chat/prompt.py

@@ -69,6 +69,7 @@ from nucliadb_models.search import (
     RagStrategyName,
     TableImageStrategy,
     TextBlockAugmentationType,
+    TextPosition,
 )
 from nucliadb_protos import resources_pb2
 from nucliadb_protos.resources_pb2 import ExtractedText, FieldComputedMetadata
@@ -727,6 +728,7 @@ async def neighbouring_paragraphs_prompt_context(
             augmented_context.paragraphs[npid.full()] = AugmentedTextBlock(
                 id=npid.full(),
                 text=ptext,
+                position=get_text_position(npid, neighbour_index, field_extracted_metadata),
                 parent=pid.full(),
                 augmentation_type=TextBlockAugmentationType.NEIGHBOURING_PARAGRAPHS,
             )
@@ -734,6 +736,30 @@ async def neighbouring_paragraphs_prompt_context(
     metrics.set("neighbouring_paragraphs_ops", len(augmented_context.paragraphs))
 
 
+def get_text_position(
+    paragraph_id: ParagraphId, index: int, field_metadata: FieldComputedMetadata
+) -> Optional[TextPosition]:
+    if paragraph_id.field_id.subfield_id:
+        metadata = field_metadata.split_metadata[paragraph_id.field_id.subfield_id]
+    else:
+        metadata = field_metadata.metadata
+    try:
+        pmetadata = metadata.paragraphs[index]
+    except IndexError:
+        return None
+    page_number = None
+    if pmetadata.HasField("page"):
+        page_number = pmetadata.page.page
+    return TextPosition(
+        page_number=page_number,
+        index=index,
+        start=pmetadata.start,
+        end=pmetadata.end,
+        start_seconds=list(pmetadata.start_seconds),
+        end_seconds=list(pmetadata.end_seconds),
+    )
+
+
 def get_neighbouring_indices(
     index: int, before: int, after: int, field_pids: list[ParagraphId]
 ) -> list[int]:

{nucliadb-6.6.1.post4590.dist-info → nucliadb-6.6.1.post4596.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.6.1.post4590
+Version: 6.6.1.post4596
 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]>=6.6.1.post4590
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.6.1.post4590
-Requires-Dist: nucliadb-protos>=6.6.1.post4590
-Requires-Dist: nucliadb-models>=6.6.1.post4590
-Requires-Dist: nidx-protos>=6.6.1.post4590
+Requires-Dist: nucliadb-telemetry[all]>=6.6.1.post4596
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.6.1.post4596
+Requires-Dist: nucliadb-protos>=6.6.1.post4596
+Requires-Dist: nucliadb-models>=6.6.1.post4596
+Requires-Dist: nidx-protos>=6.6.1.post4596
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.43.0
 Requires-Dist: uvicorn[standard]

{nucliadb-6.6.1.post4590.dist-info → nucliadb-6.6.1.post4596.dist-info}/RECORD

@@ -167,7 +167,7 @@ nucliadb/ingest/orm/utils.py,sha256=fCQRuyecgqhaY7mcBG93oaXMkzkKb9BFjOcy4-ZiSNw,
 nucliadb/ingest/orm/processor/__init__.py,sha256=Aqd9wCNTvggkMkCY3WvoI8spdr94Jnqk-0iq9XpLs18,922
 nucliadb/ingest/orm/processor/auditing.py,sha256=TeYhXGJRyQ7ROytbb2u8R0fIh_FYi3HgTu3S1ribY3U,4623
 nucliadb/ingest/orm/processor/data_augmentation.py,sha256=v-pj4GbBWSuO8dQyahs5UDr5ghsyfhCZDS0ftKd6ZYc,5179
-nucliadb/ingest/orm/processor/pgcatalog.py,sha256=jyUn9D7tPES1f8FQqiL8kjzS4puWbgv-bqorP_PEt_A,4251
+nucliadb/ingest/orm/processor/pgcatalog.py,sha256=f0_bV_5qMGXMG7iLK1DnVwi_pXwFvezfdt3aJ0fZfaQ,4347
 nucliadb/ingest/orm/processor/processor.py,sha256=jaEBwbv--WyoC8zcdxWAyF0dAzVA5crVDJl56Bqv1eI,31444
 nucliadb/ingest/orm/processor/sequence_manager.py,sha256=uqEphtI1Ir_yk9jRl2gPf7BlzzXWovbARY5MNZSBI_8,1704
 nucliadb/ingest/service/__init__.py,sha256=LHQFUkdmNBOWqBG0Md9sMMI7g5TQZ-hLAnhw6ZblrJg,2002
@@ -266,7 +266,7 @@ nucliadb/search/search/chat/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn
 nucliadb/search/search/chat/ask.py,sha256=vJ3TSdr-cT_xh43UnoYugqxnHv_-LFSCYoU7o0NnI1M,39368
 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=QwHULUDqe_pS2HZvQH1vzqpYEHQG_-UagXCNtLLtJEI,52997
+nucliadb/search/search/chat/prompt.py,sha256=gmYRC3aK03vrDoBElJP5H5Z7OEeu79k5yTxv3FEkN0I,53866
 nucliadb/search/search/chat/query.py,sha256=3jMPNbiFEOoS0ydMOPYkSx1qVlvAv51npzadWXDwkMs,16650
 nucliadb/search/search/query_parser/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/search/search/query_parser/exceptions.py,sha256=sVl9gRNzhE-s480LBBVkiXzNRbKhYRQN5F3it5tNNp8,939
@@ -375,8 +375,8 @@ 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-6.6.1.post4590.dist-info/METADATA,sha256=Qlvlg1L0sRe2pMcr6yvh9Wq27fSWSereEoun2EcPs2U,4158
-nucliadb-6.6.1.post4590.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nucliadb-6.6.1.post4590.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.6.1.post4590.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.6.1.post4590.dist-info/RECORD,,
+nucliadb-6.6.1.post4596.dist-info/METADATA,sha256=BGQonZKHKd6s_8MEWZ2NMWLh1cw9lLxHES6QWnoVQww,4158
+nucliadb-6.6.1.post4596.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nucliadb-6.6.1.post4596.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.6.1.post4596.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.6.1.post4596.dist-info/RECORD,,