nucliadb 6.5.1.post4520__py3-none-any.whl → 6.5.1.post4525__py3-none-any.whl

nucliadb/search/search/pgcatalog.py

@@ -19,6 +19,7 @@
 #
 
 import logging
+import re
 from collections import defaultdict
 from typing import Any, Literal, Union, cast
 
@@ -38,6 +39,8 @@ from .filters import translate_label
 observer = metrics.Observer("pg_catalog_search", labels={"op": ""})
 logger = logging.getLogger(__name__)
 
+SPLIT_REGEX = re.compile(r"\W")
+
 
 def _filter_operands(operands: list[CatalogExpression]) -> tuple[list[str], list[CatalogExpression]]:
     facets = []
@@ -163,10 +166,10 @@ def _prepare_query_search(query: search_models.CatalogQuery, params: dict[str, A
         # This is doing tokenization inside the SQL server (to keep the index updated). We could move it to
         # the python code at update/query time if it ever becomes a problem but for now, a single regex
         # executed per query is not a problem.
-        params["query"] = query.query
-        return sql.SQL(
-            "regexp_split_to_array(lower(title), '\\W') @> regexp_split_to_array(lower(%(query)s), '\\W')"
-        )
+
+        # Remove zero-length words from the split
+        params["query"] = [word for word in SPLIT_REGEX.split(query.query) if word]
+        return sql.SQL("regexp_split_to_array(lower(title), '\\W') @> %(query)s")
     elif query.match == search_models.CatalogQueryMatch.Fuzzy:
         params["query"] = query.query
         # Note: the operator is %>, We use %%> for psycopg escaping

{nucliadb-6.5.1.post4520.dist-info → nucliadb-6.5.1.post4525.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.5.1.post4520
+Version: 6.5.1.post4525
 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]>=6.5.1.post4520
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.5.1.post4520
-Requires-Dist: nucliadb-protos>=6.5.1.post4520
-Requires-Dist: nucliadb-models>=6.5.1.post4520
-Requires-Dist: nidx-protos>=6.5.1.post4520
+Requires-Dist: nucliadb-telemetry[all]>=6.5.1.post4525
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.5.1.post4525
+Requires-Dist: nucliadb-protos>=6.5.1.post4525
+Requires-Dist: nucliadb-models>=6.5.1.post4525
+Requires-Dist: nidx-protos>=6.5.1.post4525
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.24.2
 Requires-Dist: uvicorn[standard]

{nucliadb-6.5.1.post4520.dist-info → nucliadb-6.5.1.post4525.dist-info}/RECORD

@@ -254,7 +254,7 @@ nucliadb/search/search/ingestion_agents.py,sha256=IK6yOPEF9rST_uoqspdVdPk0pldjDh
 nucliadb/search/search/merge.py,sha256=XiRBsxhYPshPV7lZXD-9E259KZOPIf4I2tKosY0lPo4,22470
 nucliadb/search/search/metrics.py,sha256=3I6IN0qDSmqIvUaWJmT3rt-Jyjs6LcvnKI8ZqCiuJPY,3501
 nucliadb/search/search/paragraphs.py,sha256=pNAEiYqJGGUVcEf7xf-PFMVqz0PX4Qb-WNG-_zPGN2o,7799
-nucliadb/search/search/pgcatalog.py,sha256=O_nRjSJf1Qc-XorVwcNlsDOftzy_zQLLfagkjU4YmSA,16718
+nucliadb/search/search/pgcatalog.py,sha256=_AiyW6it66UX6BsZbM3-230IQhiEG4utoKYboviyOFI,16799
 nucliadb/search/search/predict_proxy.py,sha256=T4T6c1ZAbBCgPLy-6JoSlixy-0bAh6z0rNH1bMMy3bg,8631
 nucliadb/search/search/query.py,sha256=0qIQdt548L3jtKOyKo06aGJ73SLBxAW3N38_Hc1M3Uw,11528
 nucliadb/search/search/rank_fusion.py,sha256=xZtXhbmKb_56gs73u6KkFm2efvTATOSMmpOV2wrAIqE,9613
@@ -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-6.5.1.post4520.dist-info/METADATA,sha256=4zLZ2kYpWRVlz8HgcOR7TZ9WIIgBCyd-w6wDuMJU_ww,4158
-nucliadb-6.5.1.post4520.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nucliadb-6.5.1.post4520.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.5.1.post4520.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.5.1.post4520.dist-info/RECORD,,
+nucliadb-6.5.1.post4525.dist-info/METADATA,sha256=w8Hgr9H0qA7QLKmypMcYBiX1_unYDquS3n3WWodM-gM,4158
+nucliadb-6.5.1.post4525.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nucliadb-6.5.1.post4525.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.5.1.post4525.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.5.1.post4525.dist-info/RECORD,,