nucliadb 6.9.0.post5100__py3-none-any.whl → 6.9.0.post5110__py3-none-any.whl

nucliadb/search/search/query_parser/parsers/common.py

@@ -21,7 +21,6 @@ import re
 import string
 from typing import Optional, Union
 
-from nucliadb.common.exceptions import InvalidQueryError
 from nucliadb.search import logger
 from nucliadb.search.search.query_parser.fetcher import Fetcher
 from nucliadb.search.search.query_parser.models import (
@@ -32,15 +31,20 @@ from nucliadb_models import search as search_models
 
 DEFAULT_GENERIC_SEMANTIC_THRESHOLD = 0.7
 
-# -* is an invalid query in tantivy and it won't return results but if you add some whitespaces
-# between - and *, it will actually trigger a tantivy bug and panic
-INVALID_QUERY = re.compile(r"- +\*")
 
+def validate_query_syntax(query: str) -> str:
+    """Filter some queries that panic tantivy, better than returning the 500"""
 
-def validate_query_syntax(query: str):
-    # Filter some queries that panic tantivy, better than returning the 500
+    # -* is an invalid query in tantivy and it won't return results but if you add some whitespaces
+    # between - and *, it will actually trigger a tantivy bug and panic
+    INVALID_QUERY = re.compile(r"- *\*")
     if INVALID_QUERY.search(query):
-        raise InvalidQueryError("query", "Invalid query syntax")
+        # remove the * and extra spaces, as it's probably what doesn't have
+        # meaning in both cases: -* and - *
+        fixed = re.sub(INVALID_QUERY, "- ", query)
+        query = fixed
+
+    return query
 
 
 def is_empty_query(request: search_models.BaseSearchRequest) -> bool:
@@ -85,6 +89,7 @@ async def parse_keyword_query(
     fetcher: Fetcher,
 ) -> KeywordQuery:
     query = item.query
+
     # If there was a rephrase with image, we should use the rephrased query for keyword search
     rephrased_query = await fetcher.get_rephrased_query()
     if item.query_image is not None and rephrased_query is not None:
@@ -98,6 +103,10 @@ async def parse_keyword_query(
             query = synonyms_query
             is_synonyms_query = True
 
+    # after all query transformations, pass a validator that can fix some
+    # queries that trigger a panic on the index
+    query = validate_query_syntax(query)
+
     min_score = parse_keyword_min_score(item.min_score)
 
     return KeywordQuery(

nucliadb/search/search/query_parser/parsers/find.py

@@ -57,7 +57,6 @@ from .common import (
     parse_semantic_query,
     parse_top_k,
     should_disable_vector_search,
-    validate_query_syntax,
 )
 
 
@@ -146,8 +145,6 @@ class _FindParser:
         return retrieval
 
     def _validate_request(self):
-        validate_query_syntax(self.item.query)
-
         # synonyms are not compatible with vector/graph search
         if (
             self.item.with_synonyms

nucliadb/search/search/query_parser/parsers/search.py

@@ -51,7 +51,6 @@ from .common import (
     parse_semantic_query,
     parse_top_k,
     should_disable_vector_search,
-    validate_query_syntax,
 )
 
 INDEX_SORTABLE_FIELDS = [
@@ -128,8 +127,6 @@ class _SearchParser:
         return retrieval
 
     def _validate_request(self):
-        validate_query_syntax(self.item.query)
-
         # synonyms are not compatible with vector/graph search
         if (
             self.item.with_synonyms

{nucliadb-6.9.0.post5100.dist-info → nucliadb-6.9.0.post5110.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nucliadb
-Version: 6.9.0.post5100
+Version: 6.9.0.post5110
 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.post5100
-Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.9.0.post5100
-Requires-Dist: nucliadb-protos>=6.9.0.post5100
-Requires-Dist: nucliadb-models>=6.9.0.post5100
-Requires-Dist: nidx-protos>=6.9.0.post5100
+Requires-Dist: nucliadb-telemetry[all]>=6.9.0.post5110
+Requires-Dist: nucliadb-utils[cache,fastapi,storages]>=6.9.0.post5110
+Requires-Dist: nucliadb-protos>=6.9.0.post5110
+Requires-Dist: nucliadb-models>=6.9.0.post5110
+Requires-Dist: nidx-protos>=6.9.0.post5110
 Requires-Dist: nucliadb-admin-assets>=1.0.0.post1224
 Requires-Dist: nuclia-models>=0.50.0
 Requires-Dist: uvicorn[standard]

{nucliadb-6.9.0.post5100.dist-info → nucliadb-6.9.0.post5110.dist-info}/RECORD

@@ -285,10 +285,10 @@ nucliadb/search/search/query_parser/old_filters.py,sha256=GsU3T3-WiSPvjucP7evHks
 nucliadb/search/search/query_parser/parsers/__init__.py,sha256=ySCNSdbesLXGZyR88919njulA6UE10_3PhqMG_Yj1o4,1034
 nucliadb/search/search/query_parser/parsers/ask.py,sha256=ySa3lBhUuammIchJvj7xodeGIYGkR0uyLnHuOtLfWE8,2810
 nucliadb/search/search/query_parser/parsers/catalog.py,sha256=DErVfWJ9a_F5a8Qbht1WY4Lm8r3ui9Y3C7oJMOOZOng,7474
-nucliadb/search/search/query_parser/parsers/common.py,sha256=jbQweWVufngbobr99qpHh1iiaGICOC6-e9AV33x0-Gk,6594
-nucliadb/search/search/query_parser/parsers/find.py,sha256=4xQwa0BxNucenUrW_iZ2jCGd15Dm4AKS_B91BE8sDi4,12773
+nucliadb/search/search/query_parser/parsers/common.py,sha256=GimacpFLtvvD6hIODwuQBKwW_VD7MrfBzjy9U4ja13k,6871
+nucliadb/search/search/query_parser/parsers/find.py,sha256=7z6GDscMGybvlXENAKOs7XM3p1IJmdC-AZsb4GX_O3g,12698
 nucliadb/search/search/query_parser/parsers/graph.py,sha256=s7nCB7ly_4BZWds-8zce1R-r2fHSiEhAK8P-eL14wTk,9390
-nucliadb/search/search/query_parser/parsers/search.py,sha256=78KSJ9t3I7nFVY2Qk2fMw2P1RHUdGRsWzBf59FdAeTA,10503
+nucliadb/search/search/query_parser/parsers/search.py,sha256=k9JaE9qWGi_eUxEKe_pOKrMIpx7k2wr2j3XrNqZJx9A,10428
 nucliadb/search/search/query_parser/parsers/unit_retrieval.py,sha256=xiOQ7_X6MkcZs3W_0DjdVfyk-G1AY6RBx3oG5hsq7ig,11455
 nucliadb/standalone/__init__.py,sha256=cp15ZcFnHvpcu_5-aK2A4uUyvuZVV_MJn4bIXMa20ks,835
 nucliadb/standalone/api_router.py,sha256=zRSMlaRVHUDGTYA3zC03UV_aLLn-ch-kaeWn1tEjTXw,4338
@@ -384,8 +384,8 @@ 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.post5100.dist-info/METADATA,sha256=0sGsM87JOiyKio3xqNtZEcVvuiXIuhMFywyEPrXoIeQ,4158
-nucliadb-6.9.0.post5100.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nucliadb-6.9.0.post5100.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
-nucliadb-6.9.0.post5100.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
-nucliadb-6.9.0.post5100.dist-info/RECORD,,
+nucliadb-6.9.0.post5110.dist-info/METADATA,sha256=DuN8dmjX6Byrm7_V5TMuUeXvPSJjof6pl6V3CU2TXCU,4158
+nucliadb-6.9.0.post5110.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nucliadb-6.9.0.post5110.dist-info/entry_points.txt,sha256=XqGfgFDuY3zXQc8ewXM2TRVjTModIq851zOsgrmaXx4,1268
+nucliadb-6.9.0.post5110.dist-info/top_level.txt,sha256=hwYhTVnX7jkQ9gJCkVrbqEG1M4lT2F_iPQND1fCzF80,20
+nucliadb-6.9.0.post5110.dist-info/RECORD,,