Skip to content

fix(arcadedb): work around index-scan datetime comparison bug (3 code paths) #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
agc-63 wants to merge 1 commit into from
Closed

fix(arcadedb): work around index-scan datetime comparison bug (3 code paths) #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
48 changes: 37 additions & 11 deletions graphiti_core/graphiti.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@
from graphiti_core.cross_encoder.client import CrossEncoderClient
from graphiti_core.cross_encoder.openai_reranker_client import OpenAIRerankerClient
from graphiti_core.decorators import handle_multiple_group_ids
from graphiti_core.driver.driver import GraphDriver
from graphiti_core.driver.driver import GraphDriver, GraphProvider
from graphiti_core.driver.neo4j_driver import Neo4jDriver
from graphiti_core.edges import (
CommunityEdge,
Expand Down Expand Up @@ -403,12 +403,21 @@ async def _saga_get_previous_episode_uuid(
except NotImplementedError:
pass

# ArcadeDB-only: wrap valid_at/created_at in datetime(...) to avoid a
# ClassCastException comparing native-datetime representations once a
# range index exists on the property (same class of bug fixed in
# graph_data_operations.py::retrieve_episodes()).
order_by = (
'ORDER BY datetime(e.valid_at) DESC, datetime(e.created_at) DESC'
if self.driver.provider == GraphProvider.ARCADEDB
else 'ORDER BY e.valid_at DESC, e.created_at DESC'
)
records, _, _ = await self.driver.execute_query(
"""
MATCH (s:Saga {uuid: $saga_uuid})-[:HAS_EPISODE]->(e:Episodic)
f"""
MATCH (s:Saga {{uuid: $saga_uuid}})-[:HAS_EPISODE]->(e:Episodic)
WHERE e.uuid <> $current_episode_uuid
RETURN e.uuid AS uuid
ORDER BY e.valid_at DESC, e.created_at DESC
{order_by}
LIMIT 1
""",
saga_uuid=saga_uuid,
Expand Down Expand Up @@ -485,13 +494,25 @@ async def summarize_saga(self, saga_id: str) -> SagaNode:
saga_id, since=since, limit=max_episodes
)
if episodes_data is None:
# ArcadeDB-only: see the note in _saga_get_previous_episode_uuid.
is_arcadedb = self.driver.provider == GraphProvider.ARCADEDB
if since is not None:
created_at_where = (
'WHERE datetime(e.created_at) > datetime($since)'
if is_arcadedb
else 'WHERE e.created_at > $since'
)
order_by = (
'ORDER BY datetime(e.valid_at) ASC, datetime(e.created_at) ASC'
if is_arcadedb
else 'ORDER BY e.valid_at ASC, e.created_at ASC'
)
records, _, _ = await self.driver.execute_query(
"""
MATCH (s:Saga {uuid: $saga_uuid})-[:HAS_EPISODE]->(e:Episodic)
WHERE e.created_at > $since
f"""
MATCH (s:Saga {{uuid: $saga_uuid}})-[:HAS_EPISODE]->(e:Episodic)
{created_at_where}
RETURN e.content AS content, e.valid_at AS valid_at
ORDER BY e.valid_at ASC, e.created_at ASC
{order_by}
LIMIT $limit
""",
saga_uuid=saga_id,
Expand All @@ -500,11 +521,16 @@ async def summarize_saga(self, saga_id: str) -> SagaNode:
routing_='r',
)
else:
order_by = (
'ORDER BY datetime(e.valid_at) DESC, datetime(e.created_at) DESC'
if is_arcadedb
else 'ORDER BY e.valid_at DESC, e.created_at DESC'
)
records, _, _ = await self.driver.execute_query(
"""
MATCH (s:Saga {uuid: $saga_uuid})-[:HAS_EPISODE]->(e:Episodic)
f"""
MATCH (s:Saga {{uuid: $saga_uuid}})-[:HAS_EPISODE]->(e:Episodic)
RETURN e.content AS content, e.valid_at AS valid_at
ORDER BY e.valid_at DESC, e.created_at DESC
{order_by}
LIMIT $limit
""",
saga_uuid=saga_id,
Expand Down
27 changes: 22 additions & 5 deletions graphiti_core/search/search_filters.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,8 +105,25 @@ def node_search_filter_query_constructor(


def date_filter_query_constructor(
value_name: str, param_name: str, operator: ComparisonOperator
value_name: str,
param_name: str,
operator: ComparisonOperator,
provider: GraphProvider | None = None,
) -> str:
# ArcadeDB stores a native datetime write as a zone-less ChronoLocalDateTime
# but decodes an incoming native datetime parameter as a zone-aware
# OffsetDateTime -- comparing them directly can raise a ClassCastException
# at the engine level once a range index exists on the property (same class
# of bug as retrieve_episodes(), see graph_data_operations.py). Wrapping
# both sides in datetime(...) sidesteps it; scoped to ARCADEDB only, and
# skipped for IS NULL/IS NOT NULL since there's no value to wrap there.
if provider == GraphProvider.ARCADEDB and operator not in (
ComparisonOperator.is_null,
ComparisonOperator.is_not_null,
):
value_name = f'datetime({value_name})'
param_name = f'datetime({param_name})'
Comment on lines +120 to +125

@gemini-code-assist gemini-code-assist Bot Jul 5, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

Wrapping the indexed property (value_name) in the datetime() function makes the query non-sargable. This means ArcadeDB's query planner will not be able to utilize the range index on these properties, potentially falling back to full scans and causing performance degradation on larger datasets. While this is a necessary workaround for the upstream engine bug (arcadedb#5008), please ensure there is a tracking issue or TODO to remove this workaround once the upstream fix is released.


query = '(' + value_name + ' '

if operator == ComparisonOperator.is_null or operator == ComparisonOperator.is_not_null:
Expand Down Expand Up @@ -158,7 +175,7 @@ def edge_search_filter_query_constructor(

and_filters = [
date_filter_query_constructor(
'e.valid_at', f'$valid_at_{j}', date_filter.comparison_operator
'e.valid_at', f'$valid_at_{j}', date_filter.comparison_operator, provider
)
for j, date_filter in enumerate(or_list)
]
Expand Down Expand Up @@ -189,7 +206,7 @@ def edge_search_filter_query_constructor(

and_filters = [
date_filter_query_constructor(
'e.invalid_at', f'$invalid_at_{j}', date_filter.comparison_operator
'e.invalid_at', f'$invalid_at_{j}', date_filter.comparison_operator, provider
)
for j, date_filter in enumerate(or_list)
]
Expand Down Expand Up @@ -220,7 +237,7 @@ def edge_search_filter_query_constructor(

and_filters = [
date_filter_query_constructor(
'e.created_at', f'$created_at_{j}', date_filter.comparison_operator
'e.created_at', f'$created_at_{j}', date_filter.comparison_operator, provider
)
for j, date_filter in enumerate(or_list)
]
Expand Down Expand Up @@ -251,7 +268,7 @@ def edge_search_filter_query_constructor(

and_filters = [
date_filter_query_constructor(
'e.expired_at', f'$expired_at_{j}', date_filter.comparison_operator
'e.expired_at', f'$expired_at_{j}', date_filter.comparison_operator, provider
)
for j, date_filter in enumerate(or_list)
]
Expand Down
33 changes: 26 additions & 7 deletions graphiti_core/utils/maintenance/graph_data_operations.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,6 +96,25 @@ async def retrieve_episodes(
except NotImplementedError:
pass

# ArcadeDB stores a native datetime write as a zone-less ChronoLocalDateTime
# but decodes an incoming native datetime parameter as a zone-aware
# OffsetDateTime -- comparing them directly raises a ClassCastException at
# the engine level (confirmed: reproducible whenever a range index on the
# property was created before any data existed, which is exactly what
# build_indices_and_constraints() always does). Wrapping both sides in
# datetime(...) sidesteps it; scoped to ARCADEDB only so the other
# providers' generated Cypher is untouched.
valid_at_where = (
'WHERE datetime(e.valid_at) <= datetime($reference_time)'
if driver.provider == GraphProvider.ARCADEDB
else 'WHERE e.valid_at <= $reference_time'
)
valid_at_order = (
'ORDER BY datetime(e.valid_at) DESC'
if driver.provider == GraphProvider.ARCADEDB
else 'ORDER BY e.valid_at DESC'
)

# If saga is provided, retrieve episodes from that saga only
if saga is not None:
group_id = group_ids[0] if group_ids else None
Expand All @@ -104,7 +123,7 @@ async def retrieve_episodes(
records, _, _ = await driver.execute_query(
f"""
MATCH (s:Saga {{name: $saga_name, group_id: $group_id}})-[:HAS_EPISODE]->(e:Episodic)
WHERE e.valid_at <= $reference_time
{valid_at_where}
{source_filter}
RETURN
"""
Expand All @@ -113,8 +132,8 @@ async def retrieve_episodes(
if driver.provider == GraphProvider.NEPTUNE
else EPISODIC_NODE_RETURN
)
+ """
ORDER BY e.valid_at DESC
+ f"""
{valid_at_order}
LIMIT $num_episodes
""",
saga_name=saga,
Expand All @@ -138,9 +157,9 @@ async def retrieve_episodes(
query_params['source'] = source.name

query: LiteralString = (

@gemini-code-assist gemini-code-assist Bot Jul 5, 2026

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The query variable is annotated as LiteralString, but it is constructed dynamically using f-strings and query_filter (which is also a dynamic string). In Python, f-strings with runtime variables do not satisfy the LiteralString type constraint, which will cause static type checkers like mypy or pyright to report a type mismatch error. We should change the type annotation to str.

Suggested change
query: LiteralString = (
query: str = (

"""
f"""
MATCH (e:Episodic)
WHERE e.valid_at <= $reference_time
{valid_at_where}
"""
+ query_filter
+ """
Expand All @@ -151,8 +170,8 @@ async def retrieve_episodes(
if driver.provider == GraphProvider.NEPTUNE
else EPISODIC_NODE_RETURN
)
+ """
ORDER BY e.valid_at DESC
+ f"""
{valid_at_order}
LIMIT $num_episodes
"""
)
Expand Down
Loading