feat: Add conversation variable persistence layer #4
Open
Conversation
… factory to pass the ConversationVariableUpdater factory (the only non-VariablePool dependency), plus a unit test to verify the injection path. - `api/core/workflow/nodes/variable_assigner/v2/node.py` adds a kw-only `conv_var_updater_factory` dependency (defaulting to `conversation_variable_updater_factory`) and stores it for use in `_run`. - `api/core/workflow/nodes/node_factory.py` now injects the factory when creating VariableAssigner v2 nodes. - `api/tests/unit_tests/core/workflow/nodes/variable_assigner/v2/test_variable_assigner_v2.py` adds a test asserting the factory is injected. Tests not run. Next steps (optional): 1) `make lint` 2) `make type-check` 3) `uv run --project api --dev dev/pytest/pytest_unit_tests.sh`
…ructor args. - `api/core/workflow/nodes/node_factory.py` now directly instantiates `VariableAssignerNode` with the injected dependency, and uses a direct call for all other nodes. No tests run.
Add a new command for GraphEngine to update a group of variables. This command takes a group of variable selectors and new values. When the engine receives the command, it will update the corresponding variable in the variable pool. If it does not exist, it will add it; if it does, it will overwrite it. Both behaviors should be treated the same and do not need to be distinguished.
…be-kanban 0941477f) Create a new persistence layer for the Graph Engine. This layer receives a ConversationVariableUpdater upon initialization, which is used to persist the received ConversationVariables to the database. It can retrieve the currently processing ConversationId from the engine's variable pool. It captures the successful execution event of each node and determines whether the type of this node is VariableAssigner(v1 and v2). If so, it retrieves the variable name and value that need to be updated from the node's outputs. This layer is only used in the Advanced Chat. It should be placed outside of Core.Workflow package.
…rs/conversation_variable_persist_layer.py` to satisfy SIM118 - chore(lint): run `make lint` (passes; warnings about missing RECORD during venv package uninstall) - chore(type-check): run `make type-check` (fails: 1275 errors for missing type stubs like `opentelemetry`, `click`, `sqlalchemy`, `flask`, `pydantic`, `pydantic_settings`)
…tType validation and casting - test(graph-engine): update VariableUpdate usages to include value_type in command tests
… drop common_helpers usage - refactor(variable-assigner-v2): inline updated variable payload and drop common_helpers usage Tests not run.
…n and remove value type validation - test(graph-engine): update UpdateVariablesCommand tests to pass concrete Variable instances - fix(graph-engine): align VariableUpdate values with selector before adding to VariablePool Tests not run.
…e handling for v1/v2 process_data - refactor(app-layer): read updated variables from process_data in conversation variable persistence layer - test(app-layer): adapt persistence layer tests to use common_helpers updated-variable payloads Tests not run.
…nce reads from process_data
…fter venv changes) - chore(type-check): run `make type-check` (fails: 1275 missing type stubs across dependencies) Details: - `make lint` fails with `ModuleNotFoundError: No module named 'dotenv_linter.cli'`. - `make type-check` fails with missing stubs for `opentelemetry`, `click`, `sqlalchemy`, `flask`, `pydantic`, `pydantic_settings`, etc.
…ableUnion and remove value type validation" This reverts commit 5ebc87a.
…h SegmentType validation and casting" This reverts commit 3edd525.
This reverts commit 67007f6.
…y out of core.workflow into `api/services/conversation_variable_updater.py` - refactor(app): update advanced chat app runner and conversation service to import the new updater factory Tests not run.
…-linter module missing) - chore(type-check): run `make type-check` (fails: 1275 missing type stubs) Details: - `make lint` reports: `No matches for ignored import core.workflow.nodes.variable_assigner.common.impl -> extensions.ext_database` and ends with `ModuleNotFoundError: No module named 'dotenv_linter.cli'`. - `make type-check` fails with missing type stubs for `opentelemetry`, `click`, `sqlalchemy`, `flask`, `pydantic`, `pydantic_settings`, etc.
…impl import in `api/.importlinter`
There was a problem hiding this comment.
Pull request overview
This PR moves conversation-variable persistence out of VariableAssigner nodes and into a GraphEngine layer, so updated conversation variables can be persisted based on node run results rather than via direct DB calls inside node execution.
Changes:
- Added
ConversationVariablePersistenceLayerto persist updated conversation variables onNodeRunSucceededEventforVARIABLE_ASSIGNER. - Removed conversation-variable DB persistence from VariableAssigner v1/v2 nodes and updated unit tests to assert “updated variables” in
process_data. - Updated the read-only variable pool interface to accept a selector sequence (aligning wrappers/protocols/mocks).
Reviewed changes
Copilot reviewed 14 out of 14 changed files in this pull request and generated 1 comment.
Show a summary per file
| File | Description |
|---|---|
| api/core/app/layers/conversation_variable_persist_layer.py | New GraphEngine layer that persists conversation variables based on assigner updated-variable metadata. |
| api/core/app/apps/advanced_chat/app_runner.py | Registers the new persistence layer in the advanced chat workflow execution pipeline. |
| api/core/workflow/nodes/variable_assigner/v1/node.py | Removes in-node DB persistence; relies on process_data metadata for downstream persistence. |
| api/core/workflow/nodes/variable_assigner/v2/node.py | Removes in-node DB persistence; relies on process_data metadata for downstream persistence. |
| api/services/conversation_variable_updater.py | Provides a services-level updater/factory used by the new layer and conversation service. |
| api/services/conversation_service.py | Switches to the services-level conversation_variable_updater_factory import. |
| api/core/workflow/runtime/read_only_wrappers.py | Updates read-only wrapper get() signature to take selector sequences. |
| api/core/workflow/runtime/graph_runtime_state_protocol.py | Updates protocol get() signature to take selector sequences. |
| api/tests/unit_tests/core/app/layers/test_conversation_variable_persist_layer.py | Adds unit tests for the new conversation variable persistence layer behavior. |
| api/tests/unit_tests/core/app/layers/test_pause_state_persist_layer.py | Updates test mock variable-pool interface to match selector-based get(). |
| api/tests/unit_tests/core/workflow/nodes/variable_assigner/v1/test_variable_assigner_v1.py | Updates tests to validate assigner “updated variables” in node run process_data. |
| api/tests/unit_tests/core/workflow/nodes/variable_assigner/v2/test_variable_assigner_v2.py | Adds a test ensuring node factory creates the v2 assigner node. |
| api/core/workflow/nodes/node_factory.py | Minor formatting change. |
| api/.importlinter | Removes no-longer-needed ignore for the old variable assigner DB import path. |
Comments suppressed due to low confidence (1)
api/services/conversation_variable_updater.py:23
- ConversationVariableUpdaterImpl.update() creates a SQLAlchemy Session but never closes it. This will leak DB connections over time under load. Use a context manager (
with Session(...) as session:) or ensuresession.close()runs in afinallyblock (and rollback on exceptions as appropriate).
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+38
to
+53
| for item in updated_variables: | ||
| selector = item.selector | ||
| if len(selector) < 2: | ||
| logger.warning("Conversation variable selector invalid. selector=%s", selector) | ||
| continue | ||
| if selector[0] != CONVERSATION_VARIABLE_NODE_ID: | ||
| continue | ||
| variable = self.graph_runtime_state.variable_pool.get(selector) | ||
| if not isinstance(variable, Variable): | ||
| logger.warning( | ||
| "Conversation variable not found in variable pool. selector=%s", | ||
| selector, | ||
| ) | ||
| continue | ||
| self._conversation_variable_updater.update(conversation_id=conversation_id, variable=variable) | ||
| self._conversation_variable_updater.flush() |
There was a problem hiding this comment.
flush() is called inside the per-variable loop. This defeats the batching semantics described by ConversationVariableUpdater (and can add unnecessary commits/round-trips). Consider calling update() for each conversation variable, then call flush() once after the loop (or in on_graph_end) only if at least one update occurred.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Benchmark PR from agentic-review-benchmarks#4