Hot Fix AttributeError in Neo4JStorage and MemgraphStorage when using storage specified workspace env var

[danielaskdd]

Hot Fix AttributeError in Neo4JStorage and MemgraphStorage when using storage specified workspace env var

---

Summary

This PR fixes a critical bug in Neo4JStorage and MemgraphStorage that causes an AttributeError when the NEO4J_WORKSPACE or MEMGRAPH_WORKSPACE environment variables are set.

Problem

When initializing LightRAG with Neo4J or Memgraph graph storage and the workspace environment variable is set, the application crashes with:

AttributeError: 'Neo4JStorage' object has no attribute 'workspace'

or

AttributeError: 'MemgraphStorage' object has no attribute 'workspace'

Root Cause

Both Neo4JStorage and MemgraphStorage use custom __init__ methods that attempt to access self.workspace and self.namespace attributes before calling super().__init__(). Since these attributes are defined in the parent BaseGraphStorage dataclass and initialized by the parent's __init__, they don't exist until super().__init__() is called.

The problematic code pattern was:

def __init__(self, namespace, global_config, embedding_func, workspace=None):
    neo4j_workspace = os.environ.get("NEO4J_WORKSPACE")
    if neo4j_workspace and neo4j_workspace.strip():
        logger.info(
            f"Using NEO4J_WORKSPACE environment variable: '{neo4j_workspace}' (overriding '{self.workspace}/{self.namespace}')"
            # ↑ BUG: self.workspace and self.namespace don't exist yet!
        )
        workspace = neo4j_workspace

    super().__init__(...)  # Attributes are initialized here

Solution

Move the logging statement to after super().__init__() is called, and use parameter values instead of self attributes in the log message:

def __init__(self, namespace, global_config, embedding_func, workspace=None):
    neo4j_workspace = os.environ.get("NEO4J_WORKSPACE")
    original_workspace = workspace  # Save original value for logging
    if neo4j_workspace and neo4j_workspace.strip():
        workspace = neo4j_workspace

    super().__init__(
        namespace=namespace,
        workspace=workspace,
        global_config=global_config,
        embedding_func=embedding_func,
    )
    
    # Log after super().__init__() to ensure self.workspace is initialized
    if neo4j_workspace and neo4j_workspace.strip():
        logger.info(
            f"Using NEO4J_WORKSPACE environment variable: '{neo4j_workspace}' (overriding '{original_workspace}/{namespace}')"
        )

Changes

• lightrag/kg/neo4j_impl.py: Fixed Neo4JStorage.__init__() to access attributes only after parent initialization
• lightrag/kg/memgraph_impl.py: Fixed MemgraphStorage.__init__() to access attributes only after parent initialization

Why Other Storage Implementations Are Not Affected

Other storage implementations (Redis, Milvus, PostgreSQL, MongoDB, Qdrant, etc.) use dataclass __post_init__ methods instead of custom __init__ methods. With __post_init__, the dataclass framework automatically initializes all fields (including workspace and namespace) before __post_init__ is called, so there is no issue.

Testing

1. Set NEO4J_WORKSPACE environment variable
2. Initialize LightRAG with Neo4J graph storage
3. Verify no AttributeError is raised
4. Confirm Neo4J storage initializes correctly with the specified workspace

Breaking Changes

None. This is a pure bug fix with no API changes.

Related Issues

Fixes the error:

ERROR: Failed to initialize LightRAG: 'Neo4JStorage' object has no attribute 'workspace'

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.