Paperlib MCP

Instructions

Implementation Reference

• src/paperlib_mcp/tools/graph_canonicalize.py:329-368 (handler)

  The handler function for the 'lock_entity' tool. Updates the 'is_locked' field in the 'entities' table for the given entity_id.

  @mcp.tool()
  def lock_entity(entity_id: int, is_locked: bool = True) -> dict[str, Any]:
      """锁定或解锁实体
      
      锁定的实体不会被自动规范化合并。
      
      Args:
          entity_id: 实体 ID
          is_locked: 是否锁定，默认 True
          
      Returns:
          操作结果
      """
      try:
          with get_db() as conn:
              with conn.cursor() as cur:
                  cur.execute(
                      """
                      UPDATE entities SET is_locked = %s, updated_at = now()
                      WHERE entity_id = %s
                      RETURNING entity_id
                      """,
                      (is_locked, entity_id)
                  )
                  result = cur.fetchone()
                  
                  if not result:
                      return LockEntityOut(
                          ok=False,
                          error=MCPErrorModel(code="NOT_FOUND", message=f"Entity {entity_id} not found"),
                      ).model_dump()
          
          return LockEntityOut(ok=True).model_dump()
          
      except Exception as e:
          return LockEntityOut(
              ok=False,
              error=MCPErrorModel(code="DB_CONN_ERROR", message=str(e)),
          ).model_dump()

• src/paperlib_mcp/models_graph.py:234-244 (schema)

  Pydantic schemas defining input (LockEntityIn) and output (LockEntityOut) models for the lock_entity tool.

  class LockEntityIn(BaseModel):
      """lock_entity 输入"""
      entity_id: int
      is_locked: bool = True
  
  
  class LockEntityOut(BaseModel):
      """lock_entity 输出"""
      ok: bool
      error: Optional[MCPErrorModel] = None

• src/paperlib_mcp/server.py:41-41 (registration)

  Registration of the graph_canonicalize_tools module, which includes the lock_entity tool, on the main FastMCP instance.

  register_graph_canonicalize_tools(mcp)

• src/paperlib_mcp/tools/graph_canonicalize.py:329-368 (registration)

  The @mcp.tool() decorator registers the lock_entity function directly when the register_graph_canonicalize_tools function is called.

  @mcp.tool()
  def lock_entity(entity_id: int, is_locked: bool = True) -> dict[str, Any]:
      """锁定或解锁实体
      
      锁定的实体不会被自动规范化合并。
      
      Args:
          entity_id: 实体 ID
          is_locked: 是否锁定，默认 True
          
      Returns:
          操作结果
      """
      try:
          with get_db() as conn:
              with conn.cursor() as cur:
                  cur.execute(
                      """
                      UPDATE entities SET is_locked = %s, updated_at = now()
                      WHERE entity_id = %s
                      RETURNING entity_id
                      """,
                      (is_locked, entity_id)
                  )
                  result = cur.fetchone()
                  
                  if not result:
                      return LockEntityOut(
                          ok=False,
                          error=MCPErrorModel(code="NOT_FOUND", message=f"Entity {entity_id} not found"),
                      ).model_dump()
          
          return LockEntityOut(ok=True).model_dump()
          
      except Exception as e:
          return LockEntityOut(
              ok=False,
              error=MCPErrorModel(code="DB_CONN_ERROR", message=str(e)),
          ).model_dump()