dslighting 1.3.1__tar.gz → 1.3.6__tar.gz

{dslighting-1.3.1 → dslighting-1.3.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dslighting
-Version: 1.3.1
+Version: 1.3.6
 Summary: Simplified API for Data Science Agent Automation
 Author: DSLighting Team
 License: AGPL-3.0

{dslighting-1.3.1 → dslighting-1.3.6}/dsat/services/llm.py

@@ -43,7 +43,8 @@ def _load_custom_model_pricing():
                 config = yaml.safe_load(f)
                 return config.get('custom_model_pricing', {})
         else:
-            logger.warning(f"Config file not found at {config_yaml_path}")
+            # Changed to debug to avoid confusing warnings for pip-installed packages
+            logger.debug(f"Config file not found at {config_yaml_path} (this is expected for pip-installed packages)")
             return {}
     except Exception as e:
         logger.error(f"Failed to load cost configuration: {e}")

{dslighting-1.3.1 → dslighting-1.3.6}/dsat/services/llm_single.py

@@ -43,7 +43,8 @@ def _load_custom_model_pricing():
                 config = yaml.safe_load(f)
                 return config.get('custom_model_pricing', {})
         else:
-            logger.warning(f"Config file not found at {config_yaml_path}")
+            # Changed to debug to avoid confusing warnings for pip-installed packages
+            logger.debug(f"Config file not found at {config_yaml_path} (this is expected for pip-installed packages)")
             return {}
     except Exception as e:
         logger.error(f"Failed to load cost configuration: {e}")

{dslighting-1.3.1 → dslighting-1.3.6}/dsat/services/sandbox.py

@@ -245,17 +245,46 @@ class ProcessIsolatedNotebookExecutor:
 class SandboxService:
     """
     Provides unified access to isolated script and notebook code execution environments.
+
+    Args:
+        workspace: Workspace service for managing files
+        timeout: Default timeout for script execution (seconds)
+        auto_matplotlib: Automatically inject matplotlib backend (default: False).
+                        Set to True for Web UI environments that need visualization.
+                        Set to False for standalone package usage.
     """
-    def __init__(self, workspace: WorkspaceService, timeout: int = 600):
+    def __init__(
+        self,
+        workspace: WorkspaceService,
+        timeout: int = 600,
+        auto_matplotlib: bool = False
+    ):
         self.workspace = workspace
         self.timeout = timeout
+        self.auto_matplotlib = auto_matplotlib  # Store matplotlib injection preference
         self.execution_history: List[Dict[str, Any]] = []
 
     def run_script(self, script_code: str, timeout: Optional[int] = None) -> ExecutionResult:
-        """Runs a Python script within the sandbox workspace."""
-        # Force non-interactive backend for matplotlib to prevent blocking plt.show()
-        fixed_code = "import matplotlib\nmatplotlib.use('Agg')\n" + script_code
-        
+        """
+        Runs a Python script within the sandbox workspace.
+
+        Args:
+            script_code: Python code to execute
+            timeout: Optional timeout override (uses self.timeout if None)
+
+        Returns:
+            ExecutionResult with stdout, stderr, success status, etc.
+        """
+        # Optionally inject matplotlib non-interactive backend
+        # This is only done if auto_matplotlib=True (used by Web UI for visualization)
+        if self.auto_matplotlib:
+            # Force non-interactive backend for matplotlib to prevent blocking plt.show()
+            fixed_code = "import matplotlib\nmatplotlib.use('Agg')\n" + script_code
+            logger.debug("Auto-injected matplotlib non-interactive backend")
+        else:
+            # Use code as-is without modification (default for DSLighting package)
+            fixed_code = script_code
+
         script_name = f"_sandbox_script_{uuid.uuid4().hex}.py"
         script_path = self.workspace.run_dir / script_name
         execution_id = uuid.uuid4().hex

{dslighting-1.3.1 → dslighting-1.3.6}/dslighting/__init__.py

@@ -26,7 +26,7 @@ Advanced Usage:
 For more information, see: https://github.com/usail-hkust/dslighting
 """
 
-__version__ = "1.3.1"
+__version__ = "1.3.6"
 __author__ = "DSLighting Team"
 
 # Core API classes
@@ -60,7 +60,7 @@ def load_data(source, **kwargs):
     return loader.load(source, **kwargs)
 
 
-def run_agent(data=None, task_id=None, data_dir=None, **kwargs):
+def run_agent(data=None, task_id=None, data_dir=None, keep_workspace=False, keep_workspace_on_failure=True, **kwargs):
     """
     Quick one-liner: load data and run with defaults.
 
@@ -70,6 +70,8 @@ def run_agent(data=None, task_id=None, data_dir=None, **kwargs):
         data: Optional data source (path, DataFrame, dict, etc.)
         task_id: Task/Competition identifier (e.g., "bike-sharing-demand")
         data_dir: Base data directory (default: "data/competitions")
+        keep_workspace: Keep workspace after completion (default: False)
+        keep_workspace_on_failure: Keep workspace on failure (default: True)
         **kwargs: Parameters passed to Agent.__init__ and Agent.run
 
     Returns:
@@ -90,7 +92,8 @@ def run_agent(data=None, task_id=None, data_dir=None, **kwargs):
         >>> result = dslighting.run_agent(
         ...     task_id="bike-sharing-demand",
         ...     workflow="autokaggle",
-        ...     model="gpt-4o"
+        ...     model="gpt-4o",
+        ...     keep_workspace=True  # Keep workspace for debugging
         ... )
     """
     # Extract run-specific parameters if present
@@ -112,6 +115,10 @@ def run_agent(data=None, task_id=None, data_dir=None, **kwargs):
     if data_dir is not None:
         run_kwargs['data_dir'] = data_dir
 
+    # Add workspace preservation parameters to agent
+    agent_params['keep_workspace'] = keep_workspace
+    agent_params['keep_workspace_on_failure'] = keep_workspace_on_failure
+
     # Create agent and run
     agent = Agent(**agent_params)
     return agent.run(data, **run_kwargs)

{dslighting-1.3.1 → dslighting-1.3.6}/dslighting/core/agent.py

@@ -108,6 +108,8 @@ class Agent:
         num_drafts: int = None,
         workspace_dir: str = None,
         run_name: str = None,
+        keep_workspace: bool = False,
+        keep_workspace_on_failure: bool = True,
         verbose: bool = True,
         **kwargs
     ):
@@ -129,6 +131,8 @@ class Agent:
             num_drafts: Number of drafts to generate
             workspace_dir: Custom workspace directory
             run_name: Name for this run
+            keep_workspace: Keep workspace after completion (default: False)
+            keep_workspace_on_failure: Keep workspace on failure (default: True)
             verbose: Enable verbose logging
             **kwargs: Additional parameters passed to DSATConfig
         """
@@ -148,6 +152,8 @@ class Agent:
             num_drafts=num_drafts,
             workspace_dir=workspace_dir,
             run_name=run_name,
+            keep_workspace=keep_workspace,
+            keep_workspace_on_failure=keep_workspace_on_failure,
             **kwargs
         )
 

{dslighting-1.3.1 → dslighting-1.3.6}/dslighting/core/config_builder.py

@@ -60,6 +60,8 @@ class ConfigBuilder:
         num_drafts: int = None,
         workspace_dir: str = None,
         run_name: str = None,
+        keep_workspace: bool = None,
+        keep_workspace_on_failure: bool = None,
         **kwargs
     ) -> DSATConfig:
         """
@@ -76,6 +78,8 @@ class ConfigBuilder:
             num_drafts: Number of drafts to generate
             workspace_dir: Workspace directory
             run_name: Name for this run
+            keep_workspace: Keep workspace after completion
+            keep_workspace_on_failure: Keep workspace on failure
             **kwargs: Additional parameters
 
         Returns:
@@ -100,6 +104,8 @@ class ConfigBuilder:
             num_drafts=num_drafts,
             workspace_dir=workspace_dir,
             run_name=run_name,
+            keep_workspace=keep_workspace,
+            keep_workspace_on_failure=keep_workspace_on_failure,
             **kwargs
         )
         config = self._deep_merge(config, user_config)
@@ -180,7 +186,22 @@ class ConfigBuilder:
             logger.warning("LLM_MODEL_CONFIGS must be a JSON object")
             return {}
 
-        return {k: v for k, v in parsed.items() if isinstance(k, str) and isinstance(v, dict)}
+        # Process each model config
+        result = {}
+        for k, v in parsed.items():
+            if not isinstance(k, str) or not isinstance(v, dict):
+                continue
+
+            # Handle api_key as list (take the first one)
+            if "api_key" in v and isinstance(v["api_key"], list):
+                if len(v["api_key"]) > 0:
+                    v = v.copy()  # Shallow copy to avoid mutating original
+                    v["api_key"] = v["api_key"][0]
+                    logger.debug(f"Model '{k}': using first API key from list of {len(v['api_key'])}")
+
+            result[k] = v
+
+        return result
 
     def _build_user_config(
         self,
@@ -194,6 +215,8 @@ class ConfigBuilder:
         num_drafts: int = None,
         workspace_dir: str = None,
         run_name: str = None,
+        keep_workspace: bool = None,
+        keep_workspace_on_failure: bool = None,
         **kwargs
     ) -> Dict[str, Any]:
         """Build user configuration from parameters."""
@@ -230,6 +253,12 @@ class ConfigBuilder:
         if workspace_dir is not None:
             config.setdefault("run", {}).setdefault("parameters", {})["workspace_dir"] = workspace_dir
 
+        if keep_workspace is not None:
+            config.setdefault("run", {})["keep_all_workspaces"] = keep_workspace
+
+        if keep_workspace_on_failure is not None:
+            config.setdefault("run", {})["keep_workspace_on_failure"] = keep_workspace_on_failure
+
         # Additional kwargs are added to run.parameters
         if kwargs:
             config.setdefault("run", {}).setdefault("parameters", {}).update(kwargs)

{dslighting-1.3.1 → dslighting-1.3.6}/dslighting/core/data_loader.py

@@ -297,13 +297,45 @@ class DataLoader:
                             pass
             else:
                 self.logger.warning(f"Path does not exist: {path}")
-                # Still use the path even if it doesn't exist (might be created later)
-                if path.is_dir() or not path.suffix:
+
+                # Try to find the data in common locations
+                competition_id = path.name
+
+                # Common search locations for data
+                search_locations = [
+                    # Current project: ./data/competitions/
+                    Path.cwd() / "data" / "competitions" / competition_id,
+                    # Parent dslighting: ../dslighting/data/competitions/
+                    Path.cwd().parent / "dslighting" / "data" / "competitions" / competition_id,
+                    # Parent data: ../data/competitions/
+                    Path.cwd().parent / "data" / "competitions" / competition_id,
+                    # From package location: ../../data/competitions/
+                    Path(__file__).parent.parent.parent / "data" / "competitions" / competition_id,
+                    # Absolute path fallback
+                    Path("/Users/liufan/Applications/Github/dslighting/data/competitions") / competition_id,
+                ]
+
+                for location in search_locations:
+                    self.logger.info(f"  Trying: {location}")
+                    if location.exists() and location.is_dir():
+                        data_dir = location
+                        self.logger.info(f"  ✓ Found data at: {data_dir}")
+                        break
+
+                if data_dir is None:
+                    # Last resort: use the original resolved path
+                    self.logger.warning(f"  Could not find data, using original path: {path}")
                     data_dir = path
-                    self.logger.info(f"Using non-existent path as data directory: {data_dir}")
-                else:
-                    data_dir = path.parent
-                    self.logger.info(f"Using parent of non-existent file: {data_dir}")
+
+            # Try to load description (if data_dir was found)
+            if data_dir and data_dir.exists():
+                desc_file = data_dir / "description.md"
+                if desc_file.exists():
+                    try:
+                        description = desc_file.read_text(encoding='utf-8')
+                        self.logger.info(f"Loaded description from {desc_file}")
+                    except Exception:
+                        pass
 
         # Create MLE-style detection
         from dslighting.utils.defaults import WORKFLOW_RECOMMENDATIONS

{dslighting-1.3.1 → dslighting-1.3.6}/dslighting.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dslighting
-Version: 1.3.1
+Version: 1.3.6
 Summary: Simplified API for Data Science Agent Automation
 Author: DSLighting Team
 License: AGPL-3.0

{dslighting-1.3.1 → dslighting-1.3.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "dslighting"
-version = "1.3.1"
+version = "1.3.6"
 description = "Simplified API for Data Science Agent Automation"
 readme = "README.md"
 requires-python = ">=3.10"