fix(test): Prevent creation of test_data directory during unit tests

examples/config/config_example.py

@@ -23,7 +23,7 @@ def main():
     default_setting = create_default_config()
     print(f"   Source type: {type(default_setting.source).__name__}")
     print(f"   Parser type: {type(default_setting.parser).__name__}")
-    print(f"   Data directory: {default_setting.data_dir}")
+    print(f"   Storage directory: {default_setting.storage.storage_dir}")
 
     # Example 2: Save default configuration to YAML
     print("\n2. Saving default configuration to YAML:")
@@ -64,17 +64,16 @@ def main():
     print("   - ARXIV_MAX_RESULTS=50")
     print("   - OPENAI_MODEL=gpt-3.5-turbo")
     print("   - LOG_LEVEL=DEBUG")
-    print("   - DATA_DIR=./custom_data")
 
-    env_vars = ["ARXIV_MAX_RESULTS", "OPENAI_MODEL", "LOG_LEVEL", "DATA_DIR"]
+    env_vars = ["ARXIV_MAX_RESULTS", "OPENAI_MODEL", "LOG_LEVEL"]
     for var in env_vars:
         value = os.getenv(var)
         status = "✅ Set" if value else "❌ Not set"
         print(f"   {var}: {status}" + (f" = {value}" if value else ""))
 
     # Example 5: Direct configuration creation
     print("\n5. Creating configuration programmatically:")
-    from quantmind.config import ArxivSourceConfig, PDFParserConfig, LLMConfig
+    from quantmind.config import ArxivSourceConfig, LLMConfig, PDFParserConfig
 
     custom_setting = Setting(
         source=ArxivSourceConfig(
@@ -83,7 +82,6 @@ def main():
         parser=PDFParserConfig(method="pymupdf", extract_tables=True),
         llm=LLMConfig(model="gpt-4o", temperature=0.3),
         log_level="DEBUG",
-        data_dir="./custom_data",
     )
 
     print(f"   ✅ Created custom configuration")

examples/config/sample_config.yaml

@@ -59,5 +59,3 @@ llm:
 
 # Global settings
 log_level: ${LOG_LEVEL:INFO}
-data_dir: ${DATA_DIR:./data}
-temp_dir: ${TEMP_DIR:/tmp}

examples/config_example.py

@@ -146,8 +146,6 @@ def create_sample_config():
 
     # Set global settings
     settings.log_level = "INFO"
-    settings.data_dir = "./data"
-    settings.temp_dir = "./tmp"
     settings.arxiv_max_results = 500
 
     return settings
@@ -237,8 +235,7 @@ def show_configuration_details(settings):
 
     print(f"\nGlobal Settings:")
     print(f"  Log level: {settings.log_level}")
-    print(f"  Data directory: {settings.data_dir}")
-    print(f"  Temp directory: {settings.temp_dir}")
+    print(f"  Storage directory: {settings.storage.storage_dir}")
     print(f"  ArXiv max results: {settings.arxiv_max_results}")
 
     print(f"\nSources ({len(settings.sources)}):")

quantmind/config/settings.py

@@ -9,7 +9,7 @@
 from typing import Any, Dict, Optional, Union
 
 import yaml
-from pydantic import BaseModel, Field, field_validator
+from pydantic import BaseModel, Field
 
 from quantmind.config.flows import (
     AnalyzerFlowConfig,
@@ -56,30 +56,13 @@ class Setting(BaseModel):
     log_level: str = Field(
         default="INFO", pattern=r"^(DEBUG|INFO|WARNING|ERROR|CRITICAL)$"
     )
-    data_dir: str = "./data"
-    temp_dir: str = "/tmp"
 
     class Config:
         """Pydantic model configuration."""
 
         validate_assignment = True
         extra = "forbid"
 
-    @field_validator("data_dir", "temp_dir")
-    @classmethod
-    def validate_directories(cls, v: str) -> str:
-        """Validate and create directories if they don't exist."""
-        path = Path(v).expanduser()
-        # Keep the original style for relative paths.
-        if not path.is_absolute():
-            resolved_path = path.resolve()
-            resolved_path.mkdir(parents=True, exist_ok=True)
-            return v
-        else:
-            path = path.resolve()
-            path.mkdir(parents=True, exist_ok=True)
-            return str(path)
-
     @classmethod
     def load_dotenv(cls, dotenv_path: Optional[str] = None) -> bool:
         """Load environment variables from .env file.
@@ -246,9 +229,8 @@ def _parse_config(cls, config_dict: Dict[str, Any]) -> "Setting":
             parsed["llm"] = LLMConfig(**config_dict["llm"])
 
         # Copy simple fields
-        for key in ["log_level", "data_dir", "temp_dir"]:
-            if key in config_dict:
-                parsed[key] = config_dict[key]
+        if "log_level" in config_dict:
+            parsed["log_level"] = config_dict["log_level"]
 
         return cls(**parsed)
 
@@ -266,7 +248,7 @@ def create_default(cls) -> "Setting":
                 download_pdfs=True,
                 extract_tables=True,
             ),
-            storage=LocalStorageConfig(base_dir=Path("./data")),
+            storage=LocalStorageConfig(),
         )
 
     def save_to_yaml(self, config_path: Union[str, Path]) -> None:
@@ -357,13 +339,7 @@ def serialize_component(component, component_type_map):
         config_dict["llm"] = self.llm.model_dump(exclude={"api_key"})
 
         # Export simple fields
-        config_dict.update(
-            {
-                "log_level": self.log_level,
-                "data_dir": self.data_dir,
-                "temp_dir": self.temp_dir,
-            }
-        )
+        config_dict["log_level"] = self.log_level
 
         return config_dict
 

tests/config/test_settings.py

@@ -1,6 +1,7 @@
 """Unit tests for settings configuration system."""
 
 import os
+import shutil
 import tempfile
 import unittest
 from unittest.mock import patch
@@ -25,8 +26,6 @@ def test_default_setting(self):
         setting = Setting()
 
         self.assertEqual(setting.log_level, "INFO")
-        self.assertEqual(setting.data_dir, "./data")
-        self.assertEqual(setting.temp_dir, "/tmp")
         self.assertIsNone(setting.source)
         self.assertIsNone(setting.parser)
         self.assertIsNone(setting.tagger)
@@ -73,7 +72,6 @@ def test_parse_config_with_components(self):
                 "config": {"max_tags": 8, "model": "gpt-4o"},
             },
             "log_level": "DEBUG",
-            "data_dir": "./test_data",
         }
 
         setting = Setting._parse_config(config_dict)
@@ -96,7 +94,9 @@ def test_parse_config_with_components(self):
 
         # Test simple fields
         self.assertEqual(setting.log_level, "DEBUG")
-        self.assertEqual(setting.data_dir, "./test_data")
+
+        if setting.storage.storage_dir.exists():
+            shutil.rmtree(setting.storage.storage_dir)
 
     def test_parse_config_unknown_types(self):
         """Test parsing configuration with unknown component types."""
@@ -200,8 +200,6 @@ def test_substitute_env_vars(self):
                     "max_results": "${MAX_RESULTS:50}",  # with default
                 },
             },
-            "data_dir": "${DATA_DIR:./default_data}",
-            "temp_dir": "${TEST_VAR}/temp",
         }
 
         result = Setting.substitute_env_vars(config_dict)
@@ -211,8 +209,6 @@ def test_substitute_env_vars(self):
         self.assertEqual(
             result["source"]["config"]["max_results"], "50"
         )  # default used
-        self.assertEqual(result["data_dir"], "./default_data")  # default used
-        self.assertEqual(result["temp_dir"], "test_value/temp")
 
         # Clean up
         del os.environ["TEST_VAR"]
@@ -271,7 +267,6 @@ def test_export_config(self):
             parser=PDFParserConfig(method="pdfplumber", download_pdfs=True),
             tagger=LLMTaggerConfig(max_tags=5),
             log_level="DEBUG",
-            data_dir="./test_data",
         )
 
         config_dict = setting._export_config()
@@ -291,11 +286,15 @@ def test_export_config(self):
 
         # Test simple fields
         self.assertEqual(config_dict["log_level"], "DEBUG")
-        self.assertEqual(config_dict["data_dir"], "./test_data")
 
         # Test sensitive data exclusion
         self.assertNotIn("api_key", config_dict["llm"])
 
+        assert setting.storage.storage_dir.exists()
+        if setting.storage.storage_dir.exists():
+            shutil.rmtree(setting.storage.storage_dir)
+        assert not setting.storage.storage_dir.exists()
+
     def test_save_to_yaml(self):
         """Test saving configuration to YAML file."""
         setting = Setting(