realtimex-deeptutor 0.5.0.post1__py3-none-any.whl

src/services/setup/init.py

@@ -0,0 +1,285 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+System Setup and Initialization
+Combines user directory initialization and port configuration management.
+"""
+
+import json
+import os
+from pathlib import Path
+
+from src.logging import get_logger
+from src.services.config import load_config_with_main
+
+# Initialize logger for setup operations
+_setup_logger = None
+
+
+def _get_setup_logger():
+    """Get logger for setup operations"""
+    global _setup_logger
+    if _setup_logger is None:
+        _setup_logger = get_logger("Setup")
+    return _setup_logger
+
+
+# ============================================================================
+# User Directory Initialization (from user_dir_init.py)
+# ============================================================================
+
+
+def init_user_directories(project_root: Path | None = None) -> None:
+    """
+    Initialize user data directories if they don't exist.
+
+    Creates the following directory structure:
+    data/user/
+    ├── solve/              # Problem solving outputs
+    ├── question/           # Question generation outputs
+    ├── research/           # Research outputs
+    │   ├── cache/          # Research cache
+    │   └── reports/        # Research reports
+    ├── guide/              # Guided learning outputs
+    ├── notebook/           # Notebook data
+    ├── co-writer/          # Co-writer outputs
+    │   ├── audio/          # TTS audio files
+    │   └── tool_calls/     # Tool call history
+    ├── logs/               # User logs
+    ├── run_code_workspace/ # Code execution workspace
+    └── user_history.json   # User history file
+
+    Args:
+        project_root: Project root directory (if None, will try to detect)
+    """
+    if project_root is None:
+        # Path(__file__) = src/services/setup/init.py
+        # .parent = src/services/setup/
+        # .parent.parent = src/services/
+        # .parent.parent.parent = src/
+        # .parent.parent.parent.parent = DeepTutor/ (project root)
+        project_root = Path(__file__).parent.parent.parent.parent
+
+    # Get user data directory from config
+    try:
+        config = load_config_with_main("solve_config.yaml", project_root)
+        user_data_dir = config.get("paths", {}).get("user_data_dir", "./data/user")
+
+        # Convert relative path to absolute
+        if not Path(user_data_dir).is_absolute():
+            user_data_dir = project_root / user_data_dir
+        else:
+            user_data_dir = Path(user_data_dir)
+    except Exception:
+        # Fallback to default
+        user_data_dir = project_root / "data" / "user"
+
+    # Required subdirectories (based on actual usage in the codebase)
+    required_dirs = [
+        "solve",  # Problem solving outputs
+        "question",  # Question generation outputs
+        "research",  # Research outputs (will have cache/ and reports/ subdirs)
+        "guide",  # Guided learning outputs
+        "notebook",  # Notebook data
+        "co-writer",  # Co-writer outputs
+        "logs",  # User logs
+        "run_code_workspace",  # Code execution workspace
+    ]
+
+    # Additional subdirectories for specific modules
+    co_writer_subdirs = [
+        "audio",  # TTS audio files
+        "tool_calls",  # Tool call history
+    ]
+
+    research_subdirs = [
+        "cache",  # Research cache
+        "reports",  # Research reports
+    ]
+
+    # Check if user directory exists and is empty
+    user_dir_exists = user_data_dir.exists()
+    user_dir_empty = False
+    if user_dir_exists:
+        try:
+            # Check if directory is empty (no files or subdirectories)
+            user_dir_empty = not any(user_data_dir.iterdir())
+        except (OSError, PermissionError) as e:
+            # If we can't check directory contents, assume it's not empty
+            logger = _get_setup_logger()
+            logger.warning(f"Cannot check if user directory is empty: {e}")
+            user_dir_empty = False
+
+    if not user_dir_exists or user_dir_empty:
+        logger = _get_setup_logger()
+        logger.info("\n" + "=" * 80)
+        logger.info("INITIALIZING USER DATA DIRECTORY")
+        logger.info("=" * 80)
+
+        if not user_dir_exists:
+            logger.info(f"Creating user data directory: {user_data_dir}")
+        else:
+            logger.info(f"User data directory is empty, initializing: {user_data_dir}")
+
+        # Create main user directory
+        user_data_dir.mkdir(parents=True, exist_ok=True)
+
+        # Create all required subdirectories
+        for dir_name in required_dirs:
+            dir_path = user_data_dir / dir_name
+            dir_path.mkdir(parents=True, exist_ok=True)
+            logger.success(f"Created: {dir_name}/")
+
+        # Create co-writer subdirectories
+        co_writer_dir = user_data_dir / "co-writer"
+        for subdir_name in co_writer_subdirs:
+            subdir_path = co_writer_dir / subdir_name
+            subdir_path.mkdir(parents=True, exist_ok=True)
+            logger.success(f"Created: co-writer/{subdir_name}/")
+
+        # Create research subdirectories
+        research_dir = user_data_dir / "research"
+        for subdir_name in research_subdirs:
+            subdir_path = research_dir / subdir_name
+            subdir_path.mkdir(parents=True, exist_ok=True)
+            logger.success(f"Created: research/{subdir_name}/")
+
+        # Create user_history.json if it doesn't exist
+        user_history_file = user_data_dir / "user_history.json"
+        if not user_history_file.exists():
+            initial_history = {"version": "1.0", "created_at": None, "sessions": []}
+            try:
+                with open(user_history_file, "w", encoding="utf-8") as f:
+                    json.dump(initial_history, f, indent=2, ensure_ascii=False)
+                logger.success("Created: user_history.json")
+            except Exception as e:
+                logger.warning(f"Failed to create user_history.json: {e}")
+
+        # Create interface.json in settings folder if it doesn't exist
+        settings_dir = user_data_dir / "settings"
+        settings_dir.mkdir(parents=True, exist_ok=True)
+        interface_file = settings_dir / "interface.json"
+        if not interface_file.exists():
+            initial_settings = {"theme": "light", "language": "en", "output_language": "en"}
+            try:
+                with open(interface_file, "w", encoding="utf-8") as f:
+                    json.dump(initial_settings, f, indent=2, ensure_ascii=False)
+                logger.success("Created: settings/interface.json")
+            except Exception as e:
+                logger.warning(f"Failed to create settings/interface.json: {e}")
+
+        logger.info("=" * 80)
+        logger.success("User data directory initialization complete!")
+        logger.info("=" * 80 + "\n")
+    else:
+        # Directory exists and is not empty, just ensure all subdirectories exist
+        for dir_name in required_dirs:
+            dir_path = user_data_dir / dir_name
+            dir_path.mkdir(parents=True, exist_ok=True)
+
+        # Ensure co-writer subdirectories exist
+        co_writer_dir = user_data_dir / "co-writer"
+        for subdir_name in co_writer_subdirs:
+            subdir_path = co_writer_dir / subdir_name
+            subdir_path.mkdir(parents=True, exist_ok=True)
+
+        # Ensure research subdirectories exist
+        research_dir = user_data_dir / "research"
+        for subdir_name in research_subdirs:
+            subdir_path = research_dir / subdir_name
+            subdir_path.mkdir(parents=True, exist_ok=True)
+
+        # Ensure user_history.json exists
+        user_history_file = user_data_dir / "user_history.json"
+        if not user_history_file.exists():
+            initial_history = {"version": "1.0", "created_at": None, "sessions": []}
+            try:
+                with open(user_history_file, "w", encoding="utf-8") as f:
+                    json.dump(initial_history, f, indent=2, ensure_ascii=False)
+            except Exception:
+                pass  # Silent fail if file creation fails but directory exists
+
+        # Ensure interface.json exists in settings folder
+        settings_dir = user_data_dir / "settings"
+        settings_dir.mkdir(parents=True, exist_ok=True)
+        interface_file = settings_dir / "interface.json"
+        if not interface_file.exists():
+            initial_settings = {"theme": "light", "language": "en", "output_language": "en"}
+            try:
+                with open(interface_file, "w", encoding="utf-8") as f:
+                    json.dump(initial_settings, f, indent=2, ensure_ascii=False)
+            except Exception:
+                pass  # Silent fail if file creation fails but directory exists
+
+
+# ============================================================================
+# Port Configuration Management
+# ============================================================================
+# Ports are configured via environment variables in .env file:
+#   BACKEND_PORT=8001   (default: 8001)
+#   FRONTEND_PORT=3782  (default: 3782)
+# ============================================================================
+
+
+def get_backend_port(project_root: Path | None = None) -> int:
+    """
+    Get backend port from environment variable.
+
+    Configure in .env file: BACKEND_PORT=8001
+
+    Returns:
+        Backend port number (default: 8001)
+    """
+    env_port = os.environ.get("BACKEND_PORT", "8001")
+    try:
+        return int(env_port)
+    except ValueError:
+        logger = _get_setup_logger()
+        logger.warning(f"Invalid BACKEND_PORT: {env_port}, using default 8001")
+        return 8001
+
+
+def get_frontend_port(project_root: Path | None = None) -> int:
+    """
+    Get frontend port from environment variable.
+
+    Configure in .env file: FRONTEND_PORT=3782
+
+    Returns:
+        Frontend port number (default: 3782)
+    """
+    env_port = os.environ.get("FRONTEND_PORT", "3782")
+    try:
+        return int(env_port)
+    except ValueError:
+        logger = _get_setup_logger()
+        logger.warning(f"Invalid FRONTEND_PORT: {env_port}, using default 3782")
+        return 3782
+
+
+def get_ports(project_root: Path | None = None) -> tuple[int, int]:
+    """
+    Get both backend and frontend ports from configuration.
+
+    Args:
+        project_root: Project root directory (if None, will try to detect)
+
+    Returns:
+        Tuple of (backend_port, frontend_port)
+
+    Raises:
+        SystemExit: If ports are not configured
+    """
+    backend_port = get_backend_port(project_root)
+    frontend_port = get_frontend_port(project_root)
+    return (backend_port, frontend_port)
+
+
+__all__ = [
+    # User directory initialization
+    "init_user_directories",
+    # Port configuration (from .env)
+    "get_backend_port",
+    "get_frontend_port",
+    "get_ports",
+]

src/services/tts/__init__.py

@@ -0,0 +1,16 @@
+"""
+TTS Service
+===========
+
+Text-to-Speech configuration for DeepTutor.
+
+Usage:
+    from src.services.tts import get_tts_config
+
+    config = get_tts_config()
+    # config = {"model": "tts-1", "api_key": "...", "base_url": "...", "voice": "alloy"}
+"""
+
+from .config import get_tts_config
+
+__all__ = ["get_tts_config"]

src/services/tts/config.py

@@ -0,0 +1,99 @@
+"""
+TTS Configuration
+=================
+
+Configuration management for Text-to-Speech services.
+Simplified version - loads from unified config service or falls back to .env.
+"""
+
+import logging
+import os
+from pathlib import Path
+from typing import Optional
+
+from dotenv import load_dotenv
+
+logger = logging.getLogger(__name__)
+
+# Load environment variables
+PROJECT_ROOT = Path(__file__).resolve().parent.parent.parent.parent
+load_dotenv(PROJECT_ROOT / "DeepTutor.env", override=False)
+load_dotenv(PROJECT_ROOT / ".env", override=False)
+
+
+def _strip_value(value: Optional[str]) -> Optional[str]:
+    """Remove leading/trailing whitespace and quotes from string."""
+    if value is None:
+        return None
+    return value.strip().strip("\"'")
+
+
+def get_tts_config() -> dict:
+    """
+    Return complete configuration for TTS (Text-to-Speech).
+
+    Priority:
+    1. Active configuration from unified config service
+    2. Environment variables (.env)
+
+    Returns:
+        dict: Dictionary containing the following keys:
+            - model: TTS model name
+            - api_key: TTS API key
+            - base_url: TTS API endpoint URL
+            - api_version: TTS API version (for Azure OpenAI)
+            - voice: Default voice character
+
+    Raises:
+        ValueError: If required configuration is missing
+    """
+    # 1. Try to get active config from unified config service
+    try:
+        from src.services.config import get_active_tts_config
+
+        config = get_active_tts_config()
+        if config and config.get("model"):
+            return {
+                "model": config["model"],
+                "api_key": config.get("api_key", ""),
+                "base_url": config.get("base_url", ""),
+                "api_version": config.get("api_version"),
+                "voice": config.get("voice", "alloy"),
+            }
+    except ImportError:
+        # Unified config service not yet available, fall back to env
+        pass
+    except Exception as e:
+        logger.warning(f"Failed to load from unified config: {e}")
+
+    # 2. Fallback to environment variables
+    model = _strip_value(os.getenv("TTS_MODEL"))
+    api_key = _strip_value(os.getenv("TTS_API_KEY"))
+    base_url = _strip_value(os.getenv("TTS_URL"))
+    api_version = _strip_value(os.getenv("TTS_BINDING_API_VERSION"))
+    voice = _strip_value(os.getenv("TTS_VOICE", "alloy"))
+
+    # Validate required configuration
+    if not model:
+        raise ValueError(
+            "TTS_MODEL not set. Please configure it in .env file or add a configuration in Settings"
+        )
+    if not api_key:
+        raise ValueError(
+            "TTS_API_KEY not set. Please configure it in .env file or add a configuration in Settings"
+        )
+    if not base_url:
+        raise ValueError(
+            "TTS_URL not set. Please configure it in .env file or add a configuration in Settings"
+        )
+
+    return {
+        "model": model,
+        "api_key": api_key,
+        "base_url": base_url,
+        "api_version": api_version,
+        "voice": voice,
+    }
+
+
+__all__ = ["get_tts_config"]

src/tools/__init__.py

@@ -0,0 +1,91 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Tools Package - Unified tool collection
+
+Includes:
+- rag_tool: RAG retrieval tool
+- web_search: Web search tool
+- query_item_tool: Query item tool
+- paper_search_tool: Paper search tool
+- tex_downloader: LaTeX source download tool
+- tex_chunker: LaTeX text chunking tool
+- question: Question generation tools (pdf_parser, question_extractor, exam_mimic)
+"""
+
+# Patch lightrag.utils BEFORE any imports that use lightrag
+import importlib.util
+import sys
+
+try:
+    # Directly load lightrag.utils module without triggering lightrag/__init__.py
+    _spec = importlib.util.find_spec("lightrag.utils")
+    if _spec and _spec.origin:
+        _utils = importlib.util.module_from_spec(_spec)
+        sys.modules["lightrag.utils"] = _utils
+        _spec.loader.exec_module(_utils)
+
+        # Apply patches
+        for _k, _v in {
+            "verbose_debug": lambda *args, **kwargs: None,
+            "VERBOSE_DEBUG": False,
+            "get_env_value": lambda key, default=None: default,
+            "safe_unicode_decode": lambda t: (
+                t.decode("utf-8", errors="ignore") if isinstance(t, bytes) else t
+            ),
+        }.items():
+            if not hasattr(_utils, _k):
+                setattr(_utils, _k, _v)
+
+        if not hasattr(_utils, "wrap_embedding_func_with_attrs"):
+
+            def _wrap(**attrs):
+                def dec(f):
+                    for k, v in attrs.items():
+                        setattr(f, k, v)
+                    return f
+
+                return dec
+
+            _utils.wrap_embedding_func_with_attrs = _wrap
+except Exception as e:
+    import traceback
+
+    print(f"Warning: Failed to patch lightrag.utils: {e}")
+    traceback.print_exc()
+
+from .code_executor import run_code, run_code_sync
+from .query_item_tool import query_numbered_item
+from .rag_tool import rag_search
+from .web_search import web_search
+
+# Paper research related tools
+try:
+    from .paper_search_tool import PaperSearchTool
+    from .tex_chunker import TexChunker
+    from .tex_downloader import TexDownloader, read_tex_file
+
+    __all__ = [
+        "PaperSearchTool",
+        "TexChunker",
+        "TexDownloader",
+        "query_numbered_item",
+        "rag_search",
+        "read_tex_file",
+        "run_code",
+        "run_code_sync",
+        "web_search",
+    ]
+except ImportError as e:
+    # If import fails (e.g., missing tiktoken), only export basic tools
+    print(f"⚠️  Some paper tools import failed: {e}")
+    __all__ = [
+        "query_numbered_item",
+        "rag_search",
+        "run_code",
+        "run_code_sync",
+        "web_search",
+    ]
+
+# Question generation tools (lazy import to avoid circular dependencies)
+# Access via: from src.tools.question import parse_pdf_with_mineru, etc.