feat: initial ResearchOwl

src/config.py

@@ -0,0 +1,49 @@
+from pydantic_settings import BaseSettings
+from pydantic import Field
+from typing import Optional
+
+
+class Settings(BaseSettings):
+    # Telegram
+    telegram_bot_token: str = Field(..., env="TELEGRAM_BOT_TOKEN")
+    telegram_allowed_users: str = Field("", env="TELEGRAM_ALLOWED_USERS")  # comma-separated user IDs
+
+    # Ollama
+    ollama_url: str = Field("http://ollama.chemavx.xyz", env="OLLAMA_URL")
+    ollama_model: str = Field("qwen2.5:3b", env="OLLAMA_MODEL")
+    ollama_embed_model: str = Field("qwen2.5:3b", env="OLLAMA_EMBED_MODEL")
+
+    # Claude fallback (optional)
+    anthropic_api_key: Optional[str] = Field(None, env="ANTHROPIC_API_KEY")
+    claude_model: str = Field("claude-haiku-4-5", env="CLAUDE_MODEL")
+
+    # Database
+    db_path: str = Field("/data/researchowl.db", env="DB_PATH")
+
+    # Scraping
+    max_depth: int = Field(3, env="MAX_DEPTH")                      # recursion depth
+    max_sources: int = Field(150, env="MAX_SOURCES")                # hard cap
+    max_pages_per_search: int = Field(5, env="MAX_PAGES_PER_SEARCH")
+    request_timeout: int = Field(30, env="REQUEST_TIMEOUT")
+    request_delay: float = Field(1.0, env="REQUEST_DELAY")          # seconds between requests
+    min_content_length: int = Field(200, env="MIN_CONTENT_LENGTH")  # chars
+
+    # Processing
+    chunk_size: int = Field(800, env="CHUNK_SIZE")                  # tokens per chunk
+    chunk_overlap: int = Field(100, env="CHUNK_OVERLAP")
+    quality_threshold: float = Field(0.5, env="QUALITY_THRESHOLD")  # 0-1, chunks below discarded
+
+    # App
+    log_level: str = Field("INFO", env="LOG_LEVEL")
+
+    @property
+    def allowed_user_ids(self) -> list[int]:
+        if not self.telegram_allowed_users:
+            return []
+        return [int(uid.strip()) for uid in self.telegram_allowed_users.split(",") if uid.strip()]
+
+    class Config:
+        env_file = ".env"
+
+
+settings = Settings()