feat(backend): add PDF extraction with OCR fallback and cache

backend/app/config.py

@@ -0,0 +1,83 @@
+"""Configuração do backend via variáveis de ambiente."""
+
+from __future__ import annotations
+
+from functools import lru_cache
+from pathlib import Path
+
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    """Configuração lida do ambiente ou do arquivo .env."""
+
+    model_config = SettingsConfigDict(
+        env_file=(".env", "../.env"),
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    # Caminhos
+    pdf_dir: Path = Path("/data/pdfs")
+    vector_db_dir: Path = Path("/data/vector_db")
+
+    # Provedor de LLM: ollama | openai | groq | anthropic | stub
+    llm_provider: str = "ollama"
+    ollama_base_url: str = "http://ollama:11434"
+    ollama_model: str = "qwen2.5:1.5b"
+    openai_base_url: str = "https://api.openai.com/v1"
+    openai_model: str = "gpt-4o-mini"
+    openai_api_key: str = ""
+    groq_base_url: str = "https://api.groq.com/openai/v1"
+    groq_model: str = "llama-3.3-70b-versatile"
+    groq_api_key: str = ""
+    anthropic_base_url: str = "https://api.anthropic.com"
+    anthropic_model: str = "claude-sonnet-4-5"
+    anthropic_api_key: str = ""
+
+    # Embeddings
+    embedding_model: str = "paraphrase-multilingual-MiniLM-L12-v2"
+    embedding_batch_size: int = 32
+
+    # Busca
+    top_k: int = 6
+    min_similarity: float = 0.35
+    collection_name: str = "biblia"
+
+    # OCR
+    ocr_dpi: int = 200
+    ocr_language: str = "por+eng"
+    ocr_timeout: int = 120
+    ocr_workers: int = 4
+    min_legible_ratio: float = 0.35
+
+    # API
+    cors_origins: str = "http://localhost:8080,http://127.0.0.1:8080"
+    reindex_token: str = ""
+    max_message_length: int = 4000
+    max_history_turns: int = 10
+
+    # Limites de ingestão
+    max_pdf_bytes: int = 200 * 1024 * 1024
+
+    @property
+    def ocr_cache_dir(self) -> Path:
+        return self.vector_db_dir / "ocr_cache"
+
+    @property
+    def manifest_path(self) -> Path:
+        return self.vector_db_dir / "manifest.json"
+
+    @property
+    def book_index_path(self) -> Path:
+        return self.vector_db_dir / "book_index.json"
+
+    @property
+    def cors_origin_list(self) -> list[str]:
+        return [origin.strip() for origin in self.cors_origins.split(",") if origin.strip()]
+
+
+@lru_cache(maxsize=1)
+def get_settings() -> Settings:
+    """Devolve as settings singleton do processo."""
+    return Settings()