tgit/app/config.py

"""Typed configuration from the environment / `.env` — one settings class per subsystem (ARCHITECTURE §2.11)."""

from pathlib import Path
from typing import Annotated, Literal

from pydantic import AliasChoices, Field, SecretStr, field_validator
from pydantic_settings import BaseSettings, NoDecode, SettingsConfigDict

_ENV = SettingsConfigDict(env_file=".env", env_file_encoding="utf-8", extra="ignore")


class TelegramSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="TELEGRAM_")

    bot_token: SecretStr
    # http://user:pass@host:port or socks5://… — the RF server has no route to api.telegram.org (same as meditron)
    proxy_url: SecretStr | None = None


class AccessSettings(BaseSettings):
    model_config = _ENV

    # Bootstrap admins (SPEC A-8). Comma-separated in the env: ADMIN_IDS=1,2
    admin_ids: Annotated[frozenset[int], NoDecode] = frozenset()
    invite_ttl_hours: float = Field(default=24.0, gt=0)
    # Demo QR for the audience at the talk (#55): many entrants, short life
    demo_invite_max_uses: int = Field(default=50, ge=2)
    demo_invite_ttl_hours: float = Field(default=3.0, gt=0)
    # More failed /start <token> per hour than this — silence (SPEC A-6)
    max_failed_attempts: int = Field(default=5, ge=1)

    @field_validator("admin_ids", mode="before")
    @classmethod
    def _split_ids(cls, value: object) -> object:
        if isinstance(value, str):
            return frozenset(int(part) for part in value.replace(" ", "").split(",") if part)
        return value


class LLMSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="LLM_", populate_by_name=True)

    # mock: deterministic offline answers (tests, demo without network); openai_compat: any OpenAI-style API (ADR-008),
    # e.g. OpenRouter: LLM_BASE_URL=https://openrouter.ai/api/v1
    provider: Literal["mock", "openai_compat"] = "mock"
    base_url: str = ""
    model: str = ""
    # OPENROUTER_API_KEY is accepted too — the name meditron and Vobot use for the same key
    api_key: SecretStr = Field(
        default=SecretStr(""), validation_alias=AliasChoices("LLM_API_KEY", "OPENROUTER_API_KEY")
    )
    # OpenRouter refuses RF IPs: the server goes through an HTTP/SOCKS proxy (as meditron does)
    proxy_url: SecretStr | None = None
    timeout: float = Field(default=30.0, gt=0)
    max_attempts: int = Field(default=3, ge=1)
    temperature: float = Field(default=0.0, ge=0, le=2)


class DatabaseSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="DATABASE_")

    # Compose sets a PostgreSQL URL; the default lets the bot run locally without Docker.
    url: str = "sqlite+aiosqlite:///./tgit.db"


class SourceCraftSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="SOURCECRAFT_")

    token: SecretStr = SecretStr("")
    api_url: str = "https://api.sourcecraft.tech"
    uploads_repo: str = "rogachevegorushka/tgit-uploads"
    # Web link to a CI run. The SPA answers 200 on any path, so the format is unverified — adjust if it 404s in UI.
    run_url_template: str = "https://sourcecraft.dev/{repo}/cicd/runs/{run}"
    timeout: float = Field(default=30.0, gt=0)
    max_attempts: int = Field(default=5, ge=1)


class CISettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="CI_")

    # CI of the uploads repo the bot waits for before deploying (#11). none — deploy right after push, no CI status.
    provider: Literal["none", "gitea"] = "none"
    api_url: str = "https://gitea.24studbank.ru"
    # Read access to the uploads repo's Actions is enough
    token: SecretStr = SecretStr("")
    repo: str = "HSE/tgit-uploads"
    workflow: str = "check.yml"
    poll_interval: float = Field(default=10.0, gt=0)
    max_poll_interval: float = Field(default=60.0, gt=0)  # backoff while the API is failing
    run_timeout_minutes: float = Field(default=30.0, gt=0)
    timeout: float = Field(default=30.0, gt=0)
    max_attempts: int = Field(default=5, ge=1)


class GitSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="GIT_")

    # The bot's own deploy key with write access to tgit-uploads only (#7)
    uploads_url: str = "git@ssh.sourcecraft.dev:rogachevegorushka/tgit-uploads.git"
    ssh_key_path: Path | None = None
    # Alternative for containers: the private key itself, base64 — a mounted host file is unreadable for the
    # container user (uid 10001 vs the owner's 0600). Written to a private temp file at startup.
    ssh_key_b64: SecretStr | None = None
    branch: str = "main"
    # Web link to a commit in the uploads repo, e.g. https://sourcecraft.dev/{repo}/commit/{sha}; empty — no link
    commit_url_template: str = ""
    clone_dir: Path = Path("data/uploads")
    # Browsing re-fetches at most this often; a push always fetches first (#9)
    fetch_interval: float = Field(default=10.0, ge=0)
    timeout: float = Field(default=60.0, gt=0)


class DockerSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="DOCKER_")

    # docker-socket-proxy (GET only, ADR-006); empty — the «Статус» section reports Docker as unavailable
    url: str = ""
    # Only containers with this label are shown: the host runs other projects too
    label: str = "tgit.managed=true"
    timeout: float = Field(default=10.0, gt=0)
    snapshot_interval: float = Field(default=15.0, gt=0)
    # One logs request reads at most this much (SPEC L-5: 5 MB); the newest lines win
    max_log_bytes: int = Field(default=5 * 1024 * 1024, gt=0)
    max_log_lines: int = Field(default=50_000, gt=0)


class RunnerSettings(BaseSettings):
    model_config = SettingsConfigDict(**_ENV, env_prefix="RUNNER_")

    # SSH to tgit-runner on the host (ADR-006); empty host — services are not deployed from the bot
    host: str = ""
    port: int = 22
    user: str = "tgit-deploy"
    # The bot's runner key, base64; authorized_keys pins it to `command="tgit-runner"` (infra/runner/)
    ssh_key_b64: SecretStr | None = None
    timeout: float = Field(default=300.0, gt=0)


class RuntimeSettings(BaseSettings):
    model_config = _ENV

    log_level: str = "INFO"
    # Non-secret configs (texts, thresholds, prompts). Relative to the working dir: /srv/tgit in Docker.
    config_dir: Path = Path("config")
    # The heartbeat file is touched while the event loop is alive; the Docker healthcheck reads its age.
    heartbeat_file: Path = Path("/tmp/tgit-heartbeat")
    heartbeat_interval: float = Field(default=10.0, gt=0)
    heartbeat_max_age: float = Field(default=30.0, gt=0)


class Settings(BaseSettings):
    model_config = _ENV

    telegram: TelegramSettings = Field(default_factory=TelegramSettings)
    access: AccessSettings = Field(default_factory=AccessSettings)
    database: DatabaseSettings = Field(default_factory=DatabaseSettings)
    llm: LLMSettings = Field(default_factory=LLMSettings)
    sourcecraft: SourceCraftSettings = Field(default_factory=SourceCraftSettings)
    git: GitSettings = Field(default_factory=GitSettings)
    # not `ci`: Settings has no env prefix, and CI runners export CI=true
    uploads_ci: CISettings = Field(default_factory=CISettings)
    docker: DockerSettings = Field(default_factory=DockerSettings)
    runner: RunnerSettings = Field(default_factory=RunnerSettings)
    runtime: RuntimeSettings = Field(default_factory=RuntimeSettings)