src/config.py - master - maruga/mlops

src/config.py
69 строк2 KB
s546371
modified: src/config.py
29 мар 2026, 09:34
29 мар 2026, 09:34a00128c
import os
from typing import ClassVar, Dict, List, Union


class DataMap:
    target: ClassVar[str] = "target"

    views: ClassVar[str] = "views"
    purchases: ClassVar[str] = "purchases"
    ctr: ClassVar[str] = "ctr"
    hour: ClassVar[str] = "hour"
    weekday: ClassVar[str] = "weekday"
    category_id: ClassVar[str] = "categoryid"
    available: ClassVar[str] = "available"
    features: ClassVar[List[str]] = [
        views,
        purchases,
        ctr,
        hour,
        weekday,
        category_id,
        available,
    ]


class TrainConfig:
    TEST_SIZE: float = 0.2
    RANDOM_STATE: int = 42
    MODEL_PATH = os.getenv("MODEL_PATH", "ranker.pkl")
    DATA_PATH = os.getenv("DATA_PATH", "data_for_training.parquet")


class CatBoostClsConfig:
    DEPTH = 10
    LEARNING_RATE = 0.06690992453172911
    ITERATIONS = 924
    L2_LEAF_REG = 0.0018902533338309604
    BORDER_COUNT = 175
    RANDOM_STRENGTH = 1.7074626001153735e-05


class S3Config:
    S3_ENDPOINT_URL: str = os.getenv(
        "S3_ENDPOINT_URL", "https://storage.yandexcloud.net"
    )
    S3_ACCESS_KEY: str = os.getenv("S3_ACCESS_KEY", "")
    S3_SECRET_KEY: str = os.getenv("S3_SECRET_KEY", "")
    S3_RAW_DATA_PATH = os.getenv("S3_RAW_DATA_PATH", "")
    S3_DATA_PATH = os.getenv("S3_DATA_PATH", "")
    S3_MODEL_PATH = os.getenv("S3_MODEL_PATH", "")
    BUCKET_NAME = os.getenv("S3_BUCKET", "")
    S3_COMMON_PATH = os.getenv("S3_COMMON_PATH", "")
    S3_MODEL_DIR = os.getenv("S3_MODEL_DIR", "")
    S3_DATA_DIR = os.getenv("S3_DATA_DIR", "")
    ITEM_FEATS_PATH = os.getenv("ITEM_FEATS_PATH", "")

    @property
    def storage_options(self) -> Dict[str, Union[str, Dict[str, str]]] | None:
        """Storage options для pandas"""
        if not self.S3_ACCESS_KEY and self.S3_SECRET_KEY:
            return {
                "key": self.S3_ACCESS_KEY,
                "secret": self.S3_SECRET_KEY,
                "client_kwargs": {"endpoint_url": self.S3_ENDPOINT_URL},
            }
        return None


config_s3 = S3Config()