app/models/metadata.py

"""Pydantic models for SAS file metadata."""

from pydantic import BaseModel


class ColumnMetadata(BaseModel):
    """Metadata for a single SAS dataset column. Never contains raw data values."""

    name: str
    label: str | None = None
    dtype: str  # pandas dtype string, e.g. "float64", "object"
    sas_format: str | None = None
    sas_informat: str | None = None
    length: int = 0
    position: int  # 0-based column ordinal


class DatasetMetadata(BaseModel):
    """Metadata for an entire SAS dataset. Row count is included but no row data."""

    filename: str
    dataset_label: str | None = None
    num_rows: int
    num_columns: int
    columns: list[ColumnMetadata]
    encoding: str | None = None
    created_at: str | None = None
    modified_at: str | None = None