from __future__ import annotations

import re
from typing import Any

from ingestion.common.governance import validate_classifications
from ingestion.common.models import Table


def classify_table(table: Table, rules: dict[str, Any]) -> list[str]:
    """Classify a table from source-declared rules.

    ``rules`` is the source's classification config (a baseline list applied to every
    table, name/column regex patterns, and table-name rules). The engine carries no
    source-specific defaults — an empty config yields no classifications. This lives in
    ``common`` because every extract family classifies the same way; only the *rules*
    are source-specific (see the config-driven structure principle).

    Every emitted label must be in the fixed ADR-0011 governance set; an unknown label
    (typo or invented classification) is rejected here rather than silently flowing into
    governance metadata.
    """
    names = " ".join([table.name] + [column.name for column in table.columns])
    classes = set(rules.get("baseline", []))
    for rule in rules.get("name_rules", []):
        prefix = rule.get("name_prefix")
        if prefix and table.name.startswith(prefix):
            classes.add(rule["add"])
    for classification, pattern in rules.get("patterns", {}).items():
        if re.search(pattern, names, re.IGNORECASE):
            classes.add(classification)
    return validate_classifications(sorted(classes), context=f"table {table.name}")