extraction/label_mapper.py

from __future__ import annotations

from typing import Any, Dict, List, Optional, Tuple

from .spacy_extractor import VALID_SPACY_LABELS


def validate_mappings(fields: Dict[str, Dict[str, Any]]) -> Tuple[bool, Optional[Dict]]:
    invalid: List[str] = []
    for field_name, rule in fields.items():
        if rule.get("source_type") == "entity":
            label = rule.get("label", "")
            if label not in VALID_SPACY_LABELS:
                invalid.append(label)

    if not invalid:
        return True, None

    return False, {
        "invalid_labels": invalid,
        "valid_labels": VALID_SPACY_LABELS,
        "suggestion": (
            "Use source_type 'regex' for custom patterns not covered by spaCy NER labels."
        ),
    }