# Copyright (c) 2025 Patrick Motsch
# All rights reserved.
from typing import Any, Dict, List

from modules.datamodels.datamodelExtraction import ContentPart
from ..subUtils import makeId
from ..subRegistry import Extractor


class CsvExtractor(Extractor):
    """
    Extractor for CSV files.
    
    Supported formats:
    - MIME types: text/csv
    - File extensions: .csv
    - Special handling: Treats as table data
    """
    
    def detect(self, fileName: str, mimeType: str, headBytes: bytes) -> bool:
        return mimeType == "text/csv" or (fileName or "").lower().endswith(".csv")
    
    def getSupportedExtensions(self) -> list[str]:
        """Return list of supported file extensions."""
        return [".csv"]
    
    def getSupportedMimeTypes(self) -> list[str]:
        """Return list of supported MIME types."""
        return ["text/csv"]

    def extract(self, fileBytes: bytes, context: Dict[str, Any]) -> List[ContentPart]:
        fileName = context.get("fileName")
        mimeType = context.get("mimeType") or "text/csv"
        data = fileBytes.decode("utf-8", errors="replace")
        return [ContentPart(
            id=makeId(),
            parentId=None,
            label="main",
            typeGroup="table",
            mimeType=mimeType,
            data=data,
            metadata={"size": len(fileBytes)}
        )]