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

from modules.datamodels.datamodelExtraction import ContentPart
from ..subRegistry import Chunker


class TableChunker(Chunker):
    def chunk(self, part: ContentPart, options: Dict[str, Any]) -> list[Dict[str, Any]]:
        maxBytes = int(options.get("tableChunkSize", 40000))
        chunks: List[Dict[str, Any]] = []
        current: List[str] = []
        size = 0
        for line in part.data.split('\n'):
            lineSize = len(line.encode('utf-8')) + 1
            if size + lineSize > maxBytes and current:
                data = '\n'.join(current)
                chunks.append({"data": data, "size": len(data.encode('utf-8')), "order": len(chunks)})
                current = [line]
                size = lineSize
            else:
                current.append(line)
                size += lineSize
        if current:
            data = '\n'.join(current)
            chunks.append({"data": data, "size": len(data.encode('utf-8')), "order": len(chunks)})
        return chunks