"""Chunker — splits documents into fixed-size word chunks."""
from __future__ import annotations

from dataclasses import dataclass


@dataclass
class Chunk:
  chunk_id: str
  source: str
  text: str


def chunk_text(source: str, text: str, chunk_size: int = 18) -> list[Chunk]:
  """Split *text* into non-overlapping word chunks of *chunk_size* words.

  Args:
      source: Identifier for the originating document.
      text: Raw document text to split.
      chunk_size: Maximum words per chunk (default 18).

  Returns:
      Ordered list of :class:`Chunk` objects.
  """
  words = text.split()
  chunks: list[Chunk] = []
  for index in range(0, len(words), chunk_size):
    chunk_words = words[index: index + chunk_size]
    chunks.append(
        Chunk(
            chunk_id=f"{source}-{index // chunk_size}",
            source=source,
            text=" ".join(chunk_words),
        )
    )
  return chunks


def estimate_tokens(text: str) -> int:
  """Rough token estimate: word_count × 4/3."""
  return max(1, len(text.split()) * 4 // 3)