from collections.abc import Awaitable, Callable

from .linkify import tokenize
from .own_links import parse_share_link

# Each resolved link costs one backend lookup, so a single document stays bounded.
MAX_LOOKUPS = 20

def _media_kind(content_type: str) -> str:
  value = (content_type or '').split(';')[0].strip().lower()
  for prefix in ('image', 'video', 'audio'):
    if value.startswith(prefix + '/'):
      return prefix
  if 'pdf' in value:
    return 'pdf'
  return 'other'

async def build_segments(
  text: str,
  host: str,
  root_path: str,
  lookup: Callable[[str], Awaitable[dict | None]] | None = None,
  max_lookups: int = MAX_LOOKUPS,
) -> list[dict]:
  """Turn raw text into template segments: plain text, links and resolved own files."""
  segments: list[dict] = []
  resolved: dict[str, dict | None] = {}

  for kind, value in tokenize(text):
    if kind == 'text':
      segments.append({'kind': 'text', 'text': value})
      continue

    file_id = parse_share_link(value, host=host, root_path=root_path) if lookup else None
    if file_id is None:
      segments.append({'kind': 'link', 'url': value})
      continue

    if file_id not in resolved:
      if len(resolved) >= max_lookups:
        segments.append({'kind': 'link', 'url': value})
        continue
      resolved[file_id] = await lookup(file_id)

    file_data = resolved[file_id]
    if not file_data:
      segments.append({'kind': 'link', 'url': value, 'state': 'missing'})
      continue
    if file_data.get('expired'):
      segments.append({
        'kind': 'link',
        'url': value,
        'state': 'expired',
        'label': file_data.get('file_name') or file_id,
      })
      continue

    segments.append({
      'kind': 'file',
      'url': value,
      'state': 'ok',
      'file_id': file_id,
      'label': file_data.get('file_name') or file_id,
      'media': _media_kind(file_data.get('content_type') or ''),
    })

  return segments