Coverage for src/qdrant_loader/connectors/attachments.py: 0%

1from __future__ import annotations 

2 

3from typing import Any 

4 

5from qdrant_loader.core.attachment_downloader import AttachmentMetadata 

6 

7 

8def attachment_metadata_from_dict( 

9 data: dict[str, Any], parent_document_id: str 

10) -> AttachmentMetadata: 

11 """Create AttachmentMetadata from a simple dict structure with input validation. 

12 

13 Required keys: id, download_url. 

14 Optional keys: filename, size, mime_type, created_at, updated_at, author. 

15 Missing optional keys default to sensible values. 

16 """ 

17 # Validate required fields: id and download_url 

18 raw_id = data.get("id", "") 

19 attachment_id = str(raw_id).strip() 

20 if not attachment_id: 

21 raise ValueError("Attachment 'id' is required and cannot be empty.") 

22 

23 raw_url = data.get("download_url", "") 

24 download_url = str(raw_url).strip() 

25 if not download_url: 

26 raise ValueError("Attachment 'download_url' is required and cannot be empty.") 

27 

28 # Optional fields with safe conversions/defaults 

29 # Ensure None becomes default and whitespace is trimmed 

30 filename = (data.get("filename") or "").strip() or "unknown" 

31 mime_type = (data.get("mime_type") or "").strip() or "application/octet-stream" 

32 

33 # Safely coerce size to int 

34 size_value = data.get("size", 0) 

35 try: 

36 size_int = int(size_value) if size_value not in (None, "") else 0 

37 except Exception: 

38 size_int = 0 

39 if size_int < 0: 

40 size_int = 0 

41 

42 return AttachmentMetadata( 

43 id=attachment_id, 

44 filename=filename, 

45 size=size_int, 

46 mime_type=mime_type, 

47 download_url=download_url, 

48 parent_document_id=parent_document_id, 

49 created_at=data.get("created_at"), 

50 updated_at=data.get("updated_at"), 

51 author=data.get("author"), 

52 )