Coverage for src / quber / core / parsers / __init__.py: 100%

1""" 

2Document-level parsers. 

3 

4`Parser` is the engine-agnostic ABC for converting a source file into a 

5`DoclingDocument` — the canonical in-memory document model (see core/README.md). 

6`DoclingParser` is the abstract docling layer; the docling configuration is a 

7type, not a flag: `TunedFinancialParser` (canonical default) and 

8`LegacyDoclingParser` (pre-QUE-218). Use `parser_for_preset` to pick one. 

9 

10Table-only engines (Camelot+LLM) live behind a separate `TableExtractor` 

11Protocol in `core.extractors`. The two paths intentionally do not compose 

12at this layer; composition is a deferred design pass. 

13""" 

14 

15from typing import TYPE_CHECKING, Any 

16 

17from quber.core.parsers.base import Parser 

18from quber.core.parsers.mock import MockParser 

19from quber.core.parsers.result import ( 

20 PageParse, 

21 PageScore, 

22 ParsedCell, 

23 ParseResult, 

24 TableProvenance, 

25) 

26 

27if TYPE_CHECKING: 

28 from quber.core.parsers.docling_parser import ( 

29 PARSERS_BY_PRESET, 

30 DoclingParser, 

31 LegacyDoclingParser, 

32 TunedFinancialParser, 

33 parser_for_preset, 

34 ) 

35 

36# The docling engine classes need the full `docling` package (the GPU extra). 

37# The fuse cloud job imports this package only for `ParseResult` and must run 

38# on the CPU image, so the engine names resolve lazily on first access 

39# instead of at package import. 

40ENGINE_EXPORTS = ( 

41 "PARSERS_BY_PRESET", 

42 "DoclingParser", 

43 "LegacyDoclingParser", 

44 "TunedFinancialParser", 

45 "parser_for_preset", 

46) 

47 

48 

49def __getattr__(name: str) -> Any: 

50 if name in ENGINE_EXPORTS: 

51 from quber.core.parsers import docling_parser 

52 

53 return getattr(docling_parser, name) 

54 raise AttributeError(f"module {__name__!r} has no attribute {name!r}") 

55 

56 

57__all__ = [ 

58 "PARSERS_BY_PRESET", 

59 "DoclingParser", 

60 "LegacyDoclingParser", 

61 "MockParser", 

62 "PageParse", 

63 "PageScore", 

64 "ParseResult", 

65 "ParsedCell", 

66 "Parser", 

67 "TableProvenance", 

68 "TunedFinancialParser", 

69 "parser_for_preset", 

70]