Coverage for nexusLIMS/extractors/plugins/basic_metadata.py: 100%

1"""Basic metadata extractor plugin (fallback for unknown file types).""" 

2 

3import logging 

4from datetime import datetime as dt 

5from typing import Any, ClassVar 

6 

7from nexusLIMS.extractors.base import ExtractionContext 

8from nexusLIMS.instruments import get_instr_from_filepath 

9from nexusLIMS.utils.time import current_system_tz 

10 

11_logger = logging.getLogger(__name__) 

12 

13 

14class BasicFileInfoExtractor: 

15 """ 

16 Fallback extractor for files without a specific format handler. 

17 

18 This extractor provides basic metadata (creation time, file size, etc.) 

19 for files that don't have a specialized extractor. It has the lowest 

20 priority and will only be used if no other extractor supports the file. 

21 """ 

22 

23 name = "basic_file_info_extractor" 

24 priority = 0 # Lowest priority - only used as fallback 

25 supported_extensions: ClassVar = None # Wildcard extractor - supports all files 

26 

27 def supports(self, context: ExtractionContext) -> bool: # noqa: ARG002 

28 """ 

29 Check if this extractor supports the given file. 

30 

31 This extractor always returns True since it's the fallback for all files. 

32 

33 Parameters 

34 ---------- 

35 context 

36 The extraction context containing file information 

37 

38 Returns 

39 ------- 

40 bool 

41 Always True (this is the fallback extractor) 

42 """ 

43 return True 

44 

45 def extract(self, context: ExtractionContext) -> list[dict[str, Any]]: 

46 """ 

47 Extract basic metadata from any file. 

48 

49 Provides minimal metadata such as modification time and instrument ID 

50 for files that don't have a specialized extractor. 

51 

52 Parameters 

53 ---------- 

54 context 

55 The extraction context containing file information 

56 

57 Returns 

58 ------- 

59 list[dict] 

60 List containing a single metadata dict with 'nx_meta' key 

61 """ 

62 _logger.debug( 

63 "Extracting basic metadata from file (no specialized extractor): %s", 

64 context.file_path, 

65 ) 

66 

67 mdict = {"nx_meta": {}} 

68 mdict["nx_meta"]["DatasetType"] = "Unknown" 

69 mdict["nx_meta"]["Data Type"] = "Unknown" 

70 

71 # get the modification time (as ISO format): 

72 mtime = context.file_path.stat().st_mtime 

73 # Use instrument timezone if available, otherwise fall back to system timezone 

74 tz = context.instrument.timezone if context.instrument else current_system_tz() 

75 mtime_iso = dt.fromtimestamp(mtime, tz=tz).isoformat() 

76 mdict["nx_meta"]["Creation Time"] = mtime_iso 

77 

78 return [mdict] 

79 

80 

81# Backward compatibility function for tests 

82def get_basic_metadata(filename): 

83 """ 

84 Get basic metadata from a file. 

85 

86 Returns basic metadata from a file that's not currently interpretable by NexusLIMS. 

87 

88 .. deprecated:: 

89 This function is deprecated. Use BasicFileInfoExtractor class instead. 

90 

91 Parameters 

92 ---------- 

93 filename : pathlib.Path 

94 path to a file saved in the harvested directory of the instrument 

95 

96 Returns 

97 ------- 

98 mdict : dict 

99 A description of the file in lieu of any metadata extracted from it. 

100 """ 

101 context = ExtractionContext( 

102 file_path=filename, instrument=get_instr_from_filepath(filename) 

103 ) 

104 extractor = BasicFileInfoExtractor() 

105 return extractor.extract(context)