Coverage for src/blob_dict/dict/path.py: 0%

99 statements  

« prev     ^ index     » next       coverage.py v7.8.0, created at 2025-04-07 08:32 -0700

1import shutil 

2from abc import abstractmethod 

3from collections.abc import Iterator 

4from mimetypes import guess_type 

5from pathlib import Path 

6from typing import Any, Protocol, override 

7 

8from extratools_core.typing import PathLike 

9 

10from ..blob import BytesBlob, StrBlob 

11from ..blob.json import JsonDictBlob, YamlDictBlob 

12from . import BlobDictBase 

13 

14 

15class LocalPath(Path): 

16 def rmtree(self) -> None: 

17 shutil.rmtree(self) 

18 

19 

20class ExtraPathLike(PathLike, Protocol): 

21 @abstractmethod 

22 def rmtree(self) -> None: 

23 ... 

24 

25 

26class PathBlobDict(BlobDictBase): 

27 def __init__( 

28 self, 

29 path: ExtraPathLike, 

30 *, 

31 compression: bool = False, 

32 blob_class: type[BytesBlob] = BytesBlob, 

33 blob_class_args: dict[str, Any] | None = None, 

34 ) -> None: 

35 super().__init__() 

36 

37 self.__path: ExtraPathLike = path 

38 

39 self.__compression: bool = compression 

40 

41 self.__blob_class: type[BytesBlob] = blob_class 

42 self.__blob_class_args: dict[str, Any] = blob_class_args or {} 

43 

44 def create(self) -> None: 

45 self.__path.mkdir( 

46 parents=True, 

47 exist_ok=True, 

48 ) 

49 

50 def delete(self) -> None: 

51 self.__path.rmtree() 

52 

53 @override 

54 def __contains__(self, key: str) -> bool: 

55 return (self.__path / key).is_file() 

56 

57 def __get_blob_class(self, key: str) -> type[BytesBlob]: # noqa: PLR0911 

58 mime_type: str | None 

59 mime_type, _ = guess_type(self.__path / key) 

60 

61 match mime_type: 

62 case "application/json": 

63 return JsonDictBlob 

64 case "application/octet-stream": 

65 return BytesBlob 

66 case "application/yaml": 

67 return YamlDictBlob 

68 case "audo/mpeg": 

69 # Import here as it has optional dependency 

70 from ..blob.audio import AudioBlob # noqa: PLC0415 

71 

72 return AudioBlob 

73 case "image/png": 

74 # Import here as it has optional dependency 

75 from ..blob.image import ImageBlob # noqa: PLC0415 

76 

77 return ImageBlob 

78 case ( 

79 "text/css" 

80 | "text/csv" 

81 | "text/html" 

82 | "text/javascript" 

83 | "text/markdown" 

84 | "text/plain" 

85 | "text/xml" 

86 ): 

87 return StrBlob 

88 case "video/mp4": 

89 # Import here as it has optional dependency 

90 from ..blob.video import VideoBlob # noqa: PLC0415 

91 

92 return VideoBlob 

93 case _: 

94 return self.__blob_class 

95 

96 @override 

97 def get(self, key: str, default: BytesBlob | None = None) -> BytesBlob | None: 

98 if key not in self: 

99 return default 

100 

101 blob_bytes: bytes = (self.__path / key).read_bytes() 

102 

103 blob: BytesBlob = BytesBlob.from_bytes(blob_bytes, compression=self.__compression) 

104 return blob.as_blob( 

105 self.__get_blob_class(key), 

106 self.__blob_class_args, 

107 ) 

108 

109 @override 

110 def __iter__(self) -> Iterator[str]: 

111 # The concept of relative path does not exist for `CloudPath`, 

112 # and each walked path is always absolute for `CloudPath`. 

113 # Therefore, we extract each key by removing the path prefix. 

114 # In this way, the same logic works for both absolute and relative path. 

115 prefix_len: int = ( 

116 len(str(self.__path)) 

117 # Extra 1 is for separator `/` between prefix and filename 

118 + 1 

119 ) 

120 

121 for parent, _, files in self.__path.walk(top_down=False): 

122 for filename in files: 

123 yield str(parent / filename)[prefix_len:] 

124 

125 @override 

126 def clear(self) -> None: 

127 for parent, dirs, files in self.__path.walk(top_down=False): 

128 for filename in files: 

129 (parent / filename).unlink() 

130 for dirname in dirs: 

131 (parent / dirname).rmdir() 

132 

133 def __cleanup(self, key: str) -> None: 

134 (self.__path / key).unlink() 

135 

136 for parent in (self.__path / key).parents: 

137 if parent == self.__path: 

138 return 

139 

140 if parent.is_dir() and next(iter(parent.iterdir()), None) is None: 

141 parent.rmdir() 

142 

143 @override 

144 def pop(self, key: str, default: BytesBlob | None = None) -> BytesBlob | None: 

145 blob: BytesBlob | None = self.get(key) 

146 if blob: 

147 self.__cleanup(key) 

148 

149 return blob or default 

150 

151 @override 

152 def __delitem__(self, key: str) -> None: 

153 if key not in self: 

154 raise KeyError 

155 

156 self.__cleanup(key) 

157 

158 __BAD_BLOB_CLASS_ERROR_MESSAGE: str = "Must specify blob that is instance of {blob_class}" 

159 

160 @override 

161 def __setitem__(self, key: str, blob: BytesBlob) -> None: 

162 if not isinstance(blob, self.__blob_class): 

163 raise TypeError(PathBlobDict.__BAD_BLOB_CLASS_ERROR_MESSAGE.format( 

164 blob_class=self.__blob_class, 

165 )) 

166 

167 (self.__path / key).parent.mkdir( 

168 parents=True, 

169 exist_ok=True, 

170 ) 

171 

172 blob_bytes: bytes = blob.as_bytes(compression=self.__compression) 

173 (self.__path / key).write_bytes(blob_bytes)