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

102 statements  

« prev     ^ index     » next       coverage.py v7.8.1, created at 2025-05-21 20:53 -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 if isinstance(path, Path): 

38 path = path.expanduser() 

39 

40 self.__path: ExtraPathLike = path 

41 

42 self.__compression: bool = compression 

43 

44 self.__blob_class: type[BytesBlob] = blob_class 

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

46 

47 def create(self) -> None: 

48 self.__path.mkdir( 

49 parents=True, 

50 exist_ok=True, 

51 ) 

52 

53 def delete(self) -> None: 

54 self.__path.rmtree() 

55 

56 @override 

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

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

59 

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

61 mime_type: str | None 

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

63 

64 match mime_type: 

65 case "application/json": 

66 return JsonDictBlob 

67 case "application/octet-stream": 

68 return BytesBlob 

69 case "application/yaml": 

70 return YamlDictBlob 

71 case "audo/mpeg": 

72 # Import here as it has optional dependency 

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

74 

75 return AudioBlob 

76 case "image/png": 

77 # Import here as it has optional dependency 

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

79 

80 return ImageBlob 

81 case ( 

82 "text/css" 

83 | "text/csv" 

84 | "text/html" 

85 | "text/javascript" 

86 | "text/markdown" 

87 | "text/plain" 

88 | "text/xml" 

89 ): 

90 return StrBlob 

91 case "video/mp4": 

92 # Import here as it has optional dependency 

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

94 

95 return VideoBlob 

96 case _: 

97 return self.__blob_class 

98 

99 def _get(self, key: str, blob_bytes: bytes) -> BytesBlob: 

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

101 return blob.as_blob( 

102 self.__get_blob_class(key), 

103 self.__blob_class_args, 

104 ) 

105 

106 @override 

107 def __getitem__(self, key: str, /) -> BytesBlob: 

108 if key not in self: 

109 raise KeyError 

110 

111 return self._get(key, (self.__path / key).read_bytes()) 

112 

113 @override 

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

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

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

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

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

119 prefix_len: int = ( 

120 len(str(self.__path)) 

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

122 + 1 

123 ) 

124 

125 for parent, _, files in self.__path.walk(): 

126 for filename in files: 

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

128 

129 @override 

130 def clear(self) -> None: 

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

132 for filename in files: 

133 (parent / filename).unlink() 

134 for dirname in dirs: 

135 (parent / dirname).rmdir() 

136 

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

138 (self.__path / key).unlink() 

139 

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

141 if parent == self.__path: 

142 return 

143 

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

145 parent.rmdir() 

146 

147 @override 

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

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

150 if blob: 

151 self.__cleanup(key) 

152 

153 return blob or default 

154 

155 @override 

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

157 if key not in self: 

158 raise KeyError 

159 

160 self.__cleanup(key) 

161 

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

163 

164 @override 

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

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

167 raise TypeError(PathBlobDict.__BAD_BLOB_CLASS_ERROR_MESSAGE.format( 

168 blob_class=self.__blob_class, 

169 )) 

170 

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

172 parents=True, 

173 exist_ok=True, 

174 ) 

175 

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

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