| 1 | # SPDX-License-Identifier: MIT |
| 2 | # |
| 3 | # Copyright (C) 2023 EfficiOS, inc. |
| 4 | # |
| 5 | # pyright: strict, reportTypeCommentUsage=false |
| 6 | |
| 7 | |
| 8 | import re |
| 9 | import json |
| 10 | import typing |
| 11 | from typing import ( |
| 12 | Any, |
| 13 | Dict, |
| 14 | List, |
| 15 | Type, |
| 16 | Union, |
| 17 | TextIO, |
| 18 | Generic, |
| 19 | Mapping, |
| 20 | TypeVar, |
| 21 | Optional, |
| 22 | Sequence, |
| 23 | overload, |
| 24 | ) |
| 25 | |
| 26 | # Internal type aliases and variables |
| 27 | _RawArrayT = List["_RawValT"] |
| 28 | _RawObjT = Dict[str, "_RawValT"] |
| 29 | _RawValT = Union[None, bool, int, float, str, _RawArrayT, _RawObjT] |
| 30 | _RawValTV = TypeVar("_RawValTV", bound="_RawValT") |
| 31 | _ValTV = TypeVar("_ValTV", bound="Val") |
| 32 | |
| 33 | |
| 34 | # Type of a single JSON value path element. |
| 35 | PathElemT = Union[str, int] |
| 36 | |
| 37 | |
| 38 | # A JSON value path. |
| 39 | class Path: |
| 40 | def __init__(self, elems: Optional[List[PathElemT]] = None): |
| 41 | if elems is None: |
| 42 | elems = [] |
| 43 | |
| 44 | self._elems = elems |
| 45 | |
| 46 | # Elements of this path. |
| 47 | @property |
| 48 | def elems(self): |
| 49 | return self._elems |
| 50 | |
| 51 | # Returns a new path containing the current elements plus `elem`. |
| 52 | def __truediv__(self, elem: PathElemT): |
| 53 | return Path(self._elems + [elem]) |
| 54 | |
| 55 | # Returns a valid jq filter. |
| 56 | def __str__(self): |
| 57 | s = "" |
| 58 | |
| 59 | for elem in self._elems: |
| 60 | if type(elem) is str: |
| 61 | if re.match(r"[a-zA-Z]\w*$", elem): |
| 62 | s += ".{}".format(elem) |
| 63 | else: |
| 64 | s += '."{}"'.format(elem) |
| 65 | else: |
| 66 | assert type(elem) is int |
| 67 | s += "[{}]".format(elem) |
| 68 | |
| 69 | if not s.startswith("."): |
| 70 | s = "." + s |
| 71 | |
| 72 | return s |
| 73 | |
| 74 | |
| 75 | # Base of any JSON value. |
| 76 | class Val: |
| 77 | _name = "a value" |
| 78 | |
| 79 | def __init__(self, path: Optional[Path] = None): |
| 80 | if path is None: |
| 81 | path = Path() |
| 82 | |
| 83 | self._path = path |
| 84 | |
| 85 | # Path to this JSON value. |
| 86 | @property |
| 87 | def path(self): |
| 88 | return self._path |
| 89 | |
| 90 | |
| 91 | # JSON null value. |
| 92 | class NullVal(Val): |
| 93 | _name = "a null" |
| 94 | |
| 95 | |
| 96 | # JSON scalar value. |
| 97 | class _ScalarVal(Val, Generic[_RawValTV]): |
| 98 | def __init__(self, raw_val: _RawValTV, path: Optional[Path] = None): |
| 99 | super().__init__(path) |
| 100 | self._raw_val = raw_val |
| 101 | |
| 102 | # Raw value. |
| 103 | @property |
| 104 | def val(self): |
| 105 | return self._raw_val |
| 106 | |
| 107 | |
| 108 | # JSON boolean value. |
| 109 | class BoolVal(_ScalarVal[bool]): |
| 110 | _name = "a boolean" |
| 111 | |
| 112 | def __bool__(self): |
| 113 | return self.val |
| 114 | |
| 115 | |
| 116 | # JSON integer value. |
| 117 | class IntVal(_ScalarVal[int]): |
| 118 | _name = "an integer" |
| 119 | |
| 120 | def __int__(self): |
| 121 | return self.val |
| 122 | |
| 123 | |
| 124 | # JSON floating point number value. |
| 125 | class FloatVal(_ScalarVal[float]): |
| 126 | _name = "a floating point number" |
| 127 | |
| 128 | def __float__(self): |
| 129 | return self.val |
| 130 | |
| 131 | |
| 132 | # JSON string value. |
| 133 | class StrVal(_ScalarVal[str]): |
| 134 | _name = "a string" |
| 135 | |
| 136 | def __str__(self): |
| 137 | return self.val |
| 138 | |
| 139 | |
| 140 | # JSON array value. |
| 141 | class ArrayVal(Val, Sequence[Val]): |
| 142 | _name = "an array" |
| 143 | |
| 144 | def __init__(self, raw_val: _RawArrayT, path: Optional[Path] = None): |
| 145 | super().__init__(path) |
| 146 | self._raw_val = raw_val |
| 147 | |
| 148 | # Returns the value at index `index`. |
| 149 | # |
| 150 | # Raises `TypeError` if the type of the returned value isn't |
| 151 | # `expected_elem_type`. |
| 152 | def at(self, index: int, expected_elem_type: Type[_ValTV]): |
| 153 | try: |
| 154 | elem = self._raw_val[index] |
| 155 | except IndexError: |
| 156 | raise IndexError( |
| 157 | "`{}`: array index {} out of range".format(self._path, index) |
| 158 | ) |
| 159 | |
| 160 | return wrap(elem, self._path / index, expected_elem_type) |
| 161 | |
| 162 | # Returns an iterator yielding the values of this array value. |
| 163 | # |
| 164 | # Raises `TypeError` if the type of any yielded value isn't |
| 165 | # `expected_elem_type`. |
| 166 | def iter(self, expected_elem_type: Type[_ValTV]): |
| 167 | for i in range(len(self._raw_val)): |
| 168 | yield self.at(i, expected_elem_type) |
| 169 | |
| 170 | @overload |
| 171 | def __getitem__(self, index: int) -> Val: |
| 172 | ... |
| 173 | |
| 174 | @overload |
| 175 | def __getitem__(self, index: slice) -> Sequence[Val]: |
| 176 | ... |
| 177 | |
| 178 | def __getitem__(self, index: Union[int, slice]) -> Union[Val, Sequence[Val]]: |
| 179 | if type(index) is slice: |
| 180 | raise NotImplementedError |
| 181 | |
| 182 | return self.at(index, Val) |
| 183 | |
| 184 | def __len__(self): |
| 185 | return len(self._raw_val) |
| 186 | |
| 187 | |
| 188 | # JSON object value. |
| 189 | class ObjVal(Val, Mapping[str, Val]): |
| 190 | _name = "an object" |
| 191 | |
| 192 | def __init__(self, raw_val: _RawObjT, path: Optional[Path] = None): |
| 193 | super().__init__(path) |
| 194 | self._raw_val = raw_val |
| 195 | |
| 196 | # Returns the value having the key `key`. |
| 197 | # |
| 198 | # Raises `TypeError` if the type of the returned value isn't |
| 199 | # `expected_type`. |
| 200 | def at(self, key: str, expected_type: Type[_ValTV]): |
| 201 | try: |
| 202 | val = self._raw_val[key] |
| 203 | except KeyError: |
| 204 | raise KeyError("`{}`: no value has the key `{}`".format(self._path, key)) |
| 205 | |
| 206 | return wrap(val, self._path / key, expected_type) |
| 207 | |
| 208 | def __getitem__(self, key: str) -> Val: |
| 209 | return self.at(key, Val) |
| 210 | |
| 211 | def __len__(self): |
| 212 | return len(self._raw_val) |
| 213 | |
| 214 | def __iter__(self): |
| 215 | return iter(self._raw_val) |
| 216 | |
| 217 | |
| 218 | # Raises `TypeError` if the type of `val` is not `expected_type`. |
| 219 | def _check_type(val: Val, expected_type: Type[Val]): |
| 220 | if not isinstance(val, expected_type): |
| 221 | raise TypeError( |
| 222 | "`{}`: expecting {} value, got {}".format( |
| 223 | val.path, |
| 224 | expected_type._name, # pyright: ignore [reportPrivateUsage] |
| 225 | type(val)._name, # pyright: ignore [reportPrivateUsage] |
| 226 | ) |
| 227 | ) |
| 228 | |
| 229 | |
| 230 | # Wraps the raw value `raw_val` into an equivalent instance of some |
| 231 | # `Val` subclass having the path `path` and returns it. |
| 232 | # |
| 233 | # If the resulting JSON value type isn't `expected_type`, then this |
| 234 | # function raises `TypeError`. |
| 235 | def wrap( |
| 236 | raw_val: _RawValT, path: Optional[Path] = None, expected_type: Type[_ValTV] = Val |
| 237 | ) -> _ValTV: |
| 238 | val = None |
| 239 | |
| 240 | if raw_val is None: |
| 241 | val = NullVal(path) |
| 242 | elif isinstance(raw_val, bool): |
| 243 | val = BoolVal(raw_val, path) |
| 244 | elif isinstance(raw_val, int): |
| 245 | val = IntVal(raw_val, path) |
| 246 | elif isinstance(raw_val, float): |
| 247 | val = FloatVal(raw_val, path) |
| 248 | elif isinstance(raw_val, str): |
| 249 | val = StrVal(raw_val, path) |
| 250 | elif isinstance(raw_val, list): |
| 251 | val = ArrayVal(raw_val, path) |
| 252 | else: |
| 253 | assert isinstance(raw_val, dict) |
| 254 | val = ObjVal(raw_val, path) |
| 255 | |
| 256 | assert val is not None |
| 257 | _check_type(val, expected_type) |
| 258 | return typing.cast(_ValTV, val) |
| 259 | |
| 260 | |
| 261 | # Like json.loads(), but returns a `Val` instance, raising `TypeError` |
| 262 | # if its type isn't `expected_type`. |
| 263 | def loads(s: str, expected_type: Type[_ValTV] = Val, **kwargs: Any) -> _ValTV: |
| 264 | return wrap(json.loads(s, **kwargs), Path(), expected_type) |
| 265 | |
| 266 | |
| 267 | # Like json.load(), but returns a `Val` instance, raising `TypeError` if |
| 268 | # its type isn't `expected_type`. |
| 269 | def load(fp: TextIO, expected_type: Type[_ValTV] = Val, **kwargs: Any) -> _ValTV: |
| 270 | return wrap(json.load(fp, **kwargs), Path(), expected_type) |