Coverage for src/chebpy/decorators.py: 100%

78 statements  

« prev     ^ index     » next       coverage.py v7.16.2, created at 2026-10-01 13:43 +0000

1"""Decorator functions for the ChebPy package. 

2 

3This module provides various decorators used throughout the ChebPy package to 

4implement common functionality such as caching, handling empty objects, 

5pre- and post-processing of function inputs/outputs, and type conversion. 

6These decorators help reduce code duplication and ensure consistent behavior 

7across the package. 

8""" 

9 

10from collections.abc import Callable 

11from functools import wraps 

12from typing import Any 

13 

14import numpy as np 

15 

16 

17def cache(f: Callable[..., Any]) -> Callable[..., Any]: 

18 """Object method output caching mechanism. 

19 

20 This decorator caches the output of zero-argument methods to speed up repeated 

21 execution of relatively expensive operations such as .roots(). Cached computations 

22 are stored in a dictionary called _cache which is bound to self using keys 

23 corresponding to the method name. 

24 

25 Args: 

26 f (callable): The method to be cached. Must be a zero-argument method. 

27 

28 Returns: 

29 callable: A wrapped version of the method that implements caching. 

30 

31 Note: 

32 Can be used in principle on arbitrary objects. 

33 """ 

34 

35 # TODO: look into replacing this with one of the functools cache decorators 

36 @wraps(f) 

37 def wrapper(self: Any) -> Any: 

38 """Return the cached method result, computing and storing it on first call.""" 

39 try: 

40 # f has been executed previously 

41 out = self._cache[f.__name__] # ty: ignore[unresolved-attribute] 

42 except AttributeError: 

43 # f has not been executed previously and self._cache does not exist 

44 self._cache = {} 

45 out = self._cache[f.__name__] = f(self) # ty: ignore[unresolved-attribute] 

46 except KeyError: 

47 # f has not been executed previously, but self._cache exists 

48 out = self._cache[f.__name__] = f(self) # ty: ignore[unresolved-attribute] 

49 return out 

50 

51 return wrapper 

52 

53 

54def self_empty(resultif: Any = None) -> Callable[..., Any]: 

55 """Factory method to produce a decorator for handling empty objects. 

56 

57 This factory creates a decorator that checks whether the object whose method 

58 is being wrapped is empty. If the object is empty, it returns either the supplied 

59 resultif value or a copy of the object. Otherwise, it executes the wrapped method. 

60 

61 Args: 

62 resultif: Value to return when the object is empty. If None, returns a copy 

63 of the object instead. 

64 

65 Returns: 

66 callable: A decorator function that implements the empty-checking logic. 

67 

68 Note: 

69 This decorator is primarily used in chebtech.py. 

70 """ 

71 

72 def decorator(f: Callable[..., Any]) -> Callable[..., Any]: 

73 """Wrap *f* with the empty-object short-circuit logic.""" 

74 

75 @wraps(f) 

76 def wrapper(self: Any, *args: Any, **kwargs: Any) -> Any: 

77 """Return the empty-case result if *self* is empty, else call *f*.""" 

78 if self.isempty: 

79 if resultif is not None: 

80 return resultif 

81 else: 

82 return self.copy() 

83 else: 

84 return f(self, *args, **kwargs) 

85 

86 return wrapper 

87 

88 return decorator 

89 

90 

91def preandpostprocess(f: Callable[..., Any]) -> Callable[..., Any]: 

92 """Decorator for pre- and post-processing tasks common to bary and clenshaw. 

93 

94 This decorator handles several edge cases for functions like bary and clenshaw: 

95 - Empty arrays in input arguments 

96 - Constant functions 

97 - NaN values in coefficients 

98 - Scalar vs. array inputs 

99 

100 Args: 

101 f (callable): The function to be wrapped. 

102 

103 Returns: 

104 callable: A wrapped version of the function with pre- and post-processing. 

105 """ 

106 

107 @wraps(f) 

108 def thewrapper(*args: Any, **kwargs: Any) -> Any: 

109 """Handle empty/constant/NaN/scalar edge cases around *f*.""" 

110 xx, akfk = args[:2] 

111 # are any of the first two arguments empty arrays? 

112 if (np.asarray(xx).size == 0) | (np.asarray(akfk).size == 0): 

113 return np.array([]) 

114 # is the function constant? 

115 elif akfk.size == 1: 

116 if np.isscalar(xx): 

117 return akfk[0] 

118 else: 

119 return akfk * np.ones(xx.size) 

120 # are there any NaNs in the second argument? 

121 elif np.any(np.isnan(akfk)): 

122 return np.nan * np.ones(xx.size) 

123 # convert first argument to an array if it is a scalar and then 

124 # return the first (and only) element of the result if so 

125 else: 

126 args_list = list(args) 

127 args_list[0] = np.array([xx]) if np.isscalar(xx) else args_list[0] 

128 out = f(*args_list, **kwargs) 

129 return out[0] if np.isscalar(xx) else out 

130 

131 return thewrapper 

132 

133 

134def float_argument(f: Callable[..., Any]) -> Callable[..., Any]: 

135 """Decorator to ensure consistent input/output types for Chebfun __call__ method. 

136 

137 This decorator ensures that when a Chebfun object is called with a float input, 

138 it returns a float output, and when called with an array input, it returns an 

139 array output. It handles various input formats including scalars, numpy arrays, 

140 and nested arrays. 

141 

142 Args: 

143 f (callable): The __call__ method to be wrapped. 

144 

145 Returns: 

146 callable: A wrapped version of the method that ensures type consistency. 

147 """ 

148 

149 @wraps(f) 

150 def thewrapper(self: Any, *args: Any, **kwargs: Any) -> Any: 

151 """Coerce the first argument to an array and match scalar/array output to it.""" 

152 x = args[0] 

153 xx = np.array([x]) if np.isscalar(x) else np.array(x) 

154 # discern between the array(0.1) and array([0.1]) cases 

155 if xx.size == 1 and xx.ndim == 0: 

156 xx = np.array([xx]) 

157 args_list = list(args) 

158 args_list[0] = xx 

159 out = f(self, *args_list, **kwargs) 

160 return out[0] if np.isscalar(x) else out 

161 

162 return thewrapper 

163 

164 

165def cast_arg_to_chebfun(f: Callable[..., Any]) -> Callable[..., Any]: 

166 """Decorator to cast the first argument to a chebfun object if needed. 

167 

168 This decorator attempts to convert the first argument to a chebfun object 

169 if it is not already one. Currently, only numeric types can be cast to chebfun. 

170 

171 Args: 

172 f (callable): The method to be wrapped. 

173 

174 Returns: 

175 callable: A wrapped version of the method that ensures the first argument 

176 is a chebfun object. 

177 """ 

178 

179 @wraps(f) 

180 def wrapper(self: Any, *args: Any, **kwargs: Any) -> Any: 

181 """Cast the first argument to a chebfun (if needed) before calling *f*.""" 

182 other = args[0] 

183 if not isinstance(other, self.__class__): 

184 fun = self.initconst(args[0], self.support) 

185 args_list = list(args) 

186 args_list[0] = fun 

187 return f(self, *args_list, **kwargs) 

188 return f(self, *args, **kwargs) 

189 

190 return wrapper 

191 

192 

193def cast_other(f: Callable[..., Any]) -> Callable[..., Any]: 

194 """Decorator to cast the first argument to the same type as self. 

195 

196 This generic decorator is applied to binary operator methods to ensure that 

197 the first positional argument (typically 'other') is cast to the same type 

198 as the object on which the method is called. 

199 

200 Args: 

201 f (callable): The binary operator method to be wrapped. 

202 

203 Returns: 

204 callable: A wrapped version of the method that ensures type consistency 

205 between self and the first argument. 

206 """ 

207 

208 @wraps(f) 

209 def wrapper(self: Any, *args: Any, **kwargs: Any) -> Any: 

210 """Cast the first argument to ``type(self)`` (if needed) before calling *f*.""" 

211 cls = self.__class__ 

212 other = args[0] 

213 if not isinstance(other, cls): 

214 args_list = list(args) 

215 args_list[0] = cls(other) 

216 return f(self, *args_list, **kwargs) 

217 return f(self, *args, **kwargs) 

218 

219 return wrapper