1"""
2Exceptions and Warnings
3=======================
4
5General exceptions used by NumPy. Note that some exceptions may be module
6specific, such as linear algebra errors.
7
8.. versionadded:: NumPy 1.25
9
10 The exceptions module is new in NumPy 1.25.
11
12.. currentmodule:: numpy.exceptions
13
14Warnings
15--------
16.. autosummary::
17 :toctree: generated/
18
19 ComplexWarning Given when converting complex to real.
20 VisibleDeprecationWarning Same as a DeprecationWarning, but more visible.
21 RankWarning Issued when the design matrix is rank deficient.
22
23Exceptions
24----------
25.. autosummary::
26 :toctree: generated/
27
28 AxisError Given when an axis was invalid.
29 DTypePromotionError Given when no common dtype could be found.
30 TooHardError Error specific to `numpy.shares_memory`.
31
32"""
33
34
35__all__ = [
36 "ComplexWarning", "VisibleDeprecationWarning", "ModuleDeprecationWarning",
37 "TooHardError", "AxisError", "DTypePromotionError"]
38
39
40# Disallow reloading this module so as to preserve the identities of the
41# classes defined here.
42if '_is_loaded' in globals():
43 raise RuntimeError('Reloading numpy._globals is not allowed')
44_is_loaded = True
45
46
47class ComplexWarning(RuntimeWarning):
48 """
49 The warning raised when casting a complex dtype to a real dtype.
50
51 As implemented, casting a complex number to a real discards its imaginary
52 part, but this behavior may not be what the user actually wants.
53
54 """
55 pass
56
57
58class ModuleDeprecationWarning(DeprecationWarning):
59 """Module deprecation warning.
60
61 .. warning::
62
63 This warning should not be used, since nose testing is not relevant
64 anymore.
65
66 The nose tester turns ordinary Deprecation warnings into test failures.
67 That makes it hard to deprecate whole modules, because they get
68 imported by default. So this is a special Deprecation warning that the
69 nose tester will let pass without making tests fail.
70
71 """
72 pass
73
74
75class VisibleDeprecationWarning(UserWarning):
76 """Visible deprecation warning.
77
78 By default, python will not show deprecation warnings, so this class
79 can be used when a very visible warning is helpful, for example because
80 the usage is most likely a user bug.
81
82 """
83 pass
84
85
86class RankWarning(RuntimeWarning):
87 """Matrix rank warning.
88
89 Issued by polynomial functions when the design matrix is rank deficient.
90
91 """
92 pass
93
94
95# Exception used in shares_memory()
96class TooHardError(RuntimeError):
97 """``max_work`` was exceeded.
98
99 This is raised whenever the maximum number of candidate solutions
100 to consider specified by the ``max_work`` parameter is exceeded.
101 Assigning a finite number to ``max_work`` may have caused the operation
102 to fail.
103
104 """
105 pass
106
107
108class AxisError(ValueError, IndexError):
109 """Axis supplied was invalid.
110
111 This is raised whenever an ``axis`` parameter is specified that is larger
112 than the number of array dimensions.
113 For compatibility with code written against older numpy versions, which
114 raised a mixture of :exc:`ValueError` and :exc:`IndexError` for this
115 situation, this exception subclasses both to ensure that
116 ``except ValueError`` and ``except IndexError`` statements continue
117 to catch ``AxisError``.
118
119 Parameters
120 ----------
121 axis : int or str
122 The out of bounds axis or a custom exception message.
123 If an axis is provided, then `ndim` should be specified as well.
124 ndim : int, optional
125 The number of array dimensions.
126 msg_prefix : str, optional
127 A prefix for the exception message.
128
129 Attributes
130 ----------
131 axis : int, optional
132 The out of bounds axis or ``None`` if a custom exception
133 message was provided. This should be the axis as passed by
134 the user, before any normalization to resolve negative indices.
135
136 .. versionadded:: 1.22
137 ndim : int, optional
138 The number of array dimensions or ``None`` if a custom exception
139 message was provided.
140
141 .. versionadded:: 1.22
142
143
144 Examples
145 --------
146 >>> import numpy as np
147 >>> array_1d = np.arange(10)
148 >>> np.cumsum(array_1d, axis=1)
149 Traceback (most recent call last):
150 ...
151 numpy.exceptions.AxisError: axis 1 is out of bounds for array of dimension 1
152
153 Negative axes are preserved:
154
155 >>> np.cumsum(array_1d, axis=-2)
156 Traceback (most recent call last):
157 ...
158 numpy.exceptions.AxisError: axis -2 is out of bounds for array of dimension 1
159
160 The class constructor generally takes the axis and arrays'
161 dimensionality as arguments:
162
163 >>> print(np.exceptions.AxisError(2, 1, msg_prefix='error'))
164 error: axis 2 is out of bounds for array of dimension 1
165
166 Alternatively, a custom exception message can be passed:
167
168 >>> print(np.exceptions.AxisError('Custom error message'))
169 Custom error message
170
171 """
172
173 __slots__ = ("_msg", "axis", "ndim")
174
175 def __init__(self, axis, ndim=None, msg_prefix=None):
176 if ndim is msg_prefix is None:
177 # single-argument form: directly set the error message
178 self._msg = axis
179 self.axis = None
180 self.ndim = None
181 else:
182 self._msg = msg_prefix
183 self.axis = axis
184 self.ndim = ndim
185
186 def __str__(self):
187 axis = self.axis
188 ndim = self.ndim
189
190 if axis is ndim is None:
191 return self._msg
192 else:
193 msg = f"axis {axis} is out of bounds for array of dimension {ndim}"
194 if self._msg is not None:
195 msg = f"{self._msg}: {msg}"
196 return msg
197
198
199class DTypePromotionError(TypeError):
200 """Multiple DTypes could not be converted to a common one.
201
202 This exception derives from ``TypeError`` and is raised whenever dtypes
203 cannot be converted to a single common one. This can be because they
204 are of a different category/class or incompatible instances of the same
205 one (see Examples).
206
207 Notes
208 -----
209 Many functions will use promotion to find the correct result and
210 implementation. For these functions the error will typically be chained
211 with a more specific error indicating that no implementation was found
212 for the input dtypes.
213
214 Typically promotion should be considered "invalid" between the dtypes of
215 two arrays when `arr1 == arr2` can safely return all ``False`` because the
216 dtypes are fundamentally different.
217
218 Examples
219 --------
220 Datetimes and complex numbers are incompatible classes and cannot be
221 promoted:
222
223 >>> import numpy as np
224 >>> np.result_type(np.dtype("M8[s]"), np.complex128) # doctest: +IGNORE_EXCEPTION_DETAIL
225 Traceback (most recent call last):
226 ...
227 DTypePromotionError: The DType <class 'numpy.dtype[datetime64]'> could not
228 be promoted by <class 'numpy.dtype[complex128]'>. This means that no common
229 DType exists for the given inputs. For example they cannot be stored in a
230 single array unless the dtype is `object`. The full list of DTypes is:
231 (<class 'numpy.dtype[datetime64]'>, <class 'numpy.dtype[complex128]'>)
232
233 For example for structured dtypes, the structure can mismatch and the
234 same ``DTypePromotionError`` is given when two structured dtypes with
235 a mismatch in their number of fields is given:
236
237 >>> dtype1 = np.dtype([("field1", np.float64), ("field2", np.int64)])
238 >>> dtype2 = np.dtype([("field1", np.float64)])
239 >>> np.promote_types(dtype1, dtype2) # doctest: +IGNORE_EXCEPTION_DETAIL
240 Traceback (most recent call last):
241 ...
242 DTypePromotionError: field names `('field1', 'field2')` and `('field1',)`
243 mismatch.
244
245 """ # noqa: E501
246 pass