scipy/removed-in-scipy
| Rule | removed-in-scipy |
|---|---|
| Pack | SciPy |
| Severity | error |
| Fix | unsafe, with xpd check --fix --unsafe-fixes |
| Library versions | every version |
These names were removed from SciPy and raise AttributeError on SciPy 1.18.1. Use the replacement that the message names.
from scipy import integrate, signal, statsfrom scipy.misc import combarea = integrate.trapz(y, x)window = signal.hann(64)p = stats.binom_test(k, n)angle = scipy.pi * fracfrom scipy.special import combarea = integrate.trapezoid(y, x)window = signal.windows.hann(64)p = stats.binomtest(k, n).pvalueangle = np.pi * fracWhat the rule reports
Section titled “What the rule reports”The rule reports SciPy names that raise on SciPy 1.18.1. Its message says what to write instead, or says that SciPy has no public replacement.
| Old names | On SciPy 1.18.1 |
|---|---|
| Removed functions for integration, windows, wavelets, Legendre polynomials, spherical Bessel functions, statistics and linear algebra | The name is gone. |
interp2d |
A call raises NotImplementedError. |
Everything in scipy.misc |
The module is empty and warns when imported. |
NumPy functions and constants on the scipy namespace, such as scipy.pi |
The name is gone. |
The rule leaves alone code under a SciPy version check, a hasattr check, or a try that
catches ImportError or AttributeError. Such code exists to support old SciPy.
How to fix it
Section titled “How to fix it”xpd check --fix --unsafe-fixes applies the fixes of this rule. Check the result, because an unsafe fix can change what the code does.
The safe fix writes trapezoid, cumulative_trapezoid and simpson for trapz, cumtrapz and simps, in the module the code already names or in an import as trapezoid as trapz, scipy.signal.windows.<name> for the 19 windows that scipy.signal held, such as hann, and np.<name> for the 438 names that were the NumPy object on the scipy namespace, such as scipy.pi. On SciPy 1.13.1, 1.12.0 and 1.3.3 each new form is the same object as the old one or gives a result with an equal pickle. Unsafe, because the results differ on a SciPy that runs both forms: pinv for pinv2, whose third positional argument was another tolerance, median_abs_deviation for median_absolute_deviation, which had another default scale, gibrat for gilbrat, which is another class, stats.mstats.find_repeats, which returns a plain tuple, impulse, lsim and step for impulse2, lsim2 and step2, betainc and betaincinv for btdtr and btdtri, each also in an import as new as old, np.kron, np.tri, np.tril and np.triu for the scipy.linalg names, np.<name> for sqrt, log and the six other numpy.lib.scimath functions on the scipy namespace, and scipy.datasets for face, ascent and electrocardiogram on scipy.misc, which raises ImportError without the pooch package. Unsafe, because no SciPy that runs here holds the old form: eye for speye, stats.mstats.kendalltau_seasonal, expm for expm2 and expm3, stats.chi2.sf for stats.chisqprob, scipy.special.betainc and erfc for scipy.stats.betai and erfc, np.linalg.slogdet, sparse.csgraph.connected_components for sparse.cs_graph_components, scipy.special for comb, factorial and the other names on scipy.misc or the scipy namespace, six windows such as taylor, and 46 NumPy names such as np.round. The other forms have no fix, because the name has no replacement that takes the same arguments, the message names two, or the new module is one that the code does not name, as in from scipy import misc or an import from the old module.
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
$OBJ.$N was removed from SciPy and raises AttributeError on SciPy 1.18.1. Write $OBJ.$NEW.Browse the other rules of the SciPy pack.