Add a pre-commit configuration to conform to new pydata conventions.

This patch also reformats code using the `ruff` code formatter.

Author: matthewwardrop

.gitignore

@@ -83,3 +83,7 @@ doc/cdoc/build
 ehthumbs.db
 Icon?
 Thumbs.db
+
+# Test generated files #
+########################
+.python-version

.pre-commit-config.yaml

@@ -0,0 +1,23 @@
+repos:
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+    - id: check-yaml
+    - id: end-of-file-fixer
+    - id: trailing-whitespace
+    - id: fix-byte-order-marker
+    - id: destroyed-symlinks
+    - id: fix-encoding-pragma
+      args: ["--remove"]
+    - id: mixed-line-ending
+    - id: name-tests-test
+      args: ["--pytest-test-first"]
+    - id: pretty-format-json
+      args: ["--autofix", "--no-ensure-ascii"]
+      exclude: ".ipynb"
+
+- repo: https://github.com/astral-sh/ruff-pre-commit
+  rev: v0.7.3
+  hooks:
+    - id: ruff-format
+      types_or: [ python, pyi, jupyter ]

TODO

@@ -104,7 +104,7 @@ the cheap trick way of doing it is:
   def arima(n, m):
     return ArimaModelType(n, m)
 and then in the factor type sniffing code detect these things and
-separate them out from "real" factors. 
+separate them out from "real" factors.
 
 * make sure that pickling works
 - And make sure that if we allow it at all, then it's sustainable!

doc/R-comparison.rst

@@ -105,7 +105,7 @@ Differences from R:
     # R:
     > qr(model.matrix(~ 1 + a:b))$rank
     [1] 4
-  
+
   However, the matrix produced for this formula has 5 columns, meaning
   that it contains redundant overspecification:
 
@@ -149,7 +149,7 @@ Differences from R:
   use a full-rank encoding for ``b``. Therefore, we *should* use a
   full-rank encoding for ``b``, and produce a model matrix with 6
   columns. But in fact, R gives us only 4:
-  
+
   .. code-block:: rconsole
 
     # R:

doc/_examples/example_lm.py

@@ -1,9 +1,11 @@
 import numpy as np
 from patsy import dmatrices, build_design_matrices
 
+
 class LM(object):
     """An example ordinary least squares linear model class, analogous to R's
     lm() function. Don't use this in real life, it isn't properly tested."""
+
     def __init__(self, formula_like, data={}):
         y, x = dmatrices(formula_like, data, 1)
         self.nobs = x.shape[0]
@@ -12,27 +14,27 @@ def __init__(self, formula_like, data={}):
         self._x_design_info = x.design_info
 
     def __repr__(self):
-        summary = ("Ordinary least-squares regression\n"
-                   "  Model: %s ~ %s\n"
-                   "  Regression (beta) coefficients:\n"
-                   % (self._y_design_info.describe(),
-                      self._x_design_info.describe()))
+        summary = (
+            "Ordinary least-squares regression\n"
+            "  Model: %s ~ %s\n"
+            "  Regression (beta) coefficients:\n"
+            % (self._y_design_info.describe(), self._x_design_info.describe())
+        )
         for name, value in zip(self._x_design_info.column_names, self.betas):
             summary += "    %s:  %0.3g\n" % (name, value[0])
         return summary
 
     def predict(self, new_data):
-        (new_x,) = build_design_matrices([self._x_design_info],
-                                         new_data)
+        (new_x,) = build_design_matrices([self._x_design_info], new_data)
         return np.dot(new_x, self.betas)
 
     def loglik(self, new_data):
-        (new_y, new_x) = build_design_matrices([self._y_design_info,
-                                                self._x_design_info],
-                                               new_data)
+        (new_y, new_x) = build_design_matrices(
+            [self._y_design_info, self._x_design_info], new_data
+        )
         new_pred = np.dot(new_x, self.betas)
         sigma2 = self.rss / self.nobs
         # It'd be more elegant to use scipy.stats.norm.logpdf here, but adding
         # a dependency on scipy makes the docs build more complicated:
         Z = -0.5 * np.log(2 * np.pi * sigma2)
-        return Z + -0.5 * (new_y - new_x) ** 2/sigma2
+        return Z + -0.5 * (new_y - new_x) ** 2 / sigma2

doc/_examples/example_treatment.py

@@ -1,18 +1,26 @@
 import numpy as np
 
+
 class MyTreat(object):
     def __init__(self, reference=0):
         self.reference = reference
 
     def code_with_intercept(self, levels):
-        return ContrastMatrix(np.eye(len(levels)),
-                              ["[My.%s]" % (level,) for level in levels])
+        return ContrastMatrix(
+            np.eye(len(levels)), ["[My.%s]" % (level,) for level in levels]
+        )
 
     def code_without_intercept(self, levels):
         eye = np.eye(len(levels) - 1)
-        contrasts = np.vstack((eye[:self.reference, :],
-                               np.zeros((1, len(levels) - 1)),
-                               eye[self.reference:, :]))
-        suffixes = ["[MyT.%s]" % (level,) for level in
-                    levels[:self.reference] + levels[self.reference + 1:]]
+        contrasts = np.vstack(
+            (
+                eye[: self.reference, :],
+                np.zeros((1, len(levels) - 1)),
+                eye[self.reference :, :],
+            )
+        )
+        suffixes = [
+            "[MyT.%s]" % (level,)
+            for level in levels[: self.reference] + levels[self.reference + 1 :]
+        ]
         return ContrastMatrix(contrasts, suffixes)

doc/_static/facebox.css

@@ -77,4 +77,4 @@
 .facebox_overlayBG {
   background-color: #000;
   z-index: 99;
-}
+}

doc/_static/show-code.js

@@ -25,13 +25,13 @@ function scrapeText(codebox){
     return newlines.join('\\n');
 }
 
-$(document).ready(            
+$(document).ready(
         function() {
     // grab all code boxes
     var ipythoncode = $(".highlight-ipython");
     $.each(ipythoncode, function() {
         var code = scrapeText($(this).text());
-        // give them a facebox pop-up with plain text code   
+        // give them a facebox pop-up with plain text code
         $(this).append('<span style="text-align:right; display:block; margin-top:-10px; margin-left:10px; font-size:60%"><a href="javascript: jQuery.facebox(\'<textarea cols=80 rows=10 readonly style=margin:5px onmouseover=javascript:this.select();>'+htmlescape(htmlescape(code))+'</textarea>\');">View Code</a></span>');
         $(this,"textarea").select();
     });

doc/categorical-coding.rst

@@ -78,7 +78,7 @@ As an example, here's a simplified version of the built-in
 :class:`Treatment` coding object:
 
 .. literalinclude:: _examples/example_treatment.py
-                                 
+
 .. ipython:: python
    :suppress: