Add codespell config, pre-commit hook, and GH action (to catch new typos) + make it fix some it finds

.pre-commit-config.yaml

@@ -18,3 +18,10 @@ repos:
     hooks:
     -   id: ruff
         args: ["--fix"]
+-   repo: https://github.com/codespell-project/codespell
+    # Configuration for codespell is in pyproject.toml
+    rev: v2.2.6
+    hooks:
+    -   id: codespell
+        additional_dependencies:
+        -   tomli

asv/step_detect.py

@@ -754,7 +754,7 @@ def sigma_star(rights, values, rho):
 def solve_potts_approx(y, w, gamma=None, min_size=1, **kw):
     """
     Fit penalized stepwise constant function (Potts model) to data
-    approximatively, in linear time.
+    approximately, in linear time.
 
     Do this by running the exact solver using a small maximum interval
     size, and then combining consecutive intervals together if it

asv/util.py

@@ -689,7 +689,7 @@ def load_json(path, api_version=None, js_comments=False):
     path : str
         File name
     api_version : str or None
-        API version indentifier
+        API version identifier
     js_comments : bool, optional
         Whether to allow nonstandard javascript-style comments
         in the file. Note that this slows down the loading

docs/source/benchmarks.rst

@@ -116,7 +116,7 @@ benchmark types:
 
   For the purposes of identifying benchmarks in the UI, ``repr()`` is called
   on the elements of ``params``. In the event these strings contain memory
-  addresses, those adresses are stripped to allow comparison across runs.
+  addresses, those addresses are stripped to allow comparison across runs.
   Additionally, if this results in a non-unique mapping, each duplicated
   element will be suffixed with a distinct integer identifier corresponding
   to order of appearance.
@@ -154,7 +154,7 @@ Timing benchmarks
   ``number`` times, and after that ``teardown`` runs.
 
 - ``sample_time``: ``asv`` will automatically select ``number`` so that
-  each sample takes approximatively ``sample_time`` seconds.  If not
+  each sample takes approximately ``sample_time`` seconds.  If not
   specified, ``sample_time`` defaults to 10 milliseconds.
 
 - ``min_run_count``: the function is run at least this many times during

pyproject.toml

@@ -135,3 +135,10 @@ archs = ['AMD64']
 where = ["."]
 exclude = ["wheelhouse"]
 namespaces = true
+
+[tool.codespell]
+# Ref: https://github.com/codespell-project/codespell#using-a-config-file
+skip = '.git,*.css'
+check-hidden = true
+ignore-regex = '- Boris Feld\b'
+# ignore-words-list = ''

test/test_environment.py

@@ -745,7 +745,7 @@ def get_env():
     assert not os.listdir(cache_dir)
     env._set_installed_commit_hash(None)
 
-    # It should succed with nonzero cache size
+    # It should succeed with nonzero cache size
     conf.build_cache_size = 1
     env = get_env()
     env.install_project(conf, repo, commit_hash)

test/test_run.py

@@ -90,7 +90,7 @@ def _test_run(range_spec, branches, expected_commits):
         assert set(result_files) == expected
 
     for branches, expected_commits in (
-        # Without branches in config, shoud just use main
+        # Without branches in config, should just use main
         ([None], [initial_commit, main_commit]),
 
         # With one branch in config, should just use that branch

test/test_statistics.py

@@ -57,7 +57,7 @@ def test_is_different():
 
 def _check_ci(estimator, sampler, nsamples=300):
     """
-    Check whether confidence behaves as advertized.
+    Check whether confidence behaves as advertised.
 
     Draw random samples from a distribution and check how often the
     true value (assumed 0) falls in the estimated CI.

test/test_util.py

@@ -49,7 +49,7 @@ def test_parallelfailure():
     assert str(exc) == "Exception: test\n    something"
 
     # Check multiprocessing does not hang (it would hang on Python
-    # 2.7.8 if the 'raise utill.ParallelFailure ...' above is changed
+    # 2.7.8 if the 'raise util.ParallelFailure ...' above is changed
     # to just 'raise')
     pool = multiprocessing.Pool(4)
     try: