GH-109408: Move the C file whitespace check from patchcheck to pre-commit

.pre-commit-config.yaml

@@ -19,6 +19,16 @@ repos:
       - id: trailing-whitespace
         types_or: [c, python, rst]
 
+  - repo: local
+    hooks:
+      - id: c-file-whitespace
+        name: "Check C file whitespace"
+        entry: "python Tools/patchcheck/untabify.py"
+        language: "system"
+        types_or: ['c', 'c++']
+        # Don't check the style of copies directories containing external libraries
+        exclude: '^Modules/(_decimal/libmpdec|expat|zlib)/.*$'
+
   - repo: https://github.com/sphinx-contrib/sphinx-lint
     rev: v0.6.8
     hooks:

Tools/patchcheck/patchcheck.py

@@ -8,16 +8,8 @@
 import sysconfig
 
 import reindent
-import untabify
 
 
-# Excluded directories which are copies of external libraries:
-# don't check their coding style
-EXCLUDE_DIRS = [
-    os.path.join('Modules', '_decimal', 'libmpdec'),
-    os.path.join('Modules', 'expat'),
-    os.path.join('Modules', 'zlib'),
-    ]
 SRCDIR = sysconfig.get_config_var('srcdir')
 
 
@@ -147,15 +139,8 @@ def changed_files(base_branch=None):
     else:
         sys.exit('need a git checkout to get modified files')
 
-    filenames2 = []
-    for filename in filenames:
-        # Normalize the path to be able to match using .startswith()
-        filename = os.path.normpath(filename)
-        if any(filename.startswith(path) for path in EXCLUDE_DIRS):
-            # Exclude the file
-            continue
-        filenames2.append(filename)
-
+    # Normalize the path to be able to match using .startswith()
+    filenames2 = list(map(os.path.normpath, filenames))
     return filenames2
 
 
@@ -191,20 +176,6 @@ def normalize_whitespace(file_paths):
     return fixed
 
 
-@status("Fixing C file whitespace", info=report_modified_files)
-def normalize_c_whitespace(file_paths):
-    """Report if any C files """
-    fixed = []
-    for path in file_paths:
-        abspath = os.path.join(SRCDIR, path)
-        with open(abspath, 'r') as f:
-            if '\t' not in f.read():
-                continue
-        untabify.process(abspath, 8, verbose=False)
-        fixed.append(path)
-    return fixed
-
-
 ws_re = re.compile(br'\s+(\r?\n)$')
 
 @status("Fixing docs whitespace", info=report_modified_files)
@@ -267,12 +238,10 @@ def ci(pull_request):
     base_branch = get_base_branch()
     file_paths = changed_files(base_branch)
     python_files = [fn for fn in file_paths if fn.endswith('.py')]
-    c_files = [fn for fn in file_paths if fn.endswith(('.c', '.h'))]
     doc_files = [fn for fn in file_paths if fn.startswith('Doc') and
                  fn.endswith(('.rst', '.inc'))]
     fixed = []
     fixed.extend(normalize_whitespace(python_files))
-    fixed.extend(normalize_c_whitespace(c_files))
     fixed.extend(normalize_docs_whitespace(doc_files))
     if not fixed:
         print('No whitespace issues found')
@@ -285,14 +254,11 @@ def main():
     base_branch = get_base_branch()
     file_paths = changed_files(base_branch)
     python_files = [fn for fn in file_paths if fn.endswith('.py')]
-    c_files = [fn for fn in file_paths if fn.endswith(('.c', '.h'))]
     doc_files = [fn for fn in file_paths if fn.startswith('Doc') and
                  fn.endswith(('.rst', '.inc'))]
     misc_files = {p for p in file_paths if p.startswith('Misc')}
     # PEP 8 whitespace rules enforcement.
     normalize_whitespace(python_files)
-    # C rules enforcement.
-    normalize_c_whitespace(c_files)
     # Doc whitespace enforcement.
     normalize_docs_whitespace(doc_files)
     # Docs updated.
@@ -307,10 +273,12 @@ def main():
     regenerated_pyconfig_h_in(file_paths)
 
     # Test suite run and passed.
-    if python_files or c_files:
-        end = " and check for refleaks?" if c_files else "?"
-        print()
-        print("Did you run the test suite" + end)
+    has_c_files = any(fn for fn in file_paths if fn.endswith(('.c', '.h')))
+    print()
+    if has_c_files:
+        print("Did you run the test suite and check for refleaks?")
+    elif python_files:
+        print("Did you run the test suite?")
 
 
 if __name__ == '__main__':

Tools/patchcheck/untabify.py

@@ -32,10 +32,10 @@ def process(filename, tabsize, verbose=True):
             encoding = f.encoding
     except IOError as msg:
         print("%r: I/O error: %s" % (filename, msg))
-        return
+        return 2
     newtext = text.expandtabs(tabsize)
     if newtext == text:
-        return
+        return 0
     backup = filename + "~"
     try:
         os.unlink(backup)
@@ -49,7 +49,8 @@ def process(filename, tabsize, verbose=True):
         f.write(newtext)
     if verbose:
         print(filename)
+    return 1
 
 
 if __name__ == '__main__':
-    main()
+    raise SystemExit(main())