From 71d81e4a6041b9da86e6cae0e48e1332b3b5b5ed Mon Sep 17 00:00:00 2001
From: Matt Clay <matt@mystile.com>
Date: Wed, 17 Jan 2024 11:19:37 -0800
Subject: [PATCH] Add no-unwanted-characters sanity test for core (#82557)

---
 .../code-smell/no-unwanted-characters.json    |  4 +++
 .../code-smell/no-unwanted-characters.py      | 27 +++++++++++++++++++
 test/sanity/ignore.txt                        |  4 +++
 3 files changed, 35 insertions(+)
 create mode 100644 test/sanity/code-smell/no-unwanted-characters.json
 create mode 100644 test/sanity/code-smell/no-unwanted-characters.py

diff --git a/test/sanity/code-smell/no-unwanted-characters.json b/test/sanity/code-smell/no-unwanted-characters.json
new file mode 100644
index 00000000000..5648429eb04
--- /dev/null
+++ b/test/sanity/code-smell/no-unwanted-characters.json
@@ -0,0 +1,4 @@
+{
+    "text": true,
+    "output": "path-line-column-message"
+}
diff --git a/test/sanity/code-smell/no-unwanted-characters.py b/test/sanity/code-smell/no-unwanted-characters.py
new file mode 100644
index 00000000000..26e5912c546
--- /dev/null
+++ b/test/sanity/code-smell/no-unwanted-characters.py
@@ -0,0 +1,27 @@
+"""Disallow use of unwanted Unicode characters."""
+from __future__ import annotations
+
+import re
+import sys
+
+
+def main():
+    """Main entry point."""
+    for path in sys.argv[1:] or sys.stdin.read().splitlines():
+        with open(path, 'rb') as path_fd:
+            for line, text in enumerate(path_fd.readlines()):
+                try:
+                    text = text.decode('utf-8')
+                except UnicodeDecodeError as ex:
+                    print('%s:%d:%d: UnicodeDecodeError: %s' % (path, line + 1, ex.start + 1, ex))
+                    continue
+
+                match = re.search('(\u00a0)', text)
+
+                if match:
+                    print('%s:%d:%d: use an ASCII space instead of a Unicode no-break space' % (
+                        path, line + 1, match.start(1) + 1))
+
+
+if __name__ == '__main__':
+    main()
diff --git a/test/sanity/ignore.txt b/test/sanity/ignore.txt
index 57730176bfa..fe7a4ea358b 100644
--- a/test/sanity/ignore.txt
+++ b/test/sanity/ignore.txt
@@ -114,15 +114,19 @@ test/integration/targets/incidental_win_reboot/templates/post_reboot.ps1 pslint!
 test/integration/targets/json_cleanup/library/bad_json shebang
 test/integration/targets/lookup_csvfile/files/crlf.csv line-endings
 test/integration/targets/lookup_ini/lookup-8859-15.ini no-smart-quotes
+test/integration/targets/lookup_ini/lookup-8859-15.ini no-unwanted-characters
 test/integration/targets/module_precedence/lib_with_extension/a.ini shebang
 test/integration/targets/module_precedence/lib_with_extension/ping.ini shebang
 test/integration/targets/module_precedence/roles_with_extension/foo/library/a.ini shebang
 test/integration/targets/module_precedence/roles_with_extension/foo/library/ping.ini shebang
 test/integration/targets/old_style_modules_posix/library/helloworld.sh shebang
 test/integration/targets/template/files/encoding_1252_utf-8.expected no-smart-quotes
+test/integration/targets/template/files/encoding_1252_utf-8.expected no-unwanted-characters
 test/integration/targets/template/files/encoding_1252_windows-1252.expected no-smart-quotes
+test/integration/targets/template/files/encoding_1252_windows-1252.expected no-unwanted-characters
 test/integration/targets/template/files/foo.dos.txt line-endings
 test/integration/targets/template/templates/encoding_1252.j2 no-smart-quotes
+test/integration/targets/template/templates/encoding_1252.j2 no-unwanted-characters
 test/integration/targets/unicode/unicode.yml no-smart-quotes
 test/integration/targets/windows-minimal/library/win_ping_syntax_error.ps1 pslint!skip
 test/integration/targets/win_exec_wrapper/library/test_fail.ps1 pslint:PSCustomUseLiteralPath