From 7a42a3a850ea6eb9fb5db92261b5e4d626d13a9d Mon Sep 17 00:00:00 2001
From: yap <yapisaac0@gmail.com>
Date: Mon, 24 Aug 2026 11:36:56 +0800
Subject: [PATCH] gh-156185: Ignore all ASCII whitespace in re \p{} property
 names

The docs promise property and value names are matched loosely
(whitespace not significant), but _normalize() only removed spaces.
Now tab/newline/cr/formfeed/vertical-tab are ignored as well.
---
 Lib/re/_properties.py                                     | 8 ++++----
 Lib/test/test_re.py                                       | 6 ++++++
 .../2026-08-24-10-30-00.gh-issue-156185.a1b2c3.rst        | 3 +++
 3 files changed, 13 insertions(+), 4 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2026-08-24-10-30-00.gh-issue-156185.a1b2c3.rst

diff --git a/Lib/re/_properties.py b/Lib/re/_properties.py
index 6310aa7fa88f955..09f03c05a88eaf2 100644
--- a/Lib/re/_properties.py
+++ b/Lib/re/_properties.py
@@ -209,11 +209,11 @@ def _analytic_ranges():
 
 
 def _normalize(name):
-    # Unicode property and value names are matched loosely: case, spaces,
-    # hyphens and underscores are not significant, and an initial "is" prefix
-    # is ignored (UAX #44 5.9, "Matching Rules", UAX44-LM3;
+    # Unicode property and value names are matched loosely: case, whitespace
+    # and hyphens and underscores are not significant, and an initial "is"
+    # prefix is ignored (UAX #44 5.9, "Matching Rules", UAX44-LM3;
     # https://www.unicode.org/reports/tr44/).
-    name = name.lower().replace("_", "").replace("-", "").replace(" ", "")
+    name = "".join(ch for ch in name.lower() if ch not in " \t\n\r\f\v_-")
     # Strip a leading "is", unless "is" is the whole name and so not a prefix
     # (e.g. the Line_Break value lb=IS).
     if name != "is":
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
index d086fd521e73b59..f7d89f652ec2bc6 100644
--- a/Lib/test/test_re.py
+++ b/Lib/test/test_re.py
@@ -929,6 +929,12 @@ def test_property_escapes(self):
         self.assertTrue(re.fullmatch(r'\p{General_Category=Lu}+', 'ABC'))
         self.assertTrue(re.fullmatch(r'\p{ lu }+', 'ABC'))
         self.assertTrue(re.fullmatch(r'\p{LU}+', 'ABC'))
+        # All ASCII whitespace is ignored, not just spaces (UAX44-LM3;
+        # gh-156185).
+        for ws in " \t\n\r\f\v":
+            with self.subTest(ws=ws):
+                self.assertTrue(re.fullmatch(r'\p{L%su}+' % ws, 'ABC'))
+                self.assertTrue(re.fullmatch(r'\p{gc%s=L%su}+' % (ws, ws), 'ABC'))
         # An initial "is" prefix is ignored (UAX44-LM3), on the property name
         # and on a gc value; "is" alone is not a prefix (cf. lb=IS).
         self.assertTrue(re.fullmatch(r'\p{isLu}+', 'ABC'))
diff --git a/Misc/NEWS.d/next/Library/2026-08-24-10-30-00.gh-issue-156185.a1b2c3.rst b/Misc/NEWS.d/next/Library/2026-08-24-10-30-00.gh-issue-156185.a1b2c3.rst
new file mode 100644
index 000000000000000..60a2b10f4c9d2cc
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-08-24-10-30-00.gh-issue-156185.a1b2c3.rst
@@ -0,0 +1,3 @@
+``\p{property}`` and ``\p{property=value}`` regex escapes now ignore all
+ASCII whitespace in property and value names, as documented, instead of
+only spaces.