From fd471b8785e81041493db16f3feba0a9dfce111e Mon Sep 17 00:00:00 2001
From: Eli Schwartz <eschwartz@gentoo.org>
Date: Wed, 15 Jan 2025 19:19:25 -0500
Subject: [PATCH] gh-118761: Reduce import time of gettext.py by delaying re
 import

gettext is often imported in programs that may not end up translating
anything. In fact, the `struct` module already has a delayed import when
parsing GNUTranslations to speed up the no .mo files case. The re module
is also used in the same situation, but behind a function chain only
called by GNUTranslations.

cache the compiled regex globally the first time it is used. The
finditer function can be converted to a method call on the compiled
object (it always could) which is slightly more efficient and necessary
for the conditional re import.
---
 Lib/gettext.py                                | 33 ++++++++++---------
 ...-01-15-19-16-50.gh-issue-118761.cbW2ZL.rst |  3 ++
 2 files changed, 21 insertions(+), 15 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2025-01-15-19-16-50.gh-issue-118761.cbW2ZL.rst

diff --git a/Lib/gettext.py b/Lib/gettext.py
index a0d81cf846a05c..4c1f9427459b14 100644
--- a/Lib/gettext.py
+++ b/Lib/gettext.py
@@ -48,7 +48,6 @@
 
 import operator
 import os
-import re
 import sys
 
 
@@ -70,22 +69,26 @@
 # https://www.gnu.org/software/gettext/manual/gettext.html#Plural-forms
 # http://git.savannah.gnu.org/cgit/gettext.git/tree/gettext-runtime/intl/plural.y
 
-_token_pattern = re.compile(r"""
-        (?P<WHITESPACES>[ \t]+)                    | # spaces and horizontal tabs
-        (?P<NUMBER>[0-9]+\b)                       | # decimal integer
-        (?P<NAME>n\b)                              | # only n is allowed
-        (?P<PARENTHESIS>[()])                      |
-        (?P<OPERATOR>[-*/%+?:]|[><!]=?|==|&&|\|\|) | # !, *, /, %, +, -, <, >,
-                                                     # <=, >=, ==, !=, &&, ||,
-                                                     # ? :
-                                                     # unary and bitwise ops
-                                                     # not allowed
-        (?P<INVALID>\w+|.)                           # invalid token
-    """, re.VERBOSE|re.DOTALL)
-
+_token_pattern = None
 
 def _tokenize(plural):
-    for mo in re.finditer(_token_pattern, plural):
+    global _token_pattern
+    if _token_pattern is None:
+        import re
+        _token_pattern = re.compile(r"""
+                (?P<WHITESPACES>[ \t]+)                    | # spaces and horizontal tabs
+                (?P<NUMBER>[0-9]+\b)                       | # decimal integer
+                (?P<NAME>n\b)                              | # only n is allowed
+                (?P<PARENTHESIS>[()])                      |
+                (?P<OPERATOR>[-*/%+?:]|[><!]=?|==|&&|\|\|) | # !, *, /, %, +, -, <, >,
+                                                             # <=, >=, ==, !=, &&, ||,
+                                                             # ? :
+                                                             # unary and bitwise ops
+                                                             # not allowed
+                (?P<INVALID>\w+|.)                           # invalid token
+            """, re.VERBOSE|re.DOTALL)
+
+    for mo in _token_pattern.finditer(plural):
         kind = mo.lastgroup
         if kind == 'WHITESPACES':
             continue
diff --git a/Misc/NEWS.d/next/Library/2025-01-15-19-16-50.gh-issue-118761.cbW2ZL.rst b/Misc/NEWS.d/next/Library/2025-01-15-19-16-50.gh-issue-118761.cbW2ZL.rst
new file mode 100644
index 00000000000000..0eef8777512dd8
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2025-01-15-19-16-50.gh-issue-118761.cbW2ZL.rst
@@ -0,0 +1,3 @@
+Reduce import time of :mod:`gettext` by up to ten times, by importing
+:mod:`re` on demand. In particular, ``re`` is no longer implicitly
+exposed as ``gettext.re``. Patch by Eli Schwartz.