From 4ec403ccbe32e965c24e93d8ec42d3ec83a2bd79 Mon Sep 17 00:00:00 2001
From: sundeep8967 <sundeep8967@gmail.com>
Date: Thu, 27 Aug 2026 13:16:32 +0530
Subject: [PATCH] gh-156379: Avoid materializing large sequence ranges in
 mailbox.MH.get_sequences

Intersect declared ranges directly with all_keys instead of enumerating the entire interval into a set.
---
 Lib/mailbox.py                                      | 13 +++++++------
 Lib/test/test_mailbox.py                            |  9 +++++++++
 .../2026-08-27-08-00-00.gh-issue-156379.abcdef.rst  |  1 +
 3 files changed, 17 insertions(+), 6 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2026-08-27-08-00-00.gh-issue-156379.abcdef.rst

diff --git a/Lib/mailbox.py b/Lib/mailbox.py
index 99426220154360..bad0776a267937 100644
--- a/Lib/mailbox.py
+++ b/Lib/mailbox.py
@@ -1236,14 +1236,15 @@ def get_sequences(self):
                     keys = set()
                     for spec in contents.split():
                         if spec.isdigit():
-                            keys.add(int(spec))
+                            key = int(spec)
+                            if key in all_keys:
+                                keys.add(key)
                         else:
                             start, stop = (int(x) for x in spec.split('-'))
-                            keys.update(range(start, stop + 1))
-                    results[name] = [key for key in sorted(keys) \
-                                         if key in all_keys]
-                    if len(results[name]) == 0:
-                        del results[name]
+                            if start <= stop:
+                                keys.update(k for k in all_keys if start <= k <= stop)
+                    if keys:
+                        results[name] = sorted(keys)
                 except ValueError:
                     raise FormatError('Invalid sequence specification: %s' %
                                       line.rstrip())
diff --git a/Lib/test/test_mailbox.py b/Lib/test/test_mailbox.py
index 019c699bff55c4..488a443555af1d 100644
--- a/Lib/test/test_mailbox.py
+++ b/Lib/test/test_mailbox.py
@@ -1408,6 +1408,15 @@ def test_sequences(self):
         self._box.set_sequences({'foo':[key0]})
         self.assertEqual(self._box.get_sequences(), {'foo':[key0]})
 
+    def test_large_sequence_range(self):
+        # gh-156379: Verify large sequence ranges do not materialize large intervals
+        msg0 = mailbox.MHMessage(self._template % 0)
+        key0 = self._box.add(msg0)
+        seq_path = os.path.join(self._path, '.mh_sequences')
+        with open(seq_path, 'w', encoding='ASCII') as f:
+            f.write(f'unseen: 1-10000000\nempty: 50-100\n')
+        self.assertEqual(self._box.get_sequences(), {'unseen': [key0]})
+
     def test_no_dot_mh_sequences_file(self):
         path = os.path.join(self._path, 'foo.bar')
         os.mkdir(path)
diff --git a/Misc/NEWS.d/next/Library/2026-08-27-08-00-00.gh-issue-156379.abcdef.rst b/Misc/NEWS.d/next/Library/2026-08-27-08-00-00.gh-issue-156379.abcdef.rst
new file mode 100644
index 00000000000000..509ae1f52eb204
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-08-27-08-00-00.gh-issue-156379.abcdef.rst
@@ -0,0 +1 @@
+Avoid materializing large sequence ranges in :meth:`mailbox.MH.get_sequences`.