liblore 0.8.1__tar.gz → 0.8.2__tar.gz

{liblore-0.8.1 → liblore-0.8.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: liblore
-Version: 0.8.1
+Version: 0.8.2
 Summary: Shared library for public-inbox / lore.kernel.org access
 Author-email: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
 License-Expression: GPL-2.0-or-later

{liblore-0.8.1 → liblore-0.8.2}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "liblore"
-version = "0.8.1"
+version = "0.8.2"
 description = "Shared library for public-inbox / lore.kernel.org access"
 readme = "README.md"
 requires-python = ">=3.9"

{liblore-0.8.1 → liblore-0.8.2}/src/liblore/__init__.py

@@ -6,7 +6,7 @@ import email.charset
 import email.policy
 from email.message import EmailMessage
 
-__version__ = '0.8.1'
+__version__ = '0.8.2'
 
 # Email policy used for parsing and serialising messages
 emlpolicy = email.policy.EmailPolicy(

{liblore-0.8.1 → liblore-0.8.2}/src/liblore/utils.py

@@ -394,6 +394,13 @@ def get_strict_thread(
                 for hval in msg.get_all(hdr_name, []):
                     raw_refs += re.findall(r'<([^>]+)>', str(hval))
 
+            # Some tooling emits messages that reference their own Message-ID.
+            # A self-reference is never meaningful for threading, so drop it
+            # before it can be mistaken for a parent.
+            if c_msgid in raw_refs:
+                logger.debug('Ignoring self-reference in %s', c_msgid)
+                raw_refs = [ref for ref in raw_refs if ref != c_msgid]
+
             # If noparent, pretend our target message has no references
             if noparent and msgid == c_msgid:
                 logger.info('Breaking thread to remove parents of %s', msgid)

{liblore-0.8.1 → liblore-0.8.2}/src/liblore.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: liblore
-Version: 0.8.1
+Version: 0.8.2
 Summary: Shared library for public-inbox / lore.kernel.org access
 Author-email: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
 License-Expression: GPL-2.0-or-later

{liblore-0.8.1 → liblore-0.8.2}/tests/test_thread.py

@@ -47,6 +47,57 @@ class TestGetStrictThread:
         assert 'child@x.com' in ids
         assert 'grandchild@x.com' in ids
 
+    def test_noparent_self_referencing_cover(self, make_msg: MsgFactory) -> None:
+        # Some tooling emits a cover letter whose In-Reply-To/References point
+        # at its own Message-ID.  Breaking the thread there must not treat the
+        # cover's own msgid as a reference to ignore, or every patch that
+        # legitimately replies to the cover gets dropped.
+        cover = make_msg(
+            msgid='cover@x.com',
+            subject='[PATCH 0/2] Something',
+            in_reply_to='cover@x.com',
+            references=['cover@x.com'],
+        )
+        patch1 = make_msg(
+            msgid='p1@x.com',
+            subject='[PATCH 1/2] First',
+            in_reply_to='cover@x.com',
+            references=['cover@x.com'],
+        )
+        patch2 = make_msg(
+            msgid='p2@x.com',
+            subject='[PATCH 2/2] Second',
+            in_reply_to='cover@x.com',
+            references=['cover@x.com'],
+        )
+        review = make_msg(
+            msgid='review@x.com',
+            subject='Re: [PATCH 2/2] Second',
+            in_reply_to='p2@x.com',
+            references=['cover@x.com', 'p2@x.com'],
+        )
+        want = {'cover@x.com', 'p1@x.com', 'p2@x.com', 'review@x.com'}
+        # Cover first in the list is the worst case: without the fix only the
+        # cover survives.  Ordering must not matter, so check both.
+        for msgs in ([cover, patch1, patch2, review], [patch1, cover, patch2, review]):
+            result = get_strict_thread(msgs, 'cover@x.com', noparent=True)
+            assert result is not None
+            assert {get_clean_msgid(m) for m in result} == want
+
+    def test_self_reference_is_not_a_parent(self, make_msg: MsgFactory) -> None:
+        # A self-referencing message must not be treated as a reply to itself,
+        # even without noparent: it is a thread root, and starting the walk at
+        # its child still has to pull it in.
+        root = make_msg(
+            msgid='root@x.com',
+            in_reply_to='root@x.com',
+            references=['root@x.com'],
+        )
+        reply = make_msg(msgid='reply@x.com', in_reply_to='root@x.com')
+        result = get_strict_thread([root, reply], 'reply@x.com')
+        assert result is not None
+        assert {get_clean_msgid(m) for m in result} == {'root@x.com', 'reply@x.com'}
+
     def test_references_chain(self, make_msg: MsgFactory) -> None:
         msg1 = make_msg(msgid='a@x.com')
         msg2 = make_msg(msgid='b@x.com', references=['a@x.com'])