diff --git a/CHANGES.md b/CHANGES.md index a58ab7c25..e74544f48 100644 --- a/CHANGES.md +++ b/CHANGES.md @@ -1,8 +1,17 @@ -Synapse 1.46.0rc1 (2021-10-26) -============================== +Synapse 1.46.0 (2021-11-02) +=========================== The cause of the [performance regression affecting Synapse 1.44](https://github.com/matrix-org/synapse/issues/11049) has been identified and fixed. ([\#11177](https://github.com/matrix-org/synapse/issues/11177)) +Bugfixes +-------- + +- Fix a bug introduced in v1.46.0rc1 where URL previews of some XML documents would fail. ([\#11196](https://github.com/matrix-org/synapse/issues/11196)) + + +Synapse 1.46.0rc1 (2021-10-27) +============================== + Features -------- diff --git a/debian/changelog b/debian/changelog index ea96676f7..06e7a0862 100644 --- a/debian/changelog +++ b/debian/changelog @@ -1,3 +1,13 @@ +matrix-synapse-py3 (1.46.0) stable; urgency=medium + + [ Richard van der Hoff ] + * Compress debs with xz, to fix incompatibility of impish debs with reprepro. + + [ Synapse Packaging team ] + * New synapse release 1.46.0. + + -- Synapse Packaging team Tue, 02 Nov 2021 13:22:53 +0000 + matrix-synapse-py3 (1.46.0~rc1) stable; urgency=medium * New synapse release 1.46.0~rc1. diff --git a/debian/rules b/debian/rules index b9d490adc..5baf2475f 100755 --- a/debian/rules +++ b/debian/rules @@ -51,5 +51,11 @@ override_dh_shlibdeps: override_dh_virtualenv: ./debian/build_virtualenv +override_dh_builddeb: + # force the compression to xzip, to stop dpkg-deb on impish defaulting to zstd + # (which requires reprepro 5.3.0-1.3, which is currently only in 'experimental' in Debian: + # https://metadata.ftp-master.debian.org/changelogs/main/r/reprepro/reprepro_5.3.0-1.3_changelog) + dh_builddeb -- -Zxz + %: dh $@ --with python-virtualenv diff --git a/synapse/__init__.py b/synapse/__init__.py index 355b36fc6..5ef34bce4 100644 --- a/synapse/__init__.py +++ b/synapse/__init__.py @@ -47,7 +47,7 @@ try: except ImportError: pass -__version__ = "1.46.0rc1" +__version__ = "1.46.0" if bool(os.environ.get("SYNAPSE_TEST_PATCH_LOG_CONTEXTS", False)): # We import here so that we don't have to install a bunch of deps when diff --git a/synapse/rest/media/v1/preview_url_resource.py b/synapse/rest/media/v1/preview_url_resource.py index 9fb179fe0..99f7bee54 100644 --- a/synapse/rest/media/v1/preview_url_resource.py +++ b/synapse/rest/media/v1/preview_url_resource.py @@ -718,9 +718,12 @@ def decode_body( if not body: return None + # The idea here is that multiple encodings are tried until one works. + # Unfortunately the result is never used and then LXML will decode the string + # again with the found encoding. for encoding in get_html_media_encodings(body, content_type): try: - body_str = body.decode(encoding) + body.decode(encoding) except Exception: pass else: @@ -732,11 +735,11 @@ def decode_body( from lxml import etree # Create an HTML parser. - parser = etree.HTMLParser(recover=True, encoding="utf-8") + parser = etree.HTMLParser(recover=True, encoding=encoding) # Attempt to parse the body. Returns None if the body was successfully # parsed, but no tree was found. - return etree.fromstring(body_str, parser) + return etree.fromstring(body, parser) def _calc_og(tree: "etree.Element", media_uri: str) -> Dict[str, Optional[str]]: diff --git a/tests/test_preview.py b/tests/test_preview.py index 9a576f9a4..40b89fb2e 100644 --- a/tests/test_preview.py +++ b/tests/test_preview.py @@ -277,6 +277,21 @@ class CalcOgTestCase(unittest.TestCase): tree = decode_body(html, "http://example.com/test.html") self.assertIsNone(tree) + def test_xml(self): + """Test decoding XML and ensure it works properly.""" + # Note that the strip() call is important to ensure the xml tag starts + # at the initial byte. + html = b""" + + + + + FooSome text. + """.strip() + tree = decode_body(html, "http://example.com/test.html") + og = _calc_og(tree, "http://example.com/test.html") + self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."}) + def test_invalid_encoding(self): """An invalid character encoding should be ignored and treated as UTF-8, if possible.""" html = b"""