mirror of
https://github.com/element-hq/synapse.git
synced 2026-10-06 16:47:36 +00:00
4f5524043c
Use `beautifulsoup4` instead of `lxml` for URL previews. This offers some nicer APIs when parsing HTML and avoids using `libxml`, [which is unmaintained](https://gitlab.gnome.org/GNOME/libxml2/-/commit/9c80a89af2fdf4f853892f84e46580f4902658ba). I haven’t done a full regression against commonly previewed sites, but I expect this will give similar (or better) results. beautiulsoup also handles decoding the charset for us, which is less custom code. --------- Co-authored-by: Andrew Morgan <andrew@amorgan.xyz> Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>
589 lines
22 KiB
Python
589 lines
22 KiB
Python
#
|
||
# This file is licensed under the Affero General Public License (AGPL) version 3.
|
||
#
|
||
# Copyright 2014-2016 OpenMarket Ltd
|
||
# Copyright (C) 2023 New Vector, Ltd
|
||
#
|
||
# This program is free software: you can redistribute it and/or modify
|
||
# it under the terms of the GNU Affero General Public License as
|
||
# published by the Free Software Foundation, either version 3 of the
|
||
# License, or (at your option) any later version.
|
||
#
|
||
# See the GNU Affero General Public License for more details:
|
||
# <https://www.gnu.org/licenses/agpl-3.0.html>.
|
||
#
|
||
# Originally licensed under the Apache License, Version 2.0:
|
||
# <http://www.apache.org/licenses/LICENSE-2.0>.
|
||
#
|
||
# [This file includes modifications made by New Vector Limited]
|
||
#
|
||
#
|
||
from unittest.mock import patch
|
||
|
||
from synapse.media.preview_html import (
|
||
decode_body,
|
||
parse_html_to_open_graph,
|
||
summarize_paragraphs,
|
||
)
|
||
|
||
from tests import unittest
|
||
|
||
try:
|
||
import bs4
|
||
except ImportError:
|
||
bs4 = None # type: ignore[assignment]
|
||
|
||
|
||
class SummarizeTestCase(unittest.TestCase):
|
||
if not bs4:
|
||
skip = "url preview feature requires beautifulsoup4"
|
||
|
||
def test_long_summarize(self) -> None:
|
||
example_paras = [
|
||
"""Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:
|
||
Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in
|
||
Troms county, Norway. The administrative centre of the municipality is
|
||
the city of Tromsø. Outside of Norway, Tromso and Tromsö are
|
||
alternative spellings of the city.Tromsø is considered the northernmost
|
||
city in the world with a population above 50,000. The most populous town
|
||
north of it is Alta, Norway, with a population of 14,272 (2013).""",
|
||
"""Tromsø lies in Northern Norway. The municipality has a population of
|
||
(2015) 72,066, but with an annual influx of students it has over 75,000
|
||
most of the year. It is the largest urban area in Northern Norway and the
|
||
third largest north of the Arctic Circle (following Murmansk and Norilsk).
|
||
Most of Tromsø, including the city centre, is located on the island of
|
||
Tromsøya, 350 kilometres (217 mi) north of the Arctic Circle. In 2012,
|
||
Tromsøya had a population of 36,088. Substantial parts of the urban area
|
||
are also situated on the mainland to the east, and on parts of Kvaløya—a
|
||
large island to the west. Tromsøya is connected to the mainland by the Tromsø
|
||
Bridge and the Tromsøysund Tunnel, and to the island of Kvaløya by the
|
||
Sandnessund Bridge. Tromsø Airport connects the city to many destinations
|
||
in Europe. The city is warmer than most other places located on the same
|
||
latitude, due to the warming effect of the Gulf Stream.""",
|
||
"""The city centre of Tromsø contains the highest number of old wooden
|
||
houses in Northern Norway, the oldest house dating from 1789. The Arctic
|
||
Cathedral, a modern church from 1965, is probably the most famous landmark
|
||
in Tromsø. The city is a cultural centre for its region, with several
|
||
festivals taking place in the summer. Some of Norway's best-known
|
||
musicians, Torbjørn Brundtland and Svein Berge of the electronica duo
|
||
Röyksopp and Lene Marlin grew up and started their careers in Tromsø.
|
||
Noted electronic musician Geir Jenssen also hails from Tromsø.""",
|
||
]
|
||
|
||
desc = summarize_paragraphs(example_paras, min_size=200, max_size=500)
|
||
|
||
self.assertEqual(
|
||
desc,
|
||
"Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:"
|
||
" Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in"
|
||
" Troms county, Norway. The administrative centre of the municipality is"
|
||
" the city of Tromsø. Outside of Norway, Tromso and Tromsö are"
|
||
" alternative spellings of the city.Tromsø is considered the northernmost"
|
||
" city in the world with a population above 50,000. The most populous town"
|
||
" north of it is Alta, Norway, with a population of 14,272 (2013).",
|
||
)
|
||
|
||
desc = summarize_paragraphs(example_paras[1:], min_size=200, max_size=500)
|
||
|
||
self.assertEqual(
|
||
desc,
|
||
"Tromsø lies in Northern Norway. The municipality has a population of"
|
||
" (2015) 72,066, but with an annual influx of students it has over 75,000"
|
||
" most of the year. It is the largest urban area in Northern Norway and the"
|
||
" third largest north of the Arctic Circle (following Murmansk and Norilsk)."
|
||
" Most of Tromsø, including the city centre, is located on the island of"
|
||
" Tromsøya, 350 kilometres (217 mi) north of the Arctic Circle. In 2012,"
|
||
" Tromsøya had a population of 36,088. Substantial parts of the urban…",
|
||
)
|
||
|
||
def test_short_summarize(self) -> None:
|
||
example_paras = [
|
||
"Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:"
|
||
" Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in"
|
||
" Troms county, Norway.",
|
||
"Tromsø lies in Northern Norway. The municipality has a population of"
|
||
" (2015) 72,066, but with an annual influx of students it has over 75,000"
|
||
" most of the year.",
|
||
"The city centre of Tromsø contains the highest number of old wooden"
|
||
" houses in Northern Norway, the oldest house dating from 1789. The Arctic"
|
||
" Cathedral, a modern church from 1965, is probably the most famous landmark"
|
||
" in Tromsø.",
|
||
]
|
||
|
||
desc = summarize_paragraphs(example_paras, min_size=200, max_size=500)
|
||
|
||
self.assertEqual(
|
||
desc,
|
||
"Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:"
|
||
" Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in"
|
||
" Troms county, Norway.\n"
|
||
"\n"
|
||
"Tromsø lies in Northern Norway. The municipality has a population of"
|
||
" (2015) 72,066, but with an annual influx of students it has over 75,000"
|
||
" most of the year.",
|
||
)
|
||
|
||
def test_small_then_large_summarize(self) -> None:
|
||
example_paras = [
|
||
"Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:"
|
||
" Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in"
|
||
" Troms county, Norway.",
|
||
"Tromsø lies in Northern Norway. The municipality has a population of"
|
||
" (2015) 72,066, but with an annual influx of students it has over 75,000"
|
||
" most of the year."
|
||
" The city centre of Tromsø contains the highest number of old wooden"
|
||
" houses in Northern Norway, the oldest house dating from 1789. The Arctic"
|
||
" Cathedral, a modern church from 1965, is probably the most famous landmark"
|
||
" in Tromsø.",
|
||
]
|
||
|
||
desc = summarize_paragraphs(example_paras, min_size=200, max_size=500)
|
||
self.assertEqual(
|
||
desc,
|
||
"Tromsø (Norwegian pronunciation: [ˈtrʊmsœ] ( listen); Northern Sami:"
|
||
" Romsa; Finnish: Tromssa[2] Kven: Tromssa) is a city and municipality in"
|
||
" Troms county, Norway.\n"
|
||
"\n"
|
||
"Tromsø lies in Northern Norway. The municipality has a population of"
|
||
" (2015) 72,066, but with an annual influx of students it has over 75,000"
|
||
" most of the year. The city centre of Tromsø contains the highest number"
|
||
" of old wooden houses in Northern Norway, the oldest house dating from"
|
||
" 1789. The Arctic Cathedral, a modern church from…",
|
||
)
|
||
|
||
|
||
class OpenGraphFromHtmlTestCase(unittest.TestCase):
|
||
if not bs4:
|
||
skip = "url preview feature requires beautifulsoup4"
|
||
|
||
def test_simple(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<head><title>Foo</title></head>
|
||
<body>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
|
||
|
||
def test_comment(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<head><title>Foo</title></head>
|
||
<body>
|
||
<!-- HTML comment -->
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
|
||
|
||
def test_comment2(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<head><title>Foo</title></head>
|
||
<body>
|
||
Some text.
|
||
<!-- HTML comment -->
|
||
Some more text.
|
||
<p>Text</p>
|
||
More text
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": "Foo",
|
||
"og:description": "Some text.\n\nSome more text.\n\nText\n\nMore text",
|
||
},
|
||
)
|
||
|
||
def test_script(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<head><title>Foo</title></head>
|
||
<body>
|
||
<script> (function() {})() </script>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
|
||
|
||
def test_missing_title(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<body>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": None, "og:description": "Some text."})
|
||
|
||
# Another variant is a title with no content.
|
||
html = b"""
|
||
<html>
|
||
<head><title></title></head>
|
||
<body>
|
||
<h1>Title</h1>
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
|
||
self.assertEqual(og, {"og:title": "Title", "og:description": "Title"})
|
||
|
||
def test_h1_as_title(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<meta property="og:description" content="Some text."/>
|
||
<body>
|
||
<h1>Title</h1>
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": "Title", "og:description": "Some text."})
|
||
|
||
def test_empty_description(self) -> None:
|
||
"""Description tags with empty content should be ignored."""
|
||
html = b"""
|
||
<html>
|
||
<meta property="og:description" content=""/>
|
||
<meta property="og:description"/>
|
||
<meta name="description" content=""/>
|
||
<meta name="description"/>
|
||
<meta name="description" content="Finally!"/>
|
||
<body>
|
||
<h1>Title</h1>
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
|
||
self.assertEqual(og, {"og:title": "Title", "og:description": "Finally!"})
|
||
|
||
def test_missing_title_and_broken_h1(self) -> None:
|
||
html = b"""
|
||
<html>
|
||
<body>
|
||
<h1><a href="foo"/></h1>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
|
||
self.assertEqual(og, {"og:title": None, "og:description": "Some text."})
|
||
|
||
def test_empty(self) -> None:
|
||
"""Test a body with no data in it."""
|
||
html = b""
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
self.assertIsNone(soup)
|
||
|
||
def test_no_soup(self) -> None:
|
||
"""A valid body with no tree in it."""
|
||
with patch(
|
||
"bs4.BeautifulSoup",
|
||
side_effect=bs4.ParserRejectedMarkup("Invalid markup"),
|
||
):
|
||
self.assertIsNone(decode_body(b"<html></html>", "https://example.com/"))
|
||
|
||
def test_xml(self) -> None:
|
||
"""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"""
|
||
<?xml version="1.0" encoding="UTF-8"?>
|
||
|
||
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
||
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
|
||
<head><title>Foo</title></head><body>Some text.</body></html>
|
||
""".strip()
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
|
||
|
||
def test_invalid_encoding2(self) -> None:
|
||
"""A body which doesn't match the sent character encoding."""
|
||
# Note that this contains an invalid UTF-8 sequence in the title.
|
||
html = b"""
|
||
<html>
|
||
<head><title>\xff\xff Foo</title></head>
|
||
<body>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
self.assertEqual(og, {"og:title": "˙˙ Foo", "og:description": "Some text."})
|
||
|
||
def test_windows_1252(self) -> None:
|
||
"""A body which uses cp1252, but doesn't declare that."""
|
||
html = b"""
|
||
<html>
|
||
<head><title>\xf3</title></head>
|
||
<body>
|
||
Some text.
|
||
</body>
|
||
</html>
|
||
"""
|
||
soup = decode_body(html, "http://example.com/test.html")
|
||
assert soup is not None
|
||
og = parse_html_to_open_graph(soup)
|
||
self.assertIn("og:title", og)
|
||
og.pop("og:title")
|
||
self.assertEqual(og, {"og:description": "Some text."})
|
||
|
||
def test_image(self) -> None:
|
||
"""Test that ensures an image can be pulled from the HTML."""
|
||
|
||
# Tags is a list of two-element tuples: the HTML tag and the expected image which
|
||
# is chosen.
|
||
#
|
||
# They're in a particular order such that we can prove "higher" priority tags
|
||
# are parsed out first, e.g. OpenGraph, then meta tags, then images of a certain
|
||
# height/width, favicons, etc.
|
||
tags = [
|
||
(
|
||
b"""<meta property="og:image" content="https://example.com/meta-prop.png">""",
|
||
"meta-prop",
|
||
),
|
||
(
|
||
b"""<meta itemprop="IMAGE" content="https://example.com/meta-IMAGE.png">""",
|
||
"meta-IMAGE",
|
||
),
|
||
(
|
||
b"""<meta itemprop="image" content="https://example.com/meta-image.png">""",
|
||
"meta-image",
|
||
),
|
||
(b"""<img src="https://example.com/img-no-width-no-height.png">""", "img"),
|
||
(
|
||
b"""<img src="https://example.com/img-no-height.png" width="100">""",
|
||
"img",
|
||
),
|
||
(
|
||
b"""<img src="https://example.com/img-no-width.png" height="100">""",
|
||
"img",
|
||
),
|
||
(
|
||
b"""<img src="https://example.com/img-small.png" width="100" height="100">""",
|
||
"img",
|
||
),
|
||
(
|
||
b"""<img src="https://example.com/img.png" width="200" height="100">""",
|
||
"img",
|
||
),
|
||
# Put this image again since if it is the *only* image it will be used.
|
||
(
|
||
b"""<img src="https://example.com/img-no-width-no-height.png">""",
|
||
"img-no-width-no-height",
|
||
),
|
||
(
|
||
b"""<link rel="icon" href="https://example.com/favicon.png">""",
|
||
"favicon",
|
||
),
|
||
]
|
||
|
||
while tags:
|
||
html = b"<html>" + b"".join(t[0] for t in tags) + b"</html>"
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": None,
|
||
"og:image": f"https://example.com/{tags[0][1]}.png",
|
||
},
|
||
)
|
||
|
||
# Remove the highest remaining priority item.
|
||
tags.pop(0)
|
||
|
||
def test_image_bad_height_width(self) -> None:
|
||
"""A bad height/width should be ignored."""
|
||
|
||
html = b"""<html>
|
||
<img src="http://example.com/no-height-width.png">
|
||
<img src="http://example.com/bad-height-width.png" height="a" width="a">
|
||
</html>"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": None,
|
||
"og:image": "http://example.com/no-height-width.png",
|
||
},
|
||
)
|
||
|
||
def test_twitter_tag(self) -> None:
|
||
"""Twitter card tags should be used if nothing else is available."""
|
||
html = b"""
|
||
<html>
|
||
<meta name="twitter:card" content="summary">
|
||
<meta name="twitter:description" content="Description">
|
||
<meta name="twitter:site" content="@matrixdotorg">
|
||
<meta name="twitter:image" content="https://example.com/test.png">
|
||
</html>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "Description",
|
||
"og:site_name": "@matrixdotorg",
|
||
"og:image": "https://example.com/test.png",
|
||
},
|
||
)
|
||
|
||
# But they shouldn't override Open Graph values.
|
||
html = b"""
|
||
<html>
|
||
<meta name="twitter:card" content="summary">
|
||
<meta name="twitter:description" content="Description">
|
||
<meta property="og:description" content="Real Description">
|
||
<meta name="twitter:site" content="@matrixdotorg">
|
||
<meta property="og:site_name" content="matrix.org">
|
||
<meta name="twitter:image" content="https://example.com/bad.png">
|
||
<meta property="og:image" content="https://example.com/good.png">
|
||
</html>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "Real Description",
|
||
"og:site_name": "matrix.org",
|
||
"og:image": "https://example.com/good.png",
|
||
},
|
||
)
|
||
|
||
def test_extended_opengraph(self) -> None:
|
||
"""Ensure we pull in profile and article data from opengraph."""
|
||
html = b"""
|
||
<html>
|
||
<meta property="og:description" content="My description" />
|
||
<meta property="profile:username" content="myname">
|
||
<meta property="article:published_time" content="2026-04-07T10:07:37Z">
|
||
</html>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "My description",
|
||
"profile:username": "myname",
|
||
"article:published_time": "2026-04-07T10:07:37Z",
|
||
},
|
||
)
|
||
|
||
def test_nested_nodes(self) -> None:
|
||
"""A body with some nested nodes. Tests that we iterate over children
|
||
in the right order (and don't reverse the order of the text)."""
|
||
html = b"""
|
||
<a href="somewhere">Welcome <b>the bold <u>and underlined text <svg>
|
||
with a cheeky SVG</svg></u> and <strong>some</strong> tail text</b></a>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "Welcome\n\nthe bold\n\nand underlined text\n\nand\n\nsome\n\ntail text",
|
||
},
|
||
)
|
||
|
||
def test_ignored_tags(self) -> None:
|
||
"""Test ignored elements."""
|
||
html = b"""
|
||
<header>Foo bar</header>
|
||
<div>A real description</div>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "A real description",
|
||
},
|
||
)
|
||
|
||
def test_aria_tags(self) -> None:
|
||
"""Test ignored elements."""
|
||
html = b"""
|
||
<div role="menu">Foo bar</div>
|
||
<div>A real description</div>
|
||
"""
|
||
tree = decode_body(html, "http://example.com/test.html")
|
||
assert tree is not None
|
||
og = parse_html_to_open_graph(tree)
|
||
self.assertEqual(
|
||
og,
|
||
{
|
||
"og:title": None,
|
||
"og:description": "A real description",
|
||
},
|
||
)
|