1"""Tests for `chives.urls`."""
2
3from pathlib import Path
4
5import pytest
6from vcr.cassette import Cassette
7
8from chives.urls import (
9    clean_youtube_url,
10    is_mastodon_host,
11    is_url_safe,
12    parse_mastodon_post_url,
13    parse_tumblr_post_url,
14)
15
16
17@pytest.mark.parametrize(
18    "url, cleaned_url",
19    [
20        (
21            "https://www.youtube.com/watch?v=2OHPPSew2nY&list=WL&index=6&t=193s",
22            "https://www.youtube.com/watch?v=2OHPPSew2nY",
23        ),
24        (
25            "https://www.youtube.com/watch?v=2OHPPSew2nY",
26            "https://www.youtube.com/watch?v=2OHPPSew2nY",
27        ),
28        (
29            "https://www.youtube.com/watch?v=WiIi7STG3e0&start_radio=1",
30            "https://www.youtube.com/watch?v=WiIi7STG3e0",
31        ),
32        (
33            "https://www.youtube.com/watch?app=desktop&v=vQ4ATsyD4hA",
34            "https://www.youtube.com/watch?v=vQ4ATsyD4hA",
35        ),
36    ],
37)
38def test_clean_youtube_url(url: str, cleaned_url: str) -> None:
39    """
40    All the query parameters get stripped from YouTube URLs correctly.
41    """
42    assert clean_youtube_url(url) == cleaned_url
43
44
45@pytest.mark.parametrize(
46    "url, server, acct, post_id",
47    [
48        pytest.param(
49            "https://iconfactory.world/@Iconfactory/115650922400392083",
50            "iconfactory.world",
51            "Iconfactory",
52            "115650922400392083",
53            id="iconfactory",
54        ),
55        pytest.param(
56            "https://social.alexwlchan.net/@chris__martin@functional.cafe/113369395383537892",
57            "functional.cafe",
58            "chris__martin",
59            "113369395366414375",
60            id="alexwlchan_redirect",
61        ),
62        pytest.param(
63            "https://social.alexwlchan.net/@alex/116300317590482708",
64            "social.alexwlchan.net",
65            "alex",
66            "116300317590482708",
67            id="alexwlchan_self",
68        ),
69    ],
70)
71def test_parse_mastodon_post_url(
72    vcr_cassette: Cassette, url: str, server: str, acct: str, post_id: str
73) -> None:
74    """
75    Mastodon post URLs are parsed correctly.
76    """
77    assert parse_mastodon_post_url(url) == (server, acct, post_id)
78
79
80@pytest.mark.parametrize(
81    "url, error",
82    [
83        pytest.param(
84            "https://mastodon.social/", "Cannot parse Mastodon URL", id="no_path"
85        ),
86        pytest.param(
87            "https://mastodon.social/about",
88            "Cannot parse Mastodon URL",
89            id="no_post_id",
90        ),
91        pytest.param(
92            "https://mastodon.social/about/subdir", "Cannot find `acct`", id="no_acct"
93        ),
94        pytest.param(
95            "https://mastodon.social/@example/about",
96            "Mastodon post ID is not numeric",
97            id="non_numeric_post_id",
98        ),
99        pytest.param(
100            "https://social.alexwlchan.net/@does@not.exist/123",
101            "Cannot parse Mastodon URL",
102            id="alexwlchan_does_not_exist",
103        ),
104    ],
105)
106def test_parse_mastodon_post_url_errors(
107    vcr_cassette: Cassette, url: str, error: str
108) -> None:
109    """
110    parse_mastodon_post_url returns a useful error if it can't parse the URL.
111    """
112    with pytest.raises(ValueError, match=error):
113        parse_mastodon_post_url(url)
114
115
116@pytest.mark.parametrize(
117    "url, blog_identifier, post_id",
118    [
119        (
120            "https://www.tumblr.com/kynvillingur/792473255236796416/",
121            "kynvillingur",
122            "792473255236796416",
123        ),
124        (
125            "https://cut3panda.tumblr.com/post/94093772689/for-some-people-the-more-you-get-to-know-them",
126            "cut3panda",
127            "94093772689",
128        ),
129    ],
130)
131def test_parse_tumblr_post_url(url: str, blog_identifier: str, post_id: str) -> None:
132    """
133    Tumblr URLs are parsed correctly.
134    """
135    assert parse_tumblr_post_url(url) == (blog_identifier, post_id)
136
137
138@pytest.mark.parametrize(
139    "url",
140    [
141        "https://www.tumblr.com/",
142        "https://www.tumblr.com/staff/",
143        "https://staff.tumblr.com/",
144        "https://www.example.com/",
145    ],
146)
147def test_parse_bad_tumblr_url(url: str) -> None:
148    """
149    Parsing a non-Tumblr URL throws a ValueError.
150    """
151    with pytest.raises(ValueError, match="Cannot parse Tumblr URL"):
152        parse_tumblr_post_url(url)
153
154
155class TestIsMastodonHost:
156    """
157    Tests for `is_mastodon_host`.
158    """
159
160    @pytest.mark.parametrize(
161        "host", ["mastodon.social", "hachyderm.io", "social.jvns.ca"]
162    )
163    def test_mastodon_servers(self, host: str, vcr_cassette: Cassette) -> None:
164        """
165        It correctly identifies real Mastodon servers.
166        """
167        assert is_mastodon_host(host)
168
169    @pytest.mark.parametrize(
170        "host",
171        [
172            # These are regular Internet websites which don't expose
173            # the /.well-known/nodeinfo endpoint
174            "example.com",
175            "tailscale.com",
176            #
177            # PeerTube exposes /.well-known/nodeinfo, but it's running
178            # different software.
179            "peertube.tv",
180            #
181            # A website with a known bad SSL certificate, which is assumed
182            # not to be a Mastodon host because we can't connect to it.
183            "expired.badssl.com",
184        ],
185    )
186    def test_non_mastodon_servers(self, host: str, vcr_cassette: Cassette) -> None:
187        """
188        Other websites are not Mastodon servers.
189        """
190        assert not is_mastodon_host(host)
191
192
193class TestIsUrlSafe:
194    """
195    Tests for `is_url_safe`.
196    """
197
198    @pytest.mark.parametrize("path", ["example.txt", Path("a/b/cat.jpg")])
199    def test_safe(self, path: str | Path) -> None:
200        """Paths which are URL safe."""
201        assert is_url_safe(path)
202
203    @pytest.mark.parametrize("path", ["is it?", Path("cat%c.jpg"), "a#b"])
204    def test_unsafe(self, path: str | Path) -> None:
205        """Paths which are not URL safe."""
206        assert not is_url_safe(path)