feat: metadata enrichment for alerts and subscription plugins

Alert backends now populate structured `extra` field with engagement
metrics (views, stars, votes, etc.) instead of embedding them in titles.
Subscription plugins show richer announcements: Twitch viewer counts,
YouTube views/likes/dates, RSS published dates.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
user
2026-02-19 10:00:17 +01:00
parent c3b19feb0f
commit 1fe7da9ed8
10 changed files with 614 additions and 52 deletions

View File

@@ -19,6 +19,7 @@ _spec.loader.exec_module(_mod)
from plugins.alert import ( # noqa: E402
_MAX_SEEN,
_compact_num,
_delete,
_errors,
_extract_videos,
@@ -27,6 +28,7 @@ from plugins.alert import ( # noqa: E402
_pollers,
_restore,
_save,
_save_result,
_search_searx,
_search_twitch,
_search_youtube,
@@ -179,6 +181,10 @@ class _FakeBot:
async def reply(self, message, text: str) -> None:
self.replied.append(text)
async def long_reply(self, message, lines, *, label: str = "") -> None:
for line in lines:
self.replied.append(line)
def _is_admin(self, message) -> bool:
return self._admin
@@ -223,9 +229,9 @@ def _fake_tw(keyword):
"""Fake Twitch backend returning two results (keyword in title)."""
return [
{"id": "stream:tw1", "title": "TW test Stream 1",
"url": "https://twitch.tv/user1", "extra": ""},
"url": "https://twitch.tv/user1", "extra": "500 viewers"},
{"id": "vod:tw2", "title": "TW test VOD 1",
"url": "https://twitch.tv/videos/tw2", "extra": ""},
"url": "https://twitch.tv/videos/tw2", "extra": "1k views"},
]
@@ -322,6 +328,36 @@ class TestTruncate:
assert not result.endswith(" ...")
# ---------------------------------------------------------------------------
# TestCompactNum
# ---------------------------------------------------------------------------
class TestCompactNum:
def test_zero(self):
assert _compact_num(0) == "0"
def test_small(self):
assert _compact_num(999) == "999"
def test_one_k(self):
assert _compact_num(1000) == "1k"
def test_one_point_five_k(self):
assert _compact_num(1500) == "1.5k"
def test_one_m(self):
assert _compact_num(1000000) == "1M"
def test_two_point_five_m(self):
assert _compact_num(2500000) == "2.5M"
def test_exact_boundary(self):
assert _compact_num(10000) == "10k"
def test_large_millions(self):
assert _compact_num(12300000) == "12.3M"
# ---------------------------------------------------------------------------
# TestExtractVideos
# ---------------------------------------------------------------------------
@@ -873,6 +909,13 @@ class TestPollOnce:
assert "[poll/yt/" in actions[0]
assert "[poll/tw/" in actions[2]
assert "[poll/sx/" in actions[4]
# Twitch fakes have extra metadata; verify it appears in titles
tw_titles = [s for t, s in bot.sent if t == "#test" and "TW test" in s]
assert any("| 500 viewers" in t for t in tw_titles)
assert any("| 1k views" in t for t in tw_titles)
# YouTube fakes have no extra; verify no pipe suffix
yt_titles = [s for t, s in bot.sent if t == "#test" and "YT test" in s]
assert all("|" not in t for t in yt_titles)
asyncio.run(inner())
@@ -1239,3 +1282,103 @@ class TestSearchSearx:
with patch("urllib.request.urlopen", side_effect=ConnectionError("fail")):
results = _search_searx("test")
assert results == []
# ---------------------------------------------------------------------------
# TestExtraInHistory
# ---------------------------------------------------------------------------
class TestExtraInHistory:
def test_history_shows_extra(self):
"""History output includes | extra suffix when extra is non-empty."""
_clear()
bot = _FakeBot()
data = {
"keyword": "test", "name": "hist", "channel": "#test",
"interval": 300, "seen": {}, "last_poll": "", "last_error": "",
}
_save(bot, "#test:hist", data)
# Insert a result with extra metadata
_save_result("#test", "hist", "hn", {
"id": "h1", "title": "Cool HN Post", "url": "https://hn.example.com/1",
"date": "2026-01-15", "extra": "42pt 10c",
})
async def inner():
await cmd_alert(bot, _msg("!alert history hist"))
assert len(bot.replied) >= 1
found = any("| 42pt 10c" in line for line in bot.replied)
assert found, f"Expected extra in history, got: {bot.replied}"
asyncio.run(inner())
def test_history_no_extra(self):
"""History output has no pipe when extra is empty."""
_clear()
bot = _FakeBot()
data = {
"keyword": "test", "name": "hist2", "channel": "#test",
"interval": 300, "seen": {}, "last_poll": "", "last_error": "",
}
_save(bot, "#test:hist2", data)
_save_result("#test", "hist2", "yt", {
"id": "y1", "title": "Plain Video", "url": "https://yt.example.com/1",
"date": "", "extra": "",
})
async def inner():
await cmd_alert(bot, _msg("!alert history hist2"))
assert len(bot.replied) >= 1
assert all("|" not in line for line in bot.replied)
asyncio.run(inner())
# ---------------------------------------------------------------------------
# TestExtraInInfo
# ---------------------------------------------------------------------------
class TestExtraInInfo:
def test_info_shows_extra(self):
"""Info output includes | extra suffix when extra is non-empty."""
_clear()
bot = _FakeBot()
data = {
"keyword": "test", "name": "inf", "channel": "#test",
"interval": 300, "seen": {}, "last_poll": "", "last_error": "",
}
_save(bot, "#test:inf", data)
short_id = _save_result("#test", "inf", "gh", {
"id": "g1", "title": "cool/repo: A cool project",
"url": "https://github.com/cool/repo",
"date": "2026-01-10", "extra": "42* 5fk",
})
async def inner():
await cmd_alert(bot, _msg(f"!alert info {short_id}"))
assert len(bot.replied) >= 1
assert "| 42* 5fk" in bot.replied[0]
asyncio.run(inner())
def test_info_no_extra(self):
"""Info output has no pipe when extra is empty."""
_clear()
bot = _FakeBot()
data = {
"keyword": "test", "name": "inf2", "channel": "#test",
"interval": 300, "seen": {}, "last_poll": "", "last_error": "",
}
_save(bot, "#test:inf2", data)
short_id = _save_result("#test", "inf2", "yt", {
"id": "y2", "title": "Some Video",
"url": "https://youtube.com/watch?v=y2",
"date": "", "extra": "",
})
async def inner():
await cmd_alert(bot, _msg(f"!alert info {short_id}"))
assert len(bot.replied) >= 1
assert "|" not in bot.replied[0]
asyncio.run(inner())

View File

@@ -25,6 +25,7 @@ from plugins.rss import ( # noqa: E402
_feeds,
_load,
_parse_atom,
_parse_date,
_parse_feed,
_parse_rss,
_poll_once,
@@ -52,16 +53,19 @@ RSS_FEED = b"""\
<guid>item-1</guid>
<title>First Post</title>
<link>https://example.com/1</link>
<pubDate>Mon, 10 Feb 2026 09:00:00 +0000</pubDate>
</item>
<item>
<guid>item-2</guid>
<title>Second Post</title>
<link>https://example.com/2</link>
<pubDate>Tue, 11 Feb 2026 14:30:00 +0000</pubDate>
</item>
<item>
<guid>item-3</guid>
<title>Third Post</title>
<link>https://example.com/3</link>
<pubDate>Wed, 12 Feb 2026 08:00:00 +0000</pubDate>
</item>
</channel>
</rss>
@@ -88,11 +92,13 @@ ATOM_FEED = b"""\
<id>atom-1</id>
<title>Atom First</title>
<link href="https://example.com/a1"/>
<published>2026-02-15T10:00:00Z</published>
</entry>
<entry>
<id>atom-2</id>
<title>Atom Second</title>
<link href="https://example.com/a2"/>
<published>2026-02-16T15:30:00Z</published>
</entry>
</feed>
"""
@@ -333,6 +339,20 @@ class TestParseRSS:
assert items[0]["title"] == "First Post"
assert items[0]["link"] == "https://example.com/1"
def test_parses_pubdate(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(RSS_FEED)
_, items = _parse_rss(root)
assert items[0]["date"] == "2026-02-10"
assert items[1]["date"] == "2026-02-11"
assert items[2]["date"] == "2026-02-12"
def test_no_pubdate_empty_string(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(RSS_NO_GUID)
_, items = _parse_rss(root)
assert items[0]["date"] == ""
def test_fallback_to_link_as_id(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(RSS_NO_GUID)
@@ -364,6 +384,19 @@ class TestParseAtom:
assert items[0]["title"] == "Atom First"
assert items[0]["link"] == "https://example.com/a1"
def test_parses_published_date(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(ATOM_FEED)
_, items = _parse_atom(root)
assert items[0]["date"] == "2026-02-15"
assert items[1]["date"] == "2026-02-16"
def test_no_published_empty_string(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(ATOM_NO_ID)
_, items = _parse_atom(root)
assert items[0]["date"] == ""
def test_fallback_to_link_as_id(self):
import xml.etree.ElementTree as ET
root = ET.fromstring(ATOM_NO_ID)
@@ -755,6 +788,9 @@ class TestCmdRssCheck:
assert len(announcements) == 2
assert "[news]" in announcements[0]
assert "Second Post" in announcements[0]
# Verify date suffix
assert "| 2026-02-11" in announcements[0]
assert "| 2026-02-12" in announcements[1]
asyncio.run(inner())
@@ -1073,3 +1109,27 @@ class TestCmdRssUsage:
bot = _FakeBot()
asyncio.run(cmd_rss(bot, _msg("!rss foobar")))
assert "Usage:" in bot.replied[0]
# ---------------------------------------------------------------------------
# TestParseDate
# ---------------------------------------------------------------------------
class TestParseDate:
def test_iso_format(self):
assert _parse_date("2026-02-15T10:00:00Z") == "2026-02-15"
def test_iso_with_offset(self):
assert _parse_date("2026-02-15T10:00:00+00:00") == "2026-02-15"
def test_rfc2822_format(self):
assert _parse_date("Mon, 10 Feb 2026 09:00:00 +0000") == "2026-02-10"
def test_empty_string(self):
assert _parse_date("") == ""
def test_garbage(self):
assert _parse_date("not a date") == ""
def test_date_only(self):
assert _parse_date("2026-01-01") == "2026-01-01"

View File

@@ -17,6 +17,7 @@ sys.modules[_spec.name] = _mod
_spec.loader.exec_module(_mod)
from plugins.twitch import ( # noqa: E402
_compact_num,
_delete,
_errors,
_load,
@@ -652,6 +653,17 @@ class TestCmdTwitchList:
assert "broken (error)" in bot.replied[0]
def test_list_shows_live(self):
_clear()
bot = _FakeBot()
_save(bot, "#test:xqc", {
"name": "xqc", "channel": "#test",
"last_error": "", "was_live": True,
"last_viewers": 50000,
})
asyncio.run(cmd_twitch(bot, _msg("!twitch list")))
assert "xqc (live, 50k)" in bot.replied[0]
def test_list_shows_live_no_viewers(self):
_clear()
bot = _FakeBot()
_save(bot, "#test:xqc", {
@@ -725,8 +737,10 @@ class TestCmdTwitchCheck:
assert len(announcements) == 1
assert "[xqc] is live" in announcements[0]
assert "Fortnite" in announcements[0]
# Check reply shows live status
assert "| 50k viewers" in announcements[0]
# Check reply shows live status with viewers
assert "xqc: live" in bot.replied[0]
assert "| 50k viewers" in bot.replied[0]
asyncio.run(inner())
@@ -794,6 +808,7 @@ class TestPollOnce:
assert "[xqc] is live" in messages[0]
assert "Playing games" in messages[0]
assert "Fortnite" in messages[0]
assert "| 50k viewers" in messages[0]
assert "https://twitch.tv/xqc" in messages[0]
updated = _load(bot, key)
assert updated["was_live"] is True
@@ -941,6 +956,7 @@ class TestPollOnce:
assert len(messages) == 1
assert "Just chatting" in messages[0]
assert "(" not in messages[0] # No game parenthetical
assert "| 100 viewers" in messages[0]
asyncio.run(inner())
@@ -1132,3 +1148,33 @@ class TestCmdTwitchUsage:
bot = _FakeBot()
asyncio.run(cmd_twitch(bot, _msg("!twitch foobar")))
assert "Usage:" in bot.replied[0]
# ---------------------------------------------------------------------------
# TestCompactNum
# ---------------------------------------------------------------------------
class TestCompactNum:
def test_zero(self):
assert _compact_num(0) == "0"
def test_small(self):
assert _compact_num(999) == "999"
def test_one_k(self):
assert _compact_num(1000) == "1k"
def test_fractional_k(self):
assert _compact_num(1500) == "1.5k"
def test_one_m(self):
assert _compact_num(1_000_000) == "1M"
def test_fractional_m(self):
assert _compact_num(2_500_000) == "2.5M"
def test_fifty_k(self):
assert _compact_num(50000) == "50k"
def test_hundred(self):
assert _compact_num(100) == "100"

View File

@@ -19,6 +19,7 @@ _spec.loader.exec_module(_mod)
from plugins.youtube import ( # noqa: E402
_MAX_ANNOUNCE,
_channels,
_compact_num,
_delete,
_derive_name,
_errors,
@@ -44,26 +45,48 @@ from plugins.youtube import ( # noqa: E402
YT_ATOM_FEED = b"""\
<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns:yt="http://www.youtube.com/xml/schemas/2015"
xmlns="http://www.w3.org/2005/Atom">
xmlns="http://www.w3.org/2005/Atom"
xmlns:media="http://search.yahoo.com/mrss/">
<title>3Blue1Brown - Videos</title>
<author><name>3Blue1Brown</name></author>
<entry>
<id>yt:video:abc123</id>
<yt:videoId>abc123</yt:videoId>
<title>Linear Algebra</title>
<published>2026-01-15T12:00:00+00:00</published>
<link rel="alternate" href="https://www.youtube.com/watch?v=abc123"/>
<media:group>
<media:community>
<media:statistics views="1500000"/>
<media:starRating count="45000"/>
</media:community>
</media:group>
</entry>
<entry>
<id>yt:video:def456</id>
<yt:videoId>def456</yt:videoId>
<title>Calculus</title>
<published>2026-02-01T08:30:00+00:00</published>
<link rel="alternate" href="https://www.youtube.com/watch?v=def456"/>
<media:group>
<media:community>
<media:statistics views="820000"/>
<media:starRating count="32000"/>
</media:community>
</media:group>
</entry>
<entry>
<id>yt:video:ghi789</id>
<yt:videoId>ghi789</yt:videoId>
<title>Neural Networks</title>
<published>2026-02-10T14:00:00+00:00</published>
<link rel="alternate" href="https://www.youtube.com/watch?v=ghi789"/>
<media:group>
<media:community>
<media:statistics views="250000"/>
<media:starRating count="12000"/>
</media:community>
</media:group>
</entry>
</feed>
"""
@@ -362,6 +385,28 @@ class TestParseFeed:
channel_name, _ = _parse_feed(YT_ATOM_FEED)
assert channel_name == "3Blue1Brown"
def test_parses_published_date(self):
_, items = _parse_feed(YT_ATOM_FEED)
assert items[0]["date"] == "2026-01-15"
assert items[1]["date"] == "2026-02-01"
assert items[2]["date"] == "2026-02-10"
def test_parses_views(self):
_, items = _parse_feed(YT_ATOM_FEED)
assert items[0]["views"] == 1500000
assert items[1]["views"] == 820000
def test_parses_likes(self):
_, items = _parse_feed(YT_ATOM_FEED)
assert items[0]["likes"] == 45000
assert items[1]["likes"] == 32000
def test_no_media_defaults_zero(self):
_, items = _parse_feed(YT_ATOM_NO_VIDEOID)
assert items[0]["views"] == 0
assert items[0]["likes"] == 0
assert items[0]["date"] == ""
# ---------------------------------------------------------------------------
# TestResolveChannel
@@ -789,6 +834,11 @@ class TestCmdYtCheck:
assert len(announcements) == 2
assert "[news]" in announcements[0]
assert "Calculus" in announcements[0]
# Verify metadata suffix (views, likes, date)
assert "| " in announcements[0]
assert "820kv" in announcements[0]
assert "32klk" in announcements[0]
assert "2026-02-01" in announcements[0]
asyncio.run(inner())
@@ -1103,3 +1153,27 @@ class TestCmdYtUsage:
bot = _FakeBot()
asyncio.run(cmd_yt(bot, _msg("!yt foobar")))
assert "Usage:" in bot.replied[0]
# ---------------------------------------------------------------------------
# TestCompactNum
# ---------------------------------------------------------------------------
class TestCompactNum:
def test_zero(self):
assert _compact_num(0) == "0"
def test_small(self):
assert _compact_num(999) == "999"
def test_one_k(self):
assert _compact_num(1000) == "1k"
def test_fractional_k(self):
assert _compact_num(1500) == "1.5k"
def test_one_m(self):
assert _compact_num(1_000_000) == "1M"
def test_fractional_m(self):
assert _compact_num(2_500_000) == "2.5M"