Fork xyz.maubot.rss v0.4.1: whitelist match-filter op titel+summary

- plugin id nl.eddydevink.rss, module buzzrss (uniek, voorkomt botsing met
  xyz.maubot.rss die dezelfde 'rss'-module importeert)
- match_filter (v6-migratie, hernoemd van title_exclude_filter): post wordt
  alleen verstuurd als de regex matcht op titel OF (HTML-ontdaan) summary
- config-default whitelist voor ov-only: gvb/ebs/connexxion/transdev/keolis/
  ret/gvu/bus/metro/tram/trein(verwanten)/ov/ov-staking/staking
- set-filter: lege filter valt terug op config-default

base-config.yaml

@@ -0,0 +1,24 @@
+# Feed update interval in minutes
+update_interval: 60
+# Maximum backoff in minutes when failing to fetch feeds (defaults to 5 days)
+max_backoff: 7200
+# How many feeds to poll in parallel? Set to 0 to disable limit.
+poll_parallelism_limit: 10
+# The time to sleep between send requests when broadcasting a new feed entry.
+# Set to 0 to disable sleep or -1 to run all requests asynchronously at once.
+spam_sleep: 2
+# The prefix for all commands
+# It has to be prefixed with ! in matrix to be recognised
+command_prefix: "rss"
+# Default post notification template for new subscriptions
+notification_template: "New post in $feed_title: [$title]($link)"
+# Should non-admin users be allowed to set regex filters to drop posts from feeds?
+# Regexes are dangerous and may lead to DoS attacks.
+allow_filter: false
+# Per-subscription match filter (whitelist): set with `!rss set-filter <feed-id> <regex>`.
+# A post is ONLY sent when the regex matches the title OR the (HTML-stripped) summary.
+# Empty = fall back to the default below.
+match_filter: "(?i)\\b(?:gvb|ebs|connexxion|cxx|transdev|keolis|ret|gvu|bus|metro|tram|ov)\\b|trein(?:en|verkeer|stel(?:len)?|staking|ongeluk)?|\\bov-staking|staking"
+# Users who can bypass room permission checks
+admins:
+- "@user:example.com"

buzzrss/__init__.py

@@ -0,0 +1 @@
+from .bot import RSSBot

buzzrss/bot.py

@@ -0,0 +1,548 @@
+# rss - A maubot plugin to subscribe to RSS/Atom feeds.
+# Copyright (C) 2022 Tulir Asokan
+#
+# 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.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <https://www.gnu.org/licenses/>.
+from __future__ import annotations
+
+from typing import Any, Iterable
+from datetime import datetime, timezone
+from string import Template
+from time import mktime, monotonic, time
+import asyncio
+import hashlib
+import html
+import io
+import re
+
+import aiohttp
+import attr
+import feedparser
+
+from maubot import MessageEvent, Plugin, __version__ as maubot_version
+from maubot.handlers import command, event
+from mautrix.types import (
+    EventID,
+    EventType,
+    MessageType,
+    PowerLevelStateEventContent,
+    RoomID,
+    StateEvent,
+)
+from mautrix.util.async_db import UpgradeTable
+from mautrix.util.config import BaseProxyConfig, ConfigUpdateHelper
+
+from .db import DBManager, Entry, Feed, Subscription
+from .migrations import upgrade_table
+
+rss_change_level = EventType.find("xyz.maubot.rss", t_class=EventType.Class.STATE)
+
+
+DEFAULT_MATCH_FILTER = (
+    r"(?i)\b(?:gvb|ebs|connexxion|cxx|transdev|keolis|ret|gvu|bus|metro|tram|ov)\b"
+    r"|trein(?:en|verkeer|stel(?:len)?|staking|ongeluk)?"
+    r"|\bov-staking|staking"
+)
+
+
+class Config(BaseProxyConfig):
+    def do_update(self, helper: ConfigUpdateHelper) -> None:
+        helper.copy("update_interval")
+        helper.copy("max_backoff")
+        helper.copy("poll_parallelism_limit")
+        helper.copy("spam_sleep")
+        helper.copy("command_prefix")
+        helper.copy("notification_template")
+        helper.copy("allow_filter")
+        helper.copy("match_filter")
+        helper.copy("admins")
+
+
+class BoolArgument(command.Argument):
+    def __init__(self, name: str, label: str = None, *, required: bool = False) -> None:
+        super().__init__(name, label, required=required, pass_raw=False)
+
+    def match(self, val: str, **kwargs) -> tuple[str, Any]:
+        part = val.split(" ")[0].lower()
+        if part in ("f", "false", "n", "no", "0"):
+            res = False
+        elif part in ("t", "true", "y", "yes", "1"):
+            res = True
+        else:
+            raise ValueError("invalid boolean")
+        return val[len(part) :], res
+
+
+class RSSBot(Plugin):
+    dbm: DBManager
+    poll_task: asyncio.Future
+    poll_sema: asyncio.Semaphore | None
+    http: aiohttp.ClientSession
+    power_level_cache: dict[RoomID, tuple[int, PowerLevelStateEventContent, StateEvent]]
+
+    @classmethod
+    def get_config_class(cls) -> type[BaseProxyConfig]:
+        return Config
+
+    @classmethod
+    def get_db_upgrade_table(cls) -> UpgradeTable:
+        return upgrade_table
+
+    async def start(self) -> None:
+        await super().start()
+        self.on_external_config_update()
+        self.dbm = DBManager(self.database)
+        self.http = self.client.api.session
+        self.power_level_cache = {}
+        self.poll_task = asyncio.create_task(self.poll_feeds())
+
+    def on_external_config_update(self) -> None:
+        self.config.load_and_update()
+        poll_parallelism_limit = self.config["poll_parallelism_limit"]
+        self.poll_sema = (
+            asyncio.Semaphore(poll_parallelism_limit) if poll_parallelism_limit > 0 else None
+        )
+
+    async def stop(self) -> None:
+        await super().stop()
+        self.poll_task.cancel()
+
+    async def poll_feeds(self) -> None:
+        try:
+            await self._poll_feeds()
+        except asyncio.CancelledError:
+            self.log.debug("Polling stopped")
+        except Exception:
+            self.log.exception("Fatal error while polling feeds")
+
+    async def _send(self, feed: Feed, entry: Entry, sub: Subscription) -> EventID | None:
+        match_filter = sub.match_filter or self.config["match_filter"]
+        if match_filter:
+            haystack = f"{entry.title}\n{html.unescape(re.sub(r'<[^>]+>', '', entry.summary))}"
+            if not re.search(match_filter, haystack):
+                return None
+
+        message = sub.notification_template.safe_substitute(
+            {
+                "feed_url": feed.url,
+                "feed_title": feed.title,
+                "feed_subtitle": feed.subtitle,
+                "feed_link": feed.link,
+                **attr.asdict(entry),
+            }
+        )
+        msgtype = MessageType.NOTICE if sub.send_notice else MessageType.TEXT
+        try:
+            return await self.client.send_markdown(
+                sub.room_id, message, msgtype=msgtype, allow_html=True
+            )
+        except Exception as e:
+            self.log.warning(f"Failed to send {entry.id} of {feed.id} to {sub.room_id}: {e}")
+
+    async def _broadcast(
+        self, feed: Feed, entry: Entry, subscriptions: list[Subscription]
+    ) -> None:
+        self.log.debug(f"Broadcasting {entry.id} of {feed.id}")
+        spam_sleep = self.config["spam_sleep"]
+        tasks = [self._send(feed, entry, sub) for sub in subscriptions]
+        if spam_sleep >= 0:
+            for task in tasks:
+                await task
+                await asyncio.sleep(spam_sleep)
+        else:
+            await asyncio.gather(*tasks)
+
+    async def _poll_once(self) -> None:
+        subs = await self.dbm.get_feeds()
+        if not subs:
+            return
+        now = int(time())
+        start = monotonic()
+        tasks = [self.try_parse_feed(feed=feed) for feed in subs if feed.next_retry < now]
+        feed: Feed
+        entries: Iterable[Entry]
+        self.log.info(f"Polling {len(tasks)} feeds")
+        for res in asyncio.as_completed(tasks):
+            feed, entries = await res
+            self.log.trace(
+                f"Fetching {feed.id} (backoff: {feed.error_count} / {feed.next_retry}) "
+                f"success: {bool(entries)}"
+            )
+            if not entries:
+                error_count = feed.error_count + 1
+                next_retry_delay = self.config["update_interval"] * 60 * error_count
+                next_retry_delay = min(next_retry_delay, self.config["max_backoff"] * 60)
+                next_retry = int(time() + next_retry_delay)
+                self.log.debug(f"Setting backoff of {feed.id} to {error_count} / {next_retry}")
+                await self.dbm.set_backoff(feed, error_count, next_retry)
+                continue
+            elif feed.error_count > 0:
+                self.log.debug(f"Resetting backoff of {feed.id}")
+                await self.dbm.set_backoff(feed, error_count=0, next_retry=0)
+            try:
+                new_entries = {entry.id: entry for entry in entries}
+            except Exception:
+                self.log.exception(f"Weird error in items of {feed.url}")
+                continue
+            await self.dbm.filter_entries(feed.id, new_entries)
+            self.log.trace(
+                f"Feed {feed.id} had {len(new_entries)} new entries (of {len(entries)} total)"
+            )
+            new_entry_list: list[Entry] = list(new_entries.values())
+            new_entry_list.sort(key=lambda entry: (entry.date, entry.id))
+            await self.dbm.add_entries(new_entry_list)
+            for entry in new_entry_list:
+                await self._broadcast(feed, entry, feed.subscriptions)
+        duration = monotonic() - start
+        self.log.info(f"Finished polling {len(tasks)} feeds in {duration:.2f} seconds")
+
+    async def _poll_feeds(self) -> None:
+        self.log.debug("Polling started")
+        while True:
+            try:
+                await self._poll_once()
+            except asyncio.CancelledError:
+                self.log.debug("Polling stopped")
+            except Exception:
+                self.log.exception("Error while polling feeds")
+            await asyncio.sleep(self.config["update_interval"] * 60)
+
+    async def try_parse_feed(self, feed: Feed | None = None) -> tuple[Feed, list[Entry]]:
+        if self.poll_sema is not None:
+            async with self.poll_sema:
+                return await self._try_parse_feed(feed)
+        else:
+            return await self._try_parse_feed(feed)
+
+    async def _try_parse_feed(self, feed: Feed | None = None) -> tuple[Feed, list[Entry]]:
+        try:
+            self.log.trace(
+                f"Trying to fetch {feed.id} / {feed.url} "
+                f"(backoff: {feed.error_count} / {feed.next_retry})"
+            )
+            return await self.parse_feed(feed=feed)
+        except Exception as e:
+            self.log.warning(f"Failed to parse feed {feed.id} / {feed.url}: {e}")
+            return feed, []
+
+    @property
+    def _feed_get_headers(self) -> dict[str, str]:
+        return {
+            "User-Agent": f"maubot/{maubot_version} +https://github.com/maubot/rss",
+        }
+
+    async def parse_feed(
+        self, *, feed: Feed | None = None, url: str | None = None
+    ) -> tuple[Feed, list[Entry]]:
+        if feed is None:
+            if url is None:
+                raise ValueError("Either feed or url must be set")
+            feed = Feed(id=-1, url=url, title="", subtitle="", link="")
+        elif url is not None:
+            raise ValueError("Only one of feed or url must be set")
+        resp = await self.http.get(feed.url, headers=self._feed_get_headers)
+        ct = resp.headers["Content-Type"].split(";")[0].strip()
+        if ct == "application/json" or ct == "application/feed+json":
+            return await self._parse_json(feed, resp)
+        else:
+            return await self._parse_rss(feed, resp)
+
+    @classmethod
+    async def _parse_json(
+        cls, feed: Feed, resp: aiohttp.ClientResponse
+    ) -> tuple[Feed, list[Entry]]:
+        content = await resp.json()
+        if content["version"] not in (
+            "https://jsonfeed.org/version/1",
+            "https://jsonfeed.org/version/1.1",
+        ):
+            raise ValueError("Unsupported JSON feed version")
+        if not isinstance(content["items"], list):
+            raise ValueError("Feed is not a valid JSON feed (items is not a list)")
+        feed.title = content["title"]
+        feed.subtitle = content.get("subtitle", "")
+        feed.link = content.get("home_page_url", "")
+        return feed, [cls._parse_json_entry(feed.id, entry) for entry in content["items"]]
+
+    @classmethod
+    def _parse_json_entry(cls, feed_id: int, entry: dict[str, Any]) -> Entry:
+        try:
+            date = datetime.fromisoformat(entry["date_published"])
+        except (ValueError, KeyError):
+            date = datetime.now(timezone.utc)
+        title = entry.get("title", "")
+        summary = (
+            entry.get("summary") or entry.get("content_html") or entry.get("content_text") or ""
+        ).strip()
+        id = str(entry["id"])
+        link = entry.get("url") or id
+        return Entry(feed_id=feed_id, id=id, date=date, title=title, summary=summary, link=link)
+
+    @classmethod
+    async def _parse_rss(
+        cls, feed: Feed, resp: aiohttp.ClientResponse
+    ) -> tuple[Feed, list[Entry]]:
+        content = await resp.read()
+        headers = {"Content-Location": feed.url, **resp.headers, "Content-Encoding": "identity"}
+        parsed_data = feedparser.parse(io.BytesIO(content), response_headers=headers)
+        if parsed_data.bozo:
+            if not isinstance(parsed_data.bozo_exception, feedparser.ThingsNobodyCaresAboutButMe):
+                raise parsed_data.bozo_exception
+        feed_data = parsed_data.get("feed", {})
+        feed.title = feed_data.get("title", feed.url)
+        feed.subtitle = feed_data.get("description", "")
+        feed.link = feed_data.get("link", "")
+        return feed, [cls._parse_rss_entry(feed.id, entry) for entry in parsed_data.entries]
+
+    @classmethod
+    def _parse_rss_entry(cls, feed_id: int, entry: Any) -> Entry:
+        return Entry(
+            feed_id=feed_id,
+            id=(
+                getattr(entry, "id", None)
+                or getattr(entry, "link", None)
+                or hashlib.sha1(
+                    " ".join(
+                        [
+                            getattr(entry, "title", ""),
+                            getattr(entry, "description", ""),
+                        ]
+                    ).encode("utf-8")
+                ).hexdigest()
+            ),
+            date=cls._parse_rss_date(entry),
+            title=getattr(entry, "title", ""),
+            summary=getattr(entry, "description", "").strip(),
+            link=getattr(entry, "link", ""),
+        )
+
+    @staticmethod
+    def _parse_rss_date(entry: Any) -> datetime:
+        try:
+            return datetime.fromtimestamp(mktime(entry["published_parsed"]))
+        except (KeyError, TypeError, ValueError):
+            pass
+        try:
+            return datetime.fromtimestamp(mktime(entry["date_parsed"]))
+        except (KeyError, TypeError, ValueError):
+            pass
+        return datetime.now(timezone.utc)
+
+    async def get_power_levels(
+        self, room_id: RoomID
+    ) -> tuple[PowerLevelStateEventContent, StateEvent]:
+        try:
+            expiry, pls, create = self.power_level_cache[room_id]
+            if expiry < int(time()):
+                return pls, create
+        except KeyError:
+            pass
+        pls = await self.client.get_state_event(room_id, EventType.ROOM_POWER_LEVELS)
+        create = await self.client.get_state_event(room_id, EventType.ROOM_CREATE, format="event")
+        self.power_level_cache[room_id] = (int(time()) + 5 * 60, pls, create)
+        return pls, create
+
+    async def can_manage(self, evt: MessageEvent) -> bool:
+        if evt.sender in self.config["admins"]:
+            return True
+        levels, create = await self.get_power_levels(evt.room_id)
+        user_level = levels.get_user_level(evt.sender, create=create)
+        state_level = levels.get_event_level(rss_change_level)
+        if not isinstance(state_level, int):
+            state_level = 50
+        if user_level < state_level:
+            await evt.reply(
+                "You don't have the permission to manage the subscriptions of this room."
+            )
+            return False
+        return True
+
+    @command.new(
+        name=lambda self: self.config["command_prefix"],
+        help="Manage this RSS bot",
+        require_subcommand=True,
+    )
+    async def rss(self) -> None:
+        pass
+
+    @rss.subcommand("subscribe", aliases=("s", "sub"), help="Subscribe this room to a feed.")
+    @command.argument("url", "feed URL", pass_raw=True)
+    async def subscribe(self, evt: MessageEvent, url: str) -> None:
+        if not await self.can_manage(evt):
+            return
+        feed = await self.dbm.get_feed_by_url(url)
+        if not feed:
+            try:
+                info, entries = await self.parse_feed(url=url)
+            except Exception as e:
+                await evt.reply(f"Failed to load feed: {e}")
+                return
+            feed = await self.dbm.create_feed(info)
+            await self.dbm.add_entries(entries, override_feed_id=feed.id)
+        elif feed.error_count > 0:
+            await self.dbm.set_backoff(feed, error_count=feed.error_count, next_retry=0)
+        feed_info = f"feed ID {feed.id}: [{feed.title}]({feed.url})"
+        sub, _ = await self.dbm.get_subscription(feed.id, evt.room_id)
+        if sub is not None:
+            subscriber = (
+                "You"
+                if sub.user_id == evt.sender
+                else f"[{sub.user_id}](https://matrix.to/#/{sub.user_id})"
+            )
+            await evt.reply(f"{subscriber} had already subscribed this room to {feed_info}")
+        else:
+            await self.dbm.subscribe(
+                feed.id, evt.room_id, evt.sender, self.config["notification_template"]
+            )
+            await evt.reply(f"Subscribed to {feed_info}")
+
+    @rss.subcommand(
+        "set-filter", aliases=("f", "filter"), help="Set match filter (whitelist regex) for a specific feed."
+    )
+    @command.argument("feed_id", "feed ID", parser=int)
+    @command.argument("match_filter", "match filter", pass_raw=True)
+    async def set_filter(self, evt: MessageEvent, feed_id: int, match_filter: str) -> None:
+        if evt.sender in self.config["admins"]:
+            pass
+        elif not self.config["allow_filter"]:
+            await evt.reply("This bot does not allow non-admin users to set title filters")
+            return
+        elif not await self.can_manage(evt):
+            return
+        sub, feed = await self.dbm.get_subscription(feed_id, evt.room_id)
+        if not sub:
+            await evt.reply("This room is not subscribed to that feed")
+            return
+        try:
+            re.compile(match_filter)
+        except re.error:
+            await evt.reply(f"Filter is not a valid regular expression")
+            return
+        await self.dbm.update_match_filter(feed.id, evt.room_id, match_filter)
+        if match_filter:
+            await evt.reply(
+                f"Feed {feed_id} will now only send posts matching: `{match_filter}`"
+            )
+        else:
+            await evt.reply(
+                f"Removed match filter from feed {feed_id} "
+                f"(falling back to default: `{self.config['match_filter']}`)"
+            )
+
+    @rss.subcommand(
+        "unsubscribe", aliases=("u", "unsub"), help="Unsubscribe this room from a feed."
+    )
+    @command.argument("feed_id", "feed ID", parser=int)
+    async def unsubscribe(self, evt: MessageEvent, feed_id: int) -> None:
+        if not await self.can_manage(evt):
+            return
+        sub, feed = await self.dbm.get_subscription(feed_id, evt.room_id)
+        if not sub:
+            await evt.reply("This room is not subscribed to that feed")
+            return
+        await self.dbm.unsubscribe(feed.id, evt.room_id)
+        await evt.reply(f"Unsubscribed from feed ID {feed.id}: [{feed.title}]({feed.url})")
+
+    @rss.subcommand(
+        "template",
+        aliases=("t", "tpl"),
+        help="Change the notification template for a subscription in this room",
+    )
+    @command.argument("feed_id", "feed ID", parser=int)
+    @command.argument("template", "new template", pass_raw=True, required=False)
+    async def command_template(self, evt: MessageEvent, feed_id: int, template: str) -> None:
+        if not await self.can_manage(evt):
+            return
+        sub, feed = await self.dbm.get_subscription(feed_id, evt.room_id)
+        if not sub:
+            await evt.reply("This room is not subscribed to that feed")
+            return
+        if not template:
+            await evt.reply(
+                '<p>Current template in this room:</p><pre><code language="markdown">'
+                f"{html.escape(sub.notification_template.template)}"
+                "</code></pre>",
+                allow_html=True,
+                markdown=False,
+            )
+            return
+        await self.dbm.update_template(feed.id, evt.room_id, template)
+        sub = Subscription(
+            feed_id=feed.id,
+            room_id=sub.room_id,
+            user_id=sub.user_id,
+            notification_template=Template(template),
+            send_notice=sub.send_notice,
+        )
+        sample_entry = Entry(
+            feed_id=feed.id,
+            id="SAMPLE",
+            date=datetime.now(timezone.utc),
+            title="Sample entry",
+            summary="This is a sample entry to demonstrate your new template",
+            link="http://example.com",
+        )
+        await evt.reply(f"Template for feed ID {feed.id} updated. Sample notification:")
+        await self._send(feed, sample_entry, sub)
+
+    @rss.subcommand(
+        "notice", aliases=("n",), help="Set whether or not the bot should send updates as m.notice"
+    )
+    @command.argument("feed_id", "feed ID", parser=int)
+    @BoolArgument("setting", "true/false")
+    async def command_notice(self, evt: MessageEvent, feed_id: int, setting: bool) -> None:
+        if not await self.can_manage(evt):
+            return
+        sub, feed = await self.dbm.get_subscription(feed_id, evt.room_id)
+        if not sub:
+            await evt.reply("This room is not subscribed to that feed")
+            return
+        await self.dbm.set_send_notice(feed.id, evt.room_id, setting)
+        send_type = "m.notice" if setting else "m.text"
+        await evt.reply(f"Updates for feed ID {feed.id} will now be sent as `{send_type}`")
+
+    @staticmethod
+    def _format_subscription(feed: Feed, subscriber: str, match_filter: str) -> str:
+        msg = (
+            f"* {feed.id} - [{feed.title}]({feed.url}) "
+            f"(subscribed by [{subscriber}](https://matrix.to/#/{subscriber}))"
+        )
+        if match_filter:
+            msg += f" (only posts matching: {match_filter})"
+        if feed.error_count > 1:
+            msg += f"  \n  ⚠️ The last {feed.error_count} attempts to fetch the feed have failed!"
+        return msg
+
+    @rss.subcommand(
+        "subscriptions",
+        aliases=("ls", "list", "subs"),
+        help="List the subscriptions in the current room.",
+    )
+    async def command_subscriptions(self, evt: MessageEvent) -> None:
+        subscriptions = await self.dbm.get_feeds_by_room(evt.room_id)
+        if len(subscriptions) == 0:
+            await evt.reply("There are no RSS subscriptions in this room")
+            return
+        await evt.reply(
+            "**Subscriptions in this room:**\n\n"
+            + "\n".join(
+                self._format_subscription(feed, subscriber, match_filter)
+                for feed, subscriber, match_filter in subscriptions
+            )
+        )
+
+    @event.on(EventType.ROOM_TOMBSTONE)
+    async def tombstone(self, evt: StateEvent) -> None:
+        if not evt.content.replacement_room:
+            return
+        await self.dbm.update_room_id(evt.room_id, evt.content.replacement_room)

buzzrss/db.py

@@ -0,0 +1,243 @@
+# rss - A maubot plugin to subscribe to RSS/Atom feeds.
+# Copyright (C) 2022 Tulir Asokan
+#
+# 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.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <https://www.gnu.org/licenses/>.
+from __future__ import annotations
+
+from datetime import datetime, timezone
+from string import Template
+
+from asyncpg import Record
+from attr import dataclass
+import attr
+
+from mautrix.types import RoomID, UserID
+from mautrix.util.async_db import Database, Scheme
+
+
+@dataclass
+class Subscription:
+    feed_id: int
+    room_id: RoomID
+    user_id: UserID
+    notification_template: Template
+    send_notice: bool
+    match_filter: str
+
+    @classmethod
+    def from_row(cls, row: Record | None) -> Subscription | None:
+        if not row:
+            return None
+        feed_id = row["id"]
+        room_id = row["room_id"]
+        user_id = row["user_id"]
+        if not room_id or not user_id:
+            return None
+        send_notice = bool(row["send_notice"])
+        tpl = Template(row["notification_template"])
+        match_filter = row["match_filter"]
+        return cls(
+            feed_id=feed_id,
+            room_id=room_id,
+            user_id=user_id,
+            notification_template=tpl,
+            send_notice=send_notice,
+            match_filter=match_filter,
+        )
+
+
+@dataclass
+class Feed:
+    id: int
+    url: str
+    title: str
+    subtitle: str
+    link: str
+
+    next_retry: int = 0
+    error_count: int = 0
+
+    subscriptions: list[Subscription] = attr.ib(factory=lambda: [])
+
+    @classmethod
+    def from_row(cls, row: Record | None) -> Feed | None:
+        if not row:
+            return None
+        data = {**row}
+        data.pop("room_id", None)
+        data.pop("user_id", None)
+        data.pop("send_notice", None)
+        data.pop("notification_template", None)
+        data.pop("match_filter", None)
+        return cls(**data, subscriptions=[])
+
+
+date_fmt = "%Y-%m-%d %H:%M:%S"
+date_fmt_microseconds = "%Y-%m-%d %H:%M:%S.%f"
+
+
+@dataclass
+class Entry:
+    feed_id: int
+    id: str
+    date: datetime
+    title: str
+    summary: str
+    link: str
+
+    def __attrs_post_init__(self):
+        if self.date.tzinfo is None:
+            self.date = self.date.replace(tzinfo=timezone.utc)
+
+    @classmethod
+    def from_row(cls, row: Record | None) -> Entry | None:
+        if not row:
+            return None
+        data = {**row}
+        date = data.pop("date")
+        if not isinstance(date, datetime):
+            try:
+                date = datetime.strptime(date, date_fmt_microseconds if "." in date else date_fmt)
+            except ValueError:
+                date = datetime.now(timezone.utc)
+        return cls(**data, date=date)
+
+
+class DBManager:
+    db: Database
+
+    def __init__(self, db: Database) -> None:
+        self.db = db
+
+    async def get_feeds(self) -> list[Feed]:
+        q = """
+        SELECT id, url, title, subtitle, link, next_retry, error_count,
+               room_id, user_id, notification_template, send_notice, match_filter
+        FROM feed INNER JOIN subscription ON feed.id = subscription.feed_id
+        """
+        rows = await self.db.fetch(q)
+        feeds: dict[int, Feed] = {}
+        for row in rows:
+            try:
+                feed = feeds[row["id"]]
+            except KeyError:
+                feed = feeds[row["id"]] = Feed.from_row(row)
+            feed.subscriptions.append(Subscription.from_row(row))
+        return list(feeds.values())
+
+    async def get_feeds_by_room(self, room_id: RoomID) -> list[tuple[Feed, UserID, str]]:
+        q = """
+        SELECT id, url, title, subtitle, link, next_retry, error_count, user_id, match_filter FROM feed
+        INNER JOIN subscription ON feed.id = subscription.feed_id AND subscription.room_id = $1
+        """
+        rows = await self.db.fetch(q, room_id)
+        return [(Feed.from_row(row), row["user_id"], row["match_filter"]) for row in rows]
+
+    async def filter_entries(self, feed_id: int, entries: dict[str, Entry]) -> None:
+        entry_keys = list(entries.keys())
+        if not entry_keys:
+            return
+        async with self.db.acquire() as conn:
+            for key in entry_keys:
+                exists = await conn.fetchval(
+                    "SELECT EXISTS(SELECT 1 FROM entry WHERE feed_id = $1 AND id = $2)",
+                    feed_id,
+                    key,
+                )
+                if exists:
+                    del entries[key]
+
+    async def add_entries(self, entries: list[Entry], override_feed_id: int | None = None) -> None:
+        if not entries:
+            return
+        if override_feed_id:
+            for entry in entries:
+                entry.feed_id = override_feed_id
+        records = [attr.astuple(entry) for entry in entries]
+        columns = ("feed_id", "id", "date", "title", "summary", "link")
+        async with self.db.acquire() as conn:
+            if self.db.scheme == Scheme.POSTGRES:
+                await conn.copy_records_to_table("entry", records=records, columns=columns)
+            else:
+                q = (
+                    "INSERT INTO entry (feed_id, id, date, title, summary, link) "
+                    "VALUES ($1, $2, $3, $4, $5, $6)"
+                )
+                await conn.executemany(q, records)
+
+    async def get_feed_by_url(self, url: str) -> Feed | None:
+        q = "SELECT id, url, title, subtitle, link, next_retry, error_count FROM feed WHERE url=$1"
+        return Feed.from_row(await self.db.fetchrow(q, url))
+
+    async def get_subscription(
+        self, feed_id: int, room_id: RoomID
+    ) -> tuple[Subscription | None, Feed | None]:
+        q = """
+        SELECT id, url, title, subtitle, link, next_retry, error_count,
+               room_id, user_id, notification_template, send_notice, match_filter
+        FROM feed LEFT JOIN subscription ON feed.id = subscription.feed_id AND room_id = $2
+        WHERE feed.id = $1
+        """
+        row = await self.db.fetchrow(q, feed_id, room_id)
+        return Subscription.from_row(row), Feed.from_row(row)
+
+    async def update_room_id(self, old: RoomID, new: RoomID) -> None:
+        await self.db.execute("UPDATE subscription SET room_id = $1 WHERE room_id = $2", new, old)
+
+    async def create_feed(self, info: Feed) -> Feed:
+        q = (
+            "INSERT INTO feed (url, title, subtitle, link, next_retry) "
+            "VALUES ($1, $2, $3, $4, $5) RETURNING (id)"
+        )
+        info.id = await self.db.fetchval(
+            q, info.url, info.title, info.subtitle, info.link, info.next_retry
+        )
+        return info
+
+    async def set_backoff(self, info: Feed, error_count: int, next_retry: int) -> None:
+        q = "UPDATE feed SET error_count = $2, next_retry = $3 WHERE id = $1"
+        await self.db.execute(q, info.id, error_count, next_retry)
+
+    async def subscribe(
+        self,
+        feed_id: int,
+        room_id: RoomID,
+        user_id: UserID,
+        template: str | None = None,
+        send_notice: bool = True,
+    ) -> None:
+        q = """
+        INSERT INTO subscription (feed_id, room_id, user_id, notification_template, send_notice)
+        VALUES ($1, $2, $3, $4, $5)
+        """
+        template = template or "New post in $feed_title: [$title]($link)"
+        await self.db.execute(q, feed_id, room_id, user_id, template, send_notice)
+
+    async def unsubscribe(self, feed_id: int, room_id: RoomID) -> None:
+        q = "DELETE FROM subscription WHERE feed_id = $1 AND room_id = $2"
+        await self.db.execute(q, feed_id, room_id)
+
+    async def update_template(self, feed_id: int, room_id: RoomID, template: str) -> None:
+        q = "UPDATE subscription SET notification_template=$3 WHERE feed_id=$1 AND room_id=$2"
+        await self.db.execute(q, feed_id, room_id, template)
+
+    async def set_send_notice(self, feed_id: int, room_id: RoomID, send_notice: bool) -> None:
+        q = "UPDATE subscription SET send_notice=$3 WHERE feed_id=$1 AND room_id=$2"
+        await self.db.execute(q, feed_id, room_id, send_notice)
+
+    async def update_match_filter(
+        self, feed_id: int, room_id: RoomID, match_filter: str
+    ) -> None:
+        q = "UPDATE subscription SET match_filter=$3 WHERE feed_id=$1 AND room_id=$2"
+        await self.db.execute(q, feed_id, room_id, match_filter)

buzzrss/migrations.py

@@ -0,0 +1,91 @@
+# rss - A maubot plugin to subscribe to RSS/Atom feeds.
+# Copyright (C) 2022 Tulir Asokan
+#
+# 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.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Affero General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <https://www.gnu.org/licenses/>.
+from mautrix.util.async_db import Connection, Scheme, UpgradeTable
+
+upgrade_table = UpgradeTable()
+
+
+@upgrade_table.register(description="Latest revision", upgrades_to=5)
+async def upgrade_latest(conn: Connection, scheme: Scheme) -> None:
+    gen = "GENERATED ALWAYS AS IDENTITY" if scheme != Scheme.SQLITE else ""
+    await conn.execute(f"""
+        CREATE TABLE IF NOT EXISTS feed (
+            id       INTEGER {gen},
+            url      TEXT NOT NULL,
+            title    TEXT NOT NULL,
+            subtitle TEXT NOT NULL,
+            link     TEXT NOT NULL,
+
+            next_retry  BIGINT DEFAULT 0,
+            error_count BIGINT DEFAULT 0,
+
+            PRIMARY KEY (id),
+            UNIQUE (url)
+        )
+    """)
+    await conn.execute("""
+        CREATE TABLE IF NOT EXISTS subscription (
+            feed_id INTEGER,
+            room_id TEXT,
+            user_id TEXT NOT NULL,
+
+            notification_template TEXT,
+            send_notice           BOOLEAN DEFAULT true,
+            title_exclude_filter  TEXT DEFAULT '',
+
+            PRIMARY KEY (feed_id, room_id),
+            FOREIGN KEY (feed_id) REFERENCES feed (id)
+        )
+    """)
+    await conn.execute("""
+        CREATE TABLE IF NOT EXISTS entry (
+            feed_id INTEGER,
+            id      TEXT,
+            date    timestamp with time zone NOT NULL,
+            title   TEXT NOT NULL,
+            summary TEXT NOT NULL,
+            link    TEXT NOT NULL,
+            PRIMARY KEY (feed_id, id),
+            FOREIGN KEY (feed_id) REFERENCES feed (id)
+        )
+    """)
+
+
+@upgrade_table.register(description="Add send_notice field to subscriptions")
+async def upgrade_v2(conn: Connection) -> None:
+    await conn.execute("ALTER TABLE subscription ADD COLUMN send_notice BOOLEAN DEFAULT true")
+
+
+@upgrade_table.register(description="Add error counts to feeds")
+async def upgrade_v3(conn: Connection) -> None:
+    await conn.execute("ALTER TABLE feed ADD COLUMN next_retry BIGINT DEFAULT 0")
+    await conn.execute("ALTER TABLE feed ADD COLUMN error_count BIGINT DEFAULT 0")
+
+
+@upgrade_table.register(description="Change entry date to timezone-aware")
+async def upgrade_v4(conn: Connection, scheme: Scheme) -> None:
+    if scheme == Scheme.POSTGRES:
+        await conn.execute("ALTER TABLE entry ALTER COLUMN date TYPE timestamp with time zone")
+
+
+@upgrade_table.register(description="Add title exclude filter to subscriptions")
+async def upgrade_v5(conn: Connection) -> None:
+    await conn.execute("ALTER TABLE subscription ADD COLUMN title_exclude_filter TEXT DEFAULT ''")
+
+
+@upgrade_table.register(description="Rename title exclude filter to match filter (whitelist mode)")
+async def upgrade_v6(conn: Connection) -> None:
+    await conn.execute("ALTER TABLE subscription RENAME COLUMN title_exclude_filter TO match_filter")

maubot.yaml

@@ -0,0 +1,16 @@
+id: nl.eddydevink.rss
+version: 1.0.1
+modules:
+- buzzrss
+main_class: RSSBot
+maubot: 0.3.0
+database: true
+database_type: asyncpg
+config: false
+webapp: false
+license: AGPL-3.0-or-later
+extra_files:
+- base-config.yaml
+dependencies:
+- feedparser>=5.1
+soft_dependencies: []