allow to filter entries by title per subscription

Signed-off-by: Sebastian Hoß <seb@xn--ho-hia.de>

rss/bot.py

@@ -23,6 +23,7 @@ import asyncio
 import hashlib
 import html
 import io
+import re
 
 import aiohttp
 import attr
@@ -106,7 +107,12 @@ class RSSBot(Plugin):
         except Exception:
             self.log.exception("Fatal error while polling feeds")
 
-    async def _send(self, feed: Feed, entry: Entry, sub: Subscription) -> EventID:
+    async def _send(self, feed: Feed, entry: Entry, sub: Subscription) -> EventID | None:
+        title_exclude_filter = sub.title_exclude_filter
+        if title_exclude_filter:
+            if re.search(title_exclude_filter, entry.title):
+                return None
+
         message = sub.notification_template.safe_substitute(
             {
                 "feed_url": feed.url,
@@ -368,6 +374,31 @@ class RSSBot(Plugin):
             )
             await evt.reply(f"Subscribed to {feed_info}")
 
+    @rss.subcommand(
+        "set-filter", aliases=("f", "filter"), help="Set title exclude filter for a specific feed."
+    )
+    @command.argument("feed_id", "feed ID", parser=int)
+    @command.argument("title_exclude_filter", "title exclude filter", pass_raw=True)
+    async def set_filter(self, evt: MessageEvent, feed_id: int, title_exclude_filter: 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
+        try:
+            re.compile(title_exclude_filter)
+        except re.error:
+            await evt.reply(f"Filter is not a valid regular expression")
+            return
+        await self.dbm.update_title_filter(feed.id, evt.room_id, title_exclude_filter)
+        if title_exclude_filter:
+            await evt.reply(
+                f"Feed {feed_id} will now exclude titles matching: {title_exclude_filter}"
+            )
+        else:
+            await evt.reply(f"Removed title exclude filter from feed {feed_id}")
+
     @rss.subcommand(
         "unsubscribe", aliases=("u", "unsub"), help="Unsubscribe this room from a feed."
     )
@@ -441,11 +472,13 @@ class RSSBot(Plugin):
         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) -> str:
+    def _format_subscription(feed: Feed, subscriber: str, title_exclude_filter: str) -> str:
         msg = (
             f"* {feed.id} - [{feed.title}]({feed.url}) "
             f"(subscribed by [{subscriber}](https://matrix.to/#/{subscriber}))"
         )
+        if title_exclude_filter:
+            msg += f" (excludes titles matching: {title_exclude_filter})"
         if feed.error_count > 1:
             msg += f"  \n  ⚠️ The last {feed.error_count} attempts to fetch the feed have failed!"
         return msg
@@ -463,7 +496,8 @@ class RSSBot(Plugin):
         await evt.reply(
             "**Subscriptions in this room:**\n\n"
             + "\n".join(
-                self._format_subscription(feed, subscriber) for feed, subscriber in subscriptions
+                self._format_subscription(feed, subscriber, title_exclude_filter)
+                for feed, subscriber, title_exclude_filter in subscriptions
             )
         )
 

rss/db.py

@@ -39,6 +39,7 @@ class Subscription:
     user_id: UserID
     notification_template: Template
     send_notice: bool
+    title_exclude_filter: str
 
     @classmethod
     def from_row(cls, row: Record | None) -> Subscription | None:
@@ -51,12 +52,14 @@ class Subscription:
             return None
         send_notice = bool(row["send_notice"])
         tpl = Template(row["notification_template"])
+        exclude_filter = row["title_exclude_filter"]
         return cls(
             feed_id=feed_id,
             room_id=room_id,
             user_id=user_id,
             notification_template=tpl,
             send_notice=send_notice,
+            title_exclude_filter=exclude_filter,
         )
 
 
@@ -82,6 +85,7 @@ class Feed:
         data.pop("user_id", None)
         data.pop("send_notice", None)
         data.pop("notification_template", None)
+        data.pop("title_exclude_filter", None)
         return cls(**data, subscriptions=[])
 
 
@@ -121,7 +125,7 @@ class DBManager:
     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
+               room_id, user_id, notification_template, send_notice, title_exclude_filter
         FROM feed INNER JOIN subscription ON feed.id = subscription.feed_id
         """
         rows = await self.db.fetch(q)
@@ -134,13 +138,14 @@ class DBManager:
             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]]:
+    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 FROM feed
+        SELECT id, url, title, subtitle, link, next_retry, error_count, user_id, title_exclude_filter FROM feed
         INNER JOIN subscription ON feed.id = subscription.feed_id AND subscription.room_id = $1
+        ORDER BY id
         """
         rows = await self.db.fetch(q, room_id)
-        return [(Feed.from_row(row), row["user_id"]) for row in rows]
+        return [(Feed.from_row(row), row["user_id"], row["title_exclude_filter"]) for row in rows]
 
     async def get_entries(self, feed_id: int) -> list[Entry]:
         q = "SELECT feed_id, id, date, title, summary, link FROM entry WHERE feed_id = $1"
@@ -173,7 +178,7 @@ class DBManager:
     ) -> tuple[Subscription | None, Feed | None]:
         q = """
         SELECT id, url, title, subtitle, link, next_retry, error_count,
-               room_id, user_id, notification_template, send_notice
+               room_id, user_id, notification_template, send_notice, title_exclude_filter
         FROM feed LEFT JOIN subscription ON feed.id = subscription.feed_id AND room_id = $2
         WHERE feed.id = $1
         """
@@ -238,3 +243,9 @@ class DBManager:
     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_title_filter(
+        self, feed_id: int, room_id: RoomID, title_exclude_filter: str
+    ) -> None:
+        q = "UPDATE subscription SET title_exclude_filter=$3 WHERE feed_id=$1 AND room_id=$2"
+        await self.db.execute(q, feed_id, room_id, title_exclude_filter)

rss/migrations.py

@@ -18,7 +18,7 @@ from mautrix.util.async_db import Connection, Scheme, UpgradeTable
 upgrade_table = UpgradeTable()
 
 
-@upgrade_table.register(description="Latest revision", upgrades_to=3)
+@upgrade_table.register(description="Latest revision", upgrades_to=4)
 async def upgrade_latest(conn: Connection, scheme: Scheme) -> None:
     gen = "GENERATED ALWAYS AS IDENTITY" if scheme != Scheme.SQLITE else ""
     await conn.execute(
@@ -44,6 +44,7 @@ async def upgrade_latest(conn: Connection, scheme: Scheme) -> None:
 
             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)
@@ -72,3 +73,8 @@ async def upgrade_v2(conn: Connection) -> None:
 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="Add title exclude filter to subscriptions")
+async def upgrade_v4(conn: Connection) -> None:
+    await conn.execute("ALTER TABLE subscription ADD COLUMN title_exclude_filter TEXT DEFAULT ''")