Enabled all Ruff linter rules and fixed resulting violations.

crabstero/__main__.py

@@ -106,7 +106,7 @@ def _parse_args(argv: list[str] | None = None) -> argparse.Namespace:
         host, sep, port_str = args.listen_metrics.rpartition(":")
         if not sep or not host:
             parser.error(
-                "--listen-metrics must be in HOST:PORT format (e.g. 127.0.0.1:9090)"
+                "--listen-metrics must be in HOST:PORT format (e.g. 127.0.0.1:9090)",
             )
         try:
             args.listen_metrics = (host, int(port_str))
@@ -117,7 +117,7 @@ def _parse_args(argv: list[str] | None = None) -> argparse.Namespace:
         parser.error(
             "a Discord bot token is required via --token,"
             " the TOKEN environment variable,"
-            " or a systemd credential named 'token'"
+            " or a systemd credential named 'token'",
         )
 
     return args

crabstero/bot.py

@@ -58,6 +58,7 @@ class Crabstero(commands.Bot):
         self,
         token: str,
         database_path: str,
+        *,
         ingest_only: bool = False,
         metrics_address: tuple[str, int] | None = None,
     ) -> None:
@@ -104,7 +105,8 @@ class Crabstero(commands.Bot):
         :raises RuntimeError: If accessed before :meth:`setup_hook` has run.
         """
         if self._db is None:
-            raise RuntimeError("Database is not initialized")
+            msg = "Database is not initialized"
+            raise RuntimeError(msg)
         return self._db
 
     @property
@@ -123,7 +125,11 @@ class Crabstero(commands.Bot):
             await server.start()
             self._metrics_server = server
 
-        from crabstero.listeners import interaction, message, server_events
+        from crabstero.listeners import (  # noqa: PLC0415
+            interaction,
+            message,
+            server_events,
+        )
 
         if not self._ingest_only:
             await interaction.setup(self)
@@ -133,7 +139,8 @@ class Crabstero(commands.Bot):
 
         @self.tree.error
         async def on_app_command_error(
-            interaction: discord.Interaction, error: AppCommandError
+            interaction: discord.Interaction,
+            error: AppCommandError,
         ) -> None:
             original = (
                 error.original if isinstance(error, CommandInvokeError) else error
@@ -203,14 +210,18 @@ class Crabstero(commands.Bot):
 
     @override
     async def on_error(
-        self, event_method: str, /, *args: object, **kwargs: object
+        self,
+        event_method: str,
+        /,
+        *args: object,
+        **kwargs: object,
     ) -> None:
         """Increment the global error counter for event listener exceptions.
 
         :param event_method: The name of the event that raised the exception.
         """
         ERRORS.labels(source=event_method).inc()
-        logger.error("Unhandled exception in %s.", event_method, exc_info=True)
+        logger.error("Unhandled exception in %s.", event_method, exc_info=True)  # noqa: LOG014
 
     async def on_ready(self) -> None:
         """Log that the bot has started successfully."""
@@ -250,7 +261,8 @@ class Crabstero(commands.Bot):
         if channel.id in self._ingestion_tasks:
             return
         task = asyncio.create_task(
-            self._ingest_one(channel), name=f"ingest-{channel.id}"
+            self._ingest_one(channel),
+            name=f"ingest-{channel.id}",
         )
         self._ingestion_tasks[channel.id] = task
         task.add_done_callback(lambda t: self._on_ingestion_done(channel.id, t))

crabstero/cache.py

@@ -103,7 +103,8 @@ class IngestCache:
         if self._task is not None:
             return
         self._task = asyncio.create_task(
-            self._cleanup_loop(), name="ingest-cache-cleanup"
+            self._cleanup_loop(),
+            name="ingest-cache-cleanup",
         )
 
     async def stop(self) -> None:

crabstero/database.py

@@ -261,7 +261,9 @@ class Database:
             return row[0] if row else None
 
     async def get_random_completing_next_word(
-        self, channel_id: int, word: str
+        self,
+        channel_id: int,
+        word: str,
     ) -> str | None:
         """Return a random sentence-ending next word for a given word in a channel.
 
@@ -352,7 +354,10 @@ class Database:
             )
 
     async def clear_flag(
-        self, entity_type: str, entity_id: str, flag_name: str
+        self,
+        entity_type: str,
+        entity_id: str,
+        flag_name: str,
     ) -> None:
         """Clear a flag on a given entity. If the flag is not set, this is a no-op.
 
@@ -370,7 +375,10 @@ class Database:
             )
 
     async def is_flag_set(
-        self, entity_type: str, entity_id: str, flag_name: str
+        self,
+        entity_type: str,
+        entity_id: str,
+        flag_name: str,
     ) -> bool:
         """Check whether a flag is set on a given entity.
 

crabstero/listeners/interaction.py

@@ -39,7 +39,10 @@ class ForgetMeView(TrackedView):
     """Confirmation view with Confirm/Cancel buttons for /forgetme."""
 
     def __init__(
-        self, user_id: int, db: Database, interaction: discord.Interaction
+        self,
+        user_id: int,
+        db: Database,
+        interaction: discord.Interaction,
     ) -> None:
         """Create a new ForgetMeView.
 
@@ -64,12 +67,14 @@ class ForgetMeView(TrackedView):
 
     @discord.ui.button(label="Confirm", style=discord.ButtonStyle.danger)
     async def confirm(
-        self, interaction: discord.Interaction, button: discord.ui.Button[ForgetMeView]
+        self,
+        interaction: discord.Interaction,
+        _button: discord.ui.Button[ForgetMeView],
     ) -> None:
         """Delete all user data, clear flags, and set noIngest.
 
         :param interaction: The interaction event.
-        :param button: The button that was pressed.
+        :param _button: The button that was pressed.
         """
         self._responded = True
         await self._db.forget_user(self._user_id, Flag.NO_INGEST)
@@ -85,12 +90,14 @@ class ForgetMeView(TrackedView):
 
     @discord.ui.button(label="Cancel", style=discord.ButtonStyle.secondary)
     async def cancel(
-        self, interaction: discord.Interaction, button: discord.ui.Button[ForgetMeView]
+        self,
+        interaction: discord.Interaction,
+        _button: discord.ui.Button[ForgetMeView],
     ) -> None:
         """Cancel the /forgetme action.
 
         :param interaction: The interaction event.
-        :param button: The button that was pressed.
+        :param _button: The button that was pressed.
         """
         self._responded = True
         metrics.FORGETME.labels(outcome="cancelled").inc()
@@ -145,10 +152,16 @@ class InteractionCog(commands.Cog):
         :param interaction: The interaction event.
         """
         if await flags.is_flag_set(
-            self.bot.db, interaction.user, EntityType.USER, Flag.ALLOW_PINGS
+            self.bot.db,
+            interaction.user,
+            EntityType.USER,
+            Flag.ALLOW_PINGS,
         ):
             await flags.clear_flag(
-                self.bot.db, interaction.user, EntityType.USER, Flag.ALLOW_PINGS
+                self.bot.db,
+                interaction.user,
+                EntityType.USER,
+                Flag.ALLOW_PINGS,
             )
             metrics.PINGME.labels(outcome="opted_out").inc()
             await interaction.response.send_message(
@@ -159,7 +172,10 @@ class InteractionCog(commands.Cog):
             )
         else:
             await flags.set_flag(
-                self.bot.db, interaction.user, EntityType.USER, Flag.ALLOW_PINGS
+                self.bot.db,
+                interaction.user,
+                EntityType.USER,
+                Flag.ALLOW_PINGS,
             )
             metrics.PINGME.labels(outcome="opted_in").inc()
             await interaction.response.send_message(
@@ -179,7 +195,10 @@ class InteractionCog(commands.Cog):
         :param interaction: The interaction event.
         """
         if await flags.is_flag_set(
-            self.bot.db, interaction.user, EntityType.USER, Flag.NO_INGEST
+            self.bot.db,
+            interaction.user,
+            EntityType.USER,
+            Flag.NO_INGEST,
         ):
             metrics.FORGETME.labels(outcome="already_forgotten").inc()
             await interaction.response.send_message(

crabstero/listeners/message.py

@@ -49,7 +49,8 @@ class MessageCog(commands.Cog):
         channel = message.channel
 
         if not isinstance(
-            channel, (discord.TextChannel, discord.Thread, discord.VoiceChannel)
+            channel,
+            (discord.TextChannel, discord.Thread, discord.VoiceChannel),
         ):
             return
 
@@ -61,7 +62,8 @@ class MessageCog(commands.Cog):
         if self.bot.ingest_only:
             # In ingest-only mode, never reply — only ingest eligible messages.
             if message.type == discord.MessageType.default and not isinstance(
-                channel, discord.Thread
+                channel,
+                discord.Thread,
             ):
                 await ingest_message(self.bot.db, message, self.bot.ingest_cache)
             return
@@ -72,13 +74,15 @@ class MessageCog(commands.Cog):
 
         # Only ingest non-thread messages; threads share their parent channel's chain.
         if message.type == discord.MessageType.default and not isinstance(
-            channel, discord.Thread
+            channel,
+            discord.Thread,
         ):
             await ingest_message(self.bot.db, message, self.bot.ingest_cache)
 
     @commands.Cog.listener()
     async def on_raw_message_delete(
-        self, payload: discord.RawMessageDeleteEvent
+        self,
+        payload: discord.RawMessageDeleteEvent,
     ) -> None:
         """Uningest a recently ingested message when it is deleted.
 
@@ -88,7 +92,8 @@ class MessageCog(commands.Cog):
 
     @commands.Cog.listener()
     async def on_raw_bulk_message_delete(
-        self, payload: discord.RawBulkMessageDeleteEvent
+        self,
+        payload: discord.RawBulkMessageDeleteEvent,
     ) -> None:
         """Uningest recently ingested messages when they are bulk-deleted.
 

crabstero/listeners/server_events.py

@@ -55,7 +55,8 @@ class ServerEventsCog(commands.Cog):
         logger.info('Joined new server! (Name: "%s" | ID: %s)', guild.name, guild.id)
 
         embed = discord.Embed(
-            title="Joined New Server", color=discord.Color.from_rgb(255, 165, 0)
+            title="Joined New Server",
+            color=discord.Color.from_rgb(255, 165, 0),
         )
         embed.add_field(
             name=f"{guild.name} ({guild.id})",
@@ -84,7 +85,9 @@ class ServerEventsCog(commands.Cog):
 
     @commands.Cog.listener()
     async def on_guild_role_update(
-        self, before: discord.Role, after: discord.Role
+        self,
+        before: discord.Role,
+        after: discord.Role,
     ) -> None:
         """Queue all text channels for ingestion when role permissions change.
 
@@ -101,7 +104,9 @@ class ServerEventsCog(commands.Cog):
 
     @commands.Cog.listener()
     async def on_guild_channel_update(
-        self, before: discord.abc.GuildChannel, after: discord.abc.GuildChannel
+        self,
+        before: discord.abc.GuildChannel,
+        after: discord.abc.GuildChannel,
     ) -> None:
         """Queue all text channels for ingestion on permission change.
 

crabstero/markov.py

@@ -73,6 +73,9 @@ def _split_sentences(paragraph: str) -> list[str]:
     return _SENTENCE_SPLIT.split(normalized)
 
 
+_MIN_WORDS_FOR_START = 2
+
+
 def _tokenize_sentence(
     sentence: str,
 ) -> tuple[list[str], list[tuple[str, str]]]:
@@ -85,7 +88,7 @@ def _tokenize_sentence(
         sentence += DEFAULT_SENTENCE_END
     words = sentence.split()
 
-    start_words: list[str] = [words[0]] if len(words) >= 2 else []
+    start_words: list[str] = [words[0]] if len(words) >= _MIN_WORDS_FOR_START else []
     transitions: list[tuple[str, str]] = list(itertools.pairwise(words))
     return start_words, transitions
 
@@ -106,7 +109,10 @@ async def ingest(db: Database, channel_id: int, user_id: int, paragraph: str) ->
 
 
 async def _ingest_sentence(
-    db: Database, channel_id: int, user_id: int, sentence: str
+    db: Database,
+    channel_id: int,
+    user_id: int,
+    sentence: str,
 ) -> None:
     """Ingest a single sentence into the Markov chain for a given channel.
 
@@ -137,7 +143,10 @@ async def uningest(db: Database, channel_id: int, user_id: int, paragraph: str)
 
 
 async def _uningest_sentence(
-    db: Database, channel_id: int, user_id: int, sentence: str
+    db: Database,
+    channel_id: int,
+    user_id: int,
+    sentence: str,
 ) -> None:
     """Remove a single sentence's Markov data from the chain.
 
@@ -153,7 +162,10 @@ async def _uningest_sentence(
 
 
 async def generate(
-    db: Database, channel_id: int, soft_limit: int = 750, hard_limit: int = 1000
+    db: Database,
+    channel_id: int,
+    soft_limit: int = 750,
+    hard_limit: int = 1000,
 ) -> str:
     """Generate a new sentence from previously ingested words for a channel.
 

crabstero/messages.py

@@ -79,7 +79,10 @@ async def reply_to_message(db: Database, message: discord.Message) -> None:
         embed = discord.Embed(
             title=await markov.generate(db, channel_id, soft_limit=200, hard_limit=300),
             description=await markov.generate(
-                db, channel_id, soft_limit=300, hard_limit=500
+                db,
+                channel_id,
+                soft_limit=300,
+                hard_limit=500,
             ),
         )
 
@@ -117,8 +120,30 @@ async def reply_to_message(db: Database, message: discord.Message) -> None:
     metrics.REPLIES_SENT.inc()
 
 
+def _extract_embed_data(
+    embeds: list[discord.Embed],
+) -> tuple[list[str], list[str]]:
+    """Extract text and image URLs from message embeds.
+
+    :param embeds: The embeds to process.
+    :return: A tuple of (texts, image_urls).
+    """
+    texts: list[str] = []
+    image_urls: list[str] = []
+    for embed in embeds:
+        if embed.title:
+            texts.append(embed.title)
+        if embed.description:
+            texts.append(embed.description)
+        if embed.image and embed.image.url:
+            image_urls.append(embed.image.url)
+    return texts, image_urls
+
+
 async def ingest_message(
-    db: Database, message: discord.Message, cache: IngestCache | None = None
+    db: Database,
+    message: discord.Message,
+    cache: IngestCache | None = None,
 ) -> None:
     """Ingest a given message into its channel's Markov chain.
 
@@ -150,26 +175,18 @@ async def ingest_message(
     if not message.content and not message.embeds:
         return
 
-    embed_texts: list[str] = []
-    image_urls: list[str] = []
+    embed_texts, image_urls = _extract_embed_data(message.embeds)
 
     with metrics.MESSAGE_INGESTION_DURATION.time():
         if message.content:
             await markov.ingest(db, channel_id, user_id, message.content)
 
-        for embed in message.embeds:
-            if embed.title:
-                embed_texts.append(embed.title)
-                await markov.ingest(db, channel_id, user_id, embed.title)
-            if embed.description:
-                embed_texts.append(embed.description)
-                await markov.ingest(db, channel_id, user_id, embed.description)
-            if embed.image and embed.image.url:
-                image_urls.append(embed.image.url)
+        for text in embed_texts:
+            await markov.ingest(db, channel_id, user_id, text)
 
         if image_urls:
             await db.add_images(
-                [ChannelImage(channel_id, user_id, url) for url in image_urls]
+                [ChannelImage(channel_id, user_id, url) for url in image_urls],
             )
 
     if cache is not None:
@@ -210,7 +227,7 @@ async def uningest_message(db: Database, cache: IngestCache, message_id: int) ->
             [
                 ChannelImage(entry.channel_id, entry.user_id, url)
                 for url in entry.image_urls
-            ]
+            ],
         )
 
     metrics.MESSAGES_UNINGESTED.inc()

crabstero/metrics.py

@@ -148,7 +148,8 @@ class MetricsServer:
     async def start(self) -> None:
         """Create the aiohttp application and start listening."""
         if self._runner is not None:
-            raise RuntimeError("Metrics server is already running")
+            msg = "Metrics server is already running"
+            raise RuntimeError(msg)
         app = web.Application()
         app.router.add_get("/metrics", make_aiohttp_handler())
         self._runner = web.AppRunner(app)