Revert "Apply filtering to text attachments; offer to auto-upload text attachments to paste bin"

bot/exts/filtering/_filter_lists/extension.py

@@ -14,6 +14,18 @@
 if typing.TYPE_CHECKING:
     from bot.exts.filtering.filtering import Filtering
 
+PASTE_URL = "https://paste.pythondiscord.com"
+PY_EMBED_DESCRIPTION = (
+    "It looks like you tried to attach a Python file - "
+    f"please use a code-pasting service such as {PASTE_URL}"
+)
+
+TXT_LIKE_FILES = {".txt", ".csv", ".json"}
+TXT_EMBED_DESCRIPTION = (
+    "You either uploaded a `{blocked_extension}` file or entered a message that was too long. "
+    f"Please use our [paste bin]({PASTE_URL}) instead."
+)
+
 DISALLOWED_EMBED_DESCRIPTION = (
     "It looks like you tried to attach file type(s) that we do not allow ({joined_blacklist}). "
     "We currently allow the following file types: **{joined_whitelist}**.\n\n"
@@ -75,23 +87,30 @@ async def actions_for(
         not_allowed = {ext: filename for ext, filename in all_ext if ext not in allowed_ext}
 
         if ctx.event == Event.SNEKBOX:
-            not_allowed = dict(not_allowed.items())
+            not_allowed = {ext: filename for ext, filename in not_allowed.items() if ext not in TXT_LIKE_FILES}
 
         if not not_allowed:  # Yes, it's a double negative. Meaning all attachments are allowed :)
             return None, [], {ListType.ALLOW: triggered}
 
         # At this point, something is disallowed.
         if ctx.event != Event.SNEKBOX:  # Don't post the embed if it's a snekbox response.
-            meta_channel = bot.instance.get_channel(Channels.meta)
-            if not self._whitelisted_description:
-                self._whitelisted_description = ", ".join(
-                    filter_.content for filter_ in self[ListType.ALLOW].filters.values()
+            if ".py" in not_allowed:
+                # Provide a pastebin link for .py files.
+                ctx.dm_embed = PY_EMBED_DESCRIPTION
+            elif txt_extensions := {ext for ext in TXT_LIKE_FILES if ext in not_allowed}:
+                # Work around Discord auto-conversion of messages longer than 2000 chars to .txt
+                ctx.dm_embed = TXT_EMBED_DESCRIPTION.format(blocked_extension=txt_extensions.pop())
+            else:
+                meta_channel = bot.instance.get_channel(Channels.meta)
+                if not self._whitelisted_description:
+                    self._whitelisted_description = ", ".join(
+                        filter_.content for filter_ in self[ListType.ALLOW].filters.values()
+                    )
+                ctx.dm_embed = DISALLOWED_EMBED_DESCRIPTION.format(
+                    joined_whitelist=self._whitelisted_description,
+                    joined_blacklist=", ".join(not_allowed),
+                    meta_channel_mention=meta_channel.mention,
                 )
-            ctx.dm_embed = DISALLOWED_EMBED_DESCRIPTION.format(
-                joined_whitelist=self._whitelisted_description,
-                joined_blacklist=", ".join(not_allowed),
-                meta_channel_mention=meta_channel.mention,
-            )
 
         ctx.matches += not_allowed.values()
         ctx.blocked_exts |= set(not_allowed)

bot/exts/filtering/_filter_lists/filter_list.py

@@ -157,7 +157,10 @@ def __hash__(self):
         return hash(id(self))
 
 
-class FilterList[T: Filter](dict[ListType, AtomicList], FieldRequiring):
+T = typing.TypeVar("T", bound=Filter)
+
+
+class FilterList(dict[ListType, AtomicList], typing.Generic[T], FieldRequiring):
     """Dispatches events to lists of _filters, and aggregates the responses into a single list of actions to take."""
 
     # Each subclass must define a name matching the filter_list name we're expecting to receive from the database.

bot/exts/filtering/_settings.py

@@ -5,17 +5,21 @@
 from abc import abstractmethod
 from copy import copy
 from functools import reduce
-from typing import Any, NamedTuple, Self
+from typing import Any, NamedTuple, Self, TypeVar
 
 from bot.exts.filtering._filter_context import FilterContext
 from bot.exts.filtering._settings_types import settings_types
 from bot.exts.filtering._settings_types.settings_entry import ActionEntry, SettingsEntry, ValidationEntry
 from bot.exts.filtering._utils import FieldRequiring
 from bot.log import get_logger
 
+TSettings = TypeVar("TSettings", bound="Settings")
+
 log = get_logger(__name__)
 
-_already_warned = set[str]()
+_already_warned: set[str] = set()
+
+T = TypeVar("T", bound=SettingsEntry)
 
 
 def create_settings(
@@ -51,7 +55,7 @@ def create_settings(
     )
 
 
-class Settings[T: SettingsEntry](FieldRequiring, dict[str, T]):
+class Settings(FieldRequiring, dict[str, T]):
     """
     A collection of settings.
 
@@ -65,7 +69,7 @@ class Settings[T: SettingsEntry](FieldRequiring, dict[str, T]):
 
     entry_type: type[T]
 
-    _already_warned = set[str]()
+    _already_warned: set[str] = set()
 
     @abstractmethod  # ABCs have to have at least once abstract method to actually count as such.
     def __init__(self, settings_data: dict, *, defaults: Settings | None = None, keep_empty: bool = False):
@@ -100,7 +104,7 @@ def overrides(self) -> dict[str, Any]:
         """Return a dictionary of overrides across all entries."""
         return reduce(operator.or_, (entry.overrides for entry in self.values() if entry), {})
 
-    def copy(self: Self) -> Self:
+    def copy(self: TSettings) -> TSettings:
         """Create a shallow copy of the object."""
         return copy(self)
 

bot/exts/filtering/filtering.py

@@ -66,14 +66,6 @@
 WEEKLY_REPORT_ISO_DAY = 3  # 1=Monday, 7=Sunday
 
 
-async def _extract_text_file_content(att: discord.Attachment) -> str:
-    """Extract up to the first 30 lines and first 2000 characters (whichever is shorter) of an attachment."""
-    file_encoding = re.search(r"charset=(\S+)", att.content_type).group(1)
-    file_lines: list[str] = (await att.read()).decode(encoding=file_encoding).splitlines()
-    first_n_lines = "\n".join(file_lines[:30])[:2_000]
-    return f"{att.filename}: {first_n_lines}"
-
-
 class Filtering(Cog):
     """Filtering and alerting for content posted on the server."""
 
@@ -88,7 +80,7 @@ class Filtering(Cog):
     def __init__(self, bot: Bot):
         self.bot = bot
         self.filter_lists: dict[str, FilterList] = {}
-        self._subscriptions = defaultdict[Event, list[FilterList]](list)
+        self._subscriptions: defaultdict[Event, list[FilterList]] = defaultdict(list)
         self.delete_scheduler = scheduling.Scheduler(self.__class__.__name__)
         self.webhook: discord.Webhook | None = None
 
@@ -231,15 +223,6 @@ async def on_message(self, msg: Message) -> None:
         self.message_cache.append(msg)
 
         ctx = FilterContext.from_message(Event.MESSAGE, msg, None, self.message_cache)
-
-        text_contents = [
-            await _extract_text_file_content(a)
-            for a in msg.attachments if "charset" in a.content_type
-        ]
-        if text_contents:
-            attachment_content = "\n\n".join(text_contents)
-            ctx = ctx.replace(content=f"{ctx.content}\n\n{attachment_content}")
-
         result_actions, list_messages, triggers = await self._resolve_action(ctx)
         self.message_cache.update(msg, metadata=triggers)
         if result_actions:

bot/exts/utils/snekbox/_cog.py

@@ -17,6 +17,7 @@
 from bot.bot import Bot
 from bot.constants import BaseURLs, Channels, Emojis, MODERATION_ROLES, Roles, URLs
 from bot.decorators import redirect_output
+from bot.exts.filtering._filter_lists.extension import TXT_LIKE_FILES
 from bot.exts.help_channels._channel import is_help_forum_post
 from bot.exts.utils.snekbox._eval import EvalJob, EvalResult
 from bot.exts.utils.snekbox._io import FileAttachment
@@ -31,8 +32,6 @@
 ANSI_REGEX = re.compile(r"\N{ESC}\[[0-9;:]*m")
 ESCAPE_REGEX = re.compile("[`\u202E\u200B]{3,}")
 
-TXT_LIKE_FILES = {".txt", ".csv", ".json", ".py"}
-
 # The timeit command should only output the very last line, so all other output should be suppressed.
 # This will be used as the setup code along with any setup code provided.
 TIMEIT_SETUP_WRAPPER = """