Merge pull request #3241 from python-discord/pastebin-auto-upload

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

6 files changed, 180 insertions, 44 deletions

diff --git a/bot/exts/filtering/_filter_lists/extension.py b/bot/exts/filtering/_filter_lists/extension.py
index d656bc6d2..e48564092 100644
--- a/bot/exts/filtering/_filter_lists/extension.py
+++ b/bot/exts/filtering/_filter_lists/extension.py
@@ -14,18 +14,6 @@ from bot.exts.filtering._settings import ActionSettings
 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"
@@ -87,30 +75,23 @@ class ExtensionsList(FilterList[ExtensionFilter]):
         not_allowed = {ext: filename for ext, filename in all_ext if ext not in allowed_ext}
 
         if ctx.event == Event.SNEKBOX:
-            not_allowed = {ext: filename for ext, filename in not_allowed.items() if ext not in TXT_LIKE_FILES}
+            not_allowed = dict(not_allowed.items())
 
         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.
-            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,
+            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.matches += not_allowed.values()
         ctx.blocked_exts |= set(not_allowed)
diff --git a/bot/exts/filtering/_filter_lists/filter_list.py b/bot/exts/filtering/_filter_lists/filter_list.py
index 2cc54e8fb..48d05c97a 100644
--- a/bot/exts/filtering/_filter_lists/filter_list.py
+++ b/bot/exts/filtering/_filter_lists/filter_list.py
@@ -157,10 +157,7 @@ class AtomicList:
         return hash(id(self))
 
 
-T = typing.TypeVar("T", bound=Filter)
-
-
-class FilterList(dict[ListType, AtomicList], typing.Generic[T], FieldRequiring):
+class FilterList[T: Filter](dict[ListType, AtomicList], 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.
diff --git a/bot/exts/filtering/_settings.py b/bot/exts/filtering/_settings.py
index 7005dd2d1..6760a3f01 100644
--- a/bot/exts/filtering/_settings.py
+++ b/bot/exts/filtering/_settings.py
@@ -5,7 +5,7 @@ import traceback
 from abc import abstractmethod
 from copy import copy
 from functools import reduce
-from typing import Any, NamedTuple, Self, TypeVar
+from typing import Any, NamedTuple, Self
 
 from bot.exts.filtering._filter_context import FilterContext
 from bot.exts.filtering._settings_types import settings_types
@@ -13,13 +13,9 @@ from bot.exts.filtering._settings_types.settings_entry import ActionEntry, Setti
 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] = set()
-
-T = TypeVar("T", bound=SettingsEntry)
+_already_warned = set[str]()
 
 
 def create_settings(
@@ -55,7 +51,7 @@ def create_settings(
     )
 
 
-class Settings(FieldRequiring, dict[str, T]):
+class Settings[T: SettingsEntry](FieldRequiring, dict[str, T]):
     """
     A collection of settings.
 
@@ -69,7 +65,7 @@ class Settings(FieldRequiring, dict[str, T]):
 
     entry_type: type[T]
 
-    _already_warned: set[str] = set()
+    _already_warned = set[str]()
 
     @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):
@@ -104,7 +100,7 @@ class Settings(FieldRequiring, dict[str, T]):
         """Return a dictionary of overrides across all entries."""
         return reduce(operator.or_, (entry.overrides for entry in self.values() if entry), {})
 
-    def copy(self: TSettings) -> TSettings:
+    def copy(self: Self) -> Self:
         """Create a shallow copy of the object."""
         return copy(self)
 
diff --git a/bot/exts/filtering/filtering.py b/bot/exts/filtering/filtering.py
index 844f2942e..377cff015 100644
--- a/bot/exts/filtering/filtering.py
+++ b/bot/exts/filtering/filtering.py
@@ -66,6 +66,14 @@ OFFENSIVE_MSG_DELETE_TIME = datetime.timedelta(days=7)
 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."""
 
@@ -80,7 +88,7 @@ class Filtering(Cog):
     def __init__(self, bot: Bot):
         self.bot = bot
         self.filter_lists: dict[str, FilterList] = {}
-        self._subscriptions: defaultdict[Event, list[FilterList]] = defaultdict(list)
+        self._subscriptions = defaultdict[Event, list[FilterList]](list)
         self.delete_scheduler = scheduling.Scheduler(self.__class__.__name__)
         self.webhook: discord.Webhook | None = None
 
@@ -223,6 +231,15 @@ class Filtering(Cog):
         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:
diff --git a/bot/exts/utils/attachment_pastebin_uploader.py b/bot/exts/utils/attachment_pastebin_uploader.py
new file mode 100644
index 000000000..805abd238
--- /dev/null
+++ b/bot/exts/utils/attachment_pastebin_uploader.py
@@ -0,0 +1,144 @@
+from __future__ import annotations
+
+import re
+
+import aiohttp
+import discord
+from discord.ext import commands
+from pydis_core.utils import paste_service
+
+from bot.bot import Bot
+from bot.constants import Emojis
+from bot.log import get_logger
+
+log = get_logger(__name__)
+
+PASTEBIN_UPLOAD_EMOJI = Emojis.check_mark
+DELETE_PASTE_EMOJI = Emojis.trashcan
+
+
+class EmbedFileHandler(commands.Cog):
+    """
+    Handles automatic uploading of attachments to the paste bin.
+
+    Whenever a user uploads one or more attachments that is text-based (py, txt, csv, etc.), this cog offers to upload
+    all the attachments to the paste bin automatically. The steps are as follows:
+    - The bot replies to the message containing the attachments, asking the user to react with a checkmark to consent
+        to having the content uploaded.
+    - If consent is given, the bot uploads the contents and edits its own message to contain the link.
+    - The bot DMs the user the delete link for the paste.
+    - The bot waits for the user to react with a trashcan emoji, in which case the bot deletes the paste and its own
+        message.
+    """
+
+    def __init__(self, bot: Bot):
+        self.bot = bot
+        self.pending_messages = set[int]()
+
+    @staticmethod
+    async def _convert_attachment(attachment: discord.Attachment) -> paste_service.PasteFile:
+        """Converts an attachment to a PasteFile, according to the attachment's file encoding."""
+        encoding = re.search(r"charset=(\S+)", attachment.content_type).group(1)
+        file_content = (await attachment.read()).decode(encoding)
+        return paste_service.PasteFile(content=file_content, name=attachment.filename)
+
+    @commands.Cog.listener()
+    async def on_message_delete(self, message: discord.Message) -> None:
+        """Allows us to know which messages with attachments have been deleted."""
+        self.pending_messages.discard(message.id)
+
+    @commands.Cog.listener()
+    async def on_message(self, message: discord.Message) -> None:
+        """Listens for messages containing attachments and offers to upload them to the pastebin."""
+        # Check if the message contains an embedded file and is not sent by a bot.
+        if message.author.bot or not any(a.content_type.startswith("text") for a in message.attachments):
+            return
+
+        log.trace(f"Offering to upload attachments for {message.author} in {message.channel}, message {message.id}")
+        self.pending_messages.add(message.id)
+
+        # Offer to upload the attachments and wait for the user's reaction.
+        bot_reply = await message.reply(
+            f"Please react with {PASTEBIN_UPLOAD_EMOJI} to upload your file(s) to our "
+            f"[paste bin](<https://paste.pythondiscord.com/>), which is more accessible for some users."
+        )
+        await bot_reply.add_reaction(PASTEBIN_UPLOAD_EMOJI)
+
+        def wait_for_upload_permission(reaction: discord.Reaction, user: discord.User) -> bool:
+            return (
+                reaction.message.id == bot_reply.id
+                and str(reaction.emoji) == PASTEBIN_UPLOAD_EMOJI
+                and user == message.author
+            )
+
+        try:
+            # Wait for the reaction with a timeout of 60 seconds.
+            await self.bot.wait_for("reaction_add", timeout=60.0, check=wait_for_upload_permission)
+        except TimeoutError:
+            # The user does not grant permission before the timeout. Exit early.
+            log.trace(f"{message.author} didn't give permission to upload {message.id} content; aborting.")
+            await bot_reply.edit(content=f"~~{bot_reply.content}~~")
+            await bot_reply.clear_reactions()
+
+        if message.id not in self.pending_messages:
+            log.trace(f"{message.author}'s message was deleted before the attachments could be uploaded; aborting.")
+            await bot_reply.delete()
+            return
+
+        # In either case, we do not want the message ID in pending_messages anymore.
+        self.pending_messages.discard(message.id)
+
+        # Extract the attachments.
+        files = [
+            await self._convert_attachment(f)
+            for f in message.attachments
+            if "charset" in f.content_type
+        ]
+
+        # Upload the files to the paste bin, exiting early if there's an error.
+        log.trace(f"Attempting to upload {len(files)} file(s) to pastebin.")
+        try:
+            async with aiohttp.ClientSession() as session:
+                paste_response = await paste_service.send_to_paste_service(files=files, http_session=session)
+        except (paste_service.PasteTooLongError, ValueError):
+            log.trace(f"{message.author}'s attachments were too long.")
+            await bot_reply.edit(content="Your paste is too long, and couldn't be uploaded.")
+            return
+        except paste_service.PasteUploadError:
+            log.trace(f"Unexpected error uploading {message.author}'s attachments.")
+            await bot_reply.edit(content="There was an error uploading your paste.")
+            return
+
+        # Send the user a DM with the delete link for the paste.
+        # The angle brackets around the remove link are required to stop Discord from visiting the URL to produce a
+        # preview, thereby deleting the paste
+        await message.author.send(content=f"[Click here](<{paste_response.removal}>) to delete your recent paste.")
+
+        # Edit the bot message to contain the link to the paste.
+        await bot_reply.edit(content=f"[Click here]({paste_response.link}) to see this code in our pastebin.")
+        await bot_reply.clear_reactions()
+        await bot_reply.add_reaction(DELETE_PASTE_EMOJI)
+
+        # Wait for the user to react with a trash can, which they can use to delete the paste.
+
+        def wait_for_delete_reaction(reaction: discord.Reaction, user: discord.User) -> bool:
+            return (
+                reaction.message.id == bot_reply.id
+                and str(reaction.emoji) == DELETE_PASTE_EMOJI
+                and user == message.author
+            )
+
+        try:
+            log.trace(f"Offering to delete {message.author}'s attachments in {message.channel}, message {message.id}")
+            await self.bot.wait_for("reaction_add", timeout=60.0 * 10, check=wait_for_delete_reaction)
+            # Delete the paste by visiting the removal URL.
+            async with aiohttp.ClientSession() as session:
+                await session.get(paste_response.removal)
+            await bot_reply.delete()
+        except TimeoutError:
+            log.trace(f"Offer to delete {message.author}'s attachments timed out.")
+
+
+async def setup(bot: Bot) -> None:
+    """Load the EmbedFileHandler cog."""
+    await bot.add_cog(EmbedFileHandler(bot))
diff --git a/bot/exts/utils/snekbox/_cog.py b/bot/exts/utils/snekbox/_cog.py
index 39f61c6e2..9e635e18a 100644
--- a/bot/exts/utils/snekbox/_cog.py
+++ b/bot/exts/utils/snekbox/_cog.py
@@ -17,7 +17,6 @@ from pydis_core.utils.regex import FORMATTED_CODE_REGEX, RAW_CODE_REGEX
 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
@@ -32,6 +31,8 @@ log = get_logger(__name__)
 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 = """