From 624d15f4b1435fc9a96d6083b6a35d5ded721121 Mon Sep 17 00:00:00 2001
From: Lukas Dobler <69309597+doluk@users.noreply.github.com>
Date: Fri, 29 Nov 2024 23:14:53 +0100
Subject: [PATCH] DiscordChatExporterPy 2.8.1
* Fallback to filename if proxy url isn't suitable to determine file type
Signed-off-by: doluk <69309597+doluk@users.noreply.github.com>
* Prevent emojis in guild or channel name to break title or meta tags
Signed-off-by: doluk <69309597+doluk@users.noreply.github.com>
* Refactor message content handling logic
Signed-off-by: doluk <69309597+doluk@users.noreply.github.com>
---------
Signed-off-by: doluk <69309597+doluk@users.noreply.github.com>
---
chat_exporter/construct/assets/attachment.py | 29 +++++++++++---------
chat_exporter/construct/message.py | 11 ++++----
chat_exporter/construct/transcript.py | 6 ++--
chat_exporter/ext/html_generator.py | 10 ++++++-
chat_exporter/html/base.html | 14 +++++-----
5 files changed, 42 insertions(+), 28 deletions(-)
diff --git a/chat_exporter/construct/assets/attachment.py b/chat_exporter/construct/assets/attachment.py
index c936630..6d18d9e 100644
--- a/chat_exporter/construct/assets/attachment.py
+++ b/chat_exporter/construct/assets/attachment.py
@@ -88,16 +88,19 @@ async def get_file_icon(self) -> str:
"arj", "pkg", "z"
)
- extension = self.attachments.proxy_url.rsplit('.', 1)[1]
- if extension in acrobat_types:
- return DiscordUtils.file_attachment_acrobat
- elif extension in webcode_types:
- return DiscordUtils.file_attachment_webcode
- elif extension in code_types:
- return DiscordUtils.file_attachment_code
- elif extension in document_types:
- return DiscordUtils.file_attachment_document
- elif extension in archive_types:
- return DiscordUtils.file_attachment_archive
- else:
- return DiscordUtils.file_attachment_unknown
+ for tmp in [self.attachments.proxy_url, self.attachments.filename]:
+ if not tmp:
+ continue
+ extension = tmp.rsplit('.', 1)[-1]
+ if extension in acrobat_types:
+ return DiscordUtils.file_attachment_acrobat
+ elif extension in webcode_types:
+ return DiscordUtils.file_attachment_webcode
+ elif extension in code_types:
+ return DiscordUtils.file_attachment_code
+ elif extension in document_types:
+ return DiscordUtils.file_attachment_document
+ elif extension in archive_types:
+ return DiscordUtils.file_attachment_archive
+
+ return DiscordUtils.file_attachment_unknown
diff --git a/chat_exporter/construct/message.py b/chat_exporter/construct/message.py
index 7bd3565..777410c 100644
--- a/chat_exporter/construct/message.py
+++ b/chat_exporter/construct/message.py
@@ -181,12 +181,8 @@ async def build_reference(self):
is_bot = _gather_user_bot(message.author)
user_colour = await self._gather_user_colour(message.author)
- if not message.content and not getattr(message, 'interaction_metadata', None) and not getattr(message, 'interaction', None):
- message.content = "Click to see attachment"
- elif not message.content and ((hasattr(message, 'interaction_metadata') and message.interaction_metadata) or message.interaction):
- message.content = "Click to see command"
-
icon = ""
+ dummy = ""
def get_interaction_status(interaction_message):
if hasattr(interaction_message, 'interaction_metadata'):
return interaction_message.interaction_metadata
@@ -195,8 +191,13 @@ def get_interaction_status(interaction_message):
interaction_status = get_interaction_status(message)
if not interaction_status and (message.embeds or message.attachments):
icon = DiscordUtils.reference_attachment_icon
+ dummy = "Click to see attachment"
elif interaction_status:
icon = DiscordUtils.interaction_command_icon
+ dummy = "Click to see command"
+
+ if not message.content:
+ message.content = dummy
_, message_edited_at = self.set_time(message)
diff --git a/chat_exporter/construct/transcript.py b/chat_exporter/construct/transcript.py
index 526a449..4ea1fea 100644
--- a/chat_exporter/construct/transcript.py
+++ b/chat_exporter/construct/transcript.py
@@ -17,7 +17,7 @@
from chat_exporter.parse.mention import pass_bot
from chat_exporter.ext.discord_utils import DiscordUtils
from chat_exporter.ext.html_generator import (
- fill_out, total, channel_topic, meta_data_temp, fancy_time, channel_subject, PARSE_MODE_NONE
+ fill_out, total, channel_topic, meta_data_temp, fancy_time, channel_subject, PARSE_MODE_NONE, PARSE_MODE_HTML_SAFE
)
@@ -167,7 +167,9 @@ async def export_transcript(self, message_html: str, meta_data: str):
("CHANNEL_ID", str(self.channel.id), PARSE_MODE_NONE),
("MESSAGE_PARTICIPANTS", str(len(meta_data)), PARSE_MODE_NONE),
("FANCY_TIME", _fancy_time, PARSE_MODE_NONE),
- ("SD", sd, PARSE_MODE_NONE)
+ ("SD", sd, PARSE_MODE_NONE),
+ ("SERVER_NAME_SAFE", f"{guild_name}", PARSE_MODE_HTML_SAFE),
+ ("CHANNEL_NAME_SAFE", f"{html.escape(self.channel.name)}", PARSE_MODE_HTML_SAFE),
])
diff --git a/chat_exporter/ext/html_generator.py b/chat_exporter/ext/html_generator.py
index d4c4515..3054d7e 100644
--- a/chat_exporter/ext/html_generator.py
+++ b/chat_exporter/ext/html_generator.py
@@ -1,3 +1,5 @@
+import html
+import json
import os
from chat_exporter.parse.mention import ParseMention
@@ -12,6 +14,7 @@
PARSE_MODE_SPECIAL_EMBED = 4
PARSE_MODE_REFERENCE = 5
PARSE_MODE_EMOJI = 6
+PARSE_MODE_HTML_SAFE = 7
async def fill_out(guild, base, replacements):
@@ -34,8 +37,13 @@ async def fill_out(guild, base, replacements):
v = await ParseMarkdown(v).message_reference_flow()
elif mode == PARSE_MODE_EMOJI:
v = await ParseMarkdown(v).special_emoji_flow()
+ elif mode == PARSE_MODE_HTML_SAFE:
+ # escape html characters
+ v = html.escape(v, quote=True)
+ # escape characters that could be used for xss
+ v = json.dumps(v, ensure_ascii=False)[1:-1]
- base = base.replace("{{" + k + "}}", v.strip())
+ base = base.replace("{{" + k + "}}", str(v or "").strip())
return base
diff --git a/chat_exporter/html/base.html b/chat_exporter/html/base.html
index cb4a671..da72cc5 100644
--- a/chat_exporter/html/base.html
+++ b/chat_exporter/html/base.html
@@ -10,22 +10,22 @@
- {{SERVER_NAME}} - {{CHANNEL_NAME}}
+ {{SERVER_NAME_SAFE}} - {{CHANNEL_NAME_SAFE}}
-
-
+
+
-
-
+
+
-
-
+
+