add custom names and avatar in backups

src/core/backup_database.py

@@ -98,9 +98,9 @@ class BackupDatabase:
                     elif table == "permissions":
                         conn.execute("CREATE TABLE permissions (id INTEGER PRIMARY KEY AUTOINCREMENT, channel_id INTEGER, target_id INTEGER, target_type TEXT, allow INTEGER, deny INTEGER)")
                     elif table == "users":
-                        conn.execute("CREATE TABLE users (id INTEGER PRIMARY KEY, username TEXT, display_name TEXT, avatar_file TEXT, avatar_url TEXT, roles TEXT)")
+                        conn.execute("CREATE TABLE users (id INTEGER PRIMARY KEY, username TEXT, display_name TEXT, avatar_file TEXT, avatar_url TEXT, roles TEXT, type INTEGER DEFAULT 0)")
                     elif table == "messages":
-                        conn.execute("CREATE TABLE messages (id INTEGER PRIMARY KEY, channel_id INTEGER, author_id INTEGER, content TEXT, timestamp TEXT, type INTEGER, message_reference INTEGER, is_pinned INTEGER, extra_data TEXT)")
+                        conn.execute("CREATE TABLE messages (id INTEGER PRIMARY KEY, channel_id INTEGER, author_id INTEGER, content TEXT, timestamp TEXT, type INTEGER, message_reference INTEGER, is_pinned INTEGER, extra_data TEXT, custom_display_name TEXT, custom_avatar_url TEXT)")
                     elif table == "attachments":
                         conn.execute("CREATE TABLE attachments (id INTEGER PRIMARY KEY, message_id INTEGER, filename TEXT, size INTEGER, url TEXT, content_type TEXT, local_hash TEXT)")
                     elif table == "embeds":
@@ -124,6 +124,25 @@ class BackupDatabase:
                     conn.execute(f"INSERT INTO {table} ({col_str}) SELECT {col_str} FROM {table}_old")
                     conn.execute(f"DROP TABLE {table}_old")
 
+            # 3. Custom Author Profile Migration
+            res = conn.execute("SELECT count(*) FROM sqlite_master WHERE type='table' AND name='messages'").fetchone()
+            if res and res[0] > 0:
+                cols = conn.execute("PRAGMA table_info(messages)").fetchall()
+                col_names = [c["name"] for c in cols]
+                if "custom_display_name" not in col_names:
+                    logger.info("Migrating messages: adding custom author profile columns")
+                    conn.execute("ALTER TABLE messages ADD COLUMN custom_display_name TEXT")
+                    conn.execute("ALTER TABLE messages ADD COLUMN custom_avatar_url TEXT")
+
+            # 4. User Type Categorization Migration
+            res = conn.execute("SELECT count(*) FROM sqlite_master WHERE type='table' AND name='users'").fetchone()
+            if res and res[0] > 0:
+                cols = conn.execute("PRAGMA table_info(users)").fetchall()
+                col_names = [c["name"] for c in cols]
+                if "type" not in col_names:
+                    logger.info("Migrating users: adding type column")
+                    conn.execute("ALTER TABLE users ADD COLUMN type INTEGER DEFAULT 0")
+            
             conn.commit()
 
     def _init_db(self):
@@ -196,7 +215,8 @@ class BackupDatabase:
                         display_name TEXT,
                         avatar_file TEXT,
                         avatar_url TEXT,
-                        roles TEXT
+                        roles TEXT,
+                        type INTEGER DEFAULT 0
                     )
                 """)
 
@@ -211,7 +231,9 @@ class BackupDatabase:
                         type INTEGER,
                         message_reference INTEGER,
                         is_pinned INTEGER,
-                        extra_data TEXT
+                        extra_data TEXT,
+                        custom_display_name TEXT,
+                        custom_avatar_url TEXT
                     )
                 """)
                 conn.execute("CREATE INDEX IF NOT EXISTS idx_messages_channel ON messages(channel_id)")
@@ -405,8 +427,8 @@ class BackupDatabase:
         """Saves users to the author cache."""
         with self._lock:
             self._conn.executemany("""
-                INSERT OR REPLACE INTO users (id, username, display_name, avatar_file, avatar_url, roles)
+                INSERT OR REPLACE INTO users (id, username, display_name, avatar_file, avatar_url, roles, type)
-                VALUES (:id, :username, :display_name, :avatar_file, :avatar_url, :roles)
+                VALUES (:id, :username, :display_name, :avatar_file, :avatar_url, :roles, :type)
             """, users)
             self._conn.commit()
 
@@ -454,8 +476,8 @@ class BackupDatabase:
             conn = self._conn
             # Insert messages
             conn.executemany("""
-                INSERT OR REPLACE INTO messages (id, channel_id, author_id, content, timestamp, type, message_reference, is_pinned, extra_data)
+                INSERT OR REPLACE INTO messages (id, channel_id, author_id, content, timestamp, type, message_reference, is_pinned, extra_data, custom_display_name, custom_avatar_url)
-                VALUES (:id, :channel_id, :author_id, :content, :timestamp, :type, :message_reference, :is_pinned, :extra_data)
+                VALUES (:id, :channel_id, :author_id, :content, :timestamp, :type, :message_reference, :is_pinned, :extra_data, :custom_display_name, :custom_avatar_url)
             """, messages)
             
             # Extract attachments, reactions, and stickers

src/core/backup_reader.py

@@ -1339,6 +1339,18 @@ class BackupReader:
         user_id = parse_snowflake(msg_data.get("author_id", 0)) or 0
         author = self._resolve_author(user_id)
         
+        # Check for custom author profile (Webhooks / Masquerade)
+        over_name = msg_data.get("custom_display_name")
+        over_avatar = msg_data.get("custom_avatar_url")
+        if over_name:
+            # Create an ephemeral author object for this message
+            author = BackupMember({
+                "id": str(user_id),
+                "username": over_name,
+                "display_name": over_name,
+                "avatar_url": over_avatar
+            })
+
         self._ensure_media_pool_loaded()
         
         channel_id = parse_snowflake(msg_data["channel_id"])

src/core/exporter.py

@@ -439,37 +439,61 @@ class DiscordExporter:
 
         return accumulated_count, accumulated_threads, accumulated_files
 
-    async def _format_user(self, user):
+    async def _format_user(self, user, is_webhook=False):
         """Formats user data for the author or a mention.
         
-        Avatar downloads are intentionally deferred to keep this off the hot
+        For Webhooks, we use a generic name and the default Discord avatar system 
-        message-formatting path.  Call _flush_pending_avatars() after each batch.
+        for the base profile in the user cache.
         """
-        user_id = str(user.id)
+        user_id_int = int(user.id)
+        user_id = str(user_id_int)
+        
         if user_id in self.user_cache:
             return None
 
+        username = user.name
+        display_name = getattr(user, "display_name", user.name)
+        avatar = user.avatar
+        avatar_url = str(user.display_avatar.url) if user.display_avatar else None
+
+        if is_webhook:
+            # For webhooks, we use the ID as the username for technical clarity,
+            # and the current name as the display name.
+            username = user_id
+            display_name = user.name
+            # Discord default avatar formula: (ID >> 22) % 5
+            default_index = (user_id_int >> 22) % 5
+            avatar_url = f"https://cdn.discordapp.com/embed/avatars/{default_index}.png"
+            avatar = None # Don't download character avatar as the "base" webhook avatar
+
         # New user discovered — schedule avatar download but don't block here
         avatar_file = None
-        if user.avatar:
+        if avatar:
             av_name = f"{user_id}.png"
             av_target = self.users_path / av_name
             avatar_file = f"users/{av_name}"
             if not av_target.exists():
                 # Queue for deferred download
-                self._pending_avatars.append((user_id, user.avatar, av_target))
+                self._pending_avatars.append((user_id, avatar, av_target))
 
-        roles = []
+        # Determine user type
-        if hasattr(user, "roles"):
+        # 0: Regular User, 1: Bot, 2: Webhook, 3: System
-            roles = [str(r.id) for r in user.roles if not r.is_default()]
+        u_type = 0
+        if is_webhook:
+            u_type = 2
+        elif getattr(user, "system", False):
+            u_type = 3
+        elif getattr(user, "bot", False):
+            u_type = 1
 
         user_data = {
             "id": user_id,
-            "username": user.name,
+            "username": username,
-            "display_name": getattr(user, "display_name", user.name),
+            "display_name": display_name,
             "avatar_file": avatar_file,
-            "avatar_url": str(user.display_avatar.url) if user.avatar else None,
+            "avatar_url": avatar_url,
-            "roles": json.dumps(roles)
+            "roles": json.dumps(roles),
+            "type": u_type
         }
         self.user_cache[user_id] = user_data
         return user_data
@@ -494,13 +518,14 @@ class DiscordExporter:
         new_users = []
 
         # 1. Author handling
-        u_data = await self._format_user(msg.author)
+        is_webhook = bool(getattr(msg, "webhook_id", None))
+        u_data = await self._format_user(msg.author, is_webhook=is_webhook)
         if u_data: new_users.append(u_data)
 
         # 1.5 Mentions handling (ensure all mentioned users are saved)
         if msg.mentions:
             for mention in msg.mentions:
-                u_ment = await self._format_user(mention)
+                u_ment = await self._format_user(mention, is_webhook=False)
                 if u_ment: new_users.append(u_ment)
 
         # 2. Attachments handling (Content-Addressable Storage)
@@ -603,6 +628,15 @@ class DiscordExporter:
             for s_emb in snapshot.embeds:
                 embeds.append(s_emb.to_dict())
 
+        # 5.6 Author Overrides (Webhooks / Masquerade)
+        custom_display_name = None
+        custom_avatar_url = None
+        
+        # Webhooks or bots with masquerade often use per-message names/avatars
+        if getattr(msg, "webhook_id", None) or (msg.author and msg.author.bot):
+            custom_display_name = msg.author.name
+            custom_avatar_url = str(msg.author.display_avatar.url) if msg.author.display_avatar else None
+
         m_data = {
             "id": str(msg.id),
             "channel_id": str(msg.channel.id),
@@ -616,7 +650,9 @@ class DiscordExporter:
             "stickers": stickers,
             "embeds": embeds,
             "reactions": reactions,
-            "extra_data": None
+            "extra_data": None,
+            "custom_display_name": custom_display_name,
+            "custom_avatar_url": custom_avatar_url
         }
 
         return m_data, new_users