Add basic support for bridging custom emojis from Telegram

CHANGELOG.md

@@ -6,6 +6,7 @@ Minimum Conduit version remains at 0.4.0.
 
 ### Added
 * Added provisioning API for resolving Telegram identifiers (like usernames).
+* Added basic bridging of Telegram custom emojis to Matrix.
 * Added option to not bridge chats with lots of members.
 * Added option to include captions in the same message as the media to
   implement [MSC2530]. Sending captions the same way is also supported and

ROADMAP.md

@@ -24,6 +24,7 @@
 * Telegram → Matrix
   * [x] Message content (text, formatting, files, etc..)
   * [ ] Advanced message content/media
+    * [x] Custom emojis
     * [x] Polls
 	* [x] Games
 	* [ ] Buttons

mautrix_telegram/db/telegram_file.py

@@ -17,10 +17,11 @@ from __future__ import annotations
 
 from typing import TYPE_CHECKING, ClassVar
 
+from asyncpg import Record
 from attr import dataclass
 
 from mautrix.types import ContentURI, EncryptedFile
-from mautrix.util.async_db import Database
+from mautrix.util.async_db import Database, Scheme
 
 fake_db = Database.create("") if TYPE_CHECKING else None
 
@@ -40,28 +41,47 @@ class TelegramFile:
     decryption_info: EncryptedFile | None
     thumbnail: TelegramFile | None = None
 
+    columns: ClassVar[str] = (
+        "id, mxc, mime_type, was_converted, timestamp, size, width, height, thumbnail, "
+        "decryption_info"
+    )
+
     @classmethod
-    async def get(cls, loc_id: str, *, _thumbnail: bool = False) -> TelegramFile | None:
+    def _from_row(cls, row: Record | None) -> TelegramFile | None:
-        q = (
-            "SELECT id, mxc, mime_type, was_converted, timestamp, size, width, height, thumbnail,"
-            "       decryption_info "
-            "FROM telegram_file WHERE id=$1"
-        )
-        row = await cls.db.fetchrow(q, loc_id)
         if row is None:
             return None
         data = {**row}
-        thumbnail_id = data.pop("thumbnail", None)
+        data.pop("thumbnail", None)
-        if _thumbnail:
-            # Don't allow more than one level of recursion
-            thumbnail_id = None
         decryption_info = data.pop("decryption_info", None)
         return cls(
             **data,
-            thumbnail=(await cls.get(thumbnail_id, _thumbnail=True)) if thumbnail_id else None,
+            thumbnail=None,
             decryption_info=EncryptedFile.parse_json(decryption_info) if decryption_info else None,
         )
 
+    @classmethod
+    async def get_many(cls, loc_ids: list[str]) -> list[TelegramFile]:
+        if cls.db.scheme in (Scheme.POSTGRES, Scheme.COCKROACH):
+            q = f"SELECT {cls.columns} FROM telegram_file WHERE id=ANY($1)"
+            rows = await cls.db.fetch(q, loc_ids)
+        else:
+            tgid_placeholders = ("?," * len(loc_ids)).rstrip(",")
+            q = f"SELECT {cls.columns} FROM telegram_file WHERE id IN ({tgid_placeholders})"
+            rows = await cls.db.fetch(q, *loc_ids)
+        return [cls._from_row(row) for row in rows]
+
+    @classmethod
+    async def get(cls, loc_id: str, *, _thumbnail: bool = False) -> TelegramFile | None:
+        q = f"SELECT {cls.columns} FROM telegram_file WHERE id=$1"
+        row = await cls.db.fetchrow(q, loc_id)
+        file = cls._from_row(row)
+        if file is None:
+            return None
+        thumbnail_id = row.get("thumbnail", None)
+        if thumbnail_id and not _thumbnail:
+            file.thumbnail = await cls.get(thumbnail_id, _thumbnail=True)
+        return file
+
     async def insert(self) -> None:
         q = (
             "INSERT INTO telegram_file (id, mxc, mime_type, was_converted, size, width, height, "

mautrix_telegram/formatter/from_telegram.py

@@ -20,7 +20,7 @@ import logging
 import re
 
 from telethon.errors import RPCError
-from telethon.helpers import add_surrogate, del_surrogate, within_surrogate
+from telethon.helpers import add_surrogate, del_surrogate
 from telethon.tl.custom import Message
 from telethon.tl.types import (
     MessageEntityBlockquote,
@@ -52,8 +52,9 @@ from telethon.tl.types import (
 from mautrix.types import Format, MessageType, TextMessageEventContent
 
 from .. import abstract_user as au, portal as po, puppet as pu, user as u
-from ..db import Message as DBMessage
+from ..db import Message as DBMessage, TelegramFile as DBTelegramFile
 from ..types import TelegramID
+from ..util.file_transfer import transfer_custom_emojis_to_matrix
 
 log: logging.Logger = logging.getLogger("mau.fmt.tg")
 
@@ -125,6 +126,27 @@ async def _add_forward_header(
     )
 
 
+class ReuploadedCustomEmoji(MessageEntityCustomEmoji):
+    file: DBTelegramFile
+
+    def __init__(self, parent: MessageEntityCustomEmoji, file: DBTelegramFile) -> None:
+        super().__init__(parent.offset, parent.length, parent.document_id)
+        self.file = file
+
+
+async def _convert_custom_emoji(
+    source: au.AbstractUser, entities: list[TypeMessageEntity]
+) -> None:
+    emoji_ids = [
+        entity.document_id for entity in entities if isinstance(entity, MessageEntityCustomEmoji)
+    ]
+    custom_emojis = await transfer_custom_emojis_to_matrix(source, emoji_ids)
+    if len(custom_emojis) > 0:
+        for i, entity in enumerate(entities):
+            if isinstance(entity, MessageEntityCustomEmoji):
+                entities[i] = ReuploadedCustomEmoji(entity, custom_emojis[entity.document_id])
+
+
 async def telegram_to_matrix(
     evt: Message | SponsoredMessage,
     source: au.AbstractUser,
@@ -138,6 +160,7 @@ async def telegram_to_matrix(
     )
     entities = override_entities or evt.entities
     if entities:
+        await _convert_custom_emoji(source, entities)
         content.format = Format.HTML
         html = await _telegram_entities_to_matrix_catch(add_surrogate(content.body), entities)
         content.formatted_body = del_surrogate(html)
@@ -166,9 +189,20 @@ async def _telegram_entities_to_matrix_catch(text: str, entities: list[TypeMessa
     return "[failed conversion in _telegram_entities_to_matrix]"
 
 
+def within_surrogate(text, index):
+    """
+    `True` if ``index`` is within a surrogate (before and after it, not at!).
+    """
+    return (
+        1 < index < len(text)  # in bounds
+        and "\ud800" <= text[index - 1] <= "\udbff"  # current is low surrogate
+        and "\udc00" <= text[index] <= "\udfff"  # previous is high surrogate
+    )
+
+
 async def _telegram_entities_to_matrix(
     text: str,
-    entities: list[TypeMessageEntity],
+    entities: list[TypeMessageEntity | ReuploadedCustomEmoji],
     offset: int = 0,
     length: int = None,
     in_codeblock: bool = False,
@@ -197,10 +231,9 @@ async def _telegram_entities_to_matrix(
         elif relative_offset < last_offset:
             continue
 
-        # TODO this breaks when there are lots of emojis in a row (e.g. custom emojis)
+        while within_surrogate(text, relative_offset):
-        # while within_surrogate(text, relative_offset, length=length):
+            relative_offset += 1
-        #     relative_offset += 1
+        while within_surrogate(text, relative_offset + entity.length):
-        while within_surrogate(text, relative_offset + entity.length, length=length):
             entity.length += 1
 
         skip_entity = False
@@ -244,8 +277,12 @@ async def _telegram_entities_to_matrix(
                 html, entity_text, entity.url if entity_type == MessageEntityTextUrl else None
             )
         elif entity_type == MessageEntityCustomEmoji:
-            # TODO support properly
             html.append(entity_text)
+        elif entity_type == ReuploadedCustomEmoji:
+            html.append(
+                f'<img data-mx-emoticon src="{escape(entity.file.mxc)}" height="32" '
+                f'alt="{entity_text}" title="{entity_text}"/>'
+            )
         elif entity_type in (
             MessageEntityBotCommand,
             MessageEntityHashtag,

mautrix_telegram/util/file_transfer.py

@@ -31,6 +31,7 @@ from telethon.errors import (
     LocationInvalidError,
     SecurityError,
 )
+from telethon.tl.functions.messages import GetCustomEmojiDocumentsRequest
 from telethon.tl.types import (
     Document,
     InputDocumentFileLocation,
@@ -45,6 +46,7 @@ import magic
 
 from mautrix.appservice import IntentAPI
 
+from .. import abstract_user as au
 from ..db import TelegramFile as DBTelegramFile
 from ..tgclient import MautrixTelegramClient
 from ..util import sane_mimetypes
@@ -212,6 +214,37 @@ transfer_locks: dict[str, asyncio.Lock] = {}
 TypeThumbnail = Optional[Union[TypeLocation, TypePhotoSize]]
 
 
+async def transfer_custom_emojis_to_matrix(
+    source: au.AbstractUser, emoji_ids: list[int]
+) -> dict[int, DBTelegramFile]:
+    emoji_ids = set(emoji_ids)
+    existing = await DBTelegramFile.get_many([str(id) for id in emoji_ids])
+    file_map = {int(file.id): file for file in existing}
+    not_existing_ids = list(emoji_ids - file_map.keys())
+    if not_existing_ids:
+        log.debug(f"Transferring custom emojis through {source.mxid}: {not_existing_ids}")
+
+        documents: list[Document] = await source.client(
+            GetCustomEmojiDocumentsRequest(document_id=not_existing_ids)
+        )
+
+        async def transfer(document: Document) -> None:
+            file_map[document.id] = await transfer_file_to_matrix(
+                source.client,
+                source.bridge.az.intent,
+                document,
+                is_sticker=True,
+                tgs_convert={"target": "png", "args": {"width": 256, "height": 256}},
+                filename=f"emoji-{document.id}",
+                # Emojis are used as inline images and can't be encrypted
+                encrypt=False,
+                async_upload=source.config["homeserver.async_media"],
+            )
+
+        await asyncio.gather(*[transfer(doc) for doc in documents])
+    return file_map
+
+
 async def transfer_file_to_matrix(
     client: MautrixTelegramClient,
     intent: IntentAPI,