Switch from SQLAlchemy to asyncpg/aiosqlite
This commit is contained in:
Tulir Asokan
@@ -1,5 +1,5 @@
|
||||
# mautrix-telegram - A Matrix-Telegram puppeting bridge
|
||||
# Copyright (C) 2019 Tulir Asokan
|
||||
# Copyright (C) 2021 Tulir Asokan
|
||||
#
|
||||
# This program is free software: you can redistribute it and/or modify
|
||||
# it under the terms of the GNU Affero General Public License as published by
|
||||
@@ -13,39 +13,77 @@
|
||||
#
|
||||
# You should have received a copy of the GNU Affero General Public License
|
||||
# along with this program. If not, see <https://www.gnu.org/licenses/>.
|
||||
from typing import Optional, List, Tuple, Callable, Pattern, Match, TYPE_CHECKING
|
||||
import re
|
||||
import logging
|
||||
from __future__ import annotations
|
||||
|
||||
from telethon.tl.types import (MessageEntityMention, MessageEntityMentionName, MessageEntityItalic,
|
||||
TypeMessageEntity, InputMessageEntityMentionName)
|
||||
import re
|
||||
|
||||
from telethon.tl.types import MessageEntityItalic, TypeMessageEntity
|
||||
from telethon.helpers import add_surrogate, del_surrogate
|
||||
from telethon import TelegramClient
|
||||
|
||||
from mautrix.types import RoomID, MessageEventContent
|
||||
from mautrix.util.logging import TraceLogger
|
||||
|
||||
from ... import puppet as pu
|
||||
from ...types import TelegramID
|
||||
from ...db import Message as DBMessage
|
||||
from .parser import ParsedMessage, parse_html
|
||||
from .parser import MatrixParser
|
||||
|
||||
if TYPE_CHECKING:
|
||||
from ...context import Context
|
||||
|
||||
log: TraceLogger = logging.getLogger("mau.fmt.mx")
|
||||
should_bridge_plaintext_highlights: bool = False
|
||||
|
||||
command_regex: Pattern = re.compile(r"^!([A-Za-z0-9@]+)")
|
||||
not_command_regex: Pattern = re.compile(r"^\\(![A-Za-z0-9@]+)")
|
||||
plain_mention_regex: Optional[Pattern] = None
|
||||
command_regex = re.compile(r"^!([A-Za-z0-9@]+)")
|
||||
not_command_regex = re.compile(r"^\\(![A-Za-z0-9@]+)")
|
||||
|
||||
MAX_LENGTH = 4096
|
||||
CUTOFF_TEXT = " [message cut]"
|
||||
CUT_MAX_LENGTH = MAX_LENGTH - len(CUTOFF_TEXT)
|
||||
|
||||
|
||||
def _cut_long_message(message: str, entities: List[TypeMessageEntity]) -> ParsedMessage:
|
||||
class FormatError(Exception):
|
||||
pass
|
||||
|
||||
|
||||
async def matrix_reply_to_telegram(
|
||||
content: MessageEventContent, tg_space: TelegramID, room_id: RoomID | None = None
|
||||
) -> TelegramID | None:
|
||||
event_id = content.get_reply_to()
|
||||
if not event_id:
|
||||
return
|
||||
content.trim_reply_fallback()
|
||||
|
||||
message = await DBMessage.get_by_mxid(event_id, room_id, tg_space)
|
||||
if message:
|
||||
return message.tgid
|
||||
return None
|
||||
|
||||
|
||||
async def matrix_to_telegram(
|
||||
client: TelegramClient, *, text: str | None = None, html: str | None = None
|
||||
) -> tuple[str, list[TypeMessageEntity]]:
|
||||
if html is not None:
|
||||
return await _matrix_html_to_telegram(client, html)
|
||||
elif text is not None:
|
||||
return _matrix_text_to_telegram(text), []
|
||||
else:
|
||||
raise ValueError("text or html must be provided to convert formatting")
|
||||
|
||||
|
||||
async def _matrix_html_to_telegram(
|
||||
client: TelegramClient, html: str
|
||||
) -> tuple[str, list[TypeMessageEntity]]:
|
||||
try:
|
||||
html = command_regex.sub(r"<command>\1</command>", html)
|
||||
html = html.replace("\t", " " * 4)
|
||||
html = not_command_regex.sub(r"\1", html)
|
||||
|
||||
parsed = await MatrixParser(client).parse(add_surrogate(html))
|
||||
text = del_surrogate(parsed.text.strip())
|
||||
text, entities = _cut_long_message(text, parsed.telegram_entities)
|
||||
|
||||
return text, entities
|
||||
except Exception as e:
|
||||
raise FormatError(f"Failed to convert Matrix format: {html}") from e
|
||||
|
||||
|
||||
def _cut_long_message(
|
||||
message: str, entities: list[TypeMessageEntity]
|
||||
) -> tuple[str, list[TypeMessageEntity]]:
|
||||
if len(message) > MAX_LENGTH:
|
||||
message = message[0:CUT_MAX_LENGTH] + CUTOFF_TEXT
|
||||
new_entities = []
|
||||
@@ -60,112 +98,8 @@ def _cut_long_message(message: str, entities: List[TypeMessageEntity]) -> Parsed
|
||||
return message, entities
|
||||
|
||||
|
||||
class FormatError(Exception):
|
||||
pass
|
||||
|
||||
|
||||
def matrix_reply_to_telegram(content: MessageEventContent, tg_space: TelegramID,
|
||||
room_id: Optional[RoomID] = None) -> Optional[TelegramID]:
|
||||
event_id = content.get_reply_to()
|
||||
if not event_id:
|
||||
return
|
||||
content.trim_reply_fallback()
|
||||
|
||||
message = DBMessage.get_by_mxid(event_id, room_id, tg_space)
|
||||
if message:
|
||||
return message.tgid
|
||||
return None
|
||||
|
||||
|
||||
async def matrix_to_telegram(client: TelegramClient, *, text: Optional[str] = None,
|
||||
html: Optional[str] = None) -> ParsedMessage:
|
||||
if html is not None:
|
||||
text, entities = _matrix_html_to_telegram(html)
|
||||
elif text is not None:
|
||||
text, entities = _matrix_text_to_telegram(text)
|
||||
else:
|
||||
raise ValueError("text or html must be provided to convert formatting")
|
||||
await _fix_name_mentions(client, entities)
|
||||
return text, entities
|
||||
|
||||
|
||||
def _matrix_html_to_telegram(html: str) -> ParsedMessage:
|
||||
try:
|
||||
html = command_regex.sub(r"<command>\1</command>", html)
|
||||
html = html.replace("\t", " " * 4)
|
||||
html = not_command_regex.sub(r"\1", html)
|
||||
if should_bridge_plaintext_highlights:
|
||||
html = plain_mention_regex.sub(_plain_mention_to_html, html)
|
||||
|
||||
text, entities = parse_html(add_surrogate(html))
|
||||
text = del_surrogate(text.strip())
|
||||
text, entities = _cut_long_message(text, entities)
|
||||
|
||||
return text, entities
|
||||
except Exception as e:
|
||||
raise FormatError(f"Failed to convert Matrix format: {html}") from e
|
||||
|
||||
|
||||
def _matrix_text_to_telegram(text: str) -> ParsedMessage:
|
||||
def _matrix_text_to_telegram(text: str) -> str:
|
||||
text = command_regex.sub(r"/\1", text)
|
||||
text = text.replace("\t", " " * 4)
|
||||
text = not_command_regex.sub(r"\1", text)
|
||||
if should_bridge_plaintext_highlights:
|
||||
entities, pmr_replacer = _plain_mention_to_text()
|
||||
text = plain_mention_regex.sub(pmr_replacer, text)
|
||||
else:
|
||||
entities = []
|
||||
return text, entities
|
||||
|
||||
|
||||
async def _fix_name_mentions(client: TelegramClient, entities: List[TypeMessageEntity]) -> None:
|
||||
for index in reversed(range(len(entities))):
|
||||
entity = entities[index]
|
||||
if isinstance(entity, (MessageEntityMentionName, InputMessageEntityMentionName)):
|
||||
try:
|
||||
user = await client.get_input_entity(entity.user_id)
|
||||
except (ValueError, TypeError) as e:
|
||||
log.trace(f"Dropping mention of {entity.user_id}: {e}")
|
||||
del entities[index]
|
||||
else:
|
||||
entities[index] = InputMessageEntityMentionName(entity.offset, entity.length, user)
|
||||
|
||||
|
||||
def _plain_mention_to_text() -> Tuple[List[TypeMessageEntity], Callable[[Match], str]]:
|
||||
entities = []
|
||||
|
||||
def replacer(match: Match) -> str:
|
||||
puppet = pu.Puppet.find_by_displayname(match.group(2))
|
||||
if puppet:
|
||||
offset = match.start()
|
||||
length = match.end() - offset
|
||||
if puppet.username:
|
||||
entity = MessageEntityMention(offset, length)
|
||||
text = f"@{puppet.username}"
|
||||
else:
|
||||
entity = MessageEntityMentionName(offset, length, user_id=puppet.tgid)
|
||||
text = puppet.displayname
|
||||
entities.append(entity)
|
||||
return text
|
||||
return "".join(match.groups())
|
||||
|
||||
return entities, replacer
|
||||
|
||||
|
||||
def _plain_mention_to_html(match: Match) -> str:
|
||||
puppet = pu.Puppet.find_by_displayname(match.group(2))
|
||||
if puppet:
|
||||
return (f"{match.group(1)}"
|
||||
f"<a href='https://matrix.to/#/{puppet.mxid}'>"
|
||||
f"{puppet.displayname}"
|
||||
"</a>")
|
||||
return "".join(match.groups())
|
||||
|
||||
|
||||
def init_mx(context: "Context") -> None:
|
||||
global plain_mention_regex, should_bridge_plaintext_highlights
|
||||
config = context.config
|
||||
dn_template = config["bridge.displayname_template"]
|
||||
dn_template = re.escape(dn_template).replace(re.escape("{displayname}"), "[^>]+")
|
||||
plain_mention_regex = re.compile(f"^({dn_template})")
|
||||
should_bridge_plaintext_highlights = config["bridge.plaintext_highlights"]
|
||||
return text
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
# mautrix-telegram - A Matrix-Telegram puppeting bridge
|
||||
# Copyright (C) 2019 Tulir Asokan
|
||||
# Copyright (C) 2021 Tulir Asokan
|
||||
#
|
||||
# This program is free software: you can redistribute it and/or modify
|
||||
# it under the terms of the GNU Affero General Public License as published by
|
||||
@@ -13,77 +13,80 @@
|
||||
#
|
||||
# You should have received a copy of the GNU Affero General Public License
|
||||
# along with this program. If not, see <https://www.gnu.org/licenses/>.
|
||||
from typing import List, Tuple, Optional
|
||||
from __future__ import annotations
|
||||
|
||||
from telethon.tl.types import TypeMessageEntity
|
||||
import logging
|
||||
|
||||
from telethon import TelegramClient
|
||||
|
||||
from mautrix.types import UserID, RoomID
|
||||
from mautrix.util.formatter import MatrixParser as BaseMatrixParser, RecursionContext
|
||||
from mautrix.util.formatter.html_reader_htmlparser import read_html, HTMLNode
|
||||
from mautrix.util.logging import TraceLogger
|
||||
|
||||
from ... import user as u, puppet as pu, portal as po
|
||||
from .telegram_message import TelegramMessage, TelegramEntityType
|
||||
|
||||
|
||||
ParsedMessage = Tuple[str, List[TypeMessageEntity]]
|
||||
|
||||
|
||||
def parse_html(input_html: str) -> ParsedMessage:
|
||||
msg = MatrixParser.parse(input_html)
|
||||
return msg.text, msg.telegram_entities
|
||||
log: TraceLogger = logging.getLogger("mau.fmt.mx")
|
||||
|
||||
|
||||
class MatrixParser(BaseMatrixParser[TelegramMessage]):
|
||||
e = TelegramEntityType
|
||||
fs = TelegramMessage
|
||||
read_html = read_html
|
||||
client: TelegramClient
|
||||
|
||||
@classmethod
|
||||
def custom_node_to_fstring(cls, node: HTMLNode, ctx: RecursionContext
|
||||
) -> Optional[TelegramMessage]:
|
||||
msg = cls.tag_aware_parse_node(node, ctx)
|
||||
def __init__(self, client: TelegramClient) -> None:
|
||||
self.client = client
|
||||
self.read_html = read_html
|
||||
|
||||
async def custom_node_to_fstring(
|
||||
self, node: HTMLNode, ctx: RecursionContext
|
||||
) -> TelegramMessage | None:
|
||||
msg = await self.tag_aware_parse_node(node, ctx)
|
||||
if node.tag == "command":
|
||||
msg.format(TelegramEntityType.COMMAND)
|
||||
return None
|
||||
|
||||
@classmethod
|
||||
def user_pill_to_fstring(cls, msg: TelegramMessage, user_id: UserID) -> TelegramMessage:
|
||||
user = (pu.Puppet.deprecated_sync_get_by_mxid(user_id)
|
||||
or u.User.get_by_mxid(user_id, create=False))
|
||||
async def user_pill_to_fstring(self, msg: TelegramMessage, user_id: UserID) -> TelegramMessage:
|
||||
user = (await pu.Puppet.get_by_mxid(user_id)
|
||||
or await u.User.get_by_mxid(user_id, create=False))
|
||||
if not user:
|
||||
return msg
|
||||
if user.username:
|
||||
return TelegramMessage(f"@{user.username}").format(TelegramEntityType.MENTION)
|
||||
if user.tg_username:
|
||||
return TelegramMessage(f"@{user.tg_username}").format(TelegramEntityType.MENTION)
|
||||
elif user.tgid:
|
||||
displayname = user.plain_displayname or msg.text
|
||||
return TelegramMessage(displayname).format(TelegramEntityType.MENTION_NAME,
|
||||
user_id=user.tgid)
|
||||
msg = TelegramMessage(displayname)
|
||||
try:
|
||||
input_entity = self.client.get_input_entity(user.tgid)
|
||||
except (ValueError, TypeError) as e:
|
||||
log.trace(f"Dropping mention of {user.tgid}: {e}")
|
||||
else:
|
||||
msg = msg.format(TelegramEntityType.MENTION_NAME, user_id=input_entity)
|
||||
return msg
|
||||
|
||||
@classmethod
|
||||
def url_to_fstring(cls, msg: TelegramMessage, url: str) -> TelegramMessage:
|
||||
async def url_to_fstring(self, msg: TelegramMessage, url: str) -> TelegramMessage:
|
||||
if url == msg.text:
|
||||
return msg.format(cls.e.URL)
|
||||
return msg.format(self.e.URL)
|
||||
else:
|
||||
return msg.format(cls.e.INLINE_URL, url=url)
|
||||
return msg.format(self.e.INLINE_URL, url=url)
|
||||
|
||||
@classmethod
|
||||
def room_pill_to_fstring(cls, msg: TelegramMessage, room_id: RoomID) -> TelegramMessage:
|
||||
async def room_pill_to_fstring(self, msg: TelegramMessage, room_id: RoomID) -> TelegramMessage:
|
||||
username = po.Portal.get_username_from_mx_alias(room_id)
|
||||
portal = po.Portal.find_by_username(username)
|
||||
portal = await po.Portal.find_by_username(username)
|
||||
if portal and portal.username:
|
||||
return TelegramMessage(f"@{portal.username}").format(TelegramEntityType.MENTION)
|
||||
|
||||
@classmethod
|
||||
def header_to_fstring(cls, node: HTMLNode, ctx: RecursionContext) -> TelegramMessage:
|
||||
children = cls.node_to_fstrings(node, ctx)
|
||||
async def header_to_fstring(self, node: HTMLNode, ctx: RecursionContext) -> TelegramMessage:
|
||||
children = await self.node_to_fstrings(node, ctx)
|
||||
length = int(node.tag[1])
|
||||
prefix = "#" * length + " "
|
||||
return TelegramMessage.join(children, "").prepend(prefix).format(TelegramEntityType.BOLD)
|
||||
|
||||
@classmethod
|
||||
def blockquote_to_fstring(cls, node: HTMLNode, ctx: RecursionContext) -> TelegramMessage:
|
||||
msg = cls.tag_aware_parse_node(node, ctx)
|
||||
async def blockquote_to_fstring(
|
||||
self, node: HTMLNode, ctx: RecursionContext
|
||||
) -> TelegramMessage:
|
||||
msg = await self.tag_aware_parse_node(node, ctx)
|
||||
children = msg.trim().split("\n")
|
||||
children = [child.prepend("> ") for child in children]
|
||||
return TelegramMessage.join(children, "\n")
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
# mautrix-telegram - A Matrix-Telegram puppeting bridge
|
||||
# Copyright (C) 2019 Tulir Asokan
|
||||
# Copyright (C) 2021 Tulir Asokan
|
||||
#
|
||||
# This program is free software: you can redistribute it and/or modify
|
||||
# it under the terms of the GNU Affero General Public License as published by
|
||||
@@ -13,7 +13,9 @@
|
||||
#
|
||||
# You should have received a copy of the GNU Affero General Public License
|
||||
# along with this program. If not, see <https://www.gnu.org/licenses/>.
|
||||
from typing import Optional, Union, Any, List, Type, Dict
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any, Type
|
||||
from enum import Enum
|
||||
|
||||
from telethon.tl.types import (MessageEntityMention as Mention, MessageEntityBotCommand as Command,
|
||||
@@ -41,7 +43,7 @@ class TelegramEntityType(Enum):
|
||||
INLINE_CODE = Code
|
||||
BLOCKQUOTE = Blockquote
|
||||
MENTION = Mention
|
||||
MENTION_NAME = MentionName
|
||||
MENTION_NAME = InputMentionName
|
||||
COMMAND = Command
|
||||
|
||||
USER_MENTION = 1
|
||||
@@ -52,15 +54,15 @@ class TelegramEntityType(Enum):
|
||||
class TelegramEntity(SemiAbstractEntity):
|
||||
internal: TypeMessageEntity
|
||||
|
||||
def __init__(self, type: Union[TelegramEntityType, Type[TypeMessageEntity]],
|
||||
offset: int, length: int, extra_info: Dict[str, Any]) -> None:
|
||||
def __init__(self, type: TelegramEntityType | Type[TypeMessageEntity],
|
||||
offset: int, length: int, extra_info: dict[str, Any]) -> None:
|
||||
if isinstance(type, TelegramEntityType):
|
||||
if isinstance(type.value, int):
|
||||
raise ValueError(f"Can't create Entity with non-Telegram EntityType {type}")
|
||||
type = type.value
|
||||
self.internal = type(offset=offset, length=length, **extra_info)
|
||||
|
||||
def copy(self) -> Optional['TelegramEntity']:
|
||||
def copy(self) -> TelegramEntity:
|
||||
extra_info = {}
|
||||
if isinstance(self.internal, Pre):
|
||||
extra_info["language"] = self.internal.language
|
||||
@@ -95,5 +97,5 @@ class TelegramMessage(EntityString[TelegramEntity, TelegramEntityType]):
|
||||
entity_class = TelegramEntity
|
||||
|
||||
@property
|
||||
def telegram_entities(self) -> List[TypeMessageEntity]:
|
||||
def telegram_entities(self) -> list[TypeMessageEntity]:
|
||||
return [entity.internal for entity in self.entities]
|
||||
|
||||
Reference in New Issue
Block a user