Skip to content

Commit 6a4decd

Browse files
authored
fix: handle UTF-16 offsets in Telegram mentions (#10367)
1 parent 7818a9c commit 6a4decd

2 files changed

Lines changed: 114 additions & 17 deletions

File tree

‎astrbot/core/platform/sources/telegram/tg_adapter.py‎

Lines changed: 16 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -472,9 +472,7 @@ def _apply_caption() -> None:
472472
if update.message.caption and update.message.caption_entities:
473473
for entity in update.message.caption_entities:
474474
if entity.type == "mention":
475-
name = update.message.caption[
476-
entity.offset + 1 : entity.offset + entity.length
477-
]
475+
name = update.message.parse_caption_entity(entity)[1:]
478476
message.message.append(Comp.At(qq=name, name=name))
479477

480478
message = AstrBotMessage()
@@ -598,6 +596,21 @@ def _apply_caption() -> None:
598596
if update.message.text:
599597
# 处理文本消息
600598
plain_text = update.message.text
599+
if update.message.entities:
600+
# Entity offsets refer to the original text in UTF-16 code units.
601+
text_utf16 = plain_text.encode("utf-16-le")
602+
text_parts = []
603+
last_end = 0
604+
for entity in update.message.entities:
605+
if entity.type == "mention":
606+
name = update.message.parse_entity(entity)[1:]
607+
message.message.append(Comp.At(qq=name, name=name))
608+
if name.lower() == context.bot.username.lower():
609+
text_parts.append(text_utf16[last_end : entity.offset * 2])
610+
last_end = (entity.offset + entity.length) * 2
611+
text_parts.append(text_utf16[last_end:])
612+
plain_text = b"".join(text_parts).decode("utf-16-le")
613+
601614
if (
602615
message.type == MessageType.GROUP_MESSAGE
603616
and update.message
@@ -618,20 +631,6 @@ def _apply_caption() -> None:
618631
f" {command_parts[1]}" if len(command_parts) > 1 else ""
619632
)
620633

621-
if update.message.entities:
622-
for entity in update.message.entities:
623-
if entity.type == "mention":
624-
name = plain_text[
625-
entity.offset + 1 : entity.offset + entity.length
626-
]
627-
message.message.append(Comp.At(qq=name, name=name))
628-
# 如果mention是当前bot则移除;否则保留
629-
if name.lower() == context.bot.username.lower():
630-
plain_text = (
631-
plain_text[: entity.offset]
632-
+ plain_text[entity.offset + entity.length :]
633-
)
634-
635634
if plain_text:
636635
message.message.append(Comp.Plain(plain_text))
637636
message.message_str = plain_text

‎tests/test_telegram_adapter.py‎

Lines changed: 98 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -5,6 +5,7 @@
55
from unittest.mock import AsyncMock, MagicMock, patch
66

77
import pytest
8+
from telegram import Message
89

910
import astrbot.api.message_components as Comp
1011
from astrbot.api.platform import Group
@@ -495,6 +496,102 @@ async def test_telegram_reply_without_quote_text_uses_full_message(quote_text):
495496
assert reply.chain[0].text == original_text
496497

497498

499+
@pytest.mark.asyncio
500+
@pytest.mark.parametrize(
501+
("text", "entities", "expected", "names"),
502+
[
503+
("😀 @test_bot hello", [(3, 9)], "😀 hello", ["test_bot"]),
504+
("😀😀 @test_bot hello", [(5, 9)], "😀😀 hello", ["test_bot"]),
505+
("文字 @test_bot hello", [(3, 9)], "文字 hello", ["test_bot"]),
506+
("@test_bot 😀 hello", [(0, 9)], " 😀 hello", ["test_bot"]),
507+
(
508+
"😀 @test_bot @alice @test_bot hello",
509+
[(3, 9), (13, 6), (20, 9)],
510+
"😀 @alice hello",
511+
["test_bot", "alice", "test_bot"],
512+
),
513+
(
514+
"/help@test_bot 😀 @test_bot hello",
515+
[(18, 9)],
516+
"/help 😀 hello",
517+
["test_bot"],
518+
),
519+
],
520+
)
521+
async def test_telegram_mentions_use_original_utf16_offsets(
522+
text, entities, expected, names
523+
):
524+
adapter = _load_telegram_adapter()(
525+
make_platform_config("telegram"), {}, asyncio.Queue()
526+
)
527+
adapter.client.username = "test_bot"
528+
sdk_message = Message.de_json(
529+
{
530+
"message_id": 1,
531+
"date": 0,
532+
"chat": {"id": -100123, "type": "supergroup"},
533+
"text": text,
534+
"entities": [
535+
{"type": "mention", "offset": offset, "length": length}
536+
for offset, length in entities
537+
],
538+
},
539+
bot=None,
540+
)
541+
update = create_mock_update(
542+
message_text=text, chat_type="supergroup", entities=sdk_message.entities
543+
)
544+
update.message.parse_entity = sdk_message.parse_entity
545+
546+
result = await adapter.convert_message(update, _build_context())
547+
548+
assert result is not None
549+
assert result.message_str == expected
550+
assert [part.qq for part in result.message if isinstance(part, Comp.At)] == names
551+
assert any(
552+
isinstance(part, Comp.At) and part.qq == result.self_id
553+
for part in result.message
554+
)
555+
556+
557+
@pytest.mark.asyncio
558+
@pytest.mark.parametrize("media_type", ["photo", "document", "video"])
559+
async def test_telegram_media_caption_mention_uses_utf16_offsets(media_type):
560+
adapter = _load_telegram_adapter()(
561+
make_platform_config("telegram"), {}, asyncio.Queue()
562+
)
563+
caption = "😀 @test_bot hello"
564+
sdk_message = Message.de_json(
565+
{
566+
"message_id": 1,
567+
"date": 0,
568+
"chat": {"id": -100123, "type": "supergroup"},
569+
"caption": caption,
570+
"caption_entities": [{"type": "mention", "offset": 3, "length": 9}],
571+
},
572+
bot=None,
573+
)
574+
media = create_mock_file()
575+
media.file_name = "example"
576+
update = create_mock_update(
577+
message_text=None,
578+
chat_type="supergroup",
579+
caption=caption,
580+
caption_entities=sdk_message.caption_entities,
581+
**{media_type: [media] if media_type == "photo" else media},
582+
)
583+
update.message.parse_caption_entity = sdk_message.parse_caption_entity
584+
585+
result = await adapter.convert_message(update, _build_context())
586+
587+
assert result is not None
588+
assert result.message_str == caption
589+
assert any(
590+
isinstance(part, Comp.At) and part.qq == result.self_id
591+
for part in result.message
592+
)
593+
594+
498595
@pytest.mark.asyncio
499596
async def test_telegram_document_caption_populates_message_text_and_plain():
500597
TelegramPlatformAdapter = _load_telegram_adapter()
@@ -512,6 +609,7 @@ async def test_telegram_document_caption_populates_message_text_and_plain():
512609
caption="@alice 请总结这份文档",
513610
caption_entities=[mention],
514611
)
612+
update.message.parse_caption_entity.return_value = "@alice"
515613

516614
result = await adapter.convert_message(update, _build_context())
517615

0 commit comments

Comments
 (0)