From 0bee2faadae2e80a4df153669464089440265b47 Mon Sep 17 00:00:00 2001 From: kldsjfas <166977656+kldsjfas@users.noreply.github.com> Date: Tue, 7 Jul 2026 00:41:13 +0800 Subject: [PATCH 1/3] fix: apply text-to-image when TTS dual output keeps text in the chain --- .../core/pipeline/result_decorate/stage.py | 12 +- tests/unit/test_result_decorate_t2i.py | 177 ++++++++++++++++++ 2 files changed, 185 insertions(+), 4 deletions(-) create mode 100644 tests/unit/test_result_decorate_t2i.py diff --git a/astrbot/core/pipeline/result_decorate/stage.py b/astrbot/core/pipeline/result_decorate/stage.py index 5956c8b5ef..8558fe6c70 100644 --- a/astrbot/core/pipeline/result_decorate/stage.py +++ b/astrbot/core/pipeline/result_decorate/stage.py @@ -346,13 +346,17 @@ async def process( result.chain = new_chain # 文本转图片 - elif ( + if ( result.use_t2i_ is None and self.ctx.astrbot_config["t2i"] ) or result.use_t2i_: parts = [] + records = [] for comp in result.chain: if isinstance(comp, Plain): parts.append("\n\n" + comp.text) + elif isinstance(comp, Record): + # TTS 双输出时语音段与文本段交错，跳过语音段并在转图后保留 + records.append(comp) else: break plain_str = "".join(parts) @@ -374,7 +378,7 @@ async def process( ) if url: if url.startswith("http"): - result.chain = [Image.fromURL(url)] + result.chain = [*records, Image.fromURL(url)] elif ( self.ctx.astrbot_config["t2i_use_file_service"] and self.ctx.astrbot_config["callback_api_base"] @@ -382,9 +386,9 @@ async def process( token = await file_token_service.register_file(url) url = f"{self.ctx.astrbot_config['callback_api_base']}/api/file/{token}" logger.debug(f"已注册：{url}") - result.chain = [Image.fromURL(url)] + result.chain = [*records, Image.fromURL(url)] else: - result.chain = [Image.fromFileSystem(url)] + result.chain = [*records, Image.fromFileSystem(url)] # 触发转发消息 if event.get_platform_name() == "aiocqhttp": diff --git a/tests/unit/test_result_decorate_t2i.py b/tests/unit/test_result_decorate_t2i.py new file mode 100644 index 0000000000..6b198263d8 --- /dev/null +++ b/tests/unit/test_result_decorate_t2i.py @@ -0,0 +1,177 @@ +from types import SimpleNamespace + +import pytest + +import astrbot.core.pipeline.result_decorate.stage as stage_module +from astrbot.core.message.components import Image, Plain, Record +from astrbot.core.message.message_event_result import ( + MessageEventResult, + ResultContentType, +) +from astrbot.core.pipeline.result_decorate.stage import ResultDecorateStage + +LONG_TEXT = "这是一段用于验证文本转图像阈值的长文本。" * 20 +SHORT_TEXT = "短文本" + + +def _build_config(*, tts_enable: bool, dual_output: bool, t2i: bool) -> dict: + return { + "platform_settings": { + "reply_prefix": "", + "reply_with_mention": False, + "reply_with_quote": False, + "forward_threshold": 999999, + "segmented_reply": { + "enable": False, + "only_llm_result": True, + "words_count_threshold": 150, + "regex": r".*?[。？！~…]+|.+$", + "content_cleanup_rule": "", + }, + }, + "t2i": t2i, + "t2i_word_threshold": 50, + "t2i_strategy": "local", + "t2i_active_template": "base", + "t2i_use_file_service": False, + "callback_api_base": "", + "provider_tts_settings": { + "enable": tts_enable, + "dual_output": dual_output, + "use_file_service": False, + "trigger_probability": 1, + }, + "content_safety": {"also_use_in_response": False}, + "provider_settings": {"display_reasoning_text": False}, + } + + +class FakeTTSProvider: + async def get_audio(self, text: str) -> str: + del text + return "/tmp/fake_tts_audio.wav" + + +class FakeEvent: + def __init__(self, result: MessageEventResult): + self._result = result + self.plugins_name = None + self.unified_msg_origin = "test:GroupMessage:1" + self.tracked_files: list[str] = [] + + def get_result(self) -> MessageEventResult: + return self._result + + def get_extra(self, key: str): + del key + return None + + def get_platform_name(self) -> str: + return "test" + + def is_stopped(self) -> bool: + return False + + def track_temporary_local_file(self, path: str) -> None: + self.tracked_files.append(path) + + +async def _run_stage( + config: dict, result: MessageEventResult, monkeypatch +) -> FakeEvent: + tts_provider = ( + FakeTTSProvider() if config["provider_tts_settings"]["enable"] else None + ) + ctx = SimpleNamespace( + astrbot_config=config, + plugin_manager=SimpleNamespace( + context=SimpleNamespace( + get_using_tts_provider=lambda umo: tts_provider, + ), + ), + ) + + async def _always_tts(event): + del event + return True + + async def _fake_render(text, return_url=True, use_network=False, template_name=""): + del text, return_url, use_network, template_name + return "http://fake.local/t2i.png" + + monkeypatch.setattr( + stage_module.SessionServiceManager, + "should_process_tts_request", + staticmethod(_always_tts), + ) + monkeypatch.setattr(stage_module.html_renderer, "render_t2i", _fake_render) + + stage = ResultDecorateStage() + await stage.initialize(ctx) + event = FakeEvent(result) + generator = stage.process(event) + if generator is not None: + async for _ in generator: + pass + return event + + +@pytest.mark.asyncio +async def test_t2i_still_applies_when_tts_dual_output(monkeypatch): + result = MessageEventResult(chain=[Plain(LONG_TEXT)]).set_result_content_type( + ResultContentType.LLM_RESULT + ) + await _run_stage( + _build_config(tts_enable=True, dual_output=True, t2i=True), + result, + monkeypatch, + ) + + assert any(isinstance(comp, Record) for comp in result.chain) + assert isinstance(result.chain[-1], Image) + assert not any(isinstance(comp, Plain) for comp in result.chain) + + +@pytest.mark.asyncio +async def test_t2i_without_tts_keeps_original_behavior(monkeypatch): + result = MessageEventResult(chain=[Plain(LONG_TEXT)]).set_result_content_type( + ResultContentType.LLM_RESULT + ) + await _run_stage( + _build_config(tts_enable=False, dual_output=False, t2i=True), + result, + monkeypatch, + ) + + assert len(result.chain) == 1 + assert isinstance(result.chain[0], Image) + + +@pytest.mark.asyncio +async def test_dual_output_below_threshold_keeps_voice_and_text(monkeypatch): + result = MessageEventResult(chain=[Plain(SHORT_TEXT)]).set_result_content_type( + ResultContentType.LLM_RESULT + ) + await _run_stage( + _build_config(tts_enable=True, dual_output=True, t2i=True), + result, + monkeypatch, + ) + + assert isinstance(result.chain[0], Record) + assert isinstance(result.chain[1], Plain) + + +@pytest.mark.asyncio +async def test_tts_without_dual_output_sends_voice_only(monkeypatch): + result = MessageEventResult(chain=[Plain(LONG_TEXT)]).set_result_content_type( + ResultContentType.LLM_RESULT + ) + await _run_stage( + _build_config(tts_enable=True, dual_output=False, t2i=True), + result, + monkeypatch, + ) + + assert len(result.chain) == 1 + assert isinstance(result.chain[0], Record) From 0aca43ce812a6cdf7e0503f50e2314d892e96f72 Mon Sep 17 00:00:00 2001 From: kldsjfas <166977656+kldsjfas@users.noreply.github.com> Date: Tue, 7 Jul 2026 00:47:42 +0800 Subject: [PATCH 2/3] fix: keep remaining components after t2i conversion --- astrbot/core/pipeline/result_decorate/stage.py | 11 +++++++---- tests/unit/test_result_decorate_t2i.py | 17 ++++++++++++++++- 2 files changed, 23 insertions(+), 5 deletions(-) diff --git a/astrbot/core/pipeline/result_decorate/stage.py b/astrbot/core/pipeline/result_decorate/stage.py index 8558fe6c70..0bb9b2e6f4 100644 --- a/astrbot/core/pipeline/result_decorate/stage.py +++ b/astrbot/core/pipeline/result_decorate/stage.py @@ -351,13 +351,16 @@ async def process( ) or result.use_t2i_: parts = [] records = [] - for comp in result.chain: + remain = [] + for i, comp in enumerate(result.chain): if isinstance(comp, Plain): parts.append("\n\n" + comp.text) elif isinstance(comp, Record): # TTS 双输出时语音段与文本段交错，跳过语音段并在转图后保留 records.append(comp) else: + # 其余消息段不参与转图，转图后追加回结果链避免丢失 + remain = result.chain[i:] break plain_str = "".join(parts) if plain_str and len(plain_str) > self.t2i_word_threshold: @@ -378,7 +381,7 @@ async def process( ) if url: if url.startswith("http"): - result.chain = [*records, Image.fromURL(url)] + result.chain = [*records, Image.fromURL(url), *remain] elif ( self.ctx.astrbot_config["t2i_use_file_service"] and self.ctx.astrbot_config["callback_api_base"] @@ -386,9 +389,9 @@ async def process( token = await file_token_service.register_file(url) url = f"{self.ctx.astrbot_config['callback_api_base']}/api/file/{token}" logger.debug(f"已注册：{url}") - result.chain = [*records, Image.fromURL(url)] + result.chain = [*records, Image.fromURL(url), *remain] else: - result.chain = [*records, Image.fromFileSystem(url)] + result.chain = [*records, Image.fromFileSystem(url), *remain] # 触发转发消息 if event.get_platform_name() == "aiocqhttp": diff --git a/tests/unit/test_result_decorate_t2i.py b/tests/unit/test_result_decorate_t2i.py index 6b198263d8..fac8b9a45e 100644 --- a/tests/unit/test_result_decorate_t2i.py +++ b/tests/unit/test_result_decorate_t2i.py @@ -3,7 +3,7 @@ import pytest import astrbot.core.pipeline.result_decorate.stage as stage_module -from astrbot.core.message.components import Image, Plain, Record +from astrbot.core.message.components import At, Image, Plain, Record from astrbot.core.message.message_event_result import ( MessageEventResult, ResultContentType, @@ -162,6 +162,21 @@ async def test_dual_output_below_threshold_keeps_voice_and_text(monkeypatch): assert isinstance(result.chain[1], Plain) +@pytest.mark.asyncio +async def test_t2i_keeps_components_after_text(monkeypatch): + result = MessageEventResult( + chain=[Plain(LONG_TEXT), At(qq="10001", name="someone")] + ).set_result_content_type(ResultContentType.LLM_RESULT) + await _run_stage( + _build_config(tts_enable=False, dual_output=False, t2i=True), + result, + monkeypatch, + ) + + assert isinstance(result.chain[0], Image) + assert isinstance(result.chain[1], At) + + @pytest.mark.asyncio async def test_tts_without_dual_output_sends_voice_only(monkeypatch): result = MessageEventResult(chain=[Plain(LONG_TEXT)]).set_result_content_type( From 993e8fdf110b342c060c4dda46513f66c4cd1de5 Mon Sep 17 00:00:00 2001 From: kldsjfas <166977656+kldsjfas@users.noreply.github.com> Date: Tue, 7 Jul 2026 00:48:51 +0800 Subject: [PATCH 3/3] style: wrap overlong chain assignment --- astrbot/core/pipeline/result_decorate/stage.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/astrbot/core/pipeline/result_decorate/stage.py b/astrbot/core/pipeline/result_decorate/stage.py index 0bb9b2e6f4..1e05a3154f 100644 --- a/astrbot/core/pipeline/result_decorate/stage.py +++ b/astrbot/core/pipeline/result_decorate/stage.py @@ -391,7 +391,11 @@ async def process( logger.debug(f"已注册：{url}") result.chain = [*records, Image.fromURL(url), *remain] else: - result.chain = [*records, Image.fromFileSystem(url), *remain] + result.chain = [ + *records, + Image.fromFileSystem(url), + *remain, + ] # 触发转发消息 if event.get_platform_name() == "aiocqhttp":