Files
AstrBot/docs/zh/dev/plugin-platform-adapter.md
Weilong Liao 7c366a708b fix: unify media reference handling (#8764)
* fix: unify media reference handling

* fix: accept bare base64 record media refs

* chore: update agents.md

* fix: unify file URI handling across media components and utilities

* fix: unify media reference type handling with MediaRefStr alias

* Potential fix for pull request finding 'CodeQL / Incomplete URL substring sanitization'

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>

* Update astrbot/core/platform/sources/discord/discord_platform_adapter.py

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

* fix: unify media handling and improve base64 decoding across components

* fix: simplify client_kwargs type definition and enhance media message handling in platform adapter

* fix: unify media utility documentation and enhance function descriptions

* perf: drop "pilk" requirement, improve audio outbound for tencent-related IM apps which using silk

* fix: unify Tencent Silk audio handling and enhance media resolver functionality

---

- Centralize media reference materialization and base64 resolution for local paths, http(s), base64://, data URIs, and legacy bare base64 payloads.
- Normalize incoming Record audio to wav and Image media to temporary jpg during preprocess, with event-scoped cleanup.
- Reuse the shared media resolver across OpenAI, Gemini, Anthropic, MiMo, DeerFlow, STT, and platform media paths while sanitizing logs and cleaning temporary conversion outputs.
- Ensure generated TTS audio is tracked for cleanup after the event finishes.

fix #8676
fix #8543
fix #7588
fix #7580
fix #8030
fix #8034
fix #7461
fix #7565
fix #6509
fix #7144
fix #7795



---------

Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2026-06-14 10:37:16 +08:00

232 lines
9.3 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
outline: deep
---
# 开发一个平台适配器
AstrBot 支持以插件的形式接入平台适配器,你可以自行接入 AstrBot 没有的平台。如飞书、钉钉甚至是哔哩哔哩私信、Minecraft。
我们以一个平台 `FakePlatform` 为例展开讲解。
首先,在插件目录下新增 `fake_platform_adapter.py``fake_platform_event.py` 文件。前者主要是平台适配器的实现,后者是平台事件的定义。
## 平台适配器
假设 FakePlatform 的客户端 SDK 是这样:
```py
import asyncio
class FakeClient():
'''模拟一个消息平台,这里 5 秒钟下发一个消息'''
def __init__(self, token: str, username: str):
self.token = token
self.username = username
# ...
async def start_polling(self):
while True:
await asyncio.sleep(5)
await getattr(self, 'on_message_received')({
'bot_id': '123',
'content': '新消息',
'username': 'zhangsan',
'userid': '123',
'message_id': 'asdhoashd',
'group_id': 'group123',
})
async def send_text(self, to: str, message: str):
print('发了消息:', to, message)
async def send_image(self, to: str, image_path: str):
print('发了消息:', to, image_path)
```
我们创建 `fake_platform_adapter.py`
```py
import asyncio
from astrbot.api.platform import Platform, AstrBotMessage, MessageMember, PlatformMetadata, MessageType
from astrbot.api.event import MessageChain
from astrbot.api.message_components import Plain, Image, Record # 消息链中的组件,可以根据需要导入
from astrbot.core.platform.astr_message_event import MessageSesion
from astrbot.api.platform import register_platform_adapter
from astrbot import logger
from .client import FakeClient
from .fake_platform_event import FakePlatformEvent
# 注册平台适配器。第一个参数为平台名,第二个为描述。第三个为默认配置。
@register_platform_adapter("fake", "fake 适配器", default_config_tmpl={
"token": "your_token",
"username": "bot_username"
})
class FakePlatformAdapter(Platform):
def __init__(self, platform_config: dict, platform_settings: dict, event_queue: asyncio.Queue) -> None:
super().__init__(event_queue)
self.config = platform_config # 上面的默认配置,用户填写后会传到这里
self.settings = platform_settings # platform_settings 平台设置。
async def send_by_session(self, session: MessageSesion, message_chain: MessageChain):
# 必须实现
await super().send_by_session(session, message_chain)
def meta(self) -> PlatformMetadata:
# 必须实现,直接像下面一样返回即可。
return PlatformMetadata(
"fake",
"fake 适配器",
)
async def run(self):
# 必须实现,这里是主要逻辑。
# FakeClient 是我们自己定义的,这里只是示例。这个是其回调函数
async def on_received(data):
logger.info(data)
abm = await self.convert_message(data=data) # 转换成 AstrBotMessage
await self.handle_msg(abm)
# 初始化 FakeClient
self.client = FakeClient(self.config['token'], self.config['username'])
self.client.on_message_received = on_received
await self.client.start_polling() # 持续监听消息,这是个堵塞方法。
async def convert_message(self, data: dict) -> AstrBotMessage:
# 将平台消息转换成 AstrBotMessage
# 这里就体现了适配程度,不同平台的消息结构不一样,这里需要根据实际情况进行转换。
abm = AstrBotMessage()
abm.type = MessageType.GROUP_MESSAGE # 还有 friend_message对应私聊。具体平台具体分析。重要
abm.group_id = data['group_id'] # 如果是私聊,这里可以不填
abm.message_str = data['content'] # 纯文本消息。重要!
abm.sender = MessageMember(user_id=data['userid'], nickname=data['username']) # 发送者。重要!
abm.message = [Plain(text=data['content'])] # 消息链。如果有其他类型的消息,直接 append 即可。重要!
abm.raw_message = data # 原始消息。
abm.self_id = data['bot_id']
abm.session_id = data['userid'] # 会话 ID。重要
abm.message_id = data['message_id'] # 消息 ID。
return abm
async def handle_msg(self, message: AstrBotMessage):
# 处理消息
message_event = FakePlatformEvent(
message_str=message.message_str,
message_obj=message,
platform_meta=self.meta(),
session_id=message.session_id,
client=self.client
)
self.commit_event(message_event) # 提交事件到事件队列。不要忘记!
```
`fake_platform_event.py`
```py
from astrbot.api.event import AstrMessageEvent, MessageChain
from astrbot.api.platform import AstrBotMessage, PlatformMetadata
from astrbot.api.message_components import Plain, Image
from .client import FakeClient
class FakePlatformEvent(AstrMessageEvent):
def __init__(self, message_str: str, message_obj: AstrBotMessage, platform_meta: PlatformMetadata, session_id: str, client: FakeClient):
super().__init__(message_str, message_obj, platform_meta, session_id)
self.client = client
async def send(self, message: MessageChain):
for i in message.chain: # 遍历消息链
if isinstance(i, Plain): # 如果是文字类型的
await self.client.send_text(to=self.get_sender_id(), message=i.text)
elif isinstance(i, Image): # 如果是图片类型的
# convert_to_file_path() resolves supported media refs through
# the shared media utilities.
img_path = await i.convert_to_file_path()
await self.client.send_image(to=self.get_sender_id(), image_path=img_path)
await super().send(message) # 需要最后加上这一段,执行父类的 send 方法。
```
## 媒体消息处理
平台适配器不需要在每个平台里重复实现媒体解析逻辑。你只需要把平台消息转换成 AstrBot 的消息组件,组件里的 `file` / `url` 可以保存以下媒体引用:
- 本地路径,例如 `/tmp/a.jpg`
- 标准 `file:` URI例如 `file:///tmp/a.jpg`
- HTTP(S) URL例如 `https://example.com/a.jpg`
- `base64://`,例如 `base64://iVBORw0KGgo...`
- Data URI例如 `data:image/png;base64,iVBORw0KGgo...`
- 历史兼容的裸 base64例如 `iVBORw0KGgo...`,但新代码不推荐主动生成这种格式
如果你手上已经是本地文件,推荐使用组件提供的构造方法,它会生成标准 `file:` URI
```py
from astrbot.api.message_components import Image, Record, Video
abm.message.append(Image.fromFileSystem("/tmp/image.png"))
abm.message.append(Record.fromFileSystem("/tmp/audio.wav"))
abm.message.append(Video.fromFileSystem("/tmp/video.mp4"))
```
如果平台只给了可访问的 URL直接放到组件里即可
```py
abm.message.append(Image(file=image_url, url=image_url))
abm.message.append(Record(file=audio_url, url=audio_url))
abm.message.append(Video(file=video_url, url=video_url))
```
进入插件和 LLM 前AstrBot 的预处理阶段会尽量把消息链里的媒体标准化:
- `Image` 会通过统一媒体处理工具落地为本地文件,并在需要时转换为 JPEG。
- `Record` 会落地为本地文件,并在需要时转换为 WAV。
- `Reply` 中的 `Image` / `Record` 也会做同样处理。
- 这些由核心创建的临时文件会挂到当前事件上,在事件结束后清理。
发送消息时,如果平台 SDK 需要本地文件路径,调用组件的 `convert_to_file_path()` 即可,不要手写 `path.startswith("file://")` 之类的判断:
```py
if isinstance(i, Image):
image_path = await i.convert_to_file_path()
await self.client.send_image(to=self.get_sender_id(), image_path=image_path)
elif isinstance(i, Record):
audio_path = await i.convert_to_file_path()
await self.client.send_audio(to=self.get_sender_id(), audio_path=audio_path)
elif isinstance(i, Video):
video_path = await i.convert_to_file_path()
await self.client.send_video(to=self.get_sender_id(), video_path=video_path)
```
如果适配器自己下载了平台媒体并写入 AstrBot 临时目录,请在创建事件后把路径登记到事件上,避免事件结束后留下临时文件:
```py
message_event.track_temporary_local_file(temp_media_path)
```
最后main.py 只需这样,在初始化的时候导入 fake_platform_adapter 模块。装饰器会自动注册。
```py
from astrbot.api.star import Context, Star
class MyPlugin(Star):
def __init__(self, context: Context):
from .fake_platform_adapter import FakePlatformAdapter # noqa
```
搞好后,运行 AstrBot
![image](https://files.astrbot.app/docs/source/images/plugin-platform-adapter/QQ_1738155926221.png)
这里出现了我们创建的 fake。
![image](https://files.astrbot.app/docs/source/images/plugin-platform-adapter/QQ_1738155982211.png)
启动后,可以看到正常工作:
![image](https://files.astrbot.app/docs/source/images/plugin-platform-adapter/QQ_1738156166893.png)
有任何疑问欢迎加群询问~