feat: 重构图片消息传递逻辑 (#957, #955)

pkg/pipeline/preproc/preproc.py

@@ -61,9 +61,9 @@ class PreProcessor(stage.PipelineStage):
                 )
             elif isinstance(me, platform_message.Image):
                 if self.ap.provider_cfg.data['enable-vision'] and (self.ap.provider_cfg.data['runner'] != 'local-agent' or query.use_model.vision_supported):
-                    if me.url is not None:
+                    if me.base64 is not None:
                         content_list.append(
-                            llm_entities.ContentElement.from_image_url(str(me.url))
+                            llm_entities.ContentElement.from_image_base64(me.base64)
                         )
 
         query.user_message = llm_entities.Message(

pkg/platform/sources/aiocqhttp.py

@@ -6,6 +6,7 @@ import time
 import datetime
 
 import aiocqhttp
+import aiohttp
 
 from .. import adapter
 from ...pipeline.longtext.strategies import forward
@@ -13,12 +14,12 @@ from ...core import app
 from ..types import message as platform_message
 from ..types import events as platform_events
 from ..types import entities as platform_entities
-
+from ...utils import image
 
 class AiocqhttpMessageConverter(adapter.MessageConverter):
 
     @staticmethod
-    def yiri2target(message_chain: platform_message.MessageChain) -> typing.Tuple[list, int, datetime.datetime]:
+    async def yiri2target(message_chain: platform_message.MessageChain) -> typing.Tuple[list, int, datetime.datetime]:
         msg_list = aiocqhttp.Message()
 
         msg_id = 0
@@ -59,7 +60,7 @@ class AiocqhttpMessageConverter(adapter.MessageConverter):
             elif type(msg) is forward.Forward:
 
                 for node in msg.node_list:
-                    msg_list.extend(AiocqhttpMessageConverter.yiri2target(node.message_chain)[0])
+                    msg_list.extend(await AiocqhttpMessageConverter.yiri2target(node.message_chain)[0])
                 
             else:
                 msg_list.append(aiocqhttp.MessageSegment.text(str(msg)))
@@ -67,7 +68,7 @@ class AiocqhttpMessageConverter(adapter.MessageConverter):
         return msg_list, msg_id, msg_time
 
     @staticmethod
-    def target2yiri(message: str, message_id: int = -1):
+    async def target2yiri(message: str, message_id: int = -1):
         message = aiocqhttp.Message(message)
 
         yiri_msg_list = []
@@ -89,7 +90,8 @@ class AiocqhttpMessageConverter(adapter.MessageConverter):
             elif msg.type == "text":
                 yiri_msg_list.append(platform_message.Plain(text=msg.data["text"]))
             elif msg.type == "image":
-                yiri_msg_list.append(platform_message.Image(url=msg.data["url"]))
+                image_base64, image_format = await image.qq_image_url_to_base64(msg.data['url'])
+                yiri_msg_list.append(platform_message.Image(base64=f"data:image/{image_format};base64,{image_base64}"))
 
         chain = platform_message.MessageChain(yiri_msg_list)
 
@@ -99,9 +101,9 @@ class AiocqhttpMessageConverter(adapter.MessageConverter):
 class AiocqhttpEventConverter(adapter.EventConverter):
 
     @staticmethod
-    def yiri2target(event: platform_events.Event, bot_account_id: int):
+    async def yiri2target(event: platform_events.Event, bot_account_id: int):
 
-        msg, msg_id, msg_time = AiocqhttpMessageConverter.yiri2target(event.message_chain)
+        msg, msg_id, msg_time = await AiocqhttpMessageConverter.yiri2target(event.message_chain)
 
         if type(event) is platform_events.GroupMessage:
             role = "member"
@@ -164,8 +166,8 @@ class AiocqhttpEventConverter(adapter.EventConverter):
             return aiocqhttp.Event.from_payload(payload)
 
     @staticmethod
-    def target2yiri(event: aiocqhttp.Event):
+    async def target2yiri(event: aiocqhttp.Event):
-        yiri_chain = AiocqhttpMessageConverter.target2yiri(
+        yiri_chain = await AiocqhttpMessageConverter.target2yiri(
             event.message, event.message_id
         )
 
@@ -242,7 +244,7 @@ class AiocqhttpAdapter(adapter.MessageSourceAdapter):
     async def send_message(
         self, target_type: str, target_id: str, message: platform_message.MessageChain
     ):
-        aiocq_msg = AiocqhttpMessageConverter.yiri2target(message)[0]
+        aiocq_msg = await AiocqhttpMessageConverter.yiri2target(message)[0]
 
         if target_type == "group":
             await self.bot.send_group_msg(group_id=int(target_id), message=aiocq_msg)
@@ -255,8 +257,8 @@ class AiocqhttpAdapter(adapter.MessageSourceAdapter):
         message: platform_message.MessageChain,
         quote_origin: bool = False,
     ):  
-        aiocq_event = AiocqhttpEventConverter.yiri2target(message_source, self.bot_account_id)
+        aiocq_event = await AiocqhttpEventConverter.yiri2target(message_source, self.bot_account_id)
-        aiocq_msg = AiocqhttpMessageConverter.yiri2target(message)[0]
+        aiocq_msg = (await AiocqhttpMessageConverter.yiri2target(message))[0]
         if quote_origin:
             aiocq_msg = aiocqhttp.MessageSegment.reply(aiocq_event.message_id) + aiocq_msg
 
@@ -276,7 +278,7 @@ class AiocqhttpAdapter(adapter.MessageSourceAdapter):
         async def on_message(event: aiocqhttp.Event):
             self.bot_account_id = event.self_id
             try:
-                return await callback(self.event_converter.target2yiri(event), self)
+                return await callback(await self.event_converter.target2yiri(event), self)
             except:
                 traceback.print_exc()
 

pkg/provider/entities.py

@@ -38,6 +38,8 @@ class ContentElement(pydantic.BaseModel):
 
     image_url: typing.Optional[ImageURLContentObject] = None
 
+    image_base64: typing.Optional[str] = None
+
     def __str__(self):
         if self.type == 'text':
             return self.text
@@ -54,6 +56,10 @@ class ContentElement(pydantic.BaseModel):
     def from_image_url(cls, image_url: str):
         return cls(type='image_url', image_url=ImageURLContentObject(url=image_url))
     
+    @classmethod
+    def from_image_base64(cls, image_base64: str):
+        return cls(type='image_base64', image_base64=image_base64)
+
 
 class Message(pydantic.BaseModel):
     """消息"""

pkg/provider/modelmgr/requester.py

@@ -48,6 +48,7 @@ class LLMAPIRequester(metaclass=abc.ABCMeta):
     @abc.abstractmethod
     async def call(
         self,
+        query: core_entities.Query,
         model: modelmgr_entities.LLMModelInfo,
         messages: typing.List[llm_entities.Message],
         funcs: typing.List[tools_entities.LLMFunction] = None,

pkg/provider/modelmgr/requesters/anthropicmsgs.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import typing
 import traceback
+import base64
 
 import anthropic
 import httpx
@@ -39,6 +40,7 @@ class AnthropicMessages(requester.LLMAPIRequester):
 
     async def call(
         self,
+        query: core_entities.Query,
         model: entities.LLMModelInfo,
         messages: typing.List[llm_entities.Message],
         funcs: typing.List[tools_entities.LLMFunction] = None,
@@ -70,28 +72,26 @@ class AnthropicMessages(requester.LLMAPIRequester):
             if isinstance(m.content, str) and m.content.strip() != "":
                 req_messages.append(m.dict(exclude_none=True))
             elif isinstance(m.content, list):
-                # m.content = [
-                #     c for c in m.content if c.type == "text"
-                # ]
-
-                # if len(m.content) > 0:
-                #     req_messages.append(m.dict(exclude_none=True))
 
                 msg_dict = m.dict(exclude_none=True)
 
                 for i, ce in enumerate(m.content):
-                    if ce.type == "image_url":
+
-                        base64_image, image_format = await image.qq_image_url_to_base64(ce.image_url.url)
+                    if ce.type == "image_base64":
+                        image_b64, image_format = await image.extract_b64_and_format(ce.image_base64)
+
                         alter_image_ele = {
                             "type": "image",
                             "source": {
                                 "type": "base64",
                                 "media_type": f"image/{image_format}",
-                                "data": base64_image
+                                "data": image_b64
                             }
                         }
                         msg_dict["content"][i] = alter_image_ele
 
+                print(msg_dict)
+
                 req_messages.append(msg_dict)
 
         args["messages"] = req_messages

pkg/provider/modelmgr/requesters/chatcmpl.py

@@ -65,6 +65,7 @@ class OpenAIChatCompletions(requester.LLMAPIRequester):
 
     async def _closure(
         self,
+        query: core_entities.Query,
         req_messages: list[dict],
         use_model: entities.LLMModelInfo,
         use_funcs: list[tools_entities.LLMFunction] = None,
@@ -87,8 +88,12 @@ class OpenAIChatCompletions(requester.LLMAPIRequester):
         for msg in messages:
             if 'content' in msg and isinstance(msg["content"], list):
                 for me in msg["content"]:
-                    if me["type"] == "image_url":
+                    if me["type"] == "image_base64":
-                        me["image_url"]['url'] = await self.get_base64_str(me["image_url"]['url'])
+                        me["image_url"] = {
+                            "url": me["image_base64"]
+                        }
+                        me["type"] = "image_url"
+                        del me["image_base64"]
 
         args["messages"] = messages
 
@@ -102,6 +107,7 @@ class OpenAIChatCompletions(requester.LLMAPIRequester):
     
     async def call(
         self,
+        query: core_entities.Query,
         model: entities.LLMModelInfo,
         messages: typing.List[llm_entities.Message],
         funcs: typing.List[tools_entities.LLMFunction] = None,
@@ -118,7 +124,7 @@ class OpenAIChatCompletions(requester.LLMAPIRequester):
             req_messages.append(msg_dict)
 
         try:
-            return await self._closure(req_messages, model, funcs)
+            return await self._closure(query, req_messages, model, funcs)
         except asyncio.TimeoutError:
             raise errors.RequesterError('请求超时')
         except openai.BadRequestError as e:
@@ -134,11 +140,3 @@ class OpenAIChatCompletions(requester.LLMAPIRequester):
             raise errors.RequesterError(f'请求过于频繁或余额不足: {e.message}')
         except openai.APIError as e:
             raise errors.RequesterError(f'请求错误: {e.message}')
-
-    @async_lru.alru_cache(maxsize=128)
-    async def get_base64_str(
-        self,
-        original_url: str,
-    ) -> str:
-        base64_image, image_format = await image.qq_image_url_to_base64(original_url)
-        return f"data:image/{image_format};base64,{base64_image}"

pkg/provider/modelmgr/requesters/ollamachat.py

@@ -6,6 +6,7 @@ import typing
 from typing import Union, Mapping, Any, AsyncIterator
 import uuid
 import json
+import base64
 
 import async_lru
 import ollama
@@ -13,7 +14,7 @@ import ollama
 from .. import entities, errors, requester
 from ... import entities as llm_entities
 from ...tools import entities as tools_entities
-from ....core import app
+from ....core import app, entities as core_entities
 from ....utils import image
 
 REQUESTER_NAME: str = "ollama-chat"
@@ -43,7 +44,7 @@ class OllamaChatCompletions(requester.LLMAPIRequester):
             **args
         )
 
-    async def _closure(self, req_messages: list[dict], use_model: entities.LLMModelInfo,
+    async def _closure(self, query: core_entities.Query, req_messages: list[dict], use_model: entities.LLMModelInfo,
                        user_funcs: list[tools_entities.LLMFunction] = None) -> (
             llm_entities.Message):
         args: Any = self.request_cfg['args'].copy()
@@ -57,9 +58,9 @@ class OllamaChatCompletions(requester.LLMAPIRequester):
                 for me in msg["content"]:
                     if me["type"] == "text":
                         text_content.append(me["text"])
-                    elif me["type"] == "image_url":
+                    elif me["type"] == "image_base64":
-                        image_url = await self.get_base64_str(me["image_url"]['url'])
+                        image_urls.append(me["image_base64"])
-                        image_urls.append(image_url)
+                        
                 msg["content"] = "\n".join(text_content)
                 msg["images"] = [url.split(',')[1] for url in image_urls]
             if 'tool_calls' in msg:  # LangBot 内部以 str 存储 tool_calls 的参数，这里需要转换为 dict
@@ -109,6 +110,7 @@ class OllamaChatCompletions(requester.LLMAPIRequester):
 
     async def call(
             self,
+            query: core_entities.Query,
             model: entities.LLMModelInfo,
             messages: typing.List[llm_entities.Message],
             funcs: typing.List[tools_entities.LLMFunction] = None,
@@ -122,14 +124,6 @@ class OllamaChatCompletions(requester.LLMAPIRequester):
                     msg_dict["content"] = "\n".join(part["text"] for part in content)
             req_messages.append(msg_dict)
         try:
-            return await self._closure(req_messages, model, funcs)
+            return await self._closure(query, req_messages, model, funcs)
         except asyncio.TimeoutError:
             raise errors.RequesterError('请求超时')
-
-    @async_lru.alru_cache(maxsize=128)
-    async def get_base64_str(
-            self,
-            original_url: str,
-    ) -> str:
-        base64_image, image_format = await image.qq_image_url_to_base64(original_url)
-        return f"data:image/{image_format};base64,{base64_image}"

pkg/provider/runners/localagent.py

@@ -23,7 +23,7 @@ class LocalAgentRunner(runner.RequestRunner):
         req_messages = query.prompt.messages.copy() + query.messages.copy() + [query.user_message]
 
         # 首次请求
-        msg = await query.use_model.requester.call(query.use_model, req_messages, query.use_funcs)
+        msg = await query.use_model.requester.call(query, query.use_model, req_messages, query.use_funcs)
 
         yield msg
 
@@ -61,7 +61,7 @@ class LocalAgentRunner(runner.RequestRunner):
                     req_messages.append(err_msg)
 
             # 处理完所有调用，再次请求
-            msg = await query.use_model.requester.call(query.use_model, req_messages, query.use_funcs)
+            msg = await query.use_model.requester.call(query, query.use_model, req_messages, query.use_funcs)
 
             yield msg
 

pkg/utils/image.py

@@ -1,9 +1,11 @@
 import base64
 import typing
+import io
 from urllib.parse import urlparse, parse_qs
 import ssl
 
 import aiohttp
+import PIL.Image
 
 
 def get_qq_image_downloadable_url(image_url: str) -> tuple[str, dict]:
@@ -14,7 +16,7 @@ def get_qq_image_downloadable_url(image_url: str) -> tuple[str, dict]:
 
 
 async def get_qq_image_bytes(image_url: str) -> tuple[bytes, str]:
-    """获取QQ图片的bytes"""
+    """[弃用]获取QQ图片的bytes"""
     image_url, query = get_qq_image_downloadable_url(image_url)
     ssl_context = ssl.create_default_context()
     ssl_context.check_hostname = False
@@ -24,8 +26,11 @@ async def get_qq_image_bytes(image_url: str) -> tuple[bytes, str]:
             resp.raise_for_status()
             file_bytes = await resp.read()
             content_type = resp.headers.get('Content-Type')
-            if not content_type or not content_type.startswith('image/'):
+            if not content_type:
                 image_format = 'jpeg'
+            elif not content_type.startswith('image/'):
+                pil_img = PIL.Image.open(io.BytesIO(file_bytes))
+                image_format = pil_img.format.lower()
             else: 
                 image_format = content_type.split('/')[-1]
             return file_bytes, image_format
@@ -34,7 +39,7 @@ async def get_qq_image_bytes(image_url: str) -> tuple[bytes, str]:
 async def qq_image_url_to_base64(
     image_url: str
 ) -> typing.Tuple[str, str]:
-    """将QQ图片URL转为base64，并返回图片格式
+    """[弃用]将QQ图片URL转为base64，并返回图片格式
 
     Args:
         image_url (str): QQ图片URL
@@ -52,3 +57,13 @@ async def qq_image_url_to_base64(
     base64_str = base64.b64encode(file_bytes).decode()
 
     return base64_str, image_format
+
+async def extract_b64_and_format(image_base64_data: str) -> typing.Tuple[str, str]:
+    """提取base64编码和图片格式
+    
+    data:image/jpeg;base64,xxx
+    提取出base64编码和图片格式
+    """
+    base64_str = image_base64_data.split(',')[-1]
+    image_format = image_base64_data.split(':')[-1].split(';')[0].split('/')[-1]
+    return base64_str, image_format