修复已知bug

nonebot_plugin_marshoai/azure.py

@@ -234,15 +234,28 @@ async def send_markdown(msg: str):
     人工智能给出的回答一般不会包含 HTML 嵌入其中，但是包含图片或者 LaTeX 公式、代码块，都很正常。
     这个函数会把这些都以图片形式嵌入消息体。
     """
+
+    if not IMG_TAG_PATTERN.search(msg):  # 没有图片标签
+        await UniMessage(msg).send(reply_to=True)
+        return
+
     result_msg = UniMessage()
     code_blank_uuid_map = [
         (uuid.uuid4().hex, cbp.group()) for cbp in CODE_BLOCK_PATTERN.finditer(msg)
     ]
 
+    last_tag_index = 0
+
     # 代码块渲染麻烦，先不处理
     for rep, torep in code_blank_uuid_map:
         msg = msg.replace(torep, rep)
 
+    # for to_rep in CODE_SINGLE_PATTERN.finditer(msg):
+    #     code_blank_uuid_map.append((rep := uuid.uuid4().hex, to_rep.group()))
+    #     msg = msg.replace(to_rep.group(), rep)
+
+    print("#####################\n", msg, "\n\n")
+
     # 插入图片
     for each_img_tag in IMG_TAG_PATTERN.finditer(msg):
         img_tag = await get_back_uuidcodeblock(
@@ -254,15 +267,29 @@ async def send_markdown(msg: str):
         result_msg.append(
             TextMsg(
                 await get_back_uuidcodeblock(
-                    msg[: msg.find(img_tag)], code_blank_uuid_map
+                    msg[last_tag_index : msg.find(img_tag)], code_blank_uuid_map
                 )
             )
         )
 
-        result_msg.append(ImageMsg(url=image_url, name=image_description + ".png"))
+        last_tag_index = msg.find(img_tag) + len(img_tag)
 
+        if image_ := await get_image_raw_and_type(image_url):
+
+            result_msg.append(
+                ImageMsg(
+                    raw=image_[0], mimetype=image_[1], name=image_description + ".png"
+                )
+            )
             result_msg.append(TextMsg("（{}）".format(image_description)))
 
+        else:
+            result_msg.append(TextMsg(img_tag))
+
+    result_msg.append(
+        TextMsg(await get_back_uuidcodeblock(msg[last_tag_index:], code_blank_uuid_map))
+    )
+
     await result_msg.send(reply_to=True)
 
 

nonebot_plugin_marshoai/constants.py

@@ -37,7 +37,7 @@ https://github.com/LiteyukiStudio/marshoai-melo"""
 
 # 正则匹配代码块
 CODE_BLOCK_PATTERN = re.compile(
-    r"```(.*?)```|`(.*?)`",
+    r"```(.*?)```|`(.*?)`", re.DOTALL
 )
 # 正则匹配完整图片标签字段
 IMG_TAG_PATTERN = re.compile(r"!\[[^\]]*\]\([^()]*\)")
@@ -47,5 +47,5 @@ IMG_TAG_PATTERN = re.compile(r"!\[[^\]]*\]\([^()]*\)")
 # INTAG_TEXT_PATTERN = re.compile(r'!\[([^\]]*)\]')
 # 正则匹配 LaTeX 公式内容
 LATEX_PATTERN = re.compile(
-    r"\\begin\{equation\}(.*?)\\\end\{equation\}|(?<!\$)(\$(.*?)\$|\$\$(.*?)\$\$|\\\[(.*?)\\\]|\\\[.*?\\\]|\\\((.*?)\\\))",
+    r"\\begin\{equation\}(.*?)\\end\{equation\}|(?<!\$)(\$(.*?)\$|\$\$(.*?)\$\$|\\\[(.*?)\\\]|\\\[.*?\\\]|\\\((.*?)\\\))",
 )

nonebot_plugin_marshoai/util.py

@@ -2,7 +2,7 @@ import base64
 import mimetypes
 import os
 import json
-from typing import Any
+from typing import Any, Optional
 import httpx
 import nonebot_plugin_localstore as store
 
@@ -16,31 +16,65 @@ nickname_json = None  # 记录昵称
 praises_json = None  # 记录夸赞名单
 loaded_target_list = []  # 记录已恢复备份的上下文的列表
 
-
-async def get_image_b64(url):
 # noinspection LongLine
-    headers = {
+chromium_headers = {
     "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
 }
 
+
+async def get_image_raw_and_type(
+    url: str, timeout: int = 10
+) -> Optional[tuple[bytes, str]]:
+    """
+    获取图片的二进制数据
+
+    参数:
+        url: str 图片链接
+        timeout: int 超时时间 秒
+
+    return:
+        tuple[bytes, str]: 图片二进制数据, 图片MIME格式
+
+    """
+
     async with httpx.AsyncClient() as client:
-        response = await client.get(url, headers=headers)
+        response = await client.get(url, headers=chromium_headers, timeout=timeout)
         if response.status_code == 200:
             # 获取图片数据
-            image_data = response.content
             content_type = response.headers.get("Content-Type")
             if not content_type:
                 content_type = mimetypes.guess_type(url)[0]
             # image_format = content_type.split("/")[1] if content_type else "jpeg"
-            base64_image = base64.b64encode(image_data).decode("utf-8")
+            return response.content, str(content_type)
-            data_url = f"data:{content_type};base64,{base64_image}"
+        else:
+            return None
+
+
+async def get_image_b64(url: str, timeout: int = 10) -> Optional[str]:
+    """
+    获取图片的base64编码
+
+    参数:
+        url: 图片链接
+        timeout: 超时时间 秒
+
+    return: 图片base64编码
+    """
+
+    if data_type := await get_image_raw_and_type(url, timeout):
+        # image_format = content_type.split("/")[1] if content_type else "jpeg"
+        base64_image = base64.b64encode(data_type[0]).decode("utf-8")
+        data_url = "data:{};base64,{}".format(data_type[1], base64_image)
         return data_url
     else:
         return None
 
 
 async def make_chat(
-    client: ChatCompletionsClient, msg: list, model_name: str, tools: list = None
+    client: ChatCompletionsClient,
+    msg: list,
+    model_name: str,
+    tools: Optional[list] = None,
 ):
     """调用ai获取回复