MeUtils 2025.5.15.14.3.22__py3-none-any.whl → 2025.5.29.18.12.8__py3-none-any.whl

meutils/apis/siliconflow/videos.py

@@ -20,19 +20,17 @@ from openai import OpenAI, AsyncOpenAI
 
 check_token = partial(check_token_for_siliconflow, threshold=0.01)
 
-MODELS_MAP = {
-    "hunyuan-video": "tencent/HunyuanVideo",
-    "hunyuanvideo": "tencent/HunyuanVideo",
-    "mochi-1-preview": "genmo/mochi-1-preview",
-    "ltx-video": "Lightricks/LTX-Video",
-}
 """
+
+tencent/HunyuanVideo-HD
+
 Wan-AI/Wan2.1-T2V-14B
 Wan-AI/Wan2.1-T2V-14B-Turbo
 
 Wan-AI/Wan2.1-I2V-14B-720P
 Wan-AI/Wan2.1-I2V-14B-720P-Turbo
 
+
 16:9 👉 1280×720
 9:16 👉 720×1280
 1:1 👉 960×960
@@ -44,17 +42,12 @@ Wan-AI/Wan2.1-I2V-14B-720P-Turbo
 async def create_task(request: VideoRequest, token: Optional[str] = None):
     token = token or await get_next_token_for_polling(FEISHU_URL_FREE, check_token=check_token, from_redis=True)
 
-    if request.image:
-        request.image = await to_base64(request.image)
+    if 'Wan-AI' in request.model:
+        request.model = "Wan-AI/Wan2.1-T2V-14B-720P-Turbo"
 
-    if request.model == "Wan-AI/Wan2.1-T2V-14B":
-        request.model = "Wan-AI/Wan2.1-T2V-14B-Turbo"
-
-    elif request.model == "Wan-AI/Wan2.1-I2V-14B-720P":
-        request.model = "Wan-AI/Wan2.1-I2V-14B-720P-Turbo"
-
-    else:
-        request.model = "Wan-AI/Wan2.1-I2V-14B-720P-Turbo" if request.image else "Wan-AI/Wan2.1-T2V-14B-Turbo"
+    if request.image:
+        # request.image = await to_base64(request.image)
+        request.model = request.model.replace("-T2V-", "-I2V-")
 
     payload = request.model_dump(exclude_none=True)
 

meutils/apis/volcengine_apis/__init__.py

@@ -8,4 +8,3 @@
 # @Software     : PyCharm
 # @Description  : 
 
-from meutils.pipe import *

meutils/apis/volcengine_apis/images.py

@@ -90,14 +90,15 @@ async def generate(request: ImageRequest, token: Optional[str] = None):
 
 if __name__ == '__main__':
     token = f"""{os.getenv("VOLC_ACCESSKEY")}|{os.getenv("VOLC_SECRETKEY")}"""
+    token = "AKLTOWM5ZTc5ZDFhZWNlNDIzODkwYmZiNjEyNzYwNzE0MTI|T0RCbFpHRTJaRFEyWmpjeE5ERXpNR0ptWlRCaU16WmhPRE0wWVdKa01tTQ=="
     prompt = """
     3D魔童哪吒 c4d 搬砖 很开心， 很快乐， 精神抖擞， 背景是数不清的敖丙虚化 视觉冲击力强 大师构图 色彩鲜艳丰富 吸引人 背景用黄金色艺术字写着“搬砖挣钱” 冷暖色对比
     """
 
-    prompt = """
-    https://oss.ffire.cc/files/kling_watermark.png
-    让这个女人带上眼镜 衣服换个颜色
-    """
+    # prompt = """
+    # https://oss.ffire.cc/files/kling_watermark.png
+    # 让这个女人带上眼镜 衣服换个颜色
+    # """
 
     request = ImageRequest(
         model="high_aes_general_v30l_zt2i",

meutils/apis/volcengine_apis/tasks.py

@@ -15,13 +15,28 @@ https://www.volcengine.com/docs/6791/1384311
 import os
 
 from meutils.pipe import *
-from meutils.io.files_utils import to_url
+from meutils.caches import rcache
+
+from meutils.decorators.retry import retrying
+from meutils.db.redis_db import redis_aclient
+from meutils.config_utils.lark_utils import get_next_token_for_polling
 from meutils.schemas.jimeng_types import VideoRequest, ImageRequest
 
 from volcengine.visual.VisualService import VisualService
+from fastapi import APIRouter, File, UploadFile, Query, Form, Depends, Request, HTTPException, status, BackgroundTasks
+
+FEISHU = "https://xchatllm.feishu.cn/sheets/GYCHsvI4qhnDPNtI4VPcdw2knEd?sheet=OiHxsE"
+
 
+#  and "Access Denied" not in str(r)
+
+@retrying(max_retries=5, predicate=lambda r: "Concurrent Limit" in str(r))  # 限流
+async def create_task(request: Union[ImageRequest, VideoRequest, dict], token: Optional[str] = None):
+    """https://www.volcengine.com/docs/6791/1399614"""
+    token = token or await get_next_token_for_polling(FEISHU)
+
+    logger.debug(token)
 
-def create_task(request: Union[ImageRequest, VideoRequest], token: Optional[str] = None):
     visual_service = VisualService()
 
     if token:
@@ -30,26 +45,77 @@ def create_task(request: Union[ImageRequest, VideoRequest], token: Optional[str]
         visual_service.set_sk(sk)
 
     # request
-    payload = request.model_dump(exclude_none=True)
+    if not isinstance(request, dict):
+        request = request.model_dump(exclude_none=True)
+
+    response = visual_service.cv_sync2async_submit_task(request)  # 同步转异步
+
+    """
+    {'code': 10000,
+ 'data': {'task_id': '15106285208671192523'},
+ 'message': 'Success',
+ 'request_id': '202505291707517FC0D2B135CEE77BC4A5',
+ 'status': 10000,
+ 'time_elapsed': '150.967328ms'}
+    """
+
+    logger.debug(response)
+    if response.get('code') == 10000:
+        await redis_aclient.set(response['data']['task_id'], token, ex=7 * 24 * 3600)
+    else:
+        raise Exception(response)
+
+    return response
+
+
+#
+# @retrying(max_retries=5, predicate=lambda r: "Concurrent Limit" in str(r))  # 限流
+# @rcache(ttl=5)
+async def get_task(request: dict):
+    task_id = request.get("task_id", "")
+    token = await redis_aclient.get(task_id)  # 绑定对应的 token
+    token = token and token.decode()
+    if not token:
+        raise HTTPException(status_code=404, detail="TaskID not found")
+
+    visual_service = VisualService()
+
+    if token:
+        ak, sk = token.split('|')
+        visual_service.set_ak(ak)
+        visual_service.set_sk(sk)
+
+    response = visual_service.cv_get_result(request)  # 同步转异步
+
+    logger.debug(response)
 
-    response = visual_service.cv_submit_task(payload)
     return response
 
 
 if __name__ == '__main__':
     token = f"""{os.getenv("VOLC_ACCESSKEY")}|{os.getenv("VOLC_SECRETKEY")}"""
+    token = "AKLTOWM5ZTc5ZDFhZWNlNDIzODkwYmZiNjEyNzYwNzE0MTI|T0RCbFpHRTJaRFEyWmpjeE5ERXpNR0ptWlRCaU16WmhPRE0wWVdKa01tTQ=="
     prompt = """
     3D魔童哪吒 c4d 搬砖 很开心， 很快乐， 精神抖擞， 背景是数不清的敖丙虚化 视觉冲击力强 大师构图 色彩鲜艳丰富 吸引人 背景用黄金色艺术字写着“搬砖挣钱” 冷暖色对比
     """
 
     request = ImageRequest(
+        req_key="high_aes_general_v30l_zt2i",
         prompt=prompt,
     )
 
-    request = VideoRequest(
-        prompt=prompt
-    )
+    # request = VideoRequest(
+    #     prompt=prompt
+    # )
 
+    # arun(create_task(request, token))
+    # arun(create_task(request))
 
+    # request = {
+    #     "task_id": "141543714223689974",
+    #     "req_key": "high_aes_general_v30l_zt2i"
+    # }
+    # #
+    # arun(get_task(request))
 
-    print(create_task(request, token))
+    print(bjson(request))

meutils/config_utils/lark_utils/common.py

@@ -80,7 +80,7 @@ def get_spreadsheet_values(
 
 
 @alru_cache(ttl=600)
-@rcache(ttl=600, serializer='pickle')  # 缓存
+@rcache(ttl=300, serializer='pickle')  # 缓存
 async def aget_spreadsheet_values(
         spreadsheet_token: Optional[str] = None,
         sheet_id: Optional[str] = None,
@@ -101,10 +101,10 @@ async def aget_spreadsheet_values(
         if response.is_success:
             data = response.json()
             if to_dataframe:
-                # values = data.get('data').get('valueRange').get('values')
-                # [value | xmap(lambda x: x['text']) | xjoin('') for value in values]
+                values = data.get('data').get('valueRange').get('values')
+
+                return pd.DataFrame(values)
 
-                return pd.DataFrame(data.get('data').get('valueRange').get('values'))
             return data
         else:
             send_message(
@@ -307,7 +307,7 @@ async def get_dataframe(iloc_tuple: Optional[tuple] = None, feishu_url: Optional
 async def get_series(feishu_url: str, index: int = 0, duplicated: bool = False):  # 系统配置
     df = await aget_spreadsheet_values(feishu_url=feishu_url, to_dataframe=True)
     series = df[index]
-    values = [i for i in series if i]  # todo: 非标准字符串处理
+    values = [i for i in series if i and isinstance(i, str)]  # todo: 非标准字符串处理
     if duplicated:  # 有序去重
         values = values | xUnique
 

meutils/data/VERSION

@@ -1 +1 @@
-2025.05.15.14.03.22
+2025.05.29.18.12.08

meutils/io/files_utils.py

@@ -26,7 +26,7 @@ from httpx import AsyncClient
 def guess_mime_type(file):
     mime_type = None
     if isinstance(file, str):
-        mime_type, _ = mimetypes.guess_type(file)
+        mime_type, _ = mimetypes.guess_type(file.strip())
     return mime_type or "application/octet-stream"
 
 
@@ -280,5 +280,7 @@ if __name__ == '__main__':
 
     # arun(to_bytes("https://oss.ffire.cc/files/kling_watermark.png"))
 
-    file = "https://v3.fal.media/files/penguin/Rx-8V0MVgkVZM6PJ0RiPD_douyin.mp4"
-    arun(to_bytes(file))
+    # file = "https://v3.fal.media/files/penguin/Rx-8V0MVgkVZM6PJ0RiPD_douyin.mp4"
+    # arun(to_bytes(file))
+
+    print(guess_mime_type("http://admin.ilovechatgpt.top/file/ceshiwendangdocx_31118702.docx "))

meutils/io/openai_files.py

@@ -50,7 +50,7 @@ async def delete_files(client, threshold: int = 666):
 
 
 @rcache(ttl=7 * 24 * 3600)
-async def file_extract(file):
+async def file_extract(file, enable_reader: bool = True):
     """
 
     :param file: url bytes path
@@ -63,8 +63,9 @@ async def file_extract(file):
     filename = Path(file).name if isinstance(file, str) else 'untitled'
     mime_type = guess_mime_type(file)
 
-    if str(file).startswith("http") and mime_type in {"application/octet-stream", "text/html"}:
+    if enable_reader and str(file).startswith("http") and mime_type in {"application/octet-stream", "text/html"}:
         logger.debug(f"jina reader")
+
         markdown_content = await url_reader(file)
         return {
             'filename': filename,

meutils/llm/check_utils.py

@@ -7,6 +7,7 @@
 # @WeChat       : meutils
 # @Software     : PyCharm
 # @Description  :
+import os
 
 from meutils.pipe import *
 from meutils.decorators.retry import retrying
@@ -156,7 +157,7 @@ async def check_token_for_ppinfra(api_key, threshold: float = 1):
     try:
         client = AsyncOpenAI(base_url="https://api.ppinfra.com/v3/user", api_key=api_key)
         data = await client.get("", cast_to=object)
-        # logger.debug(data)  # credit_balance
+        logger.debug(data)  # credit_balance
         return data["credit_balance"] > threshold
     except TimeoutException as e:
         raise
@@ -166,8 +167,30 @@ async def check_token_for_ppinfra(api_key, threshold: float = 1):
         return False
 
 
+@retrying()
+async def check_token_for_sophnet(api_key, threshold: float = 1):
+    if not isinstance(api_key, str):
+        return await check_tokens(api_key, check_token_for_sophnet)
+
+    try:
+        client = AsyncOpenAI(base_url=os.getenv("SOPHNET_BASE_URL"), api_key=api_key)
+        data = await client.chat.completions.create(
+            model="DeepSeek-v3",
+            messages=[{"role": "user", "content": "hi"}],
+            stream=True,
+            max_tokens=1
+        )
+        return True
+    except TimeoutException as e:
+        raise
+
+    except Exception as e:
+        logger.error(f"Error: {e}\n{api_key}")
+        return False
+
+
 if __name__ == '__main__':
-    from meutils.config_utils.lark_utils import get_next_token_for_polling
+    from meutils.config_utils.lark_utils import get_next_token_for_polling, get_series
 
     check_valid_token = partial(check_token_for_siliconflow, threshold=-1)
 
@@ -189,4 +212,13 @@ if __name__ == '__main__':
 
     # arun(check_token_for_moonshot("sk-Qnr87vtf2Q6MEfc2mVNkVZ4qaoZg3smH9527I25QgcFe7HrT"))
 
-    arun(check_token_for_ppinfra("sk_DkIaRrPq7sTiRPevhjV9WFZN3FvLk6WhCXOj1JAwu6c"))
+    # arun(check_token_for_ppinfra("sk_DkIaRrPq7sTiRPevhjV9WFZN3FvLk6WhCXOj1JAwu6c"))
+
+    # from meutils.config_utils.lark_utils import get_next_token_for_polling, get_series
+    #
+    # arun(get_series("https://xchatllm.feishu.cn/sheets/Bmjtst2f6hfMqFttbhLcdfRJnNf?sheet=PP1PGr"))
+
+    # arun(check_token_for_sophnet(["gzHpp_zRtGaw1IjpepCiWu_ySyke3Hu5wR5VNNYMLyXwAESqZoZWUZ4T3tiWUxtac6n9Hk-kRRo4_jPQmndo-g"]))
+
+
+    arun(check_token_for_ppinfra("sk_F0kgPyCMTzmOH_-VCEJucOK8HIrbnLGYm_IWxBToHZQ"))

meutils/llm/completions/chat_spark.py

@@ -33,7 +33,7 @@ class Completions(object):
             file_url, *texts = request.last_user_content.split(maxsplit=1) + ["总结下"]
             text = texts[0]
 
-            file_content = await file_extract(file_url)
+            file_content = await file_extract(file_url, enable_reader=False)
 
             request.messages = [
                 {
@@ -73,7 +73,7 @@ class Completions(object):
 
                         text, file_url = texts[-1], file_urls[-1]
                         if file_url in image_urls:
-                            file_content = await file_extract(file_url)
+                            file_content = await file_extract(file_url, enable_reader=False)
 
                             message["content"] = f"""{json.dumps(file_content, ensure_ascii=False)}\n\n{text}"""
 
@@ -104,7 +104,7 @@ if __name__ == '__main__':
     request = CompletionRequest(
         # model="qwen-turbo-2024-11-01",
         # model="claude-3-5-sonnet-20241022",
-        model="deepsek-chat",
+        model="deepseek-chat",
 
         messages=[
             # {
@@ -133,15 +133,15 @@ if __name__ == '__main__':
             {
                 'role': 'user',
                 # "content": '你好',
-                "content": [
-                    {"type": "text", "text": "https://oss.ffire.cc/files/kling_watermark.png 描述第一张图片"},
-
-                    # {"type": "text", "text": "描述第一张图片"},
-                    #
-                    # {"type": "image_url", "image_url": "https://oss.ffire.cc/files/kling_watermark.png"},
-                    #     # {"type": "image_url", "image_url": "https://oss.ffire.cc/files/nsfw.jpg"}
-                    #
-                ],
+                # "content": [
+                #     {"type": "text", "text": "https://oss.ffire.cc/files/kling_watermark.png 描述第一张图片"},
+                #
+                #     # {"type": "text", "text": "描述第一张图片"},
+                #     #
+                #     # {"type": "image_url", "image_url": "https://oss.ffire.cc/files/kling_watermark.png"},
+                #     #     # {"type": "image_url", "image_url": "https://oss.ffire.cc/files/nsfw.jpg"}
+                #     #
+                # ],
 
                 # 'content': {
                 #     "type": "file_url",
@@ -153,7 +153,7 @@ if __name__ == '__main__':
                 # "content": "https://oss.ffire.cc/files/百炼系列手机产品介绍.docx 总结下"
                 # "content": "https://mj101-1317487292.cos.ap-shanghai.myqcloud.com/ai/test.pdf\n\n总结下"
 
-                # "content": "https://admin.ilovechatgpt.top/file/lunIMYAIzhinengzhushouduishenghuodocx_14905733.docx 总结",
+                "content": "http://admin.ilovechatgpt.top/file/4docx_86529298.docx 我无法确定你是否准确识别word里面的论文？",
                 # "content": "http://admin.ilovechatgpt.top/file/xinjianMicrosoftWordwendangdoc-9052714901036-bGSJLeKbqQdnIZZn.doc 111111234234",
 
             },
@@ -276,4 +276,22 @@ if __name__ == '__main__':
     #  {},
     #  {'role': 'user', 'content': '总结一下'}]
 
+    request = {
+        "model": "gemini-all",
+        "messages": [
+            {
+                "role": "system",
+                "content": "\\n Current date: 2025-05-21"
+            },
+            {
+                "role": "user",
+                "content": "http://admin.ilovechatgpt.top/file/ceshiwendangdocx_31118702.docx 你好"
+            }
+        ],
+        "stream": True,
+        "top_p": 0.7,
+        "temperature": 0.8,
+        "n": 1
+    }
+    request = CompletionRequest(**request)
     arun(c.create(request))

meutils/llm/completions/qwenllm.py

@@ -34,9 +34,6 @@ base_url = "https://chat.qwen.ai/api"
 from fake_useragent import UserAgent
 
 ua = UserAgent()
-cookie = storage_to_cookie(
-    "cookies.json"
-)
 
 thinking_budget_mapping = {
     "low": 1000,
@@ -44,15 +41,19 @@ thinking_budget_mapping = {
     "high": 24000
 }
 
+COOKIE = """
+cna=KP9DIEqqyjUCATrw/+LjJV8F; _bl_uid=LXmp28z7dwezpmyejeXL9wh6U1Rb; cnaui=310cbdaf-3754-461c-a3ff-9ec8005329c9; aui=310cbdaf-3754-461c-a3ff-9ec8005329c9; x-ap=ap-southeast-1; sca=43897cb0; acw_tc=0a03e53417483123807755658e597c5e3685457054f2ca60a0a8d87b657874; _gcl_au=1.1.106229673.1748312382; xlly_s=1; token=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6IjMxMGNiZGFmLTM3NTQtNDYxYy1hM2ZmLTllYzgwMDUzMjljOSIsImV4cCI6MTc1MDkwNDU2MH0.nV7I1sp6rAE0QnqXYKNm4I0vLZmA-TdOKWEHg_37_tw; SERVERID=1e5b6792fa61468bae321990103ad502|1748312579|1748312380; atpsida=0450727e7c9d8a7299a0b2bd_1748312579_5; ssxmod_itna=iqGxRDuQqWqxgDUxeKYI5q=xBDeMDWK07DzxC5750CDmxjKidKDUGQN0bFP=jhWOGxDkYCA+UQSD0HPKmDA5DnGx7YDtr4FN4SeKhDwIa78YRwwHQiGoh=LTX90w=3qflZqcj1T3xoD==hYDmeDUxD1GDDkS0PDxOPD5xDTDWeDGDD3t4DCCfYYD0RpiboIVxRhTABpDYPYPR4YDY56DAqEz4SpVaxXTDDzw4iaPf4vwDi8D7FRG0RpD7P6fmQDXaYDEAWmFk=Dv6Lh+QwI1/oYOyyDc=DoCO0Km0DTVj2qPGGiU5eiBdnNC4490i+yte+in2MWYHDDW=4=5rzqDxKGe1qC+LimweRk5yxmLhdYY4KGYqOqIheUk5ZB5x2QgohQBxN7spmxFezNiDD; ssxmod_itna2=iqGxRDuQqWqxgDUxeKYI5q=xBDeMDWK07DzxC5750CDmxjKidKDUGQN0bFP=jhWOGxDkYCA+UQmDDpU3qY6obeDLWr7pfFyiDDsa7QaIhEVl4FKFerBUoQiRQlSzS+caiWpDOqz2HrsO6xFOwpKsSOYH0Q0WwhDs0Ye1mah+b99w34Im3AYwt4oztFoQ7xBhThYqatKQWcgkRDBDCOiLViK62z51rnpzpbHH7pFpS=Y4zfHiWfozYCf+9FcGmRMsMEYFGRP+TgG9EbEi3Khm0lQmt2AL=quK6RomKnFmmpjzYzxHQ/QEK0AAa3qGPOl3crGu7DDpQnxjfEEgWD/oEHaE4l6jOpKp6DI6P=vQa39gN6h5i3R5eInP2Gob9pY7DKDr3lYiTZrC3R2Yiz/rsFIG1j5n=2=DD3=obLGPQsWnOZWSinlZ=KjKGpIwRPKn3jpohCU+2PciIEehVxTSnioeIx6xdp91zK29tHtN2Zcgi0clGNaY4jIenYlnw+/Qlapjg6Qho=3E/2v3mIjaYygren01ebwPI4EKgDY+4NyiTXYmU+o/lGSogI=GT/38rPOnA5mjuDg2Ub=+HtTjKnpYoEaTvwj3K0GI7FKARVAv=Wojxe0nHRBwhrQIa1vMpv5pfQ8LGCXGp=lZ3Q8v+6=lSfexroPjNP9MvyNVAXQhnKvyAwT/KEsgh/eOdqx0YiHaP1kwxzsu54i4eGiQDshEOQoiRlPBqiDiSRDQay2k1x4xWDhpBTjqZ0Neer0qlDK0YbDpBYKxSGDD; tfstk=gy8S5jal7pL47z_LOgc4laG38cQQOjuaeW1psBUz9aQRdW9e6MCzzbxBcpJ5eU7-ZoUBQpYrZMdRRipBCgUKqMCvlpJBzkd82-QvQ1KUpv4LH-C1yMxPLTWCRBvsgAuZ7QAl-akZQVyg4ccA76BdpJPAkw5IYeMoXQAl-SDk9fRkZW1NLUsRJpQAD1CRJkB8yjsAs1bLyMU8D-BcHwCdvTQAk61dpkIJpIhfttBdJgpdkj1n_yV1O06kNbzD-uH4ifA5hyUplsHhpQig5_8lNgW9wUwzUXf5VOdRhAjyDMIBFgTqPRfWDC9PNp0graIB2hQJRAg5kCx2es9KCk6vfUYCbUM_jTLlKd_JcxU5JaK95wQIf7f2PILGAUD7kT9DMh7kWx4BBIRwriYIC-BH41bANnGLDTId4azNCak3ASsgRs6ZGjZ3xSw1l2pVr0vc2sf50jGbeHjRis1mGjZ3xgCcN_GjG8Kh.; isg=BOrqXB6_dpCyTPX0tTuBOG9yO1aMW261hQXS_3ShLD3Op4xhWOtyxWGRN9O7V-ZN
+""".strip()
+
 
 @retrying()
-async def to_file(file, api_key):
+async def to_file(file, api_key, cookie: Optional[str] = None):
     qwen_client = AsyncOpenAI(
         base_url="https://all.chatfire.cn/qwen/v1",
         api_key=api_key,
         default_headers={
             'User-Agent': ua.random,
-            'Cookie': cookie
+            'Cookie': cookie or COOKIE
         }
     )
     filename = Path(file).name if isinstance(file, str) else 'untitled'
@@ -64,7 +65,8 @@ async def to_file(file, api_key):
     return file_object
 
 
-async def create(request: CompletionRequest, token: Optional[str] = None):  # ChatCompletionRequest 重构
+async def create(request: CompletionRequest, token: Optional[str] = None, cookie: Optional[str] = None):
+    cookie = cookie or COOKIE
 
     if request.temperature > 1:
         request.temperature = 1
@@ -79,14 +81,21 @@ async def create(request: CompletionRequest, token: Optional[str] = None):  # Ch
         default_headers={
             'User-Agent': ua.random,
             'Cookie': cookie,
-            "bx-v": "2.5.28",
-            "bx-umidtoken": "T2gAeo1Mqj6q05L65Ro3Hjf9KHuOsB63ttVeMP_-13M2-R82AvHnHe-o9nAbz7J8Si4=",
-            "x-request-id": "08cb2785-53d2-4c6d-a24c-069cf0303d1a"
         }
     )
     # qwen结构
     model = request.model.lower()
-    if any(i in model for i in ("search",)):
+    if any(i in model for i in ("research",)):  # 遇到错误 任意切换
+        request.model = np.random.choice({""})
+        request.messages[-1]['chat_type'] = "deep_research"
+
+        # request.messages["extra"] = {
+        #     "meta": {
+        #         "subChatType": "deep_thinking"
+        #     }
+        # }
+
+    elif any(i in model for i in ("search",)):
         request.model = "qwen-max-latest"
         request.messages[-1]['chat_type'] = "search"
 
@@ -120,13 +129,13 @@ async def create(request: CompletionRequest, token: Optional[str] = None):  # Ch
             for i, content in enumerate(user_content):
                 if content.get("type") == 'file_url':  # image_url file_url video_url
                     url = content.get(content.get("type")).get("url")
-                    file_object = await to_file(url, token)
+                    file_object = await to_file(url, token, cookie)
 
                     user_content[i] = {"type": "file", "file": file_object.id}
 
                 elif content.get("type") == 'image_url':
                     url = content.get(content.get("type")).get("url")
-                    file_object = await to_file(url, token)
+                    file_object = await to_file(url, token, cookie)
 
                     user_content[i] = {"type": "image", "image": file_object.id}
 
@@ -135,7 +144,7 @@ async def create(request: CompletionRequest, token: Optional[str] = None):  # Ch
 
             user_content = [{"type": "text", "text": user_content}]
 
-            file_object = await to_file(file_url, token)
+            file_object = await to_file(file_url, token, cookie)
 
             content_type = file_object.meta.get("content_type", "")
             if content_type.startswith("image"):
@@ -262,11 +271,11 @@ if __name__ == '__main__':
                 'role': 'user',
                 # 'content': '今天南京天气',
                 # 'content': "9.8 9.11哪个大",
-                # 'content': 'https://oss.ffire.cc/files/AIGC.pdf 总结下',
+                'content': 'https://oss.ffire.cc/files/AIGC.pdf 总结下',
 
-                # "chat_type": "search",
+                # "chat_type": "search", deep_research
 
-                'content': user_content,
+                # 'content': user_content,
 
                 # "content": [
                 #     {
@@ -332,8 +341,6 @@ if __name__ == '__main__':
 
     token = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6IjMxMGNiZGFmLTM3NTQtNDYxYy1hM2ZmLTllYzgwMDUzMjljOSIsImV4cCI6MTc0ODQ3OTE0M30.oAIE1K0XA0YYqlxB8Su-u0UJbY_BBZa4_tvZpFJKxGY"
 
-    # token = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6IjEwNzY1N2Y1LTgxN2ItNDg5Yi1iNjk4LWFhZjAyM2EwZTE4MyIsImV4cCI6MTc0NjI5NTAwNH0.D1uJN44NHiEt6URce4upbHvs7v73_Vd0V1s3T_JzclI"
     arun(create(request, token))
-    # token = "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6IjEwNzY1N2Y1LTgxN2ItNDg5Yi1iNjk4LWFhZjAyM2EwZTE4MyIsImV4cCI6MTc0NjI5NTAwNH0.D1uJN44NHiEt6URce4upbHvs7v73_Vd0V1s3T_JzclI"
 
     # arun(to_file("https://oss.ffire.cc/files/kling_watermark.png", token))

meutils/llm/openai_polling/chat.py

@@ -72,7 +72,7 @@ class Completions(object):
                 self.client = zhipuai_client
 
         elif "deepseek-r" in request.model:
-            request.separate_reasoning = True
+            request.separate_reasoning = True  # pp
         ###########################################################################
 
         data = to_openai_params(request)
@@ -82,7 +82,7 @@ class Completions(object):
             data.pop("frequency_penalty", None)
             data.pop("extra_body", None)
 
-            if not request.reasoning_effort:  # 默认关闭思考
+            if '2.5' in request.model and not request.reasoning_effort:  # 默认关闭思考
                 data['reasoning_effort'] = "none"
 
             if "thinking" in request.model:

meutils/office_automation/pdf.py

@@ -56,10 +56,46 @@ def doc2text(filename):
     pass
 
 
+def extract_images_from_pdf(file, output: Optional[str] = None):
+    import fitz
+
+    # 打开PDF文件
+
+    pdf_document = fitz.open(file)
+
+    # 遍历每一页
+
+    for page_number in range(pdf_document.page_count):
+
+        page = pdf_document.load_page(page_number)
+
+        image_list = page.get_images(full=True)
+
+        # 遍历每个图像
+
+        for image_index, img in enumerate(image_list):
+            xref = img[0]
+
+            base_image = pdf_document.extract_image(xref)
+
+            image_bytes = base_image["image"]
+
+            image_ext = base_image["ext"]
+
+            image_filename = f"{output or ''}/image{page_number + 1}_{image_index + 1}.{image_ext}"
+            Path(image_filename).parent.mkdir(parents=True, exist_ok=True)
+
+            # 将图像写入文件
+
+            with open(image_filename, "wb") as image_file:
+                image_file.write(image_bytes)
+
+
 if __name__ == '__main__':
     with timer():
-       r = extract_text('x.pdf')
+        # r = extract_text('x.pdf')
 
+        r = extract_images_from_pdf('《锋利的jQuery》(高清扫描版-有书签)_副本_加水印.pdf', 'images')
 
     # import tiktoken
-    # print(tiktoken.encoding_for_model('gpt-3.5-turbo'))
+    # print(tiktoken.encoding_for_model('gpt-3.5-turbo'))

meutils/pipe.py

@@ -390,4 +390,7 @@ data: [DONE]"""
 
     # print()
 
-    sse.splitlines() | xsse_parser(skip_strings=['DONE', 'TRACEID']) | xprint
+    # sse.splitlines() | xsse_parser(skip_strings=['DONE', 'TRACEID']) | xprint
+
+
+    a = [[{"a":1}]] | xgroup(2)

meutils/schemas/dreamfaceapp_types.py

@@ -0,0 +1,12 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : dreamfaceapp_types
+# @Time         : 2025/5/23 13:42
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.pipe import *
+

meutils/schemas/metaso_types.py

@@ -15,15 +15,20 @@ FEISHU_URL = "https://xchatllm.feishu.cn/sheets/Bmjtst2f6hfMqFttbhLcdfRJnNf?shee
 
 
 class MetasoRequest(BaseModel):
-    model: Optional[Literal["ds-r1",]] = None
+    model: Optional[Literal["ds-r1", "fast_thinking"]] = None
 
-    """search-mini search search-pro"""
+    """search-mini search search-pro
+    
+    model-mode
+    
+    """
     mode: Union[str, Literal["concise", "detail", "research", "strong-research"]] = "detail"  # concise detail research
 
     question: str = "Chatfire"
 
-    """全网 文库 学术 图片 播客"""
+    """全网 文库 学术 图片 播客 视频"""
     scholarSearchDomain: str = "all"
+    engineType: Optional[Literal["pdf", "scholar", "image", "podcast", "video"]] = None
 
     url: str = "https://metaso.cn/"
     lang: str = "zh"
@@ -35,6 +40,7 @@ class MetasoRequest(BaseModel):
     newEngine: str = 'true'
     enableImage: str = 'true'
 
+#
 
 # question: hi
 # mode: detail