MeUtils 2025.1.23.10.16.28__py3-none-any.whl → 2025.1.27.15.5.31__py3-none-any.whl

meutils/apis/jimeng/videos.py

@@ -13,13 +13,15 @@
 3. 上传音频+创建任务
 
 """
+import asyncio
+
 from meutils.pipe import *
 from meutils.str_utils.json_utils import json_path
 
 from meutils.schemas.jimeng_types import BASE_URL, MODELS_MAP, FEISHU_URL
 from meutils.schemas.video_types import LipsyncVideoRquest
 from meutils.schemas.task_types import TaskResponse
-from meutils.apis.jimeng.common import create_draft_content, get_headers, check_token
+from meutils.apis.jimeng.common import get_headers, check_token
 from meutils.apis.jimeng.files import upload_for_image, upload_for_video
 
 from meutils.config_utils.lark_utils import get_next_token_for_polling
@@ -29,50 +31,52 @@ from fake_useragent import UserAgent
 ua = UserAgent()
 
 
-async def create_realman_avatar(image_uri, token: str):
+async def create_realman_avatar(image_url: str, token: str):
+    if image_url.startswith("http"):
+        image_url = await upload_for_image(image_url, token)
+
     url = "/mweb/v1/create_realman_avatar"
     headers = get_headers(url, token)
 
     payload = {
         "input_list": [
             {
-                "image_uri": image_uri,
+                "image_uri": image_url,
                 "submit_id": str(uuid.uuid4()),
                 "mode": 0
             },
             {
-                "image_uri": image_uri,
+                "image_uri": image_url,
                 "submit_id": str(uuid.uuid4()),
                 "mode": 1
             }
         ]
     }
+
     async with httpx.AsyncClient(base_url=BASE_URL, headers=headers, timeout=60) as client:
         response = await client.post(url, json=payload)
         response.raise_for_status()
         data = response.json()
         logger.debug(bjson(data))  # 1914628189186
 
-        if task_ids := json_path(data, "$..task_id"):
-            return task_ids
-        else:  # task_id 解析失败
-            raise Exception("create_realman_avatar failed: task_id 解析失败")
+        response = TaskResponse(metadata=data, system_fingerprint=token)
+        if task_ids := json_path(data, "$..task_id"):  # 返回 imageurl vid
+            response.task_id = ' '.join(task_ids)
+            return response
 
-        # mget_generate_task
-        # todo: 从任务结果解析 resource_id_std, resource_id_loopy
+        else:
+            response.message = str(json_path(data, "$..message"))
+            response.status = "fail"
+            return response
 
 
-async def get_task(task_ids):
+async def get_task(task_id: str, token: str = "916fed81175f5186a2c05375699ea40d"):
     """
     $..image_to_avatar 成功： 先检测图片or视频
     :param task_ids:
     :return:
     """
-    if isinstance(task_ids, str):
-        task_ids = [task_ids]
-
-    token = "916fed81175f5186a2c05375699ea40d"
-
+    task_ids = task_id.split()
 
     url = "/mweb/v1/mget_generate_task"
     headers = get_headers(url, token)
@@ -84,19 +88,42 @@ async def get_task(task_ids):
         data = response.json()
         logger.debug(bjson(data))
 
-        if messages := json_path(data, "$..image_to_avatar.message"):
-            if message := "".join(messages):  # and "fail" in message
-                logger.error(message)
-            else:
-                logger.info("image_to_avatar success")
+        if json_path(data, "$..image_to_avatar"):  # 角色检测 create_realman_avatar
+            resource_id_std = resource_id_loopy = ""
+            if resource_id_stds := json_path(data, "$..resource_id_std"):
+                resource_id_std = "".join(resource_id_stds)
+
+            if resource_id_loopys := json_path(data, "$..resource_id_loopy"):
+                resource_id_loopy = "".join(resource_id_loopys)
+
+            task_data = {
+                "resource_id_std": resource_id_std,
+                "resource_id_loopy": resource_id_loopy
+            }
+            response = TaskResponse(task_id=task_id, data=task_data, metadata=data)
+            if resource_id_std and resource_id_loopy:
+                response.status = "success"
 
-                if resource_id_stds := json_path(data, "$..resource_id_std"):
-                    resource_id_std = "".join(resource_id_stds)
+            if (message := json_path(data, "$..image_to_avatar.message")) and "fail" in str(message).lower():
+                response.message = str(message)
+                response.status = "fail"
 
-                if resource_id_loopys := json_path(data, "$..resource_id_loopy"):
-                    resource_id_loopy = "".join(resource_id_loopys)
-                # return TaskResponse()
+            return response
 
+        else:
+            response = TaskResponse(task_id=task_id, metadata=data)
+            if (message := json_path(data, "$..fail_msg")) and "success" not in str(message).lower():
+                response.message = str(message)
+                response.status = "fail"
+                return response
+
+            if will_cost := json_path(data, "$..will_cost"):
+                response.will_cost = will_cost[0]
+
+            if video_urls := json_path(data, "$..[360p,480p,720p].video_url"):
+                response.data = [{"video": _} for _ in video_urls]
+
+            return response
 
 
 async def create_task(request: LipsyncVideoRquest, token: Optional[str] = None):
@@ -108,43 +135,121 @@ async def create_task(request: LipsyncVideoRquest, token: Optional[str] = None):
 
     headers = get_headers(url, token)
 
-    i2v_opt = {}
-    v2v_opt = {}
+    model = request.model
+    scene = "lip_sync_image"
+    image_url = await upload_for_image(request.image_url, token)
+
+    # 角色检测
+    realman_avatar_response = await create_realman_avatar(image_url, token)
+    if realman_avatar_response.status == "fail":
+        return realman_avatar_response
+
+    else:
+        for _ in range(10):
+            task_response = await get_task(realman_avatar_response.task_id, token)
+            if task_response.status == "fail":
+                logger.debug("fail")
+                return task_response
+            elif task_response.status == "success":
+                logger.debug("success")
+
+                realman_avatar_response = task_response
+                break
+            else:
+                await asyncio.sleep(3)
+                continue
+
+    audio_vid, audio_url = await upload_for_video(request.audio_url, token)
+
+    resource_id_std = realman_avatar_response.data.get("resource_id_std")
+    resource_id_loopy = realman_avatar_response.data.get("resource_id_loopy")
+
+    i2v_opt = v2v_opt = {}
     if request.video_url:
         v2v_opt = {}
 
+    # payload = {
+    #     "submit_id": "",
+    #     "task_extra": "{\"promptSource\":\"photo_lip_sync\",\"generateTimes\":1,\"lipSyncInfo\":{\"sourceType\":\"local-file\",\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\"},\"isUseAiGenPrompt\":false,\"batchNumber\":1}",
+    #     "http_common_info": {
+    #         "aid": 513695
+    #     },
+    #     "input": {
+    #         "seed": 3112889115,
+    #         "video_gen_inputs": [
+    #             {
+    #                 "v2v_opt": {},
+    #                 "i2v_opt": {
+    #                     "realman_avatar": {
+    #                         "enable": True,
+    #                         "origin_image": {
+    #                             # "width": 800,
+    #                             # "height": 1200,
+    #                             "image_uri": "tos-cn-i-tb4s082cfz/4dead1bfc8e84572a91f2e047016a351",
+    #                             "image_url": ""
+    #                         },
+    #                         "origin_audio": {
+    #                             # "duration": 9.976625,
+    #                             "vid": "v02870g10004cu8d4r7og65j2vr5opb0"
+    #                         },
+    #
+    #                         "resource_id_std": "381c534f-bcef-482e-8f17-5b30b64e41a1",
+    #                         "resource_id_loopy": "b9ac51cb-e26c-4b63-81d9-34ed24053032",
+    #                         #
+    #                         # "tts_info": "{\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\",\"source_type\":\"local-file\"}"
+    #                     }
+    #                 },
+    #                 "audio_vid": "v02870g10004cu8d4r7og65j2vr5opb0",
+    #                 "video_mode": 4
+    #             }
+    #         ]
+    #     },
+    #     "mode": "workbench",
+    #     "history_option": {},
+    #     "commerce_info": {
+    #         "resource_id": "generate_video",
+    #         "resource_id_type": "str",
+    #         "resource_sub_type": "aigc",
+    #         "benefit_type": "lip_sync_avatar_std",  # 5积分
+    #         # "benefit_type": "lip_sync_avatar_lively" # 10积分
+    #     },
+    #     "scene": "lip_sync_image",
+    #     "client_trace_data": {},
+    #     "submit_id_list": [
+    #         str(uuid.uuid4())
+    #     ]
+    # }
+
+    if request.image_url:
+        i2v_opt = {
+            "realman_avatar": {
+                "enable": True,
+                "origin_image": {
+                    "image_uri": image_url,
+                    "image_url": ""
+                },
+                "resource_id_loopy": resource_id_loopy,
+                "resource_id_std": resource_id_std,
+                "origin_audio": {
+                    "vid": audio_vid
+                },
+                # "tts_info": "{\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\",\"source_type\":\"local-file\"}"
+            }
+        }
+
     payload = {
         "submit_id": "",
-        "task_extra": "{\"promptSource\":\"photo_lip_sync\",\"generateTimes\":1,\"lipSyncInfo\":{\"sourceType\":\"local-file\",\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\"},\"isUseAiGenPrompt\":false,\"batchNumber\":1}",
+        # "task_extra": "{\"promptSource\":\"photo_lip_sync\",\"generateTimes\":1,\"lipSyncInfo\":{\"sourceType\":\"local-file\",\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\"},\"isUseAiGenPrompt\":false,\"batchNumber\":1}",
         "http_common_info": {
             "aid": 513695
         },
         "input": {
-            "seed": 3112889115,
+            "seed": 2032846910,
             "video_gen_inputs": [
                 {
-                    "v2v_opt": {},
-                    "i2v_opt": {
-                        "realman_avatar": {
-                            "enable": True,
-                            "origin_image": {
-                                # "width": 800,
-                                # "height": 1200,
-                                "image_uri": "tos-cn-i-tb4s082cfz/4dead1bfc8e84572a91f2e047016a351",
-                                "image_url": ""
-                            },
-                            "origin_audio": {
-                                # "duration": 9.976625,
-                                "vid": "v02870g10004cu8d4r7og65j2vr5opb0"
-                            },
-
-                            "resource_id_std": "381c534f-bcef-482e-8f17-5b30b64e41a1",
-                            "resource_id_loopy": "b9ac51cb-e26c-4b63-81d9-34ed24053032",
-                            #
-                            # "tts_info": "{\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\",\"source_type\":\"local-file\"}"
-                        }
-                    },
-                    "audio_vid": "v02870g10004cu8d4r7og65j2vr5opb0",
+                    "v2v_opt": v2v_opt,
+                    "i2v_opt": i2v_opt,
+                    "audio_vid": audio_vid,
                     "video_mode": 4
                 }
             ]
@@ -155,67 +260,17 @@ async def create_task(request: LipsyncVideoRquest, token: Optional[str] = None):
             "resource_id": "generate_video",
             "resource_id_type": "str",
             "resource_sub_type": "aigc",
-            "benefit_type": "lip_sync_avatar_std",  # 5积分
+            "benefit_type": model,
             # "benefit_type": "lip_sync_avatar_lively" # 10积分
         },
-        "scene": "lip_sync_image",
+        "scene": scene,
         "client_trace_data": {},
         "submit_id_list": [
             str(uuid.uuid4())
         ]
     }
 
-    # if request.image_url:
-    #     i2v_opt = {
-    #         "realman_avatar": {
-    #             "enable": True,
-    #             "origin_image": {
-    #                 "width": 800,
-    #                 "height": 1200,  ######## 没必要把
-    #                 "image_uri": request.image_url,
-    #                 "image_url": ""
-    #             },
-    #             "resource_id_loopy": "9c397499-a59f-47b5-9bfd-e1397ec62f61",
-    #             "resource_id_std": "0a8c8d72-5543-4e9e-8843-c03fe5b3a8c7",
-    #             "origin_audio": {
-    #                 "duration": 9.976625,
-    #                 "vid": "v03870g10004cu6vpgfog65nc9ivupg0"
-    #             },
-    #             "tts_info": "{\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\",\"source_type\":\"local-file\"}"
-    #         }
-    #     }
-    #
-    # payload = {
-    #     "submit_id": "",
-    #     "task_extra": "{\"promptSource\":\"photo_lip_sync\",\"generateTimes\":1,\"lipSyncInfo\":{\"sourceType\":\"local-file\",\"name\":\"vyFWygmZsIZlUO4s0nr2n.wav\"},\"isUseAiGenPrompt\":false,\"batchNumber\":1}",
-    #     "http_common_info": {
-    #         "aid": 513695
-    #     },
-    #     "input": {
-    #         "seed": 2032846910,
-    #         "video_gen_inputs": [
-    #             {
-    #                 "v2v_opt": v2v_opt,
-    #                 "i2v_opt": i2v_opt,
-    #                 "audio_vid": "v03870g10004cu6vpgfog65nc9ivupg0",
-    #                 "video_mode": 4
-    #             }
-    #         ]
-    #     },
-    #     "mode": "workbench",
-    #     "history_option": {},
-    #     "commerce_info": {
-    #         "resource_id": "generate_video",
-    #         "resource_id_type": "str",
-    #         "resource_sub_type": "aigc",
-    #         "benefit_type": "lip_sync_avatar_std"
-    #     },
-    #     "scene": "lip_sync_image",
-    #     "client_trace_data": {},
-    #     "submit_id_list": [
-    #         "4717038e-f4fd-4c1c-b5a5-39ae4118099c"
-    #     ]
-    # }
+    logger.debug(bjson(payload))
 
     async with httpx.AsyncClient(base_url=BASE_URL, headers=headers, timeout=60) as client:
         response = await client.post(url, json=payload)
@@ -229,8 +284,22 @@ async def create_task(request: LipsyncVideoRquest, token: Optional[str] = None):
 
 
 if __name__ == '__main__':
+    token = "916fed81175f5186a2c05375699ea40d"
+
+    request = LipsyncVideoRquest(
+        model="lip_sync_avatar_std",
+        image_url="https://oss.ffire.cc/files/kling_watermark.png",
+        video_url="",
+        audio_url="https://oss.ffire.cc/files/lipsync.mp3"
+    )
+
+    # with timer():
+    #     r = arun(create_realman_avatar(request.image_url, token))
+    #     arun(get_task(r.task_id))
+
     # image_uri = "tos-cn-i-tb4s082cfz/387649a361e546f89549bd3510ab926d"
     # task_ids = arun(create_realman_avatar(image_uri, token="7c5e148d9fa858e3180c42f843c20454"))
     # arun(mget_generate_task(task_ids))
-    r = arun(create_task(LipsyncVideoRquest()))
-    arun(get_task(r.task_id))
+    with timer():
+        r = arun(create_task(request))
+    # arun(get_task(r.task_id))

meutils/data/VERSION

@@ -1 +1 @@
-2025.01.23.10.16.28
+2025.01.27.15.05.31

meutils/jwt_utils/common.py

@@ -0,0 +1,46 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : common
+# @Time         : 2024/10/28 20:57
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.pipe import *
+
+import jwt
+import time
+import datetime
+
+# Header and payload
+header = {
+    "alg": "HS512",
+    "type": "JWT"
+}
+
+
+payload = {
+    "jti": "80004477",
+    "rol": "ROLE_REGISTER",
+    "iss": "OpenXLab",
+    "clientId": "lkzdx57nvy22jkpq9x2w",
+    "phone": "",
+    "uuid": "73a8d9b0-8bbf-4973-9b71-4b687ea23a78",
+    "email": "313303303@qq.com",
+
+    "iat": int(time.time()),
+    "exp": int(time.time()) + 3600
+}
+
+# Your secret key
+secret = ""
+
+# Create the JWT
+token = jwt.encode(payload, secret, algorithm="HS512", headers=header)
+
+print(token)
+
+
+

meutils/llm/clients.py

@@ -32,7 +32,10 @@ if __name__ == '__main__':
 
     # web-search-pro
 
-    arun(zhipuai_client.chat.completions.create(
+    r = arun(zhipuai_client.chat.completions.create(
         messages=[{"role": "user", "content": "中国队奥运会拿了多少奖牌"}],
         model='web-search-pro')
     )
+
+    r.model_dump_json()
+

meutils/llm/completions/agents/search.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : search
+# @Time         : 2025/1/27 13:41
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : todo: 格式匹配
+
+from meutils.pipe import *
+from meutils.llm.clients import AsyncOpenAI, zhipuai_client, moonshot_client
+from meutils.llm.openai_utils import to_openai_params
+from meutils.schemas.openai_types import chat_completion, chat_completion_chunk, ChatCompletionRequest, CompletionUsage
+
+
+class Completions(object):
+
+    def __init__(self, api_key: Optional[str] = None):
+        self.api_key = api_key
+
+    async def create(self, request: ChatCompletionRequest):
+        request.model = request.model.removesuffix("-search")
+
+        if request.model.startswith("baichuan"):
+            base_url = os.getenv("BAICHUAN_BASE_URL")
+            api_key = self.api_key or os.getenv("BAICHUAN_API_KEY")
+
+            request.tools = [
+                {
+                    "type": "web_search",
+                    "web_search": {
+                        "enable": True,
+                        "search_mode": "performance_first"
+                    }
+                }
+            ]
+            data = to_openai_params(request)
+            client = AsyncOpenAI(base_url=base_url, api_key=api_key)
+            completion = await client.chat.completions.create(**data)
+            return completion
+
+        elif request.model.startswith(("moonshot", "kimi")):
+            tool_call_name = "$web_search"
+            request.tools = [
+                {
+                    "type": "builtin_function",  # <-- 我们使用 builtin_function 来表示 Kimi 内置工具，也用于区分普通 function
+                    "function": {
+                        "name": "$web_search",
+                    },
+                },
+            ]
+
+            data = to_openai_params(request)
+            completion = await moonshot_client.chat.completions.create(**data)
+
+            tool_call = completion.choices[0].message.tool_calls[0]
+            tool_call_arguments = tool_call.function.arguments
+            print(tool_call_arguments)
+
+            request.messages.append({
+                "role": "tool",
+                "tool_call_id": tool_call.id,
+                "name": tool_call_name,
+                "content": tool_call_arguments,
+            })
+            logger.debug(request.messages)
+            data = to_openai_params(request)
+            logger.debug(data)
+            completion = await moonshot_client.chat.completions.create(**data)
+            return completion
+
+        elif request.model.startswith("doubao"):
+            base_url = os.getenv("DOUBAO_BASE_URL")
+            api_key = self.api_key or os.getenv("DOUBAO_API_KEY")
+
+            request.model = "bot-20250127143547-c8q8m"
+            request.tools = [
+                {
+                    "type": "web_search",
+                    "web_search": {
+                        "enable": True,
+                        "search_mode": "performance_first"
+                    }
+                }
+            ]
+            data = to_openai_params(request)
+            client = AsyncOpenAI(base_url=base_url, api_key=api_key)
+
+            completion = await client.chat.completions.create(**data)
+            return completion
+
+        else:  # glm
+            request.model = "web-search-pro"
+            data = to_openai_params(request)
+            completion = await zhipuai_client.chat.completions.create(**data)
+            logger.debug(completion)
+            return completion
+
+
+if __name__ == '__main__':
+    request = ChatCompletionRequest(
+        # model="baichuan4-turbo",
+        model="xx",
+        # model="moonshot-v1-8k",
+        # model="doubao",
+
+        messages=[
+            {"role": "user", "content": "南京天气如何"}
+        ],
+
+        stream=False
+    )
+
+    arun(Completions().create(request))

meutils/parsers/fileparser/mineru.py

@@ -0,0 +1,48 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : mineru
+# @Time         : 2025/1/23 13:13
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.pipe import *
+
+import requests
+
+token = """
+eyJ0eXBlIjoiSldUIiwiYWxnIjoiSFM1MTIifQ.eyJqdGkiOiI4MDAwNDQ3NyIsInJvbCI6IlJPTEVfUkVHSVNURVIiLCJpc3MiOiJPcGVuWExhYiIsImlhdCI6MTczNzYxMTQ5MiwiY2xpZW50SWQiOiJsa3pkeDU3bnZ5MjJqa3BxOXgydyIsInBob25lIjoiIiwidXVpZCI6IjczNTZjNjY1LTU4MTMtNGQxNC04ZjFiLWM0NWIyZmFhYTBhMCIsImVtYWlsIjoiMzEzMzAzMzAzQHFxLmNvbSIsImV4cCI6MTczODgyMTA5Mn0.i8CwWoRE6j5wAC_hD9z9amkWT56HdOewgXMFV4jMpg17JHB0HOY-K4o9zp06Puav2vxkuC3Lnqm_8ip7-QdxsQ
+"""
+url = 'https://mineru.net/api/v4/extract/task'
+header = {
+    'Content-Type': 'application/json',
+    "Authorization": f"Bearer {token.strip()}"
+}
+data = {
+    'url': 'https://cdn-mineru.openxlab.org.cn/demo/example.pdf',
+    'is_ocr': True,
+    'enable_formula': False,
+}
+
+res = requests.post(url, headers=header, json=data)
+print(res.status_code)
+print(res.json())
+print(res.json()["data"])
+
+# {'task_id': 'adb223f6-794b-4950-8d60-d766ebd0bf14'}
+
+task_id = 'adb223f6-794b-4950-8d60-d766ebd0bf14'
+import requests
+
+url = f'https://mineru.net/api/v4/extract/task/{task_id}'
+header = {
+    'Content-Type':'application/json',
+    "Authorization": f"Bearer {token.strip()}"
+}
+
+res = requests.get(url, headers=header)
+print(res.status_code)
+print(res.json())
+print(res.json()["data"])

meutils/schemas/image_types.py

@@ -459,7 +459,16 @@ class TextinImageProcessRequest(ImageProcessRequest):
 
 
 class ImageProcess(BaseModel):
-    model: Literal["remove-watermark", "clarity", "expand", "rmbg-2.0"]
+    model: Literal[
+        "remove-watermark",
+        "clarity",
+
+        "hunyuan-remove-watermark",
+        "hunyuan-clarity",
+
+        "expand",
+        "rmbg-2.0"
+    ]
     image: str
     mask: Optional[str] = None