MeUtils 2025.3.31.19.40.39__py3-none-any.whl → 2025.4.8.18.11.44__py3-none-any.whl

meutils/ai_audio/tts/minimax.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : minimax
+# @Time         : 2025/4/1 12:14
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.pipe import *
+import requests
+import json
+
+url = "https://api.minimax.chat/v1/text2voice"
+api_key = "请输入您的api key"
+
+payload = json.dumps({
+    "gender": "female",
+    "age": "old",
+    "voice_desc": [
+        "Kind and friendly",
+        "Kind and amiable",
+        "Kind hearted",
+        "Calm tone"
+    ],
+    "text": "真正的危险不是计算机开始像人一样思考，而是人开始像计算机一样思考"
+})
+headers = {
+    'Content-Type': 'application/json',
+    'Authorization': f'Bearer {api_key}',
+}
+
+response = requests.request("POST", url, headers=headers, data=payload)

meutils/apis/google_apis/__init__.py

@@ -0,0 +1,11 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : __init__.py
+# @Time         : 2025/4/2 10:40
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.apis.google_apis.common import GeminiClient

meutils/apis/google_apis/audios.py

@@ -0,0 +1,115 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : audios
+# @Time         : 2025/4/2 11:00
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : https://platform.openai.com/docs/guides/audio?example=audio-in
+
+from meutils.pipe import *
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : openai_siliconflow
+# @Time         : 2024/6/26 10:42
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  :
+import os
+
+from meutils.pipe import *
+from openai import OpenAI
+from openai import OpenAI, APIStatusError
+from meutils.schemas.openai_types import ChatCompletionRequest
+from meutils.llm.openai_utils import to_openai_completion_params
+from meutils.io.files_utils import to_base64
+
+base64_audio = arun(to_base64("https://oss.ffire.cc/files/lipsync.mp3"))
+
+
+client = OpenAI(
+    # api_key=os.getenv("GOOGLE_API_KEY"),
+    api_key="AIzaSyAQAt73dfL5-v3zaAHtXajZalZxfiumMOU",
+    base_url=os.getenv("GOOGLE_BASE_URL"),
+)
+
+print(client.models.list().model_dump_json(indent=4))
+
+# {
+#     "gemini-2.0-pro-exp": "models/gemini-2.0-pro-exp",
+#     "gemini-2.0-pro-exp-02-05": "models/gemini-2.0-pro-exp-02-05",
+#     "gemini-2.5-pro-exp-03-25": "models/gemini-2.5-pro-exp-03-25",
+#     "gemini-2.0-flash-thinking-exp": "models/gemini-2.0-flash-thinking-exp",
+#     "gemini-2.0-flash": "models/gemini-2.0-flash"
+#
+# }
+
+
+if __name__ == '__main__':
+    # messages = [
+    #
+    #             {
+    #                 "role": "user", "content": [
+    #                 {
+    #                     "type": "text",
+    #                     "text": "一句话总结"
+    #                 },
+    #                 {
+    #                     "type": "image_url",
+    #                     "image_url": {
+    #                         "url": "https://oss.ffire.cc/files/kling_watermark.png"
+    #                     }
+    #                 }
+    #             ]
+    #             }
+    #
+    #         ]
+    messages = [
+        {
+            "role": "user",
+            "content": [
+                {
+                    "type": "text",
+                    "text": "一句话总结",
+                },
+                {
+                    "type": "input_audio",
+                    "input_audio": {
+                        "data": base64_audio,
+                        "format": "wav"
+                    }
+                }
+            ],
+        }
+    ]
+
+
+
+    try:
+        completion = client.chat.completions.create(
+            # model="models/gemini-2.5-pro-exp-03-25",
+            model="models/gemini-2.0-flash",
+            # model="models/gemini-2.0-flash-exp-image-generation",
+            messages=messages,
+            # top_p=0.7,
+            top_p=None,
+            temperature=None,
+            # stream=True,
+            stream=False,
+
+            max_tokens=None,
+        )
+    except APIStatusError as e:
+        print(e.status_code)
+
+        print(e.response)
+        print(e.message)
+        print(e.code)
+    print(completion)
+    for chunk in completion:  # 剔除extra body
+        print(chunk)
+        if chunk.choices:
+            print(chunk.choices[0].delta.content)

meutils/apis/google_apis/common.py

@@ -0,0 +1,251 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : common
+# @Time         : 2025/4/2 13:03
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : https://ai.google.dev/gemini-api/docs/openai?hl=zh-cn
+import asyncio
+import os
+
+import shortuuid
+
+from meutils.pipe import *
+from meutils.llm.clients import AsyncOpenAI
+from meutils.io.files_utils import to_url, to_bytes, guess_mime_type
+from meutils.str_utils.regular_expression import parse_url
+
+from meutils.schemas.openai_types import chat_completion, chat_completion_chunk, CompletionRequest, CompletionUsage
+
+from meutils.config_utils.lark_utils import get_next_token_for_polling
+from google import genai
+from google.genai import types
+from google.genai.types import HttpOptions, GenerateContentConfig, Content, HarmCategory, HarmBlockThreshold, Part
+from google.genai.types import UserContent, ModelContent
+
+FEISHU_URL = "https://xchatllm.feishu.cn/sheets/Bmjtst2f6hfMqFttbhLcdfRJnNf?sheet=MJw6hI"
+
+"""
+Gemini 1.5 Pro 和 1.5 Flash 最多支持 3,600 个文档页面。文档页面必须采用以下文本数据 MIME 类型之一：
+
+PDF - application/pdf
+JavaScript - application/x-javascript、text/javascript
+Python - application/x-python、text/x-python
+TXT - text/plain
+HTML - text/html
+CSS - text/css
+Markdown - text/md
+CSV - text/csv
+XML - text/xml
+RTF - text/rtf
+
+- 小文件
+- 大文件: 需要等待处理
+"""
+
+
+class GeminiClient(object):
+    def __init__(self, api_key: Optional[str] = None):
+        self.api_key = api_key  # or await get_next_token_for_polling(feishu_url=FEISHU_URL, from_redis=True)
+        self.base_url = "https://all.chatfire.cc/genai"
+
+    async def create(self, request: CompletionRequest):
+        client = await self.get_client()
+
+        if any(i in request.model for i in ("image",)):
+            messages = await self.to_image_messages(request)
+
+            if len(messages) > 1:
+                history = messages[:-1]
+                message = messages[-1].parts
+            else:
+                history = []
+                message = messages[-1].parts
+
+            chat = client.aio.chats.create(  # todo: system_instruction
+                model=request.model,
+                config=GenerateContentConfig(
+                    response_modalities=['Text', 'Image'],
+                ),
+                history=history
+            )
+
+            logger.debug(message)
+
+            # message = [
+            #     Part.from_text(text="画条狗")
+            # ]
+
+            for i in range(5):
+                try:
+                    chunks = await chat.send_message_stream(message)
+                    async for chunk in chunks:
+
+                        if chunk.candidates:
+                            parts = chunk.candidates[0].content.parts or []
+                            for part in parts:
+                                logger.debug(part)
+                                if part.text:
+                                    yield part.text
+
+                                if part.inline_data:
+                                    image_url = await to_url(
+                                        part.inline_data.data,
+                                        mime_type=part.inline_data.mime_type
+                                    )
+                                    yield f"![image_url]({image_url})"
+                    break
+
+                except Exception as e:
+                    logger.debug(f"重试{i}: {e}")
+                    if "The model is overloaded." in str(e):
+                        await asyncio.sleep(1)
+                        continue
+                    else:
+                        yield e
+                        raise e
+
+    async def to_image_messages(self, request: CompletionRequest):
+        # 两轮即可连续编辑图片
+
+        messages = []
+        for m in request.messages:
+            contents = m.get("content")
+            if m.get("role") == "assistant":
+                assistant_content = str(contents)
+                if urls := parse_url(assistant_content):  # assistant
+                    datas = await asyncio.gather(*map(to_bytes, urls))
+
+                    parts = [
+                        Part.from_bytes(
+                            data=data,
+                            mime_type="image/png"
+                        )
+                        for data in datas
+                    ]
+                    parts += [
+                        Part.from_text(
+                            text=request.last_assistant_content
+                        ),
+                    ]
+                    messages.append(ModelContent(parts=parts))
+
+            elif m.get("role") == "user":
+                if isinstance(contents, list):
+                    parts = []
+                    for content in contents:
+                        if content.get("type") == "image_url":
+                            image_url = content.get("image_url", {}).get("url")
+                            data = await to_bytes(image_url)
+
+                            parts += [
+                                Part.from_bytes(data=data, mime_type="image/png")
+                            ]
+
+                        elif content.get("type") == "text":
+                            text = content.get("text")
+                            if text.startswith('http'):
+                                image_url, text = text.split(maxsplit=1)
+                                data = await to_bytes(image_url)
+
+                                parts += [
+                                    Part.from_bytes(data=data, mime_type="image/png"),
+                                    Part.from_text(text=text)
+                                ]
+                            else:
+                                parts += [
+                                    Part.from_text(text=text)
+                                ]
+
+                    messages.append(UserContent(parts=parts))
+
+                else:  # str
+                    if str(contents).startswith('http'):  # 修正提问格式， 兼容 url
+                        image_url, text = str(contents).split(maxsplit=1)
+                        data = await to_bytes(image_url)
+                        parts = [
+                            Part.from_bytes(
+                                data=data,
+                                mime_type="image/png"
+                            ),
+                            Part.from_text(
+                                text=str(contents)
+                            ),
+                        ]
+                    else:
+                        parts = [
+                            Part.from_text(
+                                text=str(contents)
+                            ),
+                        ]
+                    messages.append(UserContent(parts=parts))
+
+        return messages
+
+    async def upload(self, files: Union[str, List[str]]):  # => openai files
+        client = await self.get_client()
+
+        if isinstance(files, list):
+            return await asyncio.gather(*map(self.upload, files))
+
+        file_config = {"name": f"{shortuuid.random().lower()}", "mime_type": guess_mime_type(files)}
+        return await client.aio.files.upload(file=io.BytesIO(await to_bytes(files)), config=file_config)
+
+    @alru_cache()
+    async def get_client(self):
+        api_key = self.api_key or await get_next_token_for_polling(feishu_url=FEISHU_URL, from_redis=True)
+
+        logger.info(f"GeminiClient: {api_key}")
+
+        return genai.Client(
+            api_key=api_key,
+            http_options=HttpOptions(
+                base_url=self.base_url
+            )
+        )
+
+
+if __name__ == '__main__':
+    file = "https://oss.ffire.cc/files/kling_watermark.png"
+
+    api_key = os.getenv("GOOGLE_API_KEY")
+
+    # arun(GeminiClient(api_key=api_key).upload(file))
+    # arun(GeminiClient(api_key=api_key).upload([file] * 2))
+    # arun(GeminiClient(api_key=api_key).create())
+
+    content = [
+
+        {"type": "text", "text": "https://oss.ffire.cc/files/nsfw.jpg 移除右下角的水印"},
+
+        # {"type": "text", "text": "总结下"},
+        # {"type": "image_url", "image_url": {"url": url}},
+
+        # {"type": "video_url", "video_url": {"url": url}}
+
+    ]
+
+    # content = "https://oss.ffire.cc/files/nsfw.jpg 移除右下角的水印"
+
+    #
+    request = CompletionRequest(
+        # model="qwen-turbo-2024-11-01",
+        # model="gemini-all",
+        model="gemini-2.0-flash-exp-image-generation",
+        # model="qwen-plus-latest",
+
+        messages=[
+            {
+                'role': 'user',
+
+                'content': content
+            },
+
+        ],
+        stream=True,
+    )
+
+    # arun(GeminiClient(api_key=api_key).to_image_messages(request))
+    arun(GeminiClient(api_key=api_key).create(request))

meutils/apis/google_apis/files.py

@@ -0,0 +1,19 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : files
+# @Time         : 2025/4/2 10:40
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+
+from meutils.pipe import *
+
+
+file = "/Users/betterme/PycharmProjects/AI/QR.png"
+#
+# file_object = client.files.upload(file=file)
+# prompt = "一句话总结"
+
+file_object = client.aio.files.upload(file=file)

meutils/apis/google_apis/gemini_sdk.py

@@ -0,0 +1,170 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : st
+# @Time         : 2025/4/3 15:24
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  :
+
+from meutils.pipe import *
+from meutils.async_utils import async_to_sync
+
+from meutils.io.files_utils import to_url, to_bytes
+from meutils.schemas.openai_types import CompletionRequest
+
+from google import genai
+from google.genai.types import HttpOptions, GenerateContentConfig, Content, HarmCategory, HarmBlockThreshold, Part, \
+    ThinkingConfig
+from google.genai.chats import _is_part_type
+
+config = GenerateContentConfig(
+
+    temperature=0.7,
+    top_p=0.8,
+    response_modalities=['Text', 'Image'],
+)
+
+# self._http_options.base_url = 'https://generativelanguage.googleapis.com/'
+# self._http_options.api_version = 'v1beta'
+client = genai.Client(
+    api_key="AIzaSyCa8PYURpxFKz7yOtQB_O_wRfrX0gYh9L4",
+    http_options=HttpOptions(
+        base_url="https://all.chatfire.cc/genai"
+    )
+)
+
+if __name__ == '__main__':
+    model = "gemini-2.0-flash-exp-image-generation"
+
+    request = CompletionRequest(
+        model=model,
+        messages=[
+            {
+                "role": "user", "content": [
+                {
+                    "type": "text",
+                    "text": "一句话总结"
+                },
+                {
+                    "type": "image_url",
+                    "image_url": {
+
+                        "url": "https://oss.ffire.cc/files/kling_watermark.png"
+
+                    }
+                }
+            ]
+            }],
+
+    )
+
+    response = client.aio.chats.create(
+        model=model,
+        config=GenerateContentConfig(
+            system_instruction=None,
+            max_output_tokens=None,
+            response_modalities=['Text', 'Image'],
+
+            # thinking_config=ThinkingConfig(include_thoughts=True),
+
+        ),
+
+        # history=[
+        #
+        #     Content(
+        #         role="user",
+        #         parts=[
+        #             Part.from_text(
+        #                 text="画条狗"
+        #             )
+        #         ]
+        #     ),
+        #     Content(
+        #         role="model",
+        #         parts=[
+        #             Part.from_text(
+        #                 text="Ok"
+        #             ),
+        #
+        #             Part.from_bytes(
+        #                 data=async_to_sync(to_bytes)("https://oss.ffire.cc/files/kling_watermark.png"),
+        #                 mime_type="image/png"
+        #             ),
+        #         ]
+        #     )
+        #
+        # ]
+    )
+
+    # "mime_type": "image/png"
+
+    p = Part.from_uri(file_uri="https://oss.ffire.cc/files/kling_watermark.png", mime_type="image/png")
+    pp = Part.from_bytes(
+        data=async_to_sync(to_bytes)("https://oss.ffire.cc/files/kling_watermark.png"),
+        mime_type="image/png"
+    )
+
+
+    # pt = Part.from_text(text="文本")
+
+    # {'video_metadata': None,
+    #  'thought': None,
+    #  'code_execution_result': None,
+    #  'executable_code': None,
+    #  'file_data': {'file_uri': 'https://oss.ffire.cc/files/kling_watermark.png',
+    #                'mime_type': 'image/png'},
+    #  'function_call': None,
+    #  'function_response': None,
+    #  'inline_data': None,
+    #  'text': None}
+
+    #
+    async def main():
+        message = [
+            Part.from_text(text="画条狗 带个眼镜"),
+            # pp
+        ]
+
+        print(_is_part_type(message))
+        chunks = await response.send_message_stream(message)
+
+        async for chunk in chunks:
+            if chunk.candidates:
+                parts = chunk.candidates[0].content.parts
+                for part in parts or []:
+                    if len(str(part)) < 500:
+                        logger.debug(part)
+                    if part.inline_data:
+                        image_url = await to_url(part.inline_data.data, mime_type=part.inline_data.mime_type)
+                        logger.info(image_url)
+
+                    if part.text:
+                        logger.info(part.text)
+            # print(i.model_dump_json(indent=4, exclude_none=True))  # inline_data
+
+
+    # response = client.models.generate_content(
+    #     model="gemini-2.0-flash-exp-image-generation",
+    #     contents=['画条狗'],
+    #
+    #     # model="gemini-2.5-pro-exp-03-25",
+    #     # model="gemini-2.0-flash",
+    #
+    #     # contents=[
+    #     #     Part.from_uri(file_uri='https://generativelanguage.googleapis.com/v1beta/files/test', mime_type='image/png'),
+    #     #
+    #     #           "一句话总结"],
+    #     config=config
+    # )
+    #
+    # # client.aio.
+    # # client.aio.chats.create()
+    #
+    # if __name__ == '__main__':
+    #     arun(file_object)
+
+    arun(main())
+
+# ValueError: Message must be a valid part type: typing.Union[google.genai.types.File, google.genai.types.Part, PIL.Image.Image, str] or typing.Union[google.genai.types.File, google.genai.types.Part, PIL.Image.Image, str, google.genai.types.PartDict], got <class 'list'>

meutils/apis/google_apis/google2openai.py

@@ -0,0 +1,89 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# @Project      : AI.  @by PyCharm
+# @File         : google2openai
+# @Time         : 2025/4/1 13:38
+# @Author       : betterme
+# @WeChat       : meutils
+# @Software     : PyCharm
+# @Description  : 
+"""
+1. 生图 编辑图 多轮
+2. 文件上传、大文件问答
+
+注意 ：借助 File API，您最多可为每个项目存储 20 GB 的文件，每个文件的大小上限为 2 GB。文件会存储 48 小时。您可以在该时间段内使用 API 密钥访问这些数据，但无法从 API 下载这些数据。在已推出 Gemini API 的所有地区，此功能均可免费使用。
+
+"""
+from meutils.pipe import *
+from meutils.schemas.openai_types import CompletionRequest
+
+from google import genai
+from google.genai.types import HttpOptions, GenerateContentConfig, Content, HarmCategory, HarmBlockThreshold, Part
+
+# Content(role="user", parts=[Part.from_text(text=prompt)]),
+# Content(role="model", parts=[Part.from_text(text="Ok")]),
+
+config = GenerateContentConfig(
+
+    temperature=0.7,
+    top_p=0.8,
+    response_modalities=['Text', 'Image'],
+
+    # 公民诚信类别的默认屏蔽阈值为 Block none（对于别名为 gemini-2.0-flash、gemini-2.0-pro-exp-02-05 和 gemini-2.0-flash-lite 的 gemini-2.0-flash-001），适用于 Google AI Studio 和 Gemini API；仅适用于 Google AI Studio 中的所有其他模型的 Block most。
+    # safety_settings=[
+    #     SafetySetting(
+    #         category=HarmCategory.HARM_CATEGORY_HATE_SPEECH,
+    #         threshold=HarmBlockThreshold.BLOCK_LOW_AND_ABOVE,
+    #     ),
+    # ]
+)
+
+
+# self._http_options.base_url = 'https://generativelanguage.googleapis.com/'
+# self._http_options.api_version = 'v1beta'
+client = genai.Client(
+    api_key="AIzaSyAlpq4kR9ZP0NwaqQzqHtDKqiV8PLdUqnA",
+    http_options=HttpOptions(
+        base_url="https://all.chatfire.cc/genai"
+    )
+)
+
+file = "/Users/betterme/PycharmProjects/AI/QR.png"
+#
+# file_object = client.files.upload(file=file)
+# prompt = "一句话总结"
+
+# file_object = client.aio.files.upload(file=file)
+# https://generativelanguage.googleapis.com/v1beta/files/ickgffcfb9zl
+#
+# contents = ('Hi, can you create a 3d rendered image of a pig '
+#             'with wings and a top hat flying over a happy '
+#             'futuristic scifi city with lots of greenery?')
+#
+# prompt = "9.11 9.8哪个大呢"
+#
+
+# client.aio.chats.create(
+#     model="gemini-2.0-flash-exp-image-generation",
+# )
+
+response = client.models.generate_content(
+    model="gemini-2.0-flash-exp-image-generation",
+    contents=['画条狗'],
+
+    # model="gemini-2.5-pro-exp-03-25",
+    # model="gemini-2.0-flash",
+
+    # contents=[
+    #     Part.from_uri(file_uri='https://generativelanguage.googleapis.com/v1beta/files/test', mime_type='image/png'),
+    #
+    #           "一句话总结"],
+    config=config
+)
+
+# client.aio.
+# client.aio.chats.create()
+
+if __name__ == '__main__':
+    # arun(file_object)
+    pass