题意：将 OpenAI 的响应保存在 Azure 函数中到 Blob 存储

问题背景：

I used blob trigger template to create an Azure function that is triggered by new file updated in Azure blob storage. I am using python v2 to create Azure function in VScode. My issue is not able to store openai result into a new blob file. I noticed my function didn't have a function.json file. I am not sure if this is the problem for my issue. I tried redeploy couple times still didn't create a function.json.

我使用 Blob 触发器模板创建了一个 Azure 函数，该函数由 Azure Blob 存储中的新文件更新触发。我在 VSCode 中使用 Python v2 创建 Azure 函数。我的问题是无法将 OpenAI 的结果存储到一个新的 Blob 文件中。我注意到我的函数没有生成 `function.json` 文件，不确定这是否是导致问题的原因。我尝试重新部署了几次，但仍然没有生成 `function.json` 文件。

import azure.functions as func
import logging
import os
import openai
import PyPDF2
import io
openai.api_type = "azure"
openai.api_version = "2023-05-15" 
openai.api_base = os.getenv("OPENAI_API_BASE")  # Your Azure OpenAI resource's endpoint value.
openai.api_key = os.getenv("OPENAI_API_KEY")

app = func.FunctionApp()


@app.blob_trigger(arg_name="myblob", path="mypath/{name}",
                               connection="<My_BLOB_CONNECTION_SETTING>") 
@app.blob_trigger_output(arg_name="outputblob", path="openai-outputs/text.txt",
                                connection="<My_BLOB_CONNECTION_SETTING>")
def blob_trigger(myblob: func.InputStream, outputblob: func.Out[str]) -> str:
    logging.info(f"Python blob trigger function processed blob"
                f"Name: {myblob.name}"
                f"Blob Size: {myblob.length} bytes")
    pdf_reader = PyPDF2.PdfReader(io.BytesIO(myblob.read()))
    text = ""
    for page in pdf_reader.pages:
        text += page.extract_text()

    logging.info(
                f"textyyy: {text}")
    messages= [
       # my prompt
    ]

    response = openai.ChatCompletion.create(
        engine="testing3",
        messages=messages
    )
    
    logging.info(f"response1111: {response}"
                f"response['choices'][0]['message']['content']: {response['choices'][0]['message']['content']}")
    
    outputblob.set(response['choices'][0]['message']['content'])

i followed this azure documentation to write the above code. It started failed to deploy to Azure function portal.

我按照这个 [Azure 文档](https://azure.microsoft.com/documentation) 编写了上述代码，但开始无法部署到 Azure Function 门户。

问题解决：

I was able to save the response of the OpenAI to blob storage using this code.

我能够使用以下代码将 OpenAI 的响应保存到 Blob 存储中。

I also used python v2. In python V2 function.json file are not available. Binding is created using @app.blob_trigger for trigger binding, @app.blob_input for input binding,@app.blob_outputfor output binding.

我也使用了 Python v2。在 Python v2 中，`function.json` 文件不可用。绑定是通过 `@app.blob_trigger` 创建触发器绑定，通过 `@app.blob_input` 创建输入绑定，通过 `@app.blob_output` 创建输出绑定。

PyPDF2 package is used to work with .pdf files and perform operation on the file.

`PyPDF2` 包用于处理 `.pdf` 文件并对文件执行操作。

I have used @app.blob_trigger to trigger the function code, @app.blob_input to read the content of the file and @app.blob_output to modify the file.

我使用了 `@app.blob_trigger` 来触发函数代码，`@app.blob_input` 来读取文件内容，以及 `@app.blob_output` 来修改文件。

function_app.py:

import azure.functions as func
import logging
import os
import openai
openai.api_type = "azure"
openai.api_base = os.getenv("OPENAI_API_BASE")
openai.api_version = "2023-07-01-preview"
openai.api_key = os.getenv("OPENAI_API_KEY")

app = func.FunctionApp()

@app.blob_trigger(arg_name="myblob", path="response",
                               connection="BlobStorageConnectionString")
@app.blob_input(arg_name="inputblob",path="response/hello.txt",connection="BlobStorageConnectionString")
@app.blob_output(arg_name="output",path="chatgpt-conversation/conversation.txt",connection="BlobStorageConnectionString2")
def blob_trigger(myblob: func.InputStream, inputblob: str, output:func.Out[bytes]):
    logging.info(f"Python blob trigger function processed blob"
                f"Name: {myblob.name}"
                f"Blob Size: {myblob.length} bytes")
    
    content= inputblob
    logging.info(f"Contentof the file is : {content}")

    ai_responses=[]
    response = openai.ChatCompletion.create(
        engine="test_chatgpt",
        messages=[
            {"role": "system", "content": "You are an AI assistant that helps people find information."},
            {"role":"user","content": content}
        ])
    ai_responses.append(f"Input: {content}\nResponse: {response['choices'][0]['message']['content']}\n\n")
    logging.info(ai_responses)
    output.set("\n".join(ai_responses))

local.settings.json

{
  "IsEncrypted": false,
  "Values": {
    "AzureWebJobsStorage": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "FUNCTIONS_WORKER_RUNTIME": "python",
    "AzureWebJobsFeatureFlags": "EnableWorkerIndexing",
    "OPENAI_API_BASE":"https://<xxxxxx>.openai.azure.com/",
    "OPENAI_API_KEY": "xxxxxxxxxxxxxxxxxxxx",
    "BlobStorageConnectionString":"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "BlobStorageConnectionString2":"xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
  }
}