AgentSkillsCN

azure-functions

利用 Azure Functions 构建无服务器应用。创建 HTTP 触发器、队列处理器、定时器函数,以及持久化编排。适用于 Azure 上的事件驱动计算、API 后端,以及无服务器微服务。

SKILL.md
--- frontmatter
name: azure-functions
description: Build serverless applications with Azure Functions. Create HTTP triggers, queue processors, timer functions, and durable orchestrations. Use for event-driven computing, API backends, and serverless microservices on Azure.

Azure Functions Skill

Build serverless applications with Azure Functions for event-driven computing.

Triggers

Use this skill when you see:

  • azure functions, function app, serverless azure
  • http trigger, timer trigger, queue trigger
  • durable functions, orchestration
  • function binding, function.json

Instructions

Create Function App

bash
# Create storage account (required)
az storage account create \
    --name myfuncstorage \
    --resource-group mygroup \
    --location eastus \
    --sku Standard_LRS

# Create Function App
az functionapp create \
    --name myfuncapp \
    --resource-group mygroup \
    --storage-account myfuncstorage \
    --consumption-plan-location eastus \
    --runtime python \
    --runtime-version 3.11 \
    --functions-version 4

# Create with Premium plan
az functionapp plan create \
    --name mypremiumplan \
    --resource-group mygroup \
    --location eastus \
    --sku EP1

az functionapp create \
    --name myfuncapp \
    --resource-group mygroup \
    --storage-account myfuncstorage \
    --plan mypremiumplan \
    --runtime node \
    --runtime-version 20 \
    --functions-version 4

Python Functions (v2 Programming Model)

HTTP Trigger

python
import azure.functions as func
import logging
import json

app = func.FunctionApp(http_auth_level=func.AuthLevel.FUNCTION)

@app.route(route="hello")
def hello_http(req: func.HttpRequest) -> func.HttpResponse:
    logging.info('HTTP trigger function processed a request.')

    name = req.params.get('name')
    if not name:
        try:
            req_body = req.get_json()
            name = req_body.get('name')
        except ValueError:
            pass

    if name:
        return func.HttpResponse(f"Hello, {name}!")
    else:
        return func.HttpResponse(
            "Please pass a name on the query string or in the request body",
            status_code=400
        )

@app.route(route="users/{id}", methods=["GET"])
def get_user(req: func.HttpRequest) -> func.HttpResponse:
    user_id = req.route_params.get('id')
    # Fetch user from database
    return func.HttpResponse(
        json.dumps({"id": user_id, "name": "John"}),
        mimetype="application/json"
    )

Timer Trigger

python
@app.timer_trigger(schedule="0 */5 * * * *", arg_name="timer")
def timer_function(timer: func.TimerRequest) -> None:
    if timer.past_due:
        logging.info('The timer is past due!')

    logging.info('Timer trigger function executed.')
    # Run scheduled task

Queue Trigger

python
@app.queue_trigger(arg_name="msg", queue_name="myqueue",
                   connection="AzureWebJobsStorage")
def queue_processor(msg: func.QueueMessage) -> None:
    logging.info(f'Queue trigger processed: {msg.get_body().decode()}')

    data = json.loads(msg.get_body().decode())
    process_message(data)

# Output binding to queue
@app.route(route="enqueue")
@app.queue_output(arg_name="msg", queue_name="myqueue",
                  connection="AzureWebJobsStorage")
def enqueue_message(req: func.HttpRequest, msg: func.Out[str]) -> func.HttpResponse:
    message = req.get_json()
    msg.set(json.dumps(message))
    return func.HttpResponse("Message enqueued", status_code=202)

Blob Trigger

python
@app.blob_trigger(arg_name="blob", path="container/{name}",
                  connection="AzureWebJobsStorage")
def blob_processor(blob: func.InputStream) -> None:
    logging.info(f'Blob trigger: {blob.name}, Size: {blob.length} bytes')
    content = blob.read()
    process_blob(content)

# Output binding to blob
@app.route(route="upload")
@app.blob_output(arg_name="outputblob", path="container/{rand-guid}.txt",
                 connection="AzureWebJobsStorage")
def upload_blob(req: func.HttpRequest, outputblob: func.Out[str]) -> func.HttpResponse:
    content = req.get_body().decode()
    outputblob.set(content)
    return func.HttpResponse("Blob created", status_code=201)

Cosmos DB Trigger

python
@app.cosmos_db_trigger(arg_name="documents",
                       container_name="items",
                       database_name="mydb",
                       connection="CosmosDBConnection",
                       lease_container_name="leases",
                       create_lease_container_if_not_exists=True)
def cosmos_trigger(documents: func.DocumentList) -> None:
    for doc in documents:
        logging.info(f'Document id: {doc["id"]}')
        process_document(doc)

TypeScript Functions (v4 Programming Model)

typescript
import { app, HttpRequest, HttpResponseInit, InvocationContext } from "@azure/functions";

// HTTP trigger
app.http("hello", {
  methods: ["GET", "POST"],
  authLevel: "function",
  handler: async (request: HttpRequest, context: InvocationContext): Promise<HttpResponseInit> => {
    context.log(`Http function processed request for url "${request.url}"`);

    const name = request.query.get("name") || (await request.text()) || "world";

    return {
      body: `Hello, ${name}!`,
    };
  },
});

// Timer trigger
app.timer("timerTrigger", {
  schedule: "0 */5 * * * *",
  handler: async (timer: Timer, context: InvocationContext): Promise<void> => {
    context.log("Timer trigger function executed");
  },
});

// Queue trigger with output binding
app.storageQueue("queueTrigger", {
  queueName: "myqueue",
  connection: "AzureWebJobsStorage",
  handler: async (message: unknown, context: InvocationContext): Promise<void> => {
    context.log(`Queue message: ${JSON.stringify(message)}`);
  },
});

Durable Functions

python
import azure.functions as func
import azure.durable_functions as df

app = func.FunctionApp(http_auth_level=func.AuthLevel.FUNCTION)

# Orchestrator
@app.orchestration_trigger(context_name="context")
def orchestrator(context: df.DurableOrchestrationContext):
    # Fan-out/fan-in pattern
    tasks = []
    for i in range(5):
        tasks.append(context.call_activity("activity_function", i))

    results = yield context.task_all(tasks)

    # Process results
    total = sum(results)
    return total

# Activity
@app.activity_trigger(input_name="input")
def activity_function(input: int) -> int:
    return input * 2

# HTTP starter
@app.route(route="orchestrators/{functionName}")
@app.durable_client_input(client_name="client")
async def http_start(req: func.HttpRequest, client: df.DurableOrchestrationClient) -> func.HttpResponse:
    function_name = req.route_params.get('functionName')
    instance_id = await client.start_new(function_name)

    return client.create_check_status_response(req, instance_id)

Application Settings

bash
# Set application settings
az functionapp config appsettings set \
    --name myfuncapp \
    --resource-group mygroup \
    --settings "DatabaseConnection=connection-string" \
               "ApiKey=your-api-key"

# Get application settings
az functionapp config appsettings list \
    --name myfuncapp \
    --resource-group mygroup

# Reference Key Vault secrets
az functionapp config appsettings set \
    --name myfuncapp \
    --resource-group mygroup \
    --settings "Secret=@Microsoft.KeyVault(SecretUri=https://myvault.vault.azure.net/secrets/mysecret/)"

Managed Identity

python
from azure.identity import DefaultAzureCredential
from azure.keyvault.secrets import SecretClient

# Use managed identity
credential = DefaultAzureCredential()
secret_client = SecretClient(
    vault_url="https://myvault.vault.azure.net",
    credential=credential
)

secret = secret_client.get_secret("my-secret")

Deployment

bash
# Deploy from local
func azure functionapp publish myfuncapp

# Deploy with zip
az functionapp deployment source config-zip \
    --name myfuncapp \
    --resource-group mygroup \
    --src app.zip

# Continuous deployment from GitHub
az functionapp deployment source config \
    --name myfuncapp \
    --resource-group mygroup \
    --repo-url https://github.com/org/repo \
    --branch main \
    --manual-integration

Best Practices

  1. Cold Starts: Use Premium plan for latency-sensitive apps
  2. Bindings: Use input/output bindings instead of SDK calls when possible
  3. Secrets: Use Key Vault references for sensitive settings
  4. Logging: Use structured logging with Application Insights
  5. Scaling: Configure host.json for optimal scaling

Common Workflows

API Backend

  1. Create Function App with HTTP triggers
  2. Implement CRUD operations
  3. Add authentication (Azure AD, API keys)
  4. Configure CORS settings
  5. Enable Application Insights

Event Processing

  1. Set up queue/blob/Cosmos DB triggers
  2. Implement processing logic
  3. Configure dead-letter queues
  4. Add retry policies
  5. Monitor with alerts