Chapter 9 · Parametrization and Data-Driven Testing

Chapter goals: Learn to drive tests with environment variables, CSV/JSON files, and per-virtual-user independent data — avoiding hard-coded values.

9.1 Environment Variable Injection

The simplest parametrization: control target host and concurrency via environment variables:

import os
from locust import HttpUser, task, between

class ApiUser(HttpUser):
    host = os.getenv("TARGET_HOST", "https://httpbin.org")
    wait_time = between(
        float(os.getenv("MIN_WAIT", "1")),
        float(os.getenv("MAX_WAIT", "3"))
    )

    @task
    def get_data(self):
        self.client.get("/get")

Override at runtime:

TARGET_HOST=https://staging.example.com locust -f locustfile.py

9.2 CSV Data Loading

import csv
from locust import HttpUser, task, between

# Load all test data at startup
with open("users.csv") as f:
    USERS = list(csv.DictReader(f))

class LoginUser(HttpUser):
    wait_time = between(1, 3)

    def on_start(self):
        # Each virtual user pops one account from the queue (avoid duplicate logins)
        self.user_data = USERS.pop() if USERS else {"username": "default", "password": "test"}

    @task
    def login_and_browse(self):
        self.client.post("/login", json=self.user_data)
        self.client.get("/dashboard")

9.3 JSON Configuration Files

import json
from locust import HttpUser, task, between

with open("config.json") as f:
    config = json.load(f)

class ConfigurableUser(HttpUser):
    host = config["target_url"]
    wait_time = between(*config["wait_range"])

    @task
    def hit_endpoint(self):
        for ep in config["endpoints"]:
            self.client.get(ep)

9.4 Per-Virtual-User Independent Data

Use queue.Queue to ensure each virtual user gets unique test data:

import queue
from locust import HttpUser, task, between

class QueueUser(HttpUser):
    wait_time = between(1, 3)

    def on_start(self):
        # Each user pops one item — no two users share the same data
        self.account = data_queue.get()

    @task
    def login(self):
        self.client.post("/login", json=self.account)

9.5 Chapter Summary

• Environment variables for external config (host, timeouts, flags);
• CSV/JSON files for bulk test data loaded at startup;
• queue.Queue for per-user unique data allocation;
• Keep test data separate from test code for maintainability.

🧪 随堂测验

点击你认为正确的选项。答错时会展示正确答案与原因解析。

1. What is the simplest way to override the target host without changing code?

2. Why use queue.Queue for per-user data instead of a plain list?

3. Where should you load CSV/JSON test data in a locustfile?

4. What is the trade-off between CSV and JSON for test data?

🛠️ Hands-On Exercises

1. Create a users.csv with 100 accounts; write a locustfile that assigns one account per virtual user via on_start().
2. Replace the CSV approach with a JSON config file containing nested endpoint definitions; verify all endpoints are hit.
3. Use queue.Queue to allocate unique accounts; run 10 users and verify no two users share the same account.

Proceed to Chapter 10: Distributed Load Testing.