Mastering Interface Testing: Strategies for Complex Custom Data
This article explores the challenges of interface testing for a large-scale home‑decoration platform, detailing pain points, testing stages, practical tips, and real‑world case studies to help backend engineers design effective, automated API tests for complex custom data.
Background
KuJiaLe is a leading Chinese home‑decoration platform that provides design solutions, floor‑plan tools, rendering, and custom furniture. Its custom design tools rely heavily on data, making test data complexity a major challenge. Interface testing is essential to ensure data correctness and is a key release metric.
How to Design Interface Tests
Good backend test developers must know how to design interface test cases. The diagram below outlines the process.
Interface Testing Pain Points
Test data is hard to prepare – custom data is complex and scenarios are numerous.
Test data is hard to maintain – feature changes quickly invalidate existing cases, requiring frequent updates.
Result verification is imprecise – simple field checks are insufficient; JSON noise reduction is needed.
Communicating results to developers is inconvenient – raw JSON bodies are hard to interpret.
Interface Testing at Different Stages
Test Data Preparation
Test Data Maintenance
Result Verification
Communicating Results to Developers
Ask developers to run test code – unrealistic and inefficient.
Send screenshots – confusing.
Send full request/response data – labor‑intensive.
Tips:
Use small tools to automatically upload test results to OSS.
Store test data in models or plans and share the links.
Maintain a public area to record test changes for feedback.
How Custom Interface Testing Is Done
Case 1: 3D interface of the modeleditor service.
Before: Test data manually placed in post body, causing frequent breakage when fields change.
After: A model stores required scenario data; opening the model provides stable body data for comparison, allowing direct 3D interface testing.
Case 2: parameter‑model package split testing.
The split affects many services; to keep the front‑end unaware, object‑level validation and full in‑memory model checks are used to ensure consistency before and after the split.
Result:
Purpose
All non‑manual, systematic operations should be automated.
Qunhe Technology Quality Tech
Kujiale Technology Quality
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.