How a Regression Platform Transforms Construction Drawing Testing and Boosts Efficiency

Background Introduction

The core of construction‑drawing business is converting 3D design schemes into 2D drawings to guide building and decoration work. In the product flow, construction drawings are at the end of the pipeline.

Floor plan import/design: create or import house floor plans.

Hard‑decoration, MEP, lighting, custom, soft‑decoration design: complete interior details.

Rendering: generate renderings and walkthroughs.

Construction‑drawing design: convert 3D content to 2D sheets with annotations.

Export drawings: final output of construction drawings.

The key challenge is handling massive design data and accurately converting it into various 2D drawings such as layout plans, elevations, sections, detail views, material tables, and MEP system diagrams.

Example soft‑decoration drawings illustrate multiple perspectives, annotations, and markings.

A complete set also includes custom, hard‑decoration, and MEP drawings, each with different details and views, resulting in a large and complex set of files (PDF, DWG, JPG).

Ensuring drawing quality requires strict quality‑control measures.

Platform Design

Initial Core Requirements

Test scenario management: maintain data/drawing baselines for different design schemes and business lines.

Task management: manage, execute, and trace task collections.

Automated data processing: real‑time acquisition of drawing data from front‑end applications.

Data analysis and comparison: analyze large JSON data sets and support visualization.

Drawing diff marking: compare and highlight differences across PDF/JPG/DWG formats.

Platform Business Architecture

Tool layer: independent utilities for case import, issue tracing, encryption, config push, and comparison.

Interaction layer: core UI providing scenario and baseline management, test set construction, task execution, and reporting.

Engine layer: underlying services for data processing, analysis, comparison, and visualization, ensuring extensibility and modularity.

Core Module – Test Baseline

Baseline creation: batch import or manual creation of schemes and baselines, with tagging and classification.

Baseline import: one‑click generation of test cases, baselines, and test sets from real user schemes.

After defining test scenarios, the platform generates baselines, collects backend service data, and captures design‑scheme data via automated tasks dispatched to the Apollo platform.

Regression Tasks

Data collection & comparison: collect data for specified business code versions, compare drawing and data differences, and generate visual results.

Result notification: after task completion, notify stakeholders via enterprise WeChat or email for analysis and confirmation.

Regression Results

Data diff results are visualized using the viewer, turning raw JSON comparisons into marked drawings, greatly improving issue‑resolution efficiency.

Data diff and visualization.

Reports are built with BizCharts4.

Platform Benefits

Design‑scheme regression throughput increased from single digits per hour to hundreds per hour, enabling thousands of regressions per day.

Automated defect detection rate exceeds 25%.

Stability improved: high‑priority failures reduced to zero, and drawing‑export module failures eliminated.

Process Promotion – Test Left‑Shift

The regression capability moves testing earlier into development self‑test and feature testing, catching issues sooner and reducing fix costs.

Future Outlook

Custom data collection: open a configurable data‑collection module for business lines.

Full‑automation CI: integrate with front‑ and back‑end CI pipelines to trigger regression from code merge.

Automatic case selection: match regression cases to code changes for precise testing.

Continuous coverage expansion: maintain module and ticket‑driven test cases and improve code‑coverage metrics.