iOS 應用程式自動化測試與跨平台整合

Abstract

目前的手機自動化測試工具較多對於 Android 平台的設計與開發，因 iOS 系 統向來較封閉與非開源的情況下，較少開發者願意針對 iOS 平台開發相關測試工 具，然而，iOS 系統上應用程式數量是僅次於 Android 平台，數量也相當多，亦即 自動化測試的需求亦然存在，若能克服 iOS 系統之限制因素，著手開發相關工具， 勢必能造福廣大的開發者。 這篇論文針對 iOS 系統設計測試工具，開發一個負責處理 iOS 基本操作的模 組。此模組會先自動地為系統設定環境與安裝相依套件，開始執行後，可獲取 iOS 應用程式畫面之結構性資料與系統資料，用以辨別畫面中可操作之物件並將物件 解析成符合一現存測試工具規範之相對應資料結構，讓測試工具能夠對應用程式 進行後續操作，並且不會因為底層實作不同，而需重新設計測試演算法。此一測試 工具原始僅支援網頁與 Android 平台之測試，在加入本篇論文所開發之 iOS 模組 後，即能另外支援 iOS 應用程式之測試，以及產生測試報告，成為一跨平台自動化 測試工具。在跨平台整合上，我們著手避免不同平台開發可能造成的不一致性，對 於相似的物件或屬性，進行統一的轉換，使得分析後的資料以及測試報告，能維持 相同的規格及品質。 為了應證此 iOS 模組是否能夠有效執行 iOS 應用程式的操作，我們在測試工 具上設計了包含大多數基本操作的測試演算法，並實際應用於 iOS 上的熱門應用 程式。根據實驗結果，此模組可正確地在手機裝置上執行測試工具所要求的基本操 作，並成功執行 iOS 自動化測試。

Current automated testing tools are mainly developed for the Android platform. Due to the proprietary nature of the iOS platform, developers have paid little attention on test framework developments or studies for iOS. However, the number of iOS applications is only second to that of Android applications, that is, the needs of automated testing tools exist as well. If the obstacles to the development for the iOS platform could be solved, more related tools could be developed and would definitely benefit lots of iOS developers. In this thesis, we focus on testing tool development for the iOS platform that a platform module responsible for handling fundamental operations for iOS devices is developed. It first configures the dependencies for testing on iOS devices automatically. Then, during execution, the hierarchical XML data and system information can be acquired, which is used for recognizing operatable elements and parsed into data structures that conform to an existing testing framework so that it can operate on the system without redefining any top-level testing algorithms. The testing framework is originally for web and Android applications only. With the proposed design in our work, the testing framework is able to support iOS applications as well, i.e. a cross-platform testing framework. For the cross-platform integration, we want to avoid the inconsistencies during developments on different platforms, so we generalize elements or attributes that have similar characteristics by transforming them into identical words, which makes the analyzed data and testing reports have the same standard and quality. To assess the effectiveness of the iOS platform module, we implement some test algorithms for various test scenarios on the testing framework and apply it on popular iOS applications. According to the experiment result, the module can correctly perform operations on the mobile devices based on requests from the testing framework and successfully execute automated tests on iOS devices.