Stoat Prototype

Stoat (STochastic model App Tester) is a guided approach to perform stochastic model-based testing on Android Apps. The idea is to thoroughly test the functionalities of an app from its GUI model, and validate the app’s behavior by enforcing various user/system interactions.

You can find more about our work on testing/analyzing Android apps at this website.

Update

We have made all the Stoat's source code publicly available. We hope this project can benefit other researchers or practiontiners in the field of automated GUI testing of Android apps. Feel free to contact us if you have any questions and issues. We will continue to maintain this project. Thanks for your feedback.

Next step: We will integrate Jacoco with Stoat to handle gradle-based Android projects.

Publication

[1] Guided, Stochastic Model-based GUI Testing of Android Apps (FSE'2017)

@inproceedings{DBLP:conf/sigsoft/SuMCWYYPLS17,
  author    = {Ting Su and
               Guozhu Meng and
               Yuting Chen and
               Ke Wu and
               Weiming Yang and
               Yao Yao and
               Geguang Pu and
               Yang Liu and
               Zhendong Su},
  title     = {Guided, stochastic model-based {GUI} testing of Android apps},
  booktitle = {Proceedings of the 2017 11th Joint Meeting on Foundations of Software
               Engineering, {ESEC/FSE} 2017, Paderborn, Germany, September 4-8, 2017},
  pages     = {245--256}
}

[2] FSMdroid: Guided GUI Testing of Android Apps (ICSE'16, ACM SRC)

@inproceedings{FSMdroid16,
  author    = {Ting Su},
  title     = {FSMdroid: guided {GUI} testing of android apps},
  booktitle = {Proceedings of the 38th International Conference on Software Engineering,
               {ICSE} 2016, Austin, TX, USA, May 14-22, 2016 - Companion Volume},
  pages     = {689--691},
  year      = {2016}
}

[3] Large-Scale Analysis of Framework-Specific Exceptions in Android Apps (ICSE'18)

@inproceedings{FanSCMLXPS18,
  author    = {Lingling Fan and
               Ting Su and
               Sen Chen and
               Guozhu Meng and
               Yang Liu and
               Lihua Xu and
               Geguang Pu and
               Zhendong Su},
  title     = {Large-scale analysis of framework-specific exceptions in Android apps},
  booktitle = {Proceedings of the 40th International Conference on Software Engineering,
               {ICSE} 2018, Gothenburg, Sweden, May 27 - June 03, 2018},
  pages     = {408--419},
  year      = {2018}
}

Quick Review

Click this link for quick review on the approach workflow, evaluation results, demo and etc.

Setup

You can checkout Stoat from this repo.

Environment Configration

Ruby: 2.1
Nokogiri

If your default ruby version is lower than 2.1, the installation of Nokogiri will fail. In this case, please upgrade ruby to 2.1 or higher. If the installation still fails due to this issue, you need to execute "sudo apt-get install ruby2.1-dev".

Python: 2.7
Android SDK: API 18+

Android SDK recently makes an update about emulators. If you cannot start the emulator (like "Cannot launch AVD in emulator:QT library not found"), please check this post for solution. You can create an emulator before running Stoat. See this link for how to create avd using avdmanager.

E.g.

sdkmanager "system-images;android-18;google_apis;x86"
avdmanager create avd --force --name testAVD_1 --package 'system-images;android-18;google_apis;x86' --abi google_apis/x86 --sdcard 512M --device 'Nexus 7'

Ubuntu 14.04/Linux
uiautomator (A python wrapper of UIAutomator)

We strongly recommend to run Stoat on a physical machine to ensure the performance instead of ruuning on virtual machines (e.g., VirtualBox or Docker). In addition, please choose x86 image if you use Android emulators, and setup hardware acceleration for Ubuntu.

Please export ANDROID_HOME (for android sdk), PYTHON_PATH (for uiautomator), CLASSPATH (for soot)

Example:

export ANDROID_HOME="/home/XX/Android/Sdk"
export PYTHONPATH="/home/XX/uiautomator"
export CLASSPATH="/home/XX/fsmdroid/soot-github/lib/soot-develop.jar
export PATH=$PATH:${ANDROID_HOME}/build-tools/25.0.0:${ANDROID_HOME}/emulator:${ANDROID_HOME}/tools:${ANDROID_HOME}/tools/bin:${ANDROID_HOME}/platform-tools:

You may also need to modify "Stoat/CONF.txt" to set the tool path.

Usage

Stoat provides several ways to test android apps by command lines. Note before running Stoat, please disable keyboard (for emulator, add "hw.keyboard=yes" in its config.ini; for real device, please install the "com.wparam.nullkeyboard_1.apk" at "Stoat/bin/sdcard", and configure it to be the default input method) Please also add "hw.mainKeys=yes" in the config.ini to disable the soft "main" keys if you use emulators, which may affect Stoat's exploration.

Ant opens-soruce projects

ruby run_stoat_testing.rb --app_dir /home/XX/caldwell.ben.bites_4_src --avd_name testAVD_1 --avd_port 5554 --stoat_port 2000 --project_type ant
apk without instrumentation

Note this may mitigate Stoat's power due to lack of coverage info for test optimization. Otherwise, you need to instrument apk with Ella

ruby run_stoat_testing.rb --app_dir /home/XX/Bites.apk --avd_name testAVD_1 --avd_port 5554 --stoat_port 2000 (the output will be under "/home/XX/Bites-output")
Use real device, ant projects

Please open wifi, and disable keyboard before do testing on real device.

ruby run_stoat_testing.rb --app_dir /home/XX/caldwell.ben.bites_4_src/ --real_device_serial cf00b9e6 --stoat_port 2000 --project_type ant
A list of apps (If they are apks, append the option "--project_type apk")

ruby run_stoat_testing.rb --apps_dir /home/XX/test_apps/ --apps_list /home/XX/test_apps/apps_list.txt --avd_name testAVD_1 --avd_port 5554 --stoat_port 2000 --force_restart

Subject Requirement:

instrumented apps (use Emma for open-source apps and Ella for closed-source apps) should end with "-debug.apk"
closed-source/non-instrumented apk can also be tested, and its name should end with ".apk"

Output

stoat_fsm_building_output: the outputs of model construction.

 crashes/ -- crash report (include crash stack, event trace, screen shots); 
 ui/ -- ui xml files; 
 coverage/ -- coverage files during model construction; 
 FSM.txt/app.gv -- xdot model graph; 
 fsm_building_process.txt/fsm_states_edges.txt -- the model building process, mainly the increasing coverage/#states/#edges 
 CONF.txt -- configuration file

stoat_mcmc_sampling_output: the outputs of mcmc sampling.

 crashes/ -- crash report (include crash stack, event trace, screen shots); 
 MCMC_coverage/ -- the coverage data during mcmc sampling; 
 mcmc_sampling_progress.txt/mcmc_data.txt -- mcmc sampling progress data; 
 initial_markov_model.txt/optimal_markov_model.txt/mcmc_models.txt -- the initial/optimal/all mcmc sampling models; 
 mcmc_all_history_testsuites.txt -- all executed test suites for mcmc sampling; 
 test_suite_to_execute.txt -- the current test suite under execution;
 CONF.txt -- configuration file.

coverage: the all coverage data during two phases

Benchmark

Some benchmark apps used in our paper.

Notes

This implementation has been tested with Android 4.4, running on Ubuntu 14.04.
If measure statement coverage for open-sourced apps, the subjects need to be processed to support EMMA instrumentation: (Please refer to Dynodroid for details.) You can also refer to the apps in the benchmark.
This version only supports testing ant projects.

Contact

Ting Su All Copyright Reserved.

TODO

use monkey to bypass welcome page in google play apps
directly use monkey to start the app instead of using "am", the monkey way is more robust.

Papers that uses, extends or compares with Stoat

Zhe Liu, Chunyang Chen, Junjie Wang, Xing Che, Yuekai Huang, Jun Hu, Qing Wang. Efficiency Matters: Speeding Up Automated Testing with GUI Rendering Inference. ICSE 2023
Zhengwei Lv, Chao Peng, Zhao Zhang, Ting Su, Kai Liu, Ping Yang. Fastbot2: Reusable Automated Model-based GUI Testing for Android Enhanced by Reinforcement Learning. ASE 2022.
Wenyu Wang, Wing Lam, and Tao Xie. An infrastructure approach to improving effectiveness of Android UI testing tools." Proceedings of the 30th ACM SIGSOFT International Symposium on Software Testing and Analysis. ISSTA 2021.
Ting Su, Jue Wang, Zhendong Su. Benchmarking Automated GUI Testing for Android against Real-World Bugs. ESEC/FSE 2021.
Hao Zhou, Ting Chen, Haoyu Wang, Le Yu, Xiapu Luo, Ting Wang, Wei Zhang. UI Obfuscation and Its Effects on Automated UI Analysis for Android Apps. ASE 2020.
Minxue Pan, An Huang, Guoxin Wang, Tian Zhang, and Xuandong Li. Reinforcement Learning Based Curiosity-driven Testing of Android Applications. ISSTA 2020.
Yuyu He, Lei Zhang, Zhemin Yang, Yinzhi Cao, Keke Lian, Shuai Li, Wei Yang, Zhibo Zhang, Min Yang, Yuan Zhang, Haixin Duan. *TextExerciser: Feedback-driv

Stoat

Install / Use

README