Scrawler

An automated tool to crawl Android apps and detect changes of GUI at the function level.

Requirement

• Android SDK
• Android devices or emulators.
• Python 3.7.6
• All the other requirements are listed in the requirements.txt.

Connect devices

1. Connect devices via the adb tool.
2. Execute adb command in the terminal, e.g.,adb connect.

Crawl an app

Set the parameters in the script file backend/crawler.py:

• save_dir the directory to save the results, e.g., obj = Crawler(package_name, save_dir).
• max_depth
• max_time
• action_interval time interval between two adjacent actions.
• replay_try_times number of attempts to traverse back to the previous screen.
• repoen_to_first whether returning to the main screen when reopening the app.
• distinct_rate threshold to distinct different screens.
• black_elem_list element black list.
• black_screen_list screen black list.
• act_max_count maximum number of screen per activity collected.
• black_act_list activity black list. Then run the script file backend/crawler.py.

When the crawling process finished, we can get the results in the directory save_dir/result/:

• images : screenshots.
• log: which records the crawling process.
• model: containing the screens and actions.
• traverse_graph: describing the crawing process.

Besides, the GUI data of each screen is recorded in the directory save_dir/screen_name, in the form of screenshot and xml file.

Compare the GUI models

1. Set the work directory in the script file backend/model_comparison.py, e.g., obj = Comparator('C:/Users/dell/Desktop/tmp_comparisio').
2. Prepare the base and updated model of an app, and put them in the work directory, i.e., base_model and updated_model.
3. Assure that the screens recorded in the models are in your file system.

Then run the script file backend/model_comparison.py.

When the comparing process finished, we can get the results in the directory work_dir/result/:

• edges : record the removed, matched and added actions.
• screens: record the removed, matched and added screens.
• base_graph: show the results in the base version graph.
• updated_graph: show the results in the updated version graph.

GUI Test Repair

Or we can view the process as finding the new execution path of an function for the app in the updated version.

1. Set the work directory in the script file /get_new_path/find_path.py.
2. Record the event sequences of an function in the base version with coordinates and set the coordinates in the script file /get_new_path/find_path.py.
3. Replay the event sequences in the base version, e.g., obj.replay() by running the script file /get_new_path/find_path.py and we can get the scenario_model in the work directory.
4. Put the updated_model in the work directory.
5. Set the parameters in the script file /get_new_path/find_path.py:
   • distinct_rate: threshold to distinct different screens.
   • text_sim: threshold to match screens.
   • is_circle: whether the path in the results could contain circle.
   • max_candidate_num: the maximum number of candidate paths in the result.
6. Run the script file /get_new_path/find_path.py with only obj.work(), in this time we need not obj.replay().

When the repairing process finished, we can get the results in the directory work_dir/result/:

• candidate: all candidate paths.
• optimal: the best path.