Test case number inconsistent with main repos

[xeophon]

Hey! I am currently looking into NL2RepoBench and try to reproduce the original results by loading the Docker images and then simply copying the main repo over and running the tests.

Lets take pyperclip as an example: The main task from NL2RepoBench covers all platforms, while the Docker image only uses Linux and has no clipboard backend. Also, the test case count is 10, while the main repo has over 80. How were those test case counts obtained?

[YYDJZ]

Thanks for your comment. We calculate the number of test cases by calculating the nums of cases passing tests during the environment build session. Considering some of the tests may be skipped, so we don't take them into account. We are currently working on checking the number of test cases again.