Pruning dir-filename in coverage.py html report

Question

When testing some code using coverage.py I stumbled over a problem with too long filenames under windows: the python code to be tested lies pretty well nested somethwere deep in some directory. Down there I also would like to create the actual html report using

coverage.py html

which ends in some failure, that some file does not exist. The problem: the length of the directory string + lenght of filename string is just too long (nice combination of windows problem + my complicated directory structure).

My solution right now: I configure the file .coveragerc such that the html report is written somewhere at c:\tmp\report, which works fine. However, later this task should run somehwere on a Jenkins server, and there I am bound to some longer directory.

My question: why does coverage.py html need to create these very long filenames, including all the directory? Is there any way to prune these filename such that common preambles are ommited?

Answer 1

To answer your specific question, it doesn't! Not anymore, that is. Since around version 6.0, coverage.py uses a different file naming approach, which results in shorter filenames. See: https://github.com/nedbat/coveragepy/issues/580 and the commit that fixes it here: https://github.com/nedbat/coveragepy/commit/4d05ddeeded7f3f594c0614630f467e1bf3fa629

If this is still an issue for you, I recommend updating your coverage package and see if that helps.