Chnage-Prone Mining Tool built based on the approach introduced in Evaluating SZZ Implementations Through a Developer-informed Oracle (https://github.com/grosa1/pyszz)
(* algorithm used in this tool is similar but also quite different from the original pyszz leading different results)
To run PySZZ you need:
- You have to install srcML
All external tools have to be available in your system's path.
Run the following command to install the required python dependencies:
pip3 install --no-cache-dir -r app/src/main/java/hotdog/CPCMiner/pyszz/requirements.txt
To run the tool, simply execute the following command:
make clean
make build
make run args="<flag1> <argument1> <flag2> <argument2> ..."
Instead, we made shell commands to run the program partially in sequential order.
make pipe url="<GitHub URL>" wp="<clone directory>" proj="<Github project name>"
This command runs make pc command, make pyszz and make cpc
Please, go over makefile for detail.
compile multi_pipe.c and run it with...
./multi_pipe <.csv file with URL list> <clone path>
This will generate multiple processes run at the same time.
-ip input path: it could be URL of a github repo or simple csv file containing list of URLs.
-wp working path: it is a path for cloning github repo.
-l log: giving this option lets it leave a log file at data directory.
If you want to run the pyszz separately (settings are different from the original) use following command
python3 main.py /path/to/bug-fixes.json /path/to/configuration-file.yml /path/to/repo-directory
We used R-SZZ in this tool. Therefore, AG-SZZ and R-SZZ part of pyszz is modified for our usages.
Other SZZ algorithms can also be used with a custom settings.
Figure above shows the overall approach when run by multi_pipe.c