program-repair-project / bugfixer-testbench

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

bugfixer-testbench

Installation

  1. Checkout & build
git clone --recursive https://github.com/program-repair-project/bugfixer-testbench.git
make

Currently make only builds bug-localizer. Ignore warnings like

warning: Using 'getservbyport' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
  1. Install Bugzoo
pipenv install bugzoo
  1. Build ManyBugs dockers For example,
bugzoo bug build manybugs:gzip:2009-08-16-3fe0caeada-39a362ae9d

Launch a ManyBugs docker

For example,

./bin/run-docker.py gzip-2009-09-26-a1d3d4019d-f17cbd13a1  --rm

The argument --rm is used to delete the docker container automatically after it terminates.

In the docker container, directory /bugfixer contains

  • localizer: directory for localizer
  • synthesizer: directory for synthesizer
  • bug_desc.json: bug description

Run the bug localizer

In the docker container, the following command runs the localizer:

/bugfixer/localizer/main.exe .

The result is a sorted list of suspicious lines that will be stored in localizer-out/result.txt. Option -skip_compile will skip recompilation.

/bugfixer/localizer/main.exe -skip_compile . 

Run the bug localizer with script

Instead of directcly running the bug localizer inside the docker, you can run the bug localizer in the root directory of this repository with the script, "test-localization.py".

You can configure the localization execution with the three options.

-p is for the project of the target bug. Default will run all the projects.

-c is for the case ID of the target bug. Default will run all the cases in the given project.

-e is for the engine, the localization algorithm, to localize the fault.

For example, you can run

yourWorkingDirectory/bugfixer-testbench/bin/test_localization.py -p gzip -c 2009-09-26-a1d3d4019d-f17cbd13a1 -e prophet

to run bug localizer on bug case "2009-09-26-a1d3d4019d-f17cbd13a1" in project "gzip" with the engine "prophet".

About


Languages

Language:Python 62.1%Language:Shell 30.5%Language:Dockerfile 3.6%Language:Perl 2.9%Language:C 0.7%Language:Makefile 0.2%Language:Assembly 0.0%