ManySStuBs4J Dataset

The ManySStuBs4J corpus is a collection of simple fixes to Java bugs, designed for evaluating program repair techniques. We collect all bug-fixing changes using the SZZ heuristic, and then filter these to obtain a data set of small bug fix changes.

Download and unzip dataset

python

!wget https://zenodo.org/record/5845439/files/tssb_data_3M.zip?download=1

python

!mv tssb_data_3M.zip?download=1 tssb_data_3M.zip

python

!unzip tssb_data_3M.zip

Read dataset with pandas and show it as a table

python

import pandas

FILENUM = 32
table = pandas.read_json(f"tssb_data_3M/file-{FILENUM}.jsonl.gz", lines=True)
table

Getting the urls of the repositories, useful to get commits from SHA

python

urls = set()

total_rows = 0
for num in range(0, 34):
    table = pandas.read_json(f"tssb_data_3M/file-{FILENUM}.jsonl.gz", lines=True)
    for index, row in table.iterrows():
        total_rows += 1
        urls.add(row["project"])
print(len(urls))
print("total rows", total_rows)
print(urls)

Get commits from SHA TODO!

python

!pip install PyGithub

python

from github import Github

g = Github()

# TO DO, find a way to get a commmit from SHA
# 1. Use GitHub API
# 2. Download repos with their history
# 3. Web scaping

Write the prompts

python

import re

TEMPLATE = """User: Find the bug in the following code:
{}
Reply: The fixed code is:

{}



"""


def remove_starting_plus_minus(text):
    if text.startswith("+") or text.startswith("-"):
        return text[1:]
    else:
        return text


def remove_extraneous_diff_info(text):
    pattern = "@@.*@@"
    return re.sub(pattern, "", text)


def clean(text):
    return remove_extraneous_diff_info(remove_starting_plus_minus(text))


def write_prompts(num, table):
    with open(f"generated_bugfix_prompts/prompts_{num}.txt", "w+") as f:
        for index, row in table.iterrows():
            # lines starting with "+" are the "new" lines, after the modification to fix the bug
            # removing them gives us the old code, with the bug
            correct = "\n".join(clean(line) for line in row["diff"].split("\n") if not line.startswith("+"))
            # lines starting with "-" are the bugged lines, removing them gives the "new" fixed code
            wrong = "\n".join(clean(line) for line in row["diff"].split("\n") if not line.startswith("-"))
            f.write(TEMPLATE.format(correct, wrong))

python

!mkdir generated_bugfix_prompts

python

for num in range(34):
    table = pandas.read_json(f"tssb_data_3M/file-{num}.jsonl.gz", lines=True)
    write_prompts(num, table)

Zip the prompts together for easier downloading

python

!zip -r generated_bugfix_prompts.zip generated_bugfix_prompts