metadata
configs:
- config_name: default
data_files:
- split: test
path: data/test-*
dataset_info:
features:
- name: target_id
dtype: string
- name: project
dtype: string
- name: level
dtype: string
- name: target_file
dtype: string
- name: target_line
dtype: int64
- name: url
dtype: string
- name: commit
dtype: string
- name: patch
dtype: string
- name: dockerfile
dtype: string
- name: fuzzing_harness
dtype: string
- name: build_script
dtype: string
- name: harness_name
dtype: string
- name: realistic_context
dtype: string
- name: realistic_call_paths
dtype: string
splits:
- name: test
num_bytes: 125865708
num_examples: 317
download_size: 42082752
dataset_size: 125865708
Dataset Summary
dfuzzBench is a dataset tests systems' ability to generate directed inputs to reach a specific target line in the codebase automatically.
TODO: Details of the dataset
Dataset Structure
project: (str) - The name of the project.
level: (str) - The level of difficulty to reach the target based on the first hit time of several traditional undirected fuzzers.
target_file: (str) - The file path where the target line is located.
target_line: (int) - The line number of the target line in the target file.
url: (str) - The url of the project Github repository.
commit: (str) - The commit hash of the repository when we performed the target line selection.
patch: (str) - The patch which adds a print statement "HIT TARGET!" at the target line to indicate the target is reached and exit.
dockerfile: (str) - The Dockerfile of evaluation environment.
fuzzing_harness: (str) - The fuzzing harness, used by Oss-Fuzz.
build_script: (str) - The build script to install required dependencies for fuzzing campaigns.
realistic_context: (list) - The source code of programs retrieved with realistic retrieval.
realistic_call_paths: (list) - The valid call paths from fuzzing harness to the target function retrieved from static call graph.