From 4c7f30ce97848a8b21e4eb7aa2aeb4887df8b4e6 Mon Sep 17 00:00:00 2001 From: ajdesh2000 <47187312+ajdesh2000@users.noreply.github.com> Date: Sun, 15 Sep 2024 12:32:48 -0400 Subject: [PATCH] Update java_dataset.py --- repoclassbench/dataset/java_dataset.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/repoclassbench/dataset/java_dataset.py b/repoclassbench/dataset/java_dataset.py index 8ee6015..0226541 100755 --- a/repoclassbench/dataset/java_dataset.py +++ b/repoclassbench/dataset/java_dataset.py @@ -21,7 +21,9 @@ def __init__(self, specification: str, delete_relatives: bool) -> None: self.data = json.load(open("data/input/java_data.json", "r")) self._download_data() ## Extract jdk and maven - + + os.makedirs("external/java",exist_ok=True) + if not os.path.exists("external/java/jdk-17.0.6"): if not os.path.exists("external/java/jdk-17.0.6.zip"): data_url = "https://drive.google.com/uc?id=1HIJICJgQQvM_LzbSVRdBlQyiD_kY5BNc" @@ -31,7 +33,7 @@ def __init__(self, specification: str, delete_relatives: bool) -> None: if not os.path.exists("external/java/apache-maven-3.8.7"): if not os.path.exists("external/java/apache-maven-3.8.7.zip"): - data_url = "https://drive.google.com/uc?id=1JFzF2oAzS8D31fhtpn3uIWhhWJTGG" + data_url = "https://drive.google.com/uc?id=1JFzF2oAzS8D31fhtpn3uIWhhWJTGG-5i" gdown.download(data_url, "external/java/apache-maven-3.8.7.zip", quiet=False) with zipfile.ZipFile( "external/java/apache-maven-3.8.7.zip", "r" @@ -49,6 +51,8 @@ def __init__(self, specification: str, delete_relatives: bool) -> None: def _download_data(self) -> None: if os.path.exists("temp/java/original_repo"): return + + os.makedirs("temp/java",exist_ok=True) data_url = "https://drive.google.com/uc?id=16ZeWM_wKfeBfm7rvsBnksbVJZuLAZ1yo" gdown.download(data_url, "temp/java/java_repos.zip", quiet=False)