fix: repo_id scheme to use a better separator (#11)

r2e/config.yml

@@ -1,4 +1,3 @@
-bucket_dir: buckets/repoeval_bucket
 r2e_bucket_dir: buckets/r2e_bucket
 repos_dir: buckets/local_repoeval_bucket/repos
 cache_dir: cache_dir

r2e/models/repo.py

@@ -27,7 +27,7 @@ def from_file_path(cls, file_path: Path | str) -> "Repo":
         # NOTE: this is a hacky way to get the repo path
         repo_path = file_path.split(str(REPOS_DIR))[1].split("/")[1]
         try:
-            repo_org, repo_name = repo_path.split("|")
+            repo_org, repo_name = repo_path.split("___")
         except:
             repo_org = repo_path
             repo_name = repo_path

r2e/paths.py

@@ -3,8 +3,6 @@
 from pathlib import Path
 import yaml
 
-# TODO: FIXME! eventually, bucket_dir and r2e_bucket_dir will be merged
-
 # Load configuration from config.yaml
 
 current_dir = Path(__file__).parent

r2e/repo_builder/run_pycg.py

@@ -12,7 +12,7 @@
 def construct_pycg(repo: Repo):
     cgraph = CallGraphGenerator.construct_call_graph(repo.repo_path)
     GRAPHS_DIR.mkdir(parents=True, exist_ok=True)
-    with open(GRAPHS_DIR / f"{repo.repo_id}_cgraph.json", "w") as f:
+    with open(repo.callgraph_path, "w") as f:
         json.dump(cgraph, f, indent=4)
 
     CallGraphProcessor.remove_unresolvable_callers(repo)
@@ -24,7 +24,7 @@ def construct_pycg(repo: Repo):
 
     new_cgraph = {"graph": cgdict, "id2type": id2type}
 
-    with open(GRAPHS_DIR / f"{repo.repo_id}_cgraph.json", "w") as f:
+    with open(repo.callgraph_path, "w") as f:
         json.dump(new_cgraph, f, indent=4)
 
     return new_cgraph

r2e/repo_builder/setup_repos.py

@@ -19,8 +19,10 @@ def clone_and_setup_repos(repo_args: RepoArgs):
         REPOS_DIR.mkdir(parents=True, exist_ok=True)
         if repo_args.repo_url:
             SetupRepos.clone_repo_from_url(repo_args.repo_url)
+
         elif repo_args.local_repo_path:
             SetupRepos.copy_repo(repo_args.local_repo_path)
+
         elif repo_args.repo_paths_file:
             with open(repo_args.repo_paths_file) as f:
                 repo_paths: list[str] = json.load(f)
@@ -31,6 +33,7 @@ def clone_and_setup_repos(repo_args: RepoArgs):
                     isinstance(x, str) for x in repo_paths
                 ), f"Expected list of strings, got {repo_paths}"
             SetupRepos.copy_repos(repo_paths, repo_args.cloning_multiprocess)
+
         elif repo_args.repo_urls_file:
             with open(repo_args.repo_urls_file) as f:
                 repo_urls: list[str] = json.load(f)
@@ -49,7 +52,7 @@ def clone_repo_from_url(repo_url: str):
         repo_username, repo_name = (
             repo_url.rstrip("/").removesuffix(".git").split("/")[-2:]
         )
-        local_repo_clone_path = REPOS_DIR / f"{repo_username}|{repo_name}"
+        local_repo_clone_path = REPOS_DIR / f"{repo_username}___{repo_name}"
 
         if os.path.exists(local_repo_clone_path):
             print(
@@ -66,7 +69,7 @@ def copy_repo(local_repo_path: str):
         local_repo_path = str(Path(local_repo_path).resolve())
 
         local_repo_name = local_repo_path.split("/")[-1]
-        local_repo_clone_path = REPOS_DIR / f"LOCAL|{local_repo_name}"
+        local_repo_clone_path = REPOS_DIR / f"LOCAL___{local_repo_name}"
 
         if os.path.exists(local_repo_clone_path):
             print(