From fe8ee850f477b55db90016668a8d1b2db27085fd Mon Sep 17 00:00:00 2001
From: Kaiyu Yang <kaiyuy@meta.com>
Date: Tue, 3 Dec 2024 16:56:57 +0000
Subject: [PATCH] minor updates

---
 src/lean_dojo/data_extraction/lean.py | 13 +++++++++----
 tests/conftest.py                     |  2 ++
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/src/lean_dojo/data_extraction/lean.py b/src/lean_dojo/data_extraction/lean.py
index 0fc7bedb..1e9a2d20 100644
--- a/src/lean_dojo/data_extraction/lean.py
+++ b/src/lean_dojo/data_extraction/lean.py
@@ -55,7 +55,7 @@
 
 _URL_REGEX = re.compile(r"(?P<url>.*?)/*")
 
-_SSH_TO_HTTPS_REGEX = re.compile(r"^git@github\.com:(.+)/(.+)(?:\.git)?$")
+_SSH_TO_HTTPS_REGEX = re.compile(r"git@github\.com:(?P<user>.+)/(?P<repo>.+)\.git")
 
 REPO_CACHE_PREFIX = "repos"
 
@@ -70,7 +70,13 @@ def normalize_url(url: str, repo_type: RepoType = RepoType.GITHUB) -> str:
     if repo_type == RepoType.LOCAL:  # Convert to absolute path if local.
         return os.path.abspath(url)
     # Remove trailing `/`.
-    return _URL_REGEX.fullmatch(url)["url"]  # type: ignore
+    url = _URL_REGEX.fullmatch(url)["url"]  # type: ignore
+    return ssh_to_https(url)
+
+
+def ssh_to_https(url: str) -> str:
+    m = _SSH_TO_HTTPS_REGEX.fullmatch(url)
+    return f"https://github.com/{m.group('user')}/{m.group('repo')}" if m else url
 
 
 def get_repo_type(url: str) -> Optional[RepoType]:
@@ -81,8 +87,7 @@ def get_repo_type(url: str) -> Optional[RepoType]:
     Returns:
         Optional[str]: The type of the repository (None if the repo cannot be found).
     """
-    m = _SSH_TO_HTTPS_REGEX.match(url)
-    url = f"https://github.com/{m.group(1)}/{m.group(2)}" if m else url
+    url = ssh_to_https(url)
     parsed_url = urllib.parse.urlparse(url)  # type: ignore
     if parsed_url.scheme in ["http", "https"]:
         # Case 1 - GitHub URL.
diff --git a/tests/conftest.py b/tests/conftest.py
index d581c620..a735e64e 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -3,6 +3,7 @@
 from lean_dojo import *
 
 
+MINIF2F_URL = "git@github.com:yangky11/miniF2F-lean4.git"
 BATTERIES_URL = "https://github.com/leanprover-community/batteries"
 AESOP_URL = "https://github.com/leanprover-community/aesop"
 MATHLIB4_URL = "https://github.com/leanprover-community/mathlib4"
@@ -10,6 +11,7 @@
 EXAMPLE_COMMIT_HASH = "3f8c5eb303a225cdef609498b8d87262e5ef344b"
 REMOTE_EXAMPLE_URL = "https://gitee.com/rexzong/lean4-example"
 URLS = [
+    MINIF2F_URL,
     BATTERIES_URL,
     AESOP_URL,
     MATHLIB4_URL,