refactor: implement _get_status_code, adjust _check_repo_exists, and update tests

- Implemented function `_get_status_code` in repository_clone.py to extract the status code from an HTTP response
- Adjusted `_check_repo_exists` in repository_clone.py to utilize the new `_get_status_code` function
- Modified `_check_repo_exists` to return True for status codes 200 and 301, and False for 404 and 302
- Updated `test_check_repo_exists_with_redirect` in test_repository_clone.py to verify that `_check_repo_exists` returns False for status code 302
- Implemented test `test_check_repo_exists_with_permanent_redirect` in test_repository_clone.py to verify that `_check_repo_exists` returns True for status code 301

Author: filipchristiansen

src/gitingest/repository_clone.py

@@ -111,6 +111,11 @@ async def _check_repo_exists(url: str) -> bool:
     -------
     bool
         True if the repository exists, False otherwise.
+
+    Raises
+    ------
+    RuntimeError
+        If the curl command returns an unexpected status code.
     """
     proc = await asyncio.create_subprocess_exec(
         "curl",
@@ -120,11 +125,20 @@ async def _check_repo_exists(url: str) -> bool:
         stderr=asyncio.subprocess.PIPE,
     )
     stdout, _ = await proc.communicate()
+
     if proc.returncode != 0:
         return False
-    # Check if stdout contains "404" status code
-    stdout_str = stdout.decode()
-    return "HTTP/1.1 404" not in stdout_str and "HTTP/2 404" not in stdout_str
+
+    response = stdout.decode()
+    status_code = _get_status_code(response)
+
+    if status_code in (200, 301):
+        return True
+
+    if status_code in (404, 302):
+        return False
+
+    raise RuntimeError(f"Unexpected status code: {status_code}")
 
 
 async def _run_git_command(*args: str) -> tuple[bytes, bytes]:
@@ -157,3 +171,22 @@ async def _run_git_command(*args: str) -> tuple[bytes, bytes]:
         raise RuntimeError(f"Git command failed: {' '.join(args)}\nError: {error_message}")
 
     return stdout, stderr
+
+
+def _get_status_code(response: str) -> int:
+    """
+    Extract the status code from an HTTP response.
+
+    Parameters
+    ----------
+    response : str
+        The HTTP response string.
+
+    Returns
+    -------
+    int
+        The status code of the response
+    """
+    status_line = response.splitlines()[0].strip()
+    status_code = int(status_line.split(" ", 2)[1])
+    return status_code

tests/test_repository_clone.py

@@ -204,8 +204,9 @@ async def test_clone_repo_commit_without_branch() -> None:
 @pytest.mark.asyncio
 async def test_check_repo_exists_with_redirect() -> None:
     """
-    Test the `_check_repo_exists` function for handling HTTP redirects (302 Found).
-    Verifies that it correctly identifies the repository's existence.
+    Test the `_check_repo_exists` function when the repository URL returns a redirect response.
+
+    Verifies that the function returns False when a 302 Found response is received.
     """
     url = "https://github.com/user/repo"
     with patch("asyncio.create_subprocess_exec", new_callable=AsyncMock) as mock_exec:
@@ -214,4 +215,21 @@ async def test_check_repo_exists_with_redirect() -> None:
         mock_process.returncode = 0  # Simulate successful request
         mock_exec.return_value = mock_process
 
+        assert await _check_repo_exists(url) is False
+
+
+@pytest.mark.asyncio
+async def test_check_repo_exists_with_permanent_redirect() -> None:
+    """
+    Test the `_check_repo_exists` function when the repository URL returns a redirect response.
+
+    Verifies that the function returns True when a 301 Found response is received.
+    """
+    url = "https://github.com/user/repo"
+    with patch("asyncio.create_subprocess_exec", new_callable=AsyncMock) as mock_exec:
+        mock_process = AsyncMock()
+        mock_process.communicate.return_value = (b"HTTP/1.1 301 Found\n", b"")
+        mock_process.returncode = 0  # Simulate successful request
+        mock_exec.return_value = mock_process
+
         assert await _check_repo_exists(url)