From 38208c8858e517b8d9926f125fdeb9c362915093 Mon Sep 17 00:00:00 2001
From: Daniil Anfimov <anfimovdan@gmail.com>
Date: Fri, 22 May 2026 12:11:48 +0200
Subject: [PATCH] Add the ability to use multiple QA repos

---
 alts/shared/utils/git_utils.py | 45 ++++++++++++++++++++++++
 alts/worker/runners/base.py    |  8 ++---
 tests/shared/test_git_utils.py | 64 ++++++++++++++++++++++++++++++++++
 3 files changed, 113 insertions(+), 4 deletions(-)
 create mode 100644 tests/shared/test_git_utils.py

diff --git a/alts/shared/utils/git_utils.py b/alts/shared/utils/git_utils.py
index 6cda8b67..bc70fc35 100644
--- a/alts/shared/utils/git_utils.py
+++ b/alts/shared/utils/git_utils.py
@@ -1,3 +1,6 @@
+import os
+import re
+import urllib.parse
 from logging import Logger
 from pathlib import Path
 from random import randint
@@ -9,6 +12,48 @@
 from alts.shared.utils.path_utils import get_abspath
 
 
+_SCP_URL_RE = re.compile(r'^(?P<user>[^@]+@)?(?P<host>[^:/]+):(?P<path>.+)$')
+
+
+def repo_reference_subpath(repo_url: str) -> str:
+    """
+    Derive a host-aware on-disk subpath for a git repository's reference
+    mirror. Two repositories that share a basename but live on different
+    hosts (e.g. a Gerrit "QA" and a GitLab "QA") must map to different
+    subpaths to avoid clobbering each other's bare mirrors.
+
+    Examples
+    --------
+    >>> repo_reference_subpath('ssh://gerrit.cloudlinux.com:29418/QA')
+    'gerrit.cloudlinux.com/QA.git'
+    >>> repo_reference_subpath('https://gitlab.cloudlinux.com/qa/QA.git')
+    'gitlab.cloudlinux.com/qa/QA.git'
+    >>> repo_reference_subpath('git@gitlab.com:qa/QA.git')
+    'gitlab.com/qa/QA.git'
+    """
+    host = None
+    path = None
+    parsed = urllib.parse.urlparse(repo_url)
+    if parsed.hostname:
+        host = parsed.hostname
+        path = parsed.path
+    else:
+        scp_match = _SCP_URL_RE.match(repo_url)
+        if scp_match:
+            host = scp_match.group('host')
+            path = scp_match.group('path')
+    if not host or not path:
+        # Unparseable URL — fall back to basename so behaviour stays
+        # predictable; collisions across hosts remain the caller's problem
+        # in that degenerate case.
+        basename = os.path.basename(repo_url) or 'repo'
+        return basename if basename.endswith('.git') else f'{basename}.git'
+    path = path.strip('/')
+    if not path.endswith('.git'):
+        path = f'{path}.git'
+    return f'{host.lower()}/{path}'
+
+
 def checkout(
     git_ref: str,
     git_repo_path: Path,
diff --git a/alts/worker/runners/base.py b/alts/worker/runners/base.py
index 76b1e8cc..e45ddeda 100644
--- a/alts/worker/runners/base.py
+++ b/alts/worker/runners/base.py
@@ -52,6 +52,7 @@
     clone_git_repo,
     git_reset_hard,
     prepare_gerrit_command,
+    repo_reference_subpath,
 )
 from alts.shared.utils.log_utils import (
     get_temp_log_files,
@@ -1150,13 +1151,12 @@ def clone_third_party_repo(
             self._logger.debug('An unknown repository format, skipping')
             return git_repo_path
         self._logger.info('Cloning %s to %s', repo_url, self._work_dir)
-        repo_name = os.path.basename(repo_url)
-        if not repo_name.endswith('.git'):
-            repo_name += '.git'
         repo_reference_dir = None
         if CONFIG.git_reference_directory:
             repo_reference_dir = os.path.join(
-                CONFIG.git_reference_directory, repo_name)
+                CONFIG.git_reference_directory,
+                repo_reference_subpath(repo_url),
+            )
         repo_path = None
         for attempt in range(1, 6):
             try:
diff --git a/tests/shared/test_git_utils.py b/tests/shared/test_git_utils.py
new file mode 100644
index 00000000..2c8b5d07
--- /dev/null
+++ b/tests/shared/test_git_utils.py
@@ -0,0 +1,64 @@
+"""Tests for alts.shared.utils.git_utils.repo_reference_subpath."""
+import pytest
+
+from alts.shared.utils.git_utils import repo_reference_subpath
+
+
+class TestRepoReferenceSubpath:
+    @pytest.mark.parametrize(
+        'url,expected',
+        [
+            (
+                'ssh://gerrit.cloudlinux.com:29418/QA',
+                'gerrit.cloudlinux.com/QA.git',
+            ),
+            (
+                'https://gitlab.cloudlinux.com/qa-team/QA.git',
+                'gitlab.cloudlinux.com/qa-team/QA.git',
+            ),
+            (
+                'https://github.com/foo/bar.git',
+                'github.com/foo/bar.git',
+            ),
+            (
+                'git@gitlab.com:qa/QA.git',
+                'gitlab.com/qa/QA.git',
+            ),
+            (
+                'git@gitlab.com:qa/QA',
+                'gitlab.com/qa/QA.git',
+            ),
+            # SCP-style URL without a user prefix.
+            (
+                'gitlab.com:qa/QA.git',
+                'gitlab.com/qa/QA.git',
+            ),
+            (
+                'gerrit.cloudlinux.com:QA',
+                'gerrit.cloudlinux.com/QA.git',
+            ),
+        ],
+    )
+    def test_known_urls_map_to_host_aware_subpaths(self, url, expected):
+        assert repo_reference_subpath(url) == expected
+
+    def test_same_basename_different_hosts_diverge(self):
+        gerrit = repo_reference_subpath(
+            'ssh://gerrit.cloudlinux.com:29418/QA',
+        )
+        gitlab = repo_reference_subpath(
+            'https://gitlab.cloudlinux.com/qa-team/QA.git',
+        )
+        assert gerrit != gitlab
+        # Both keys preserve enough host information to be distinguishable.
+        assert 'gerrit.cloudlinux.com' in gerrit
+        assert 'gitlab.cloudlinux.com' in gitlab
+
+    def test_host_is_lowercased(self):
+        assert repo_reference_subpath(
+            'https://GITLAB.COM/foo/Bar.git',
+        ) == 'gitlab.com/foo/Bar.git'
+
+    def test_unparseable_url_falls_back_to_basename(self):
+        # No scheme, no SCP-style colon — degenerate input.
+        assert repo_reference_subpath('just-a-name') == 'just-a-name.git'