fix: mount repos in working dir, not (pvc)mount path (#2005)

* fix: mount repos in working dir, not (pvc)mount path

* fix commit sha only being checked out as anonymous user

* address comments

git_services/git_services/init/clone.py

@@ -1,8 +1,11 @@
+import logging
 import sys
+from pathlib import Path
+from typing import cast
 
 from git_services.cli.sentry import setup_sentry
 from git_services.init import errors
-from git_services.init.cloner import GitCloner
+from git_services.init.cloner import GitCloner, Repository
 from git_services.init.config import config_from_env
 
 # NOTE: register exception handler
@@ -11,13 +14,15 @@
 if __name__ == "__main__":
     config = config_from_env()
     setup_sentry(config.sentry)
+    logging.basicConfig(level=logging.INFO)
+    base_path = Path(config.mount_path)
 
     git_cloner = GitCloner(
-        repositories=config.repositories,
-        git_providers=config.git_providers,
-        workspace_mount_path=config.workspace_mount_path,
+        repositories=[Repository.from_config_repo(r, mount_path=base_path) for r in config.repositories],
+        git_providers={p.id: p for p in config.git_providers},
+        mount_path=base_path,
         user=config.user,
-        lfs_auto_fetch=config.lfs_auto_fetch,
-        is_git_proxy_enabled=config.is_git_proxy_enabled,
+        lfs_auto_fetch=cast(bool, config.lfs_auto_fetch),
+        is_git_proxy_enabled=cast(bool, config.is_git_proxy_enabled),
     )
     git_cloner.run(storage_mounts=config.storage_mounts)

git_services/git_services/init/cloner.py

@@ -2,7 +2,7 @@
 import logging
 import re
 from contextlib import contextmanager
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from pathlib import Path
 from shutil import disk_usage
 from urllib.parse import urljoin, urlparse
@@ -28,15 +28,15 @@ class Repository:
     _git_cli: GitCLI | None = None
 
     @classmethod
-    def from_config_repo(cls, data: ConfigRepo, workspace_mount_path: Path):
+    def from_config_repo(cls, data: ConfigRepo, mount_path: Path):
         dirname = data.dirname or cls._make_dirname(data.url)
         provider = data.provider
         branch = data.branch
         commit_sha = data.commit_sha
         return cls(
             url=data.url,
             dirname=dirname,
-            absolute_path=workspace_mount_path / dirname,
+            absolute_path=mount_path / dirname,
             provider=provider,
             branch=branch,
             commit_sha=commit_sha,
@@ -67,31 +67,18 @@ def _make_dirname(url: str) -> str:
         return path.rsplit("/", maxsplit=1).pop()
 
 
+@dataclass
 class GitCloner:
+    mount_path: Path
+    git_providers: dict[str, Provider]
+    user: User
+    repositories: list[Repository]
+    lfs_auto_fetch: bool = False
+    is_git_proxy_enabled: bool = False
     remote_name = "origin"
     remote_origin_prefix = f"remotes/{remote_name}"
     proxy_url = "http://localhost:8080"
-
-    def __init__(
-        self,
-        repositories: list[ConfigRepo],
-        git_providers: list[Provider],
-        workspace_mount_path: str,
-        user: User,
-        lfs_auto_fetch=False,
-        is_git_proxy_enabled=False,
-    ):
-        base_path = Path(workspace_mount_path)
-        logging.basicConfig(level=logging.INFO)
-        self.repositories: list[Repository] = [
-            Repository.from_config_repo(r, workspace_mount_path=base_path) for r in repositories
-        ]
-        self.git_providers = {p.id: p for p in git_providers}
-        self.workspace_mount_path = Path(workspace_mount_path)
-        self.user = user
-        self.lfs_auto_fetch = lfs_auto_fetch
-        self.is_git_proxy_enabled = is_git_proxy_enabled
-        self._access_tokens: dict[str, str | None] = dict()
+    _access_tokens: dict[str, str | None] = field(default_factory=dict, repr=False)
 
     def _initialize_repo(self, repository: Repository):
         logging.info("Initializing repo")
@@ -264,15 +251,13 @@ def run_helper(self, repository: Repository, *, storage_mounts: list[str]):
 
         self._initialize_repo(repository)
         try:
-            if self.user.is_anonymous:
-                self._clone(repository)
-                if repository.commit_sha:
-                    repository.git_cli.git_reset("--hard", repository.commit_sha)
-            elif git_access_token is None:
+            if self.user.is_anonymous or git_access_token is None:
                 self._clone(repository)
             else:
                 with self._temp_plaintext_credentials(repository, git_user, git_access_token):
                     self._clone(repository)
+            if repository.commit_sha:
+                repository.git_cli.git_reset("--hard", repository.commit_sha)
         except errors.GitFetchError as err:
             logging.error(msg=f"Cannot clone {repository.url}", exc_info=err)
             with open(repository.absolute_path / "ERROR", mode="w", encoding="utf-8") as f:

git_services/git_services/init/config.py

@@ -52,7 +52,6 @@ class Provider:
 @dataclass
 class Config:
     sentry: SentryConfig
-    workspace_mount_path: str
     mount_path: str
     user: User
     repositories: list[Repository] = field(default_factory=list)

git_services/tests/test_init_clone.py

@@ -6,8 +6,9 @@
 
 from git_services.cli import GitCLI
 from git_services.init import errors
-from git_services.init.clone import GitCloner
-from git_services.init.config import Repository, User
+from git_services.init.clone import GitCloner, Repository
+from git_services.init.config import Repository as ConfigRepo
+from git_services.init.config import User
 
 
 @pytest.fixture
@@ -31,11 +32,12 @@ def clone_dir(tmp_path: Path):
 def test_simple_git_clone(test_user: User, clone_dir: str, mocker):
     repo_url = "https://github.com/SwissDataScienceCenter/amalthea.git"
     mocker.patch("git_services.init.cloner.GitCloner._temp_plaintext_credentials", autospec=True)
-    repositories = [Repository(url=repo_url)]
+    mount_path = Path(clone_dir)
+    repositories = [Repository.from_config_repo(ConfigRepo(url=repo_url), mount_path=mount_path)]
     cloner = GitCloner(
         repositories=repositories,
-        git_providers=[],
-        workspace_mount_path=clone_dir,
+        git_providers={},
+        mount_path=mount_path,
         user=test_user,
     )
 
@@ -55,11 +57,12 @@ def test_lfs_size_check(test_user, clone_dir, mocker):
     mock_disk_usage = mocker.patch("git_services.init.cloner.disk_usage", autospec=True)
     mock_get_lfs_total_size_bytes.return_value = 100
     mock_disk_usage.return_value = 0, 0, 10
-    repositories = [Repository(url=repo_url)]
+    mount_path = Path(clone_dir)
+    repositories = [Repository.from_config_repo(ConfigRepo(url=repo_url), mount_path=mount_path)]
     cloner = GitCloner(
         repositories=repositories,
-        git_providers=[],
-        workspace_mount_path=clone_dir,
+        git_providers={},
+        mount_path=mount_path,
         user=test_user,
         lfs_auto_fetch=True,
     )
@@ -74,11 +77,12 @@ def test_lfs_size_check(test_user, clone_dir, mocker):
 )
 def test_lfs_output_parse(test_user, clone_dir, mocker, lfs_lfs_files_output, expected_output):
     repo_url = "https://github.com"
-    repositories = [Repository(url=repo_url)]
+    mount_path = Path(clone_dir)
+    repositories = [Repository.from_config_repo(ConfigRepo(url=repo_url), mount_path=mount_path)]
     cloner = GitCloner(
         repositories=repositories,
-        git_providers=[],
-        workspace_mount_path=clone_dir,
+        git_providers={},
+        mount_path=mount_path,
         user=test_user,
     )
 

renku_notebooks/api/amalthea_patches/init_containers.py

@@ -156,9 +156,7 @@ def certificates():
             }
         },
     )
-    volume_etc_certs = client.V1Volume(
-        name="etc-ssl-certs", empty_dir=client.V1EmptyDirVolumeSource(medium="Memory")
-    )
+    volume_etc_certs = client.V1Volume(name="etc-ssl-certs", empty_dir=client.V1EmptyDirVolumeSource(medium="Memory"))
     volume_custom_certs = client.V1Volume(
         name="custom-ca-certs",
         projected=client.V1ProjectedVolumeSource(