feat: cleanup git repository before doing project backup

weblate/trans/backups.py

@@ -278,6 +278,12 @@
         # Store VCS repo in case it is present
         if component.is_repo_link:
             return
+
+        # Compact the repository
+        with component.repository.lock:
+            component.repository.compact()
+
+        # Actually perform the backup
         self.backup_dir(
             backupzip,
             component.full_path,
@@ -316,7 +322,7 @@
             for component in project.component_set.iterator():
                 self.backup_component(backupzip, component)
 
         os.rename(part_name, self.filename)
 
     def list_components(self, zipfile):
         return [
@@ -349,7 +355,7 @@
             data = json.load(handle)
             validate_schema(data, "weblate-component.schema.json")
             if skip_linked and data["component"]["repo"].startswith("weblate:"):
                 return False
             if data["component"]["vcs"] not in VCS_REGISTRY:
                 msg = f"Component {data['component']['name']} uses unsupported VCS: {data['component']['vcs']}"
                 raise ValueError(msg)
@@ -366,12 +372,12 @@
 
             if callback is not None:
                 callback(zipfile, data)
             return True
 
     def load_components(self, zipfile, callback: Callable | None = None) -> None:
         pending: list[str] = []
         for component in self.list_components(zipfile):
             processed = self.load_component(
                 zipfile, component, callback, skip_linked=True
             )
             if not processed:

weblate/vcs/base.py

@@ -138,7 +138,7 @@
         """Ensure the configuration is periodically checked."""
         if self._config_updated:
             return
         cache_key = f"sp-config-check-{self.component.pk}"
         if cache.get(cache_key) is None:
             self.check_config()
             cache.set(cache_key, True, 86400)
@@ -217,7 +217,7 @@
         except subprocess.TimeoutExpired as error:
             raise RepositoryError(
                 0,
                 f"Subprocess didn't complete before {error.timeout} seconds\n{error.stdout}{error.stderr or ''}",
             ) from error
         cls.add_breadcrumb(
             text_cmd,
@@ -562,3 +562,6 @@
 
     def list_remote_branches(self):
         return []
+
+    def compact(self) -> None:
+        return

weblate/vcs/git.py

@@ -566,6 +566,9 @@
 
         return "\n".join(result)
 
+    def compact(self) -> None:
+        self.execute(["gc"])
+
 
 class GitWithGerritRepository(GitRepository):
     name = "Gerrit"
@@ -820,7 +823,7 @@
         self, repo: str | None = None
     ) -> tuple[str | None, str | None, str | None, str, str, str]:
         if repo is None:
             repo = self.component.push or self.component.repo
         parsed = urllib.parse.urlparse(repo)
         host = parsed.hostname
         scheme: str | None = parsed.scheme
@@ -1062,7 +1065,7 @@
 
     def get_merge_message(self):
         lines = render_template(
             self.component.pull_message.strip(), component=self.component
         ).splitlines()
         return lines[0], "\n".join(lines[1:]).strip()
 
@@ -1276,7 +1279,7 @@
         self, repo: str | None = None
     ) -> tuple[str | None, str | None, str | None, str, str, str]:
         if repo is None:
             repo = self.component.repo
 
         scheme_regex = r"^[a-z]+:\/\/.*"  # matches for example ssh://* and https://*