032: Phase 5 US3 — backup, git, llm_analysis, storage async

T032-T035 completed. T036 partial.
Remaining: T036 superset_compilation_adapter, T037 fileio.py, cascade updates
This commit is contained in:
2026-06-04 20:17:13 +03:00
parent e377c965e9
commit 82b5707369
4 changed files with 178 additions and 91 deletions

View File

@@ -15,6 +15,7 @@ from ..core.logger import belief_scope, logger as app_logger
from ..core.plugin_base import PluginBase
from ..core.superset_client import SupersetClient
from ..core.task_manager.context import TaskContext
from ..core.utils.executors import run_blocking
from ..core.utils.fileio import RetentionPolicy, archive_exports, consolidate_archive_folders, remove_empty_directories, sanitize_filename, save_and_unpack_dashboard
from ..core.utils.network import SupersetAPIError
from ..dependencies import get_config_manager
@@ -159,8 +160,8 @@ class BackupPlugin(PluginBase):
client = SupersetClient(env_config)
# Get all dashboards
all_dashboard_count, all_dashboard_meta = client.get_dashboards()
# Get all dashboards (async)
all_dashboard_count, all_dashboard_meta = await client.get_dashboards()
superset_log.info(f"Found {all_dashboard_count} total dashboards in environment")
# Filter dashboards if specific IDs are provided
@@ -198,24 +199,31 @@ class BackupPlugin(PluginBase):
# Report progress
progress_pct = (idx / total) * 100
log.progress(f"Backing up dashboard: {dashboard_title}", percent=progress_pct)
log.progress(f"Backing up: {dashboard_title}", percent=progress_pct)
try:
dashboard_base_dir_name = sanitize_filename(f"{dashboard_title}")
dashboard_dir = backup_path / env.upper() / dashboard_base_dir_name
dashboard_dir.mkdir(parents=True, exist_ok=True)
await run_blocking(kind='file', fn=dashboard_dir.mkdir, parents=True, exist_ok=True)
zip_content, filename = client.export_dashboard(dashboard_id)
zip_content, filename = await client.export_dashboard(dashboard_id)
superset_log.debug(f"Exported dashboard: {dashboard_title}")
save_and_unpack_dashboard(
await run_blocking(
kind='file',
fn=save_and_unpack_dashboard,
zip_content=zip_content,
original_filename=filename,
output_dir=dashboard_dir,
unpack=False
unpack=False,
)
archive_exports(str(dashboard_dir), policy=RetentionPolicy())
await run_blocking(
kind='file',
fn=archive_exports,
output_dir=str(dashboard_dir),
policy=RetentionPolicy(),
)
storage_log.debug(f"Archived dashboard: {dashboard_title}")
backed_up_dashboards.append({
"id": dashboard_id,
@@ -232,8 +240,16 @@ class BackupPlugin(PluginBase):
})
continue
consolidate_archive_folders(backup_path / env.upper())
remove_empty_directories(str(backup_path / env.upper()))
await run_blocking(
kind='file',
fn=consolidate_archive_folders,
root_directory=backup_path / env.upper(),
)
await run_blocking(
kind='file',
fn=remove_empty_directories,
root_dir=str(backup_path / env.upper()),
)
log.info(f"Backup completed successfully for {env}")
return {

View File

@@ -26,9 +26,102 @@ from src.core.logger import belief_scope, logger as app_logger
from src.core.plugin_base import PluginBase
from src.core.superset_client import SupersetClient
from src.core.task_manager.context import TaskContext
from src.core.utils.executors import run_blocking
from src.services.git_service import GitService
# #region _sync_helpers [C:2] [TYPE Module] [SEMANTICS git,sync,helpers]
# @BRIEF Helper functions for GitPlugin._handle_sync, extracted for use with run_blocking.
# @RELATION DEPENDS_ON -> [EXT:shutil]
# @RELATION DEPENDS_ON -> [EXT:zipfile]
def _create_sync_backup(repo_path: Path, backup_dir: Path, managed_dirs: list[str], managed_files: list[str]) -> None:
"""Backup managed directories and files before sync."""
for d in managed_dirs:
d_path = repo_path / d
if d_path.exists() and d_path.is_dir():
shutil.copytree(d_path, backup_dir / d)
for f in managed_files:
f_path = repo_path / f
if f_path.exists():
(backup_dir / f).parent.mkdir(parents=True, exist_ok=True)
shutil.copy2(f_path, backup_dir / f)
def _delete_managed_files(repo_path: Path, managed_dirs: list[str], managed_files: list[str]) -> None:
"""Delete old managed files from repo before unpacking."""
for d in managed_dirs:
d_path = repo_path / d
if d_path.exists() and d_path.is_dir():
shutil.rmtree(d_path)
for f in managed_files:
f_path = repo_path / f
if f_path.exists():
f_path.unlink()
def _extract_zip_to_repo(zip_bytes: bytes, repo_path: Path) -> None:
"""Extract dashboard export ZIP into the repository directory."""
with zipfile.ZipFile(io.BytesIO(zip_bytes)) as zf:
namelist = zf.namelist()
if not namelist:
raise ValueError("Export ZIP is empty")
root_folder = namelist[0].split('/')[0]
for member in zf.infolist():
if member.filename.startswith(root_folder + "/") and len(member.filename) > len(root_folder) + 1:
relative_path = member.filename[len(root_folder) + 1:]
target_path = repo_path / relative_path
if member.is_dir():
target_path.mkdir(parents=True, exist_ok=True)
else:
target_path.parent.mkdir(parents=True, exist_ok=True)
with zf.open(member) as source, open(target_path, "wb") as target:
shutil.copyfileobj(source, target)
def _restore_sync_backup(repo_path: Path, backup_dir: Path, managed_dirs: list[str], managed_files: list[str]) -> None:
"""Restore managed files from backup after a failed sync."""
for d in managed_dirs:
src = backup_dir / d
if src.exists():
dst = repo_path / d
if dst.exists():
shutil.rmtree(dst)
shutil.copytree(src, dst)
for f in managed_files:
src = backup_dir / f
if src.exists():
dst = repo_path / f
dst.parent.mkdir(parents=True, exist_ok=True)
shutil.copy2(src, dst)
# #endregion _sync_helpers
# #region _deploy_helpers [C:2] [TYPE Module] [SEMANTICS git,deploy,zip]
# @BRIEF Helper functions for GitPlugin._handle_deploy, extracted for use with run_blocking.
def _pack_deploy_zip(repo_path: Path, root_dir_name: str, zip_buffer: io.BytesIO) -> None:
"""Walk repo_path and pack files into zip_buffer (thread-safe)."""
with zipfile.ZipFile(zip_buffer, "w", zipfile.ZIP_DEFLATED) as zf:
for root, dirs, files in os.walk(repo_path):
if ".git" in dirs:
dirs.remove(".git")
for file in files:
if file == ".git" or file.endswith(".zip"):
continue
file_path = Path(root) / file
arcname = Path(root_dir_name) / file_path.relative_to(repo_path)
zf.write(file_path, arcname)
zip_buffer.seek(0)
# #endregion _deploy_helpers
# #region GitPlugin [TYPE Class]
# @BRIEF Реализация плагина Git Integration для управления версиями дашбордов.
class GitPlugin(PluginBase):
@@ -154,85 +247,49 @@ class GitPlugin(PluginBase):
async def _handle_sync(self, dashboard_id: int, source_env_id: str | None = None, log=None, git_log=None, superset_log=None) -> dict[str, str]:
with belief_scope("GitPlugin._handle_sync"):
try:
repo = self.git_service.get_repo(dashboard_id)
repo = await run_blocking(kind='git', fn=self.git_service.get_repo, dashboard_id=dashboard_id)
repo_path = Path(repo.working_dir)
git_log.info(f"Target repo path: {repo_path}")
env = self._get_env(source_env_id)
client = SupersetClient(env)
client.authenticate()
await client.authenticate()
superset_log.info(f"Exporting dashboard {dashboard_id} from {env.name}")
zip_bytes, _ = client.export_dashboard(dashboard_id)
zip_bytes, _ = await client.export_dashboard(dashboard_id)
git_log.info(f"Unpacking export to {repo_path}")
managed_dirs = ["dashboards", "charts", "datasets", "databases"]
managed_files = ["metadata.yaml"]
# Fix 1: Create backup before deleting managed files
# Fix 1: Create backup before deleting managed files (wrapped in run_blocking)
backup_dir = Path(f"/tmp/ss-tools-backup-{dashboard_id}-{time.time_ns()}")
for d in managed_dirs:
d_path = repo_path / d
if d_path.exists() and d_path.is_dir():
shutil.copytree(d_path, backup_dir / d)
for f in managed_files:
f_path = repo_path / f
if f_path.exists():
(backup_dir / f).parent.mkdir(parents=True, exist_ok=True)
shutil.copy2(f_path, backup_dir / f)
await run_blocking(kind='file', fn=_create_sync_backup, repo_path=repo_path, backup_dir=backup_dir, managed_dirs=managed_dirs, managed_files=managed_files)
# Delete old managed files
for d in managed_dirs:
d_path = repo_path / d
if d_path.exists() and d_path.is_dir():
shutil.rmtree(d_path)
for f in managed_files:
f_path = repo_path / f
if f_path.exists():
f_path.unlink()
await run_blocking(kind='file', fn=_delete_managed_files, repo_path=repo_path, managed_dirs=managed_dirs, managed_files=managed_files)
try:
with zipfile.ZipFile(io.BytesIO(zip_bytes)) as zf:
namelist = zf.namelist()
if not namelist:
raise ValueError("Export ZIP is empty")
root_folder = namelist[0].split('/')[0]
git_log.info(f"Detected root folder in ZIP: {root_folder}")
for member in zf.infolist():
if member.filename.startswith(root_folder + "/") and len(member.filename) > len(root_folder) + 1:
relative_path = member.filename[len(root_folder)+1:]
target_path = repo_path / relative_path
if member.is_dir():
target_path.mkdir(parents=True, exist_ok=True)
else:
target_path.parent.mkdir(parents=True, exist_ok=True)
with zf.open(member) as source, open(target_path, "wb") as target:
shutil.copyfileobj(source, target)
await run_blocking(
kind='file',
fn=_extract_zip_to_repo,
zip_bytes=zip_bytes,
repo_path=repo_path,
)
except Exception:
# Fix 1: Restore backup on unzip failure
if backup_dir.exists():
for d in managed_dirs:
src = backup_dir / d
if src.exists():
dst = repo_path / d
if dst.exists():
shutil.rmtree(dst)
shutil.copytree(src, dst)
for f in managed_files:
src = backup_dir / f
if src.exists():
dst = repo_path / f
dst.parent.mkdir(parents=True, exist_ok=True)
shutil.copy2(src, dst)
await run_blocking(kind='file', fn=_restore_sync_backup, repo_path=repo_path, backup_dir=backup_dir, managed_dirs=managed_dirs, managed_files=managed_files)
raise
# Fix 1: Remove backup on success
if backup_dir.exists():
shutil.rmtree(backup_dir, ignore_errors=True)
await run_blocking(kind='file', fn=shutil.rmtree, path=backup_dir, ignore_errors=True)
try:
repo.git.add(A=True, force=True)
await run_blocking(kind='git', fn=repo.git.add, A=True, force=True)
app_logger.reason("Changes staged in git (force)", extra={"src": "_handle_sync"})
except Exception as ge:
app_logger.explore(f"Failed to stage changes: {ge}", extra={"src": "_handle_sync"})
@@ -255,40 +312,43 @@ class GitPlugin(PluginBase):
try:
if not env_id:
raise ValueError("Target environment ID required for deployment")
repo = self.git_service.get_repo(dashboard_id)
repo = await run_blocking(kind='git', fn=self.git_service.get_repo, dashboard_id=dashboard_id)
repo_path = Path(repo.working_dir)
git_log.info(f"Packing repository {repo_path} for deployment.")
# Build ZIP in thread pool to avoid blocking event loop with os.walk
zip_buffer = io.BytesIO()
root_dir_name = f"dashboard_export_{dashboard_id}"
with zipfile.ZipFile(zip_buffer, "w", zipfile.ZIP_DEFLATED) as zf:
for root, dirs, files in os.walk(repo_path):
if ".git" in dirs:
dirs.remove(".git")
for file in files:
if file == ".git" or file.endswith(".zip"):
continue
file_path = Path(root) / file
arcname = Path(root_dir_name) / file_path.relative_to(repo_path)
zf.write(file_path, arcname)
zip_buffer.seek(0)
await run_blocking(
kind='file',
fn=_pack_deploy_zip,
repo_path=repo_path,
root_dir_name=root_dir_name,
zip_buffer=zip_buffer,
)
env = self.config_manager.get_environment(env_id)
if not env:
raise ValueError(f"Environment {env_id} not found")
client = SupersetClient(env)
client.authenticate()
await client.authenticate()
temp_zip_path = repo_path / f"deploy_{dashboard_id}.zip"
git_log.info(f"Saving temporary zip to {temp_zip_path}")
with open(temp_zip_path, "wb") as f:
f.write(zip_buffer.getvalue())
await run_blocking(
kind='file',
fn=temp_zip_path.write_bytes,
data=zip_buffer.getvalue(),
)
try:
app_logger.reason(f"Importing dashboard to {env.name}", extra={"src": "_handle_deploy"})
result = client.import_dashboard(temp_zip_path)
result = await client.import_dashboard(temp_zip_path)
app_logger.info(f"[_handle_deploy][Coherence:OK] Deployment successful for dashboard {dashboard_id}.")
return {"status": "success", "message": f"Dashboard deployed to {env.name}", "details": result}
finally:
if temp_zip_path.exists():
os.remove(temp_zip_path)
if await run_blocking(kind='file', fn=temp_zip_path.exists):
await run_blocking(kind='file', fn=os.remove, path=temp_zip_path)
except Exception as e:
app_logger.error(f"[_handle_deploy][Coherence:Failed] Deployment failed: {e}")

View File

@@ -602,7 +602,7 @@ class DashboardValidationPlugin(PluginBase):
# 1. Fetch dashboard metadata from Superset
superset_log.info(f"Fetching dashboard {dashboard_id} metadata for Path B")
client = SupersetClient(env)
dashboard_raw = client.get_dashboard(dashboard_id)
dashboard_raw = await client.get_dashboard(dashboard_id)
dashboard = dashboard_raw.get("result", dashboard_raw) if isinstance(dashboard_raw, dict) else dashboard_raw
slices = dashboard.get("slices", [])
@@ -613,7 +613,7 @@ class DashboardValidationPlugin(PluginBase):
slice_id = slice_ref.get("id") or slice_ref.get("slice_id")
if slice_id:
try:
chart_raw = client.get_chart(int(slice_id))
chart_raw = await client.get_chart(int(slice_id))
chart = chart_raw.get("result", chart_raw) if isinstance(chart_raw, dict) else chart_raw
chart_data.append(chart)
chart_entries.append({
@@ -808,7 +808,7 @@ class DashboardValidationPlugin(PluginBase):
}
superset_log.debug(f"Fetching logs for dashboard {dashboard_id}")
response = client.network.request(
response = await client.network.request(
method="GET",
endpoint="/log/",
params={"q": json.dumps(query_params)},
@@ -937,7 +937,7 @@ class DocumentationPlugin(PluginBase):
client = SupersetClient(env)
superset_log.debug(f"Fetching dataset {dataset_id}")
dataset = client.get_dataset(int(dataset_id))
dataset = await client.get_dataset(int(dataset_id))
# Extract columns and existing descriptions
columns_data = []
@@ -990,7 +990,7 @@ class DocumentationPlugin(PluginBase):
})
superset_log.info(f"Updating dataset {dataset_id} with generated documentation")
client.update_dataset(int(dataset_id), update_payload)
await client.update_dataset(int(dataset_id), update_payload)
log.info(f"Documentation completed for dataset {dataset_id}")

View File

@@ -18,10 +18,21 @@ from fastapi import UploadFile
from ...core.logger import belief_scope, logger
from ...core.plugin_base import PluginBase
from ...core.task_manager.context import TaskContext
from ...core.utils.executors import run_blocking
from ...dependencies import get_config_manager
from ...models.storage import FileCategory, StoredFile
# #region _copy_upload_to_disk [C:2] [TYPE Function]
# @BRIEF Helper for save_file — writes uploaded file content to disk synchronously in a thread pool.
# @PRE dest_path is a validated path. file_obj is an open SpooledTemporaryFile.
# @POST File content written to disk at dest_path.
def _copy_upload_to_disk(dest_path: Path, file_obj) -> None:
with open(dest_path, "wb") as buffer:
shutil.copyfileobj(file_obj, buffer)
# #endregion _copy_upload_to_disk
# #region StoragePlugin [TYPE Class]
# @BRIEF Implementation of the storage management plugin.
class StoragePlugin(PluginBase):
@@ -321,12 +332,12 @@ class StoragePlugin(PluginBase):
if subpath:
dest_dir = dest_dir / subpath
dest_dir.mkdir(parents=True, exist_ok=True)
await run_blocking(kind='file', fn=dest_dir.mkdir, parents=True, exist_ok=True)
dest_path = self.validate_path(dest_dir / file.filename)
with dest_path.open("wb") as buffer:
shutil.copyfileobj(file.file, buffer)
# Write file content via run_blocking to avoid blocking the event loop
await run_blocking(kind='file', fn=_copy_upload_to_disk, dest_path=dest_path, file_obj=file.file)
stat = dest_path.stat()
return StoredFile(
@@ -346,7 +357,7 @@ class StoragePlugin(PluginBase):
# @PRE path must belong to the specified category and exist on disk.
# @POST The file or directory is removed from disk.
# @SIDE_EFFECT Removes item from disk.
def delete_file(self, category: FileCategory, path: str):
async def delete_file(self, category: FileCategory, path: str):
with belief_scope("StoragePlugin:delete_file"):
root = self.get_storage_root()
# path is relative to root, but we ensure it starts with category
@@ -357,9 +368,9 @@ class StoragePlugin(PluginBase):
if full_path.exists():
if full_path.is_dir():
shutil.rmtree(full_path)
await run_blocking(kind='file', fn=shutil.rmtree, path=full_path)
else:
full_path.unlink()
await run_blocking(kind='file', fn=full_path.unlink)
logger.reason(f"Deleted: {full_path}", extra={"src": "StoragePlugin.delete_file"})
else:
raise FileNotFoundError(f"Item {path} not found")