async def start_web_socket(ws: WebSocket):
while True:
try:
_ = await ws.receive_text()
except WebSocketDisconnect:
log_info(
f'{ws.scope["client"]} - "WebSocket {ws.scope["path"]}" [disconnected]'
)
break
async def catch_all_no_namespace(request: Request, rest_of_path: str):
headers = generate_headers_downstream(request.headers)
segments = rest_of_path.strip('/').split('/')
namespace, name, version, resource = get_info(segments) \
if segments[0].startswith('@') \
else get_info([""] + segments)
log_info("forward request to CDN:",
namespace=namespace,
name=name,
version=version,
resource=resource)
return await get_raw_resource(namespace, name, version, resource, headers)
async def init_resources(config: Configuration):
log_info("Ensure database resources")
headers = await config.admin_headers if config.admin_headers else {}
log_info("Successfully retrieved authorization for resources creation")
await config.storage.ensure_bucket(headers=headers)
log_info("resources initialization done")
async def init_resources(config: Configuration):
log_info("Ensure database resources")
headers = await config.admin_headers if config.admin_headers else {}
log_info("Successfully retrieved authorization for resources creation")
doc_db = config.data_client.docdb
storage = config.data_client.storage
table_ok, bucket_ok = await asyncio.gather(
doc_db.ensure_table(headers=headers),
storage.ensure_bucket(headers=headers))
if not bucket_ok or not table_ok:
raise Exception(
f"Problem during resources initialisation: table ok? {table_ok}; bucket ok? {bucket_ok}"
)
log_info("resources initialization done")
async def post_metadata(
request: Request,
project_id: str,
metadata_body: EditMetadata,
configuration: Configuration = Depends(get_configuration)):
headers = generate_headers_downstream(request.headers)
doc_db, storage, assets_gtw = configuration.doc_db, configuration.storage, configuration.assets_gtw_client
owner = configuration.default_owner
req, workflow, description = await asyncio.gather(
storage.get_json(path="projects/{}/requirements.json".format(project_id), owner=owner, headers=headers),
storage.get_json(path="projects/{}/workflow.json".format(project_id), owner=owner, headers=headers),
storage.get_json(path="projects/{}/description.json".format(project_id), owner=owner, headers=headers)
)
log_info("Flux-Backend@Post metadata: got requirements and workflow")
libraries = {**req['libraries'], **metadata_body.libraries}
def get_package_id(factory_id: Union[str, Mapping[str, str]]):
return "@youwol/" + factory_id.split("@")[1] if isinstance(factory_id, str) else factory_id['pack']
used_packages = {get_package_id(m["factoryId"]) for m in workflow["modules"] + workflow["plugins"]}
log_info("Flux-Backend@Post metadata: used_packages", used_packages=used_packages)
body = {
"libraries": {name: version for name, version in libraries.items() if name in used_packages},
"using": {name: version for name, version in libraries.items()}
}
loading_graph = await assets_gtw.cdn_loading_graph(body=body, headers=headers)
flux_packs = [p['name'] for p in loading_graph['lock'] if p['type'] == 'flux-pack']
log_info("Flux-Backend@Post metadata: got loading graph", loading_graph=loading_graph)
used_libraries = {lib["name"]: lib["version"] for lib in loading_graph["lock"]}
requirements = Requirements(fluxComponents=[], fluxPacks=flux_packs,
libraries=used_libraries, loadingGraph=loading_graph)
schema_version = description['schemaVersion'] if 'schemaVersion' else '0'
coroutines = update_metadata(project_id=project_id, schema_version=schema_version,
name=metadata_body.name, description=metadata_body.description,
requirements=requirements, owner=owner, storage=storage, docdb=doc_db, headers=headers)
await asyncio.gather(*coroutines)
return {}
async def format_download_form(file_path: Path, base_path: Path, dir_path: Path, compress: bool, rename: str = None) \
-> FormData:
if compress and get_content_encoding(file_path) == "br":
path_log = "/".join(file_path.parts[2:])
start = time.time()
if which('brotli'):
log_info(f'brotlify (system) {path_log} ...')
p = await asyncio.create_subprocess_shell(
cmd=f'brotli {str(file_path)}',
stdout=asyncio.subprocess.PIPE,
stderr=asyncio.subprocess.PIPE,
shell=True)
async for f in p.stderr:
log_error(f.decode('utf-8'))
await p.communicate()
os.system(f'rm {str(file_path)}')
os.system(f'mv {str(file_path)}.br {str(file_path)}')
else:
log_info(f'brotlify (python) {path_log}')
start = time.time()
compressed = brotli.compress(file_path.read_bytes())
with file_path.open("wb") as f:
f.write(compressed)
log_info(f'...{path_log} => {time.time() - start} s')
data = open(str(file_path), 'rb').read()
path_bucket = base_path / file_path.relative_to(
dir_path) if not rename else base_path / rename
return FormData(objectName=path_bucket,
objectData=data,
owner=Configuration.owner,
objectSize=len(data),
content_type=get_content_type(file_path.name),
content_encoding=get_content_encoding(file_path.name))
async def init_resources(
config: Configuration
):
log_info("Ensure database resources")
headers = await config.admin_headers if config.admin_headers else {}
log_info("Successfully retrieved authorization for resources creation")
doc_dbs = config.doc_dbs
log_info("Ensure items_db table")
items_ok = await doc_dbs.items_db.ensure_table(headers=headers)
log_info("Ensure folders_db table")
folders_ok = await doc_dbs.folders_db.ensure_table(headers=headers)
log_info("Ensure drives_db table")
drives_ok = await doc_dbs.drives_db.ensure_table(headers=headers)
log_info("Ensure deleted_db table")
deleted_ok = await doc_dbs.deleted_db.ensure_table(headers=headers)
if not (items_ok and folders_ok and drives_ok and deleted_ok):
raise Exception(f"Problem during doc-db's table initialisation {[items_ok, folders_ok, drives_ok, deleted_ok]}")
log_info("resources initialization done")
async def resolve_loading_tree(
request: Request,
body: LoadingGraphBody,
configuration: Configuration = Depends(get_configuration)):
doc_db = configuration.doc_db
headers = generate_headers_downstream(request.headers)
libraries = {name: version for name, version in body.libraries.items()}
log_info(f"Start resolving loading graph: {libraries}")
latest_queries = [
name for name, version in libraries.items() if version == "latest"
]
versions_resp = await asyncio.gather(*[
list_versions(request=request, name=name, configuration=configuration)
for name in latest_queries
],
return_exceptions=True)
if any(isinstance(v, Exception) for v in versions_resp):
packages_error = [
f"{name}#latest" for e, name in zip(versions_resp, latest_queries)
if isinstance(e, Exception)
]
raise PackagesNotFound(
detail="Failed to retrieved latest version of package(s)",
packages=packages_error)
latest_versions = {
name: resp.versions[0]
for name, resp in zip(latest_queries, versions_resp)
}
explicit_versions = {**libraries, **latest_versions}
log_info(f"Latest versions resolved",
latest_versions=latest_versions,
explicit_versions=explicit_versions)
queries = [
doc_db.get_document(partition_keys={"library_name": name},
clustering_keys={
"version_number":
get_version_number_str(version)
},
owner=configuration.owner,
headers=headers)
for name, version in explicit_versions.items()
]
dependencies = await asyncio.gather(*queries, return_exceptions=True)
if any(isinstance(v, Exception) for v in dependencies):
packages_error = [
f"{name}#{version}"
for e, (name,
version) in zip(dependencies, explicit_versions.items())
if isinstance(e, Exception)
]
raise PackagesNotFound(
detail="Failed to retrieved explicit version of package(s)",
packages=packages_error)
dependencies_dict = {d["library_name"]: d for d in dependencies}
async def add_missing_dependencies(missing_previous_loop=None):
""" It maybe the case where some dependencies are missing in the provided body,
here we fetch using 'body.using' or the latest version of them"""
flatten_dependencies = set(
flatten([[p.split("#")[0] for p in package['dependencies']]
for package in dependencies_dict.values()]))
missing = [
d for d in flatten_dependencies if d not in dependencies_dict
]
if not missing:
return dependencies_dict
if missing_previous_loop and missing == missing_previous_loop:
raise PackagesNotFound(
detail="Indirect dependencies not found in the CDN",
packages=missing)
def get_dependency(dependency):
if dependency in body.using:
return get_query_version(configuration.doc_db, dependency,
body.using[dependency], headers)
return get_query_latest(configuration.doc_db, dependency, headers)
versions = await asyncio.gather(
*[get_dependency(dependency) for dependency in missing],
return_exceptions=True)
if any(len(v["documents"]) == 0 for v in versions):
raise PackagesNotFound(
detail="Failed to retrieve a version of indirect dependencies",
packages=[
f"{name}#{body.using.get(name,'latest')}"
for v, name in zip(versions, missing)
if len(v["documents"]) == 0
])
versions = list(flatten([d['documents'] for d in versions]))
for version in versions:
lib_name = version["library_name"]
dependencies_dict[lib_name] = version
return await add_missing_dependencies(missing_previous_loop=missing)
await add_missing_dependencies()
items_dict = {
d["library_name"]: [to_package_id(d["library_name"]),
get_url(d)]
for d in dependencies_dict.values()
}
r = loading_graph([], dependencies_dict.values(), items_dict)
lock = [
Library(name=d["library_name"],
version=d["version"],
namespace=d["namespace"],
id=to_package_id(d["library_name"]),
type=d["type"],
fingerprint=d["fingerprint"])
for d in dependencies_dict.values()
]
return LoadingGraphResponseV1(graphType="sequential-v1",
lock=lock,
definition=r)
async def init_resources(config: Configuration):
log_info("Ensure database resources")
headers = await config.admin_headers if config.admin_headers else {}
log_info("Successfully retrieved authorization for resources creation")
log_info("Ensure assets table")
table1_ok = await config.doc_db_asset.ensure_table(headers=headers)
if not table1_ok:
raise Exception("Problem during docdb_asset resources initialisation")
log_info("Ensure assets bucket")
bucket_ok = await config.storage.ensure_bucket(headers=headers)
if not bucket_ok:
raise Exception("Problem during bucket initialisation")
log_info("Ensure access policy table")
table2_ok = await asyncio.gather(
config.doc_db_access_policy.ensure_table(headers=headers))
if not table2_ok:
raise Exception(
"Problem during docdb_access_policy resources initialisation")
log_info("Ensure access history table")
table3_ok = await asyncio.gather(
config.doc_db_access_history.ensure_table(headers=headers))
if not table3_ok:
raise Exception(
"Problem during docdb_access_history resources initialisation")
log_info("resources initialization done")
async def publish_package(file: IO, filename: str, content_encoding,
configuration, headers):
if content_encoding not in ['identity', 'brotli']:
raise HTTPException(
status_code=422,
detail="Only identity and brotli encoding are accepted ")
need_compression = content_encoding == 'identity'
dir_path = Path("./tmp_zips") / str(uuid4())
zip_path = (dir_path / filename).with_suffix('.zip')
os.makedirs(dir_path)
try:
log_info("extract .zip file...")
compressed_size = extract_zip_file(file,
zip_path,
dir_path,
delete_original=False)
log_info("...zip extracted", compressed_size=compressed_size)
package_path = next(
flatten([[Path(root) / f for f in files if f == "package.json"]
for root, _, files in os.walk(dir_path)]))
package_json = json.loads(open(package_path).read())
library_id = package_json["name"].replace("@", '')
version = package_json["version"]
base_path = Path('libraries') / library_id / version
storage = configuration.storage
paths = flatten([[Path(root) / f for f in files]
for root, _, files in os.walk(dir_path)])
paths = [p for p in paths if p != zip_path]
form_original = await format_download_form(zip_path, base_path,
package_path.parent,
need_compression,
'__original.zip')
forms = await asyncio.gather(*[
format_download_form(path, base_path, package_path.parent,
need_compression) for path in paths
])
forms = list(forms) + [form_original]
# the fingerprint in the md5 checksum of the included files after having eventually being compressed
os.remove(zip_path)
md5_stamp = md5_from_folder(dir_path)
post_requests = [
storage.post_object(path=form.objectName,
content=form.objectData,
content_type=form.content_type,
owner=Configuration.owner,
headers=headers) for form in forms
]
log_info(f"Clean directory {str(base_path)}")
await storage.delete_group(prefix=base_path,
owner=Configuration.owner,
headers=headers)
log_info(f"Send {len(post_requests)} files to storage")
await asyncio.gather(*post_requests)
record = format_doc_db_record(package_path=package_path,
fingerprint=md5_stamp)
log_info("Create docdb document", record=record)
await configuration.doc_db.create_document(record,
owner=Configuration.owner,
headers=headers)
log_info("Done", md5_stamp=md5_stamp)
return PublishResponse(
name=package_json["name"],
version=version,
compressedSize=compressed_size,
id=to_package_id(package_json["name"]),
fingerprint=md5_stamp,
url=
f"{to_package_id(package_json['name'])}/{record['version']}/{record['bundle']}"
)
finally:
shutil.rmtree(dir_path)
