diff --git a/swh/deposit/api/common.py b/swh/deposit/api/common.py index b2a24959..432b64ae 100644 --- a/swh/deposit/api/common.py +++ b/swh/deposit/api/common.py @@ -1,1288 +1,1296 @@ # Copyright (C) 2017-2022 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information from abc import ABCMeta, abstractmethod import datetime import hashlib import json from typing import Any, Dict, Optional, Sequence, Tuple, Type, Union import uuid from xml.etree import ElementTree import attr from django.core.files.uploadedfile import UploadedFile from django.http import FileResponse, HttpResponse from django.shortcuts import render from django.template.loader import render_to_string from django.urls import reverse from django.utils import timezone from rest_framework import status from rest_framework.authentication import BaseAuthentication, BasicAuthentication from rest_framework.permissions import BasePermission, IsAuthenticated from rest_framework.request import Request from rest_framework.views import APIView from swh.deposit.api.checks import check_metadata, check_url_match_provider from swh.deposit.api.converters import convert_status_detail from swh.deposit.auth import HasDepositPermission, KeycloakBasicAuthentication from swh.deposit.models import DEPOSIT_METADATA_ONLY, Deposit from swh.deposit.parsers import parse_xml from swh.deposit.utils import ( NAMESPACES, compute_metadata_context, parse_swh_metadata_provenance, ) from swh.model import hashutil from swh.model.model import ( MetadataAuthority, MetadataAuthorityType, Origin, RawExtrinsicMetadata, ) from swh.model.swhids import ( ExtendedObjectType, ExtendedSWHID, QualifiedSWHID, ValidationError, ) from swh.scheduler.utils import create_oneshot_task_dict from ..config import ( ARCHIVE_KEY, ARCHIVE_TYPE, CONT_FILE_IRI, DEPOSIT_STATUS_DEPOSITED, DEPOSIT_STATUS_LOAD_SUCCESS, DEPOSIT_STATUS_PARTIAL, EDIT_IRI, EM_IRI, METADATA_TYPE, RAW_METADATA_KEY, SE_IRI, STATE_IRI, APIConfig, ) from ..errors import ( BAD_REQUEST, CHECKSUM_MISMATCH, ERROR_CONTENT, FORBIDDEN, MAX_UPLOAD_SIZE_EXCEEDED, MEDIATION_NOT_ALLOWED, METHOD_NOT_ALLOWED, NOT_FOUND, PARSING_ERROR, DepositError, ParserError, ) from ..models import DepositClient, DepositCollection, DepositRequest from ..utils import ( extended_swhid_from_qualified, parse_swh_deposit_origin, parse_swh_reference, ) ACCEPT_PACKAGINGS = ["http://purl.org/net/sword/package/SimpleZip"] ACCEPT_ARCHIVE_CONTENT_TYPES = ["application/zip", "application/x-tar"] @attr.s class ParsedRequestHeaders: content_type = attr.ib(type=str) content_length = attr.ib(type=Optional[int]) in_progress = attr.ib(type=bool) content_disposition = attr.ib(type=Optional[str]) content_md5sum = attr.ib(type=Optional[bytes]) packaging = attr.ib(type=Optional[str]) slug = attr.ib(type=Optional[str]) on_behalf_of = attr.ib(type=Optional[str]) metadata_relevant = attr.ib(type=Optional[str]) swhid = attr.ib(type=Optional[str]) @attr.s class Receipt: """Data computed while handling the request body that will be served in the Deposit Receipt.""" deposit_id = attr.ib(type=int) deposit_date = attr.ib(type=datetime.datetime) status = attr.ib(type=str) archive = attr.ib(type=Optional[str]) def _compute_md5(filehandler: UploadedFile) -> bytes: h = hashlib.md5() for chunk in filehandler: h.update(chunk) return h.digest() def get_deposit_by_id( deposit_id: int, collection_name: Optional[str] = None ) -> Deposit: """Gets an existing Deposit object if it exists, or raises `DepositError`. If `collection` is not None, also checks the deposit belongs to the collection.""" try: deposit = Deposit.objects.get(pk=deposit_id) except Deposit.DoesNotExist: raise DepositError(NOT_FOUND, f"Deposit {deposit_id} does not exist") if collection_name and deposit.collection.name != collection_name: get_collection_by_name(collection_name) # raises if does not exist raise DepositError( NOT_FOUND, f"Deposit {deposit_id} does not belong to collection {collection_name}", ) return deposit def get_collection_by_name(collection_name: str): """Gets an existing Deposit object if it exists, or raises `DepositError`.""" try: collection = DepositCollection.objects.get(name=collection_name) except DepositCollection.DoesNotExist: raise DepositError(NOT_FOUND, f"Unknown collection name {collection_name}") assert collection is not None return collection def guess_deposit_origin_url(deposit: Deposit): """Guesses an origin url for the given deposit.""" external_id = deposit.external_id if not external_id: # The client provided neither an origin_url nor a slug. That's inconvenient, # but SWORD requires we support it. So let's generate a random slug. external_id = str(uuid.uuid4()) return "%s/%s" % (deposit.client.provider_url.rstrip("/"), external_id) class APIBase(APIConfig, APIView, metaclass=ABCMeta): """Base deposit request class sharing multiple common behaviors.""" _client: Optional[DepositClient] = None def __init__(self): super().__init__() auth_provider = self.config.get("authentication_provider") if auth_provider == "basic": self.authentication_classes: Sequence[Type[BaseAuthentication]] = ( BasicAuthentication, ) self.permission_classes: Sequence[Type[BasePermission]] = (IsAuthenticated,) elif auth_provider == "keycloak": self.authentication_classes: Sequence[Type[BaseAuthentication]] = ( KeycloakBasicAuthentication, ) self.permission_classes: Sequence[Type[BasePermission]] = ( IsAuthenticated, HasDepositPermission, ) else: raise ValueError( "Configuration key 'authentication_provider' should be provided with" f"either 'basic' or 'keycloak' value not {auth_provider!r}." ) def _read_headers(self, request: Request) -> ParsedRequestHeaders: """Read and unify the necessary headers from the request (those are not stored in the same location or not properly formatted). Args: request: Input request Returns: Dictionary with the following keys (some associated values may be None): - content-type - content-length - in-progress - content-disposition - packaging - slug - on-behalf-of """ meta = request._request.META content_length = meta.get("CONTENT_LENGTH") if content_length and isinstance(content_length, str): content_length = int(content_length) # final deposit if not provided in_progress = meta.get("HTTP_IN_PROGRESS", False) if isinstance(in_progress, str): in_progress = in_progress.lower() == "true" content_md5sum = meta.get("HTTP_CONTENT_MD5") if content_md5sum: content_md5sum = bytes.fromhex(content_md5sum) return ParsedRequestHeaders( content_type=request.content_type, content_length=content_length, in_progress=in_progress, content_disposition=meta.get("HTTP_CONTENT_DISPOSITION"), content_md5sum=content_md5sum, packaging=meta.get("HTTP_PACKAGING"), slug=meta.get("HTTP_SLUG"), on_behalf_of=meta.get("HTTP_ON_BEHALF_OF"), metadata_relevant=meta.get("HTTP_METADATA_RELEVANT"), swhid=meta.get("HTTP_X_CHECK_SWHID"), ) def _deposit_put(self, deposit: Deposit, in_progress: bool = False) -> None: """Save/Update a deposit in db. Args: deposit: deposit being updated/created in_progress: deposit status """ if in_progress is False: self._complete_deposit(deposit) else: deposit.status = DEPOSIT_STATUS_PARTIAL deposit.save() def _complete_deposit(self, deposit: Deposit) -> None: """Marks the deposit as 'deposited', then schedule a check task if configured to do so.""" deposit.complete_date = timezone.now() deposit.status = DEPOSIT_STATUS_DEPOSITED deposit.save() if not deposit.origin_url: deposit.origin_url = guess_deposit_origin_url(deposit) if self.config["checks"]: scheduler = self.scheduler if deposit.status == DEPOSIT_STATUS_DEPOSITED and not deposit.check_task_id: task = create_oneshot_task_dict( "check-deposit", collection=deposit.collection.name, deposit_id=deposit.id, retries_left=3, ) check_task_id = scheduler.create_tasks([task])[0]["id"] deposit.check_task_id = check_task_id deposit.save() def _deposit_request_put( self, deposit: Deposit, deposit_request_data: Dict[str, Any], replace_metadata: bool = False, replace_archives: bool = False, ) -> DepositRequest: """Save a deposit request with metadata attached to a deposit. Args: deposit: The deposit concerned by the request deposit_request_data: The dictionary with at most 2 deposit request types (archive, metadata) to associate to the deposit replace_metadata: Flag defining if we add or update existing metadata to the deposit replace_archives: Flag defining if we add or update archives to existing deposit Returns: the DepositRequest object stored in the backend """ if replace_metadata: DepositRequest.objects.filter(deposit=deposit, type=METADATA_TYPE).delete() if replace_archives: DepositRequest.objects.filter(deposit=deposit, type=ARCHIVE_TYPE).delete() deposit_request = None archive_file = deposit_request_data.get(ARCHIVE_KEY) if archive_file: deposit_request = DepositRequest( type=ARCHIVE_TYPE, deposit=deposit, archive=archive_file ) deposit_request.save() raw_metadata = deposit_request_data.get(RAW_METADATA_KEY) if raw_metadata: deposit_request = DepositRequest( type=METADATA_TYPE, deposit=deposit, raw_metadata=raw_metadata.decode("utf-8"), ) deposit_request.save() assert deposit_request is not None return deposit_request def _delete_archives(self, collection_name: str, deposit: Deposit) -> Dict: """Delete archive references from the deposit id.""" DepositRequest.objects.filter(deposit=deposit, type=ARCHIVE_TYPE).delete() return {} def _delete_deposit(self, collection_name: str, deposit: Deposit) -> Dict: """Delete deposit reference. Args: collection_name: Client's collection deposit: The deposit to delete Returns Empty dict when ok. Dict with error key to describe the failure. """ if deposit.collection.name != collection_name: summary = "Cannot delete a deposit from another collection" description = "Deposit %s does not belong to the collection %s" % ( deposit.id, collection_name, ) raise DepositError( BAD_REQUEST, summary=summary, verbose_description=description ) DepositRequest.objects.filter(deposit=deposit).delete() deposit.delete() return {} def _check_file_length( self, filehandler: UploadedFile, content_length: Optional[int] = None, ) -> None: """Check the filehandler passed as argument has exactly the expected content_length Args: filehandler: The file to check content_length: the expected length if provided. Raises: DepositError if the actual length does not match """ max_upload_size = self.config["max_upload_size"] if content_length: length = filehandler.size if length != content_length: raise DepositError(status.HTTP_412_PRECONDITION_FAILED, "Wrong length") if filehandler.size > max_upload_size: raise DepositError( MAX_UPLOAD_SIZE_EXCEEDED, f"Upload size limit exceeded (max {max_upload_size} bytes)." "Please consider sending the archive in multiple steps.", ) def _check_file_md5sum( self, filehandler: UploadedFile, md5sum: Optional[bytes], ) -> None: """Check the filehandler passed as argument has the expected md5sum Args: filehandler: The file to check md5sum: md5 hash expected from the file's content Raises: DepositError if the md5sum does not match """ if md5sum: _md5sum = _compute_md5(filehandler) if _md5sum != md5sum: raise DepositError( CHECKSUM_MISMATCH, "Wrong md5 hash", f"The checksum sent {hashutil.hash_to_hex(md5sum)} and the actual " f"checksum {hashutil.hash_to_hex(_md5sum)} does not match.", ) def _binary_upload( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Deposit, replace_metadata: bool = False, replace_archives: bool = False, ) -> Receipt: """Binary upload routine. Other than such a request, a 415 response is returned. Args: request: the request holding information to parse and inject in db headers: parsed request headers collection_name: the associated client deposit: deposit to be updated replace_metadata: 'Update or add' request to existing deposit. If False (default), this adds new metadata request to existing ones. Otherwise, this will replace existing metadata. replace_archives: 'Update or add' request to existing deposit. If False (default), this adds new archive request to existing ones. Otherwise, this will replace existing archives. ones. Raises: - 400 (bad request) if the request is not providing an external identifier - 413 (request entity too large) if the length of the archive exceeds the max size configured - 412 (precondition failed) if the length or md5 hash provided mismatch the reality of the archive - 415 (unsupported media type) if a wrong media type is provided """ content_length = headers.content_length if not content_length: raise DepositError( BAD_REQUEST, "CONTENT_LENGTH header is mandatory", "For archive deposit, the CONTENT_LENGTH header must be sent.", ) content_disposition = headers.content_disposition if not content_disposition: raise DepositError( BAD_REQUEST, "CONTENT_DISPOSITION header is mandatory", "For archive deposit, the CONTENT_DISPOSITION header must be sent.", ) packaging = headers.packaging if packaging and packaging not in ACCEPT_PACKAGINGS: raise DepositError( BAD_REQUEST, f"Only packaging {ACCEPT_PACKAGINGS} is supported", f"The packaging provided {packaging} is not supported", ) filehandler = request.FILES["file"] assert isinstance(filehandler, UploadedFile), filehandler self._check_file_length(filehandler, content_length) self._check_file_md5sum(filehandler, headers.content_md5sum) # actual storage of data archive_metadata = filehandler self._deposit_put( deposit=deposit, in_progress=headers.in_progress, ) self._deposit_request_put( deposit, {ARCHIVE_KEY: archive_metadata}, replace_metadata=replace_metadata, replace_archives=replace_archives, ) return Receipt( deposit_id=deposit.id, deposit_date=deposit.reception_date, status=deposit.status, archive=filehandler.name, ) def _read_metadata(self, metadata_stream) -> Tuple[bytes, ElementTree.Element]: """ Given a metadata stream, reads the metadata and returns the metadata in three forms: * verbatim (as raw bytes), for archival in long-term storage * parsed as a Python dict, for archival in postgresql's jsonb type * parsed as ElementTree, to extract information immediately """ raw_metadata = metadata_stream.read() metadata_tree = parse_xml(raw_metadata) return raw_metadata, metadata_tree def _multipart_upload( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Deposit, replace_metadata: bool = False, replace_archives: bool = False, ) -> Receipt: """Multipart upload supported with exactly: - 1 archive (zip) - 1 atom entry Other than such a request, a 415 response is returned. Args: request: the request holding information to parse and inject in db headers: parsed request headers collection_name: the associated client deposit: deposit to be updated replace_metadata: 'Update or add' request to existing deposit. If False (default), this adds new metadata request to existing ones. Otherwise, this will replace existing metadata. replace_archives: 'Update or add' request to existing deposit. If False (default), this adds new archive request to existing ones. Otherwise, this will replace existing archives. ones. Raises: - 400 (bad request) if the request is not providing an external identifier - 412 (precondition failed) if the potentially md5 hash provided mismatch the reality of the archive - 413 (request entity too large) if the length of the archive exceeds the max size configured - 415 (unsupported media type) if a wrong media type is provided """ content_types_present = set() data: Dict[str, Optional[Any]] = { "application/zip": None, # expected either zip "application/x-tar": None, # or x-tar "application/atom+xml": None, } for key, value in request.FILES.items(): fh = value content_type = fh.content_type if content_type in content_types_present: raise DepositError( ERROR_CONTENT, "Only 1 application/zip (or application/x-tar) archive " "and 1 atom+xml entry is supported (as per sword2.0 " "specification)", "You provided more than 1 application/(zip|x-tar) " "or more than 1 application/atom+xml content-disposition " "header in the multipart deposit", ) content_types_present.add(content_type) assert content_type is not None data[content_type] = fh if len(content_types_present) != 2: raise DepositError( ERROR_CONTENT, "You must provide both 1 application/zip (or " "application/x-tar) and 1 atom+xml entry for multipart " "deposit", "You need to provide only 1 application/(zip|x-tar) " "and 1 application/atom+xml content-disposition header " "in the multipart deposit", ) - filehandler = data["application/zip"] - if not filehandler: - filehandler = data["application/x-tar"] + filehandler = data["application/zip"] or data["application/x-tar"] + if filehandler is None: + raise DepositError( + BAD_REQUEST, + "You must provide an archive, either as application/zip or " + "application/x-tar", + ) assert isinstance(filehandler, UploadedFile), filehandler self._check_file_length(filehandler) self._check_file_md5sum(filehandler, headers.content_md5sum) + if data["application/atom+xml"] is None: + raise DepositError( + BAD_REQUEST, "You must provide an application/atom+xml entry." + ) + try: raw_metadata, metadata_tree = self._read_metadata( data["application/atom+xml"] ) except ParserError: raise DepositError( PARSING_ERROR, "Malformed xml metadata", "The xml received is malformed. " "Please ensure your metadata file is correctly formatted.", ) self._set_deposit_origin_from_metadata(deposit, metadata_tree, headers) # actual storage of data self._deposit_put( deposit=deposit, in_progress=headers.in_progress, ) deposit_request_data = { ARCHIVE_KEY: filehandler, RAW_METADATA_KEY: raw_metadata, } self._deposit_request_put( deposit, deposit_request_data, replace_metadata, replace_archives ) - assert filehandler is not None return Receipt( deposit_id=deposit.id, deposit_date=deposit.reception_date, archive=filehandler.name, status=deposit.status, ) def _store_metadata_deposit( self, deposit: Deposit, swhid_reference: Union[str, QualifiedSWHID], metadata_tree: ElementTree.Element, raw_metadata: bytes, deposit_origin: Optional[str] = None, ) -> Tuple[ExtendedSWHID, Deposit, DepositRequest]: """When all user inputs pass the checks, this associates the raw_metadata to the swhid_reference in the raw extrinsic metadata storage. In case of any issues, a bad request response is returned to the user with the details. Checks: - metadata are technically parsable - metadata pass the functional checks - SWHID (if any) is technically valid Args: deposit: Deposit reference swhid_reference: The swhid or the origin to attach metadata information to metadata_tree: Full element tree of metadata to check for validity (parsed out of raw_metadata) raw_metadata: The actual raw metadata to send in the storage metadata deposit_origin: Optional deposit origin url to use if any (e.g. deposit update scenario provides one) Raises: DepositError in case of incorrect inputs from the deposit client (e.g. functionally invalid metadata, ...) Returns: Tuple of target swhid, deposit, and deposit request """ metadata_ok, error_details = check_metadata(metadata_tree) if not metadata_ok: assert error_details, "Details should be set when a failure occurs" raise DepositError( BAD_REQUEST, "Functional metadata checks failure", convert_status_detail(error_details), ) metadata_authority = MetadataAuthority( type=MetadataAuthorityType.DEPOSIT_CLIENT, url=deposit.client.provider_url, ) metadata_fetcher = self.swh_deposit_fetcher() # replace metadata within the deposit backend deposit_request_data = { RAW_METADATA_KEY: raw_metadata, } # actually add the metadata to the completed deposit deposit_request = self._deposit_request_put(deposit, deposit_request_data) target_swhid: ExtendedSWHID # origin URL or CoreSWHID if isinstance(swhid_reference, str): target_swhid = Origin(swhid_reference).swhid() metadata_context = {} else: metadata_context = compute_metadata_context(swhid_reference) if deposit_origin: # metadata deposit update on completed deposit metadata_context["origin"] = deposit_origin target_swhid = extended_swhid_from_qualified(swhid_reference) self._check_swhid_in_archive(target_swhid) # metadata deposited by the client metadata_object = RawExtrinsicMetadata( target=target_swhid, # core swhid or origin discovery_date=deposit_request.date, authority=metadata_authority, fetcher=metadata_fetcher, format="sword-v2-atom-codemeta", metadata=raw_metadata, **metadata_context, ) # metadata on the metadata object swh_deposit_authority = self.swh_deposit_authority() swh_deposit_fetcher = self.swh_deposit_fetcher() metametadata_object = RawExtrinsicMetadata( target=metadata_object.swhid(), discovery_date=deposit_request.date, authority=swh_deposit_authority, fetcher=swh_deposit_fetcher, format="xml-deposit-info", metadata=render_to_string( "deposit/deposit_info.xml", context={"deposit": deposit} ).encode(), ) # write to metadata storage self.storage_metadata.metadata_authority_add( [metadata_authority, swh_deposit_authority] ) self.storage_metadata.metadata_fetcher_add( [metadata_fetcher, swh_deposit_fetcher] ) self.storage_metadata.raw_extrinsic_metadata_add( [metadata_object, metametadata_object] ) return (target_swhid, deposit, deposit_request) def _check_swhid_in_archive(self, target_swhid: ExtendedSWHID) -> None: """Check the target object already exists in the archive, and raises a BAD_REQUEST if it does not.""" if target_swhid.object_type in (ExtendedObjectType.CONTENT,): if list( self.storage.content_missing_per_sha1_git([target_swhid.object_id]) ): raise DepositError( BAD_REQUEST, f"Cannot load metadata on {target_swhid}, this content " f"object does not exist in the archive (yet?).", ) elif target_swhid.object_type in ( ExtendedObjectType.DIRECTORY, ExtendedObjectType.REVISION, ExtendedObjectType.RELEASE, ExtendedObjectType.SNAPSHOT, ): target_type_name = target_swhid.object_type.name.lower() method = getattr(self.storage, target_type_name + "_missing") if list(method([target_swhid.object_id])): raise DepositError( BAD_REQUEST, f"Cannot load metadata on {target_swhid}, this {target_type_name} " f"object does not exist in the archive (yet?).", ) elif target_swhid.object_type in (ExtendedObjectType.ORIGIN,): if None in list(self.storage.origin_get_by_sha1([target_swhid.object_id])): raise DepositError( BAD_REQUEST, "Cannot load metadata on origin, it is not (yet?) known to the " "archive.", ) else: # This should not happen, because target_swhid is generated from either # a core swhid or an origin URL. # Let's just check it again so the "switch" is exhaustive. raise ValueError( f"_check_swhid_in_archive expected core SWHID or origin SWHID, " f"but got {target_swhid}." ) def _atom_entry( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Deposit, replace_metadata: bool = False, replace_archives: bool = False, ) -> Receipt: """Atom entry deposit. Args: request: the request holding information to parse and inject in db headers: parsed request headers collection_name: the associated client deposit: deposit to be updated replace_metadata: 'Update or add' request to existing deposit. If False (default), this adds new metadata request to existing ones. Otherwise, this will replace existing metadata. replace_archives: 'Update or add' request to existing deposit. If False (default), this adds new archive request to existing ones. Otherwise, this will replace existing archives. ones. Raises: - 400 (bad request) if the request is not providing an external identifier - 400 (bad request) if the request's body is empty - 415 (unsupported media type) if a wrong media type is provided """ metadata_stream = request.data empty_atom_entry_summary = "Empty body request is not supported." empty_atom_entry_desc = ( "Atom entry request is about non-empty metadata deposit." ) if not metadata_stream: raise DepositError( BAD_REQUEST, empty_atom_entry_summary, empty_atom_entry_desc ) try: raw_metadata, metadata_tree = self._read_metadata(metadata_stream) except ParserError: raise DepositError( BAD_REQUEST, "Malformed xml metadata", "The xml received is malformed. " "Please ensure your metadata file is correctly formatted.", ) if len(metadata_tree) == 0: raise DepositError( BAD_REQUEST, empty_atom_entry_summary, empty_atom_entry_desc ) self._set_deposit_origin_from_metadata(deposit, metadata_tree, headers) # Determine if we are in the metadata-only deposit case try: swhid_ref = parse_swh_reference(metadata_tree) except ValidationError as e: raise DepositError( PARSING_ERROR, "Invalid SWHID reference", str(e), ) if swhid_ref is not None and ( deposit.origin_url or deposit.parent or deposit.external_id ): raise DepositError( BAD_REQUEST, " is for metadata-only deposits and " " / / Slug are for " "code deposits, only one may be used on a given deposit.", ) if swhid_ref is not None: # It's suggested to user to provide it metadata_provenance_url = parse_swh_metadata_provenance(metadata_tree) if metadata_provenance_url: # If the provenance is provided, ensure it matches client provider url check_url_match_provider( metadata_provenance_url, deposit.client.provider_url ) deposit.save() # We need a deposit id target_swhid, depo, depo_request = self._store_metadata_deposit( deposit, swhid_ref, metadata_tree, raw_metadata ) deposit.status = DEPOSIT_STATUS_LOAD_SUCCESS if isinstance(swhid_ref, QualifiedSWHID): deposit.swhid = str(extended_swhid_from_qualified(swhid_ref)) deposit.swhid_context = str(swhid_ref) deposit.type = DEPOSIT_METADATA_ONLY deposit.complete_date = depo_request.date deposit.reception_date = depo_request.date deposit.save() return Receipt( deposit_id=deposit.id, deposit_date=depo_request.date, status=deposit.status, archive=None, ) self._deposit_put( deposit=deposit, in_progress=headers.in_progress, ) self._deposit_request_put( deposit, {RAW_METADATA_KEY: raw_metadata}, replace_metadata, replace_archives, ) return Receipt( deposit_id=deposit.id, deposit_date=deposit.reception_date, status=deposit.status, archive=None, ) def _set_deposit_origin_from_metadata(self, deposit, metadata, headers): (create_origin, add_to_origin) = parse_swh_deposit_origin(metadata) if create_origin and add_to_origin: raise DepositError( BAD_REQUEST, " and are mutually exclusive, " "as they respectively create a new origin and add to an existing " "origin.", ) if create_origin: origin_url = create_origin check_url_match_provider(origin_url, deposit.client.provider_url) deposit.origin_url = origin_url if add_to_origin: origin_url = add_to_origin check_url_match_provider(origin_url, deposit.client.provider_url) deposit.parent = ( Deposit.objects.filter( client=deposit.client, origin_url=origin_url, status=DEPOSIT_STATUS_LOAD_SUCCESS, ) .order_by("-id")[0:1] .get() ) deposit.origin_url = origin_url external_identifier_element = metadata.find( "atom:external_identifier", namespaces=NAMESPACES ) if external_identifier_element is not None: # Deprecated tag. # When clients stopped using it, this should raise an error # unconditionally if deposit.origin_url: raise DepositError( BAD_REQUEST, " is deprecated, you should only use " " and from now on.", ) if headers.slug and external_identifier_element.text != headers.slug: raise DepositError( BAD_REQUEST, "The tag and Slug header are deprecated, " " or " "should be used instead.", ) def _empty_post( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Deposit, ) -> Receipt: """Empty post to finalize a deposit. Args: request: the request holding information to parse and inject in db headers: parsed request headers collection_name: the associated client deposit: deposit to be finalized """ self._complete_deposit(deposit) assert deposit.complete_date is not None return Receipt( deposit_id=deposit.id, deposit_date=deposit.complete_date, status=deposit.status, archive=None, ) def additional_checks( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Optional[Deposit], ) -> Dict[str, Any]: """Permit the child class to enrich additional checks. Returns: dict with 'error' detailing the problem. """ return {} def get_client(self, request) -> DepositClient: # This class depends on AuthenticatedAPIView, so request.user.username # is always set username = request.user.username assert username is not None if self._client is None: try: self._client = DepositClient.objects.get( # type: ignore username=username ) except DepositClient.DoesNotExist: raise DepositError(NOT_FOUND, f"Unknown client name {username}") assert self._client.username == username return self._client def checks( self, request: Request, collection_name: str, deposit: Optional[Deposit] = None ) -> ParsedRequestHeaders: if deposit is None: collection = get_collection_by_name(collection_name) else: assert collection_name == deposit.collection.name collection = deposit.collection client = self.get_client(request) collection_id = collection.id collections = client.collections assert collections is not None if collection_id not in collections: raise DepositError( FORBIDDEN, f"Client {client.username} cannot access collection {collection_name}", ) headers = self._read_headers(request) if deposit is not None: self.restrict_access(request, headers, deposit) if headers.on_behalf_of: raise DepositError(MEDIATION_NOT_ALLOWED, "Mediation is not supported.") self.additional_checks(request, headers, collection_name, deposit) return headers def restrict_access( self, request: Request, headers: ParsedRequestHeaders, deposit: Deposit ) -> None: """Allow modifications on deposit with status 'partial' only, reject the rest.""" if request.method != "GET" and deposit.status != DEPOSIT_STATUS_PARTIAL: summary = "You can only act on deposit with status '%s'" % ( DEPOSIT_STATUS_PARTIAL, ) description = f"This deposit has status '{deposit.status}'" raise DepositError( BAD_REQUEST, summary=summary, verbose_description=description ) def _basic_not_allowed_method(self, request: Request, method: str): raise DepositError( METHOD_NOT_ALLOWED, f"{method} method is not supported on this endpoint", ) def get( self, request: Request, *args, **kwargs ) -> Union[HttpResponse, FileResponse]: return self._basic_not_allowed_method(request, "GET") def post(self, request: Request, *args, **kwargs) -> HttpResponse: return self._basic_not_allowed_method(request, "POST") def put(self, request: Request, *args, **kwargs) -> HttpResponse: return self._basic_not_allowed_method(request, "PUT") def delete(self, request: Request, *args, **kwargs) -> HttpResponse: return self._basic_not_allowed_method(request, "DELETE") class APIGet(APIBase, metaclass=ABCMeta): """Mixin for class to support GET method.""" def get( # type: ignore self, request: Request, collection_name: str, deposit_id: int ) -> Union[HttpResponse, FileResponse]: """Endpoint to create/add resources to deposit. Returns: 200 response when no error during routine occurred 400 if the deposit does not belong to the collection 404 if the deposit or the collection does not exist """ deposit = get_deposit_by_id(deposit_id, collection_name) self.checks(request, collection_name, deposit) r = self.process_get(request, collection_name, deposit) status, content, content_type = r if content_type == "swh/generator": with content as path: return FileResponse( open(path, "rb"), status=status, content_type="application/tar" ) if content_type == "application/json": return HttpResponse( json.dumps(content), status=status, content_type=content_type ) return HttpResponse(content, status=status, content_type=content_type) @abstractmethod def process_get( self, request: Request, collection_name: str, deposit: Deposit ) -> Tuple[int, Any, str]: """Routine to deal with the deposit's get processing. Returns: Tuple status, stream of content, content-type """ pass class APIPost(APIBase, metaclass=ABCMeta): """Mixin for class to support POST method.""" def post( # type: ignore self, request: Request, collection_name: str, deposit_id: Optional[int] = None ) -> HttpResponse: """Endpoint to create/add resources to deposit. Returns: 204 response when no error during routine occurred. 400 if the deposit does not belong to the collection 404 if the deposit or the collection does not exist """ if deposit_id is None: deposit = None else: deposit = get_deposit_by_id(deposit_id, collection_name) headers = self.checks(request, collection_name, deposit) status, iri_key, receipt = self.process_post( request, headers, collection_name, deposit ) return self._make_deposit_receipt( request, collection_name, status, iri_key, receipt, ) def _make_deposit_receipt( self, request, collection_name: str, status: int, iri_key: str, receipt: Receipt, ) -> HttpResponse: """Returns an HttpResponse with a SWORD Deposit receipt as content.""" # Build the IRIs in the receipt args = [collection_name, receipt.deposit_id] iris = { iri: request.build_absolute_uri(reverse(iri, args=args)) for iri in [EM_IRI, EDIT_IRI, CONT_FILE_IRI, SE_IRI, STATE_IRI] } context = { **attr.asdict(receipt), **iris, "packagings": ACCEPT_PACKAGINGS, } response = render( request, "deposit/deposit_receipt.xml", context=context, content_type="application/xml", status=status, ) response["Location"] = iris[iri_key] return response @abstractmethod def process_post( self, request, headers: ParsedRequestHeaders, collection_name: str, deposit: Optional[Deposit] = None, ) -> Tuple[int, str, Receipt]: """Routine to deal with the deposit's processing. Returns Tuple of: - response status code (200, 201, etc...) - key iri (EM_IRI, EDIT_IRI, etc...) - Receipt """ pass class APIPut(APIBase, metaclass=ABCMeta): """Mixin for class to support PUT method.""" def put( # type: ignore self, request: Request, collection_name: str, deposit_id: int ) -> HttpResponse: """Endpoint to update deposit resources. Returns: 204 response when no error during routine occurred. 400 if the deposit does not belong to the collection 404 if the deposit or the collection does not exist """ if deposit_id is None: deposit = None else: deposit = get_deposit_by_id(deposit_id, collection_name) headers = self.checks(request, collection_name, deposit) self.process_put(request, headers, collection_name, deposit) return HttpResponse(status=status.HTTP_204_NO_CONTENT) @abstractmethod def process_put( self, request: Request, headers: ParsedRequestHeaders, collection_name: str, deposit: Deposit, ) -> None: """Routine to deal with updating a deposit in some way. Returns dictionary of the processing result """ pass class APIDelete(APIBase, metaclass=ABCMeta): """Mixin for class to support DELETE method.""" def delete( # type: ignore self, request: Request, collection_name: str, deposit_id: Optional[int] = None ) -> HttpResponse: """Endpoint to delete some deposit's resources (archives, deposit). Returns: 204 response when no error during routine occurred. 400 if the deposit does not belong to the collection 404 if the deposit or the collection does not exist """ assert deposit_id is not None deposit = get_deposit_by_id(deposit_id, collection_name) self.checks(request, collection_name, deposit) self.process_delete(request, collection_name, deposit) return HttpResponse(status=status.HTTP_204_NO_CONTENT) @abstractmethod def process_delete( self, request: Request, collection_name: str, deposit: Deposit ) -> None: """Routine to delete a resource. This is mostly not allowed except for the EM_IRI (cf. .api.deposit_update.APIUpdateArchive) """ pass diff --git a/swh/deposit/tests/api/test_collection_post_multipart.py b/swh/deposit/tests/api/test_collection_post_multipart.py index 20f1bdb4..812263e9 100644 --- a/swh/deposit/tests/api/test_collection_post_multipart.py +++ b/swh/deposit/tests/api/test_collection_post_multipart.py @@ -1,393 +1,471 @@ # Copyright (C) 2017-2021 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information """Tests handling of multipart requests to POST Col-IRI.""" from io import BytesIO import uuid from django.core.files.uploadedfile import InMemoryUploadedFile from django.urls import reverse_lazy as reverse import pytest from rest_framework import status from swh.deposit.config import COL_IRI, DEPOSIT_STATUS_DEPOSITED from swh.deposit.models import Deposit, DepositRequest from swh.deposit.parsers import parse_xml from swh.deposit.tests.common import check_archive, post_multipart from swh.deposit.utils import NAMESPACES def test_post_deposit_multipart( authenticated_client, deposit_collection, atom_dataset, mocker, deposit_user, sample_archive, ): # given external_id = "foobar" origin_url = deposit_user.provider_url + external_id url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry = atom_dataset["entry-data0"] % origin_url # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry, HTTP_IN_PROGRESS="false", ) assert response.status_code == status.HTTP_201_CREATED response_content = parse_xml(response.content) deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES)) deposit = Deposit.objects.get(pk=deposit_id) assert deposit.collection == deposit_collection assert deposit.origin_url == origin_url assert deposit.status == DEPOSIT_STATUS_DEPOSITED def test_post_deposit_multipart_without_origin_url( authenticated_client, deposit_collection, atom_dataset, mocker, deposit_user, sample_archive, ): # given url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry = atom_dataset["entry-data-deposit-binary"] id_ = str(uuid.uuid4()) mocker.patch("uuid.uuid4", return_value=id_) # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry, HTTP_IN_PROGRESS="false", ) assert response.status_code == status.HTTP_201_CREATED response_content = parse_xml(response.content) deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES)) deposit = Deposit.objects.get(pk=deposit_id) assert deposit.collection == deposit_collection assert deposit.origin_url == deposit_user.provider_url + id_ assert deposit.status == DEPOSIT_STATUS_DEPOSITED def test_post_deposit_multipart_zip( authenticated_client, deposit_collection, atom_dataset, sample_archive ): """one multipart deposit (zip+xml) should be accepted""" # given url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry = atom_dataset["entry-data-deposit-binary"] external_id = "external-id" # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry, HTTP_IN_PROGRESS="false", HTTP_SLUG=external_id, ) # then assert response.status_code == status.HTTP_201_CREATED response_content = parse_xml(response.content) deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES)) deposit = Deposit.objects.get(pk=deposit_id) assert deposit.status == DEPOSIT_STATUS_DEPOSITED assert deposit.external_id == external_id assert deposit.collection == deposit_collection assert deposit.swhid is None deposit_requests = DepositRequest.objects.filter(deposit=deposit) assert len(deposit_requests) == 2 for deposit_request in deposit_requests: assert deposit_request.deposit == deposit if deposit_request.type == "archive": check_archive(sample_archive["name"], deposit_request.archive.name) assert deposit_request.raw_metadata is None else: assert ( parse_xml(deposit_request.raw_metadata).findtext( "atom:id", namespaces=NAMESPACES ) == "urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a" ) assert deposit_request.raw_metadata == data_atom_entry def test_post_deposit_multipart_tar( authenticated_client, deposit_collection, atom_dataset, sample_archive ): """one multipart deposit (tar+xml) should be accepted""" # given url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry = atom_dataset["entry-data-deposit-binary"] external_id = "external-id" # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry, HTTP_IN_PROGRESS="false", HTTP_SLUG=external_id, ) # then assert response.status_code == status.HTTP_201_CREATED response_content = parse_xml(response.content) deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES)) deposit = Deposit.objects.get(pk=deposit_id) assert deposit.status == DEPOSIT_STATUS_DEPOSITED assert deposit.external_id == external_id assert deposit.collection == deposit_collection assert deposit.swhid is None deposit_requests = DepositRequest.objects.filter(deposit=deposit) assert len(deposit_requests) == 2 for deposit_request in deposit_requests: assert deposit_request.deposit == deposit if deposit_request.type == "archive": check_archive(sample_archive["name"], deposit_request.archive.name) assert deposit_request.raw_metadata is None else: assert ( parse_xml(deposit_request.raw_metadata).findtext( "atom:id", namespaces=NAMESPACES ) == "urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a" ) assert deposit_request.raw_metadata == data_atom_entry def test_post_deposit_multipart_put_to_replace_metadata( authenticated_client, deposit_collection, atom_dataset, sample_archive ): """One multipart deposit followed by a metadata update should be accepted """ # given url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry = atom_dataset["entry-data-deposit-binary"] external_id = "external-id" # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry, HTTP_IN_PROGRESS="true", HTTP_SLUG=external_id, ) # then assert response.status_code == status.HTTP_201_CREATED response_content = parse_xml(response.content) deposit_id = int(response_content.findtext("swh:deposit_id", namespaces=NAMESPACES)) deposit = Deposit.objects.get(pk=deposit_id) assert deposit.status == "partial" assert deposit.external_id == external_id assert deposit.collection == deposit_collection assert deposit.swhid is None deposit_requests = DepositRequest.objects.filter(deposit=deposit) assert len(deposit_requests) == 2 for deposit_request in deposit_requests: assert deposit_request.deposit == deposit if deposit_request.type == "archive": check_archive(sample_archive["name"], deposit_request.archive.name) else: assert ( parse_xml(deposit_request.raw_metadata).findtext( "atom:id", namespaces=NAMESPACES ) == "urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a" ) assert deposit_request.raw_metadata == data_atom_entry replace_metadata_uri = response["location"] response = authenticated_client.put( replace_metadata_uri, content_type="application/atom+xml;type=entry", data=atom_dataset["entry-data-deposit-binary"], HTTP_IN_PROGRESS="false", ) assert response.status_code == status.HTTP_204_NO_CONTENT # deposit_id did not change deposit = Deposit.objects.get(pk=deposit_id) assert deposit.status == DEPOSIT_STATUS_DEPOSITED assert deposit.external_id == external_id assert deposit.collection == deposit_collection assert deposit.swhid is None deposit_requests = DepositRequest.objects.filter(deposit=deposit) assert len(deposit_requests) == 2 for deposit_request in deposit_requests: assert deposit_request.deposit == deposit if deposit_request.type == "archive": check_archive(sample_archive["name"], deposit_request.archive.name) else: assert ( parse_xml(deposit_request.raw_metadata).findtext( "atom:id", namespaces=NAMESPACES ) == "urn:uuid:1225c695-cfb8-4ebb-aaaa-80da344efa6a" ) assert ( deposit_request.raw_metadata == atom_dataset["entry-data-deposit-binary"] ) # FAILURE scenarios -def test_post_deposit_multipart_only_archive_and_atom_entry( +def test_post_deposit_multipart_only_one_archive_and_atom_entry( authenticated_client, deposit_collection ): """Multipart deposit only accepts one archive and one atom+xml""" # given url = reverse(COL_IRI, args=[deposit_collection.name]) archive_content = b"some content representing archive" archive = InMemoryUploadedFile( BytesIO(archive_content), field_name="archive0", name="archive0", content_type="application/x-tar", size=len(archive_content), charset=None, ) other_archive_content = b"some-other-content" other_archive = InMemoryUploadedFile( BytesIO(other_archive_content), field_name="atom0", name="atom0", content_type="application/x-tar", size=len(other_archive_content), charset="utf-8", ) # when response = authenticated_client.post( url, format="multipart", data={ "archive": archive, "atom_entry": other_archive, }, # + headers HTTP_IN_PROGRESS="false", HTTP_SLUG="external-id", ) # then assert response.status_code == status.HTTP_415_UNSUPPORTED_MEDIA_TYPE assert ( "Only 1 application/zip (or application/x-tar) archive" in response.content.decode("utf-8") ) # when archive.seek(0) response = authenticated_client.post( url, format="multipart", data={ "archive": archive, }, # + headers HTTP_IN_PROGRESS="false", HTTP_SLUG="external-id", ) # then assert response.status_code == status.HTTP_415_UNSUPPORTED_MEDIA_TYPE assert ( "You must provide both 1 application/zip (or " "application/x-tar) and 1 atom+xml entry for " "multipart deposit" in response.content.decode("utf-8") ) is True def test_post_deposit_multipart_400_when_badly_formatted_xml( authenticated_client, deposit_collection, sample_archive, atom_dataset ): # given url = reverse(COL_IRI, args=[deposit_collection.name]) data_atom_entry_ko = atom_dataset["entry-data-ko"] # when response = post_multipart( authenticated_client, url, sample_archive, data_atom_entry_ko, HTTP_IN_PROGRESS="false", HTTP_SLUG="external-id", ) assert b"Malformed xml metadata" in response.content assert response.status_code == status.HTTP_400_BAD_REQUEST def test_post_deposit_multipart_if_upload_size_limit_exceeded( authenticated_client, deposit_collection, atom_dataset, sample_archive ): # given url = reverse(COL_IRI, args=[deposit_collection.name]) archive = { **sample_archive, "data": sample_archive["data"] * 100, } data_atom_entry = atom_dataset["entry-data-deposit-binary"] external_id = "external-id" # when response = post_multipart( authenticated_client, url, archive, data_atom_entry, HTTP_IN_PROGRESS="false", HTTP_SLUG=external_id, ) # then assert response.status_code == status.HTTP_413_REQUEST_ENTITY_TOO_LARGE assert b"Upload size limit exceeded" in response.content with pytest.raises(Deposit.DoesNotExist): Deposit.objects.get(external_id=external_id) + + +def test_post_deposit_atom_400_multipart_no_atom( + authenticated_client, deposit_collection, atom_dataset, deposit_user, sample_archive +): + """Posting without an atom body should return a 400 response""" + origin_url = "http://example.org/foo" + + archive = InMemoryUploadedFile( + BytesIO(sample_archive["data"]), + field_name=sample_archive["name"], + name=sample_archive["name"], + content_type="application/x-tar", + size=sample_archive["length"], + charset=None, + ) + + atom_data = atom_dataset["entry-data0"] % origin_url + + atom_entry = InMemoryUploadedFile( + BytesIO(atom_data.encode("utf-8")), + field_name="atom0", + name="atom0", + content_type="application/x-foobar", # should be application/atom+xml + size=len(atom_data), + charset="utf-8", + ) + + response = authenticated_client.post( + reverse(COL_IRI, args=[deposit_collection.name]), + format="multipart", + data={ + "archive": archive, + "atom_entry": atom_entry, + }, + ) + + assert response.status_code == status.HTTP_400_BAD_REQUEST + assert "provide an application/atom+xml entry" in response.content.decode() + + +def test_post_deposit_atom_400_multipart_no_archive( + authenticated_client, deposit_collection, atom_dataset, deposit_user, sample_archive +): + """Posting without an atom body should return a 400 response""" + origin_url = "http://example.org/foo" + + archive = InMemoryUploadedFile( + BytesIO(sample_archive["data"]), + field_name=sample_archive["name"], + name=sample_archive["name"], + content_type="application/x-foobar", # should be application/x-tar + size=sample_archive["length"], + charset=None, + ) + + atom_data = atom_dataset["entry-data0"] % origin_url + + atom_entry = InMemoryUploadedFile( + BytesIO(atom_data.encode("utf-8")), + field_name="atom0", + name="atom0", + content_type="application/atom+xml", + size=len(atom_data), + charset="utf-8", + ) + + response = authenticated_client.post( + reverse(COL_IRI, args=[deposit_collection.name]), + format="multipart", + data={ + "archive": archive, + "atom_entry": atom_entry, + }, + ) + + assert response.status_code == status.HTTP_400_BAD_REQUEST + assert "provide an archive" in response.content.decode()