def register_document(folder: str, session_db, storage) -> None:
logger.info("Processando a Pasta %s", folder)
list_files = files.list_files(folder)
obj_xml = None
prefix = ""
xml_files = files.xml_files_list(folder)
_renditions = list(
filter(lambda file: ".pdf" in file or ".html" in file, list_files))
if len(xml_files) > 1:
raise exceptions.XMLError("Existe %s xmls no pacote SPS",
len(xml_files))
else:
try:
x_file = xml_files[0]
except IndexError as ex:
raise exceptions.XMLError("Não existe XML no pacote SPS: %s", ex)
xml_path = os.path.join(folder, x_file)
obj_xml = xml.loadToXML(xml_path)
xml_sps = SPS_Package(obj_xml)
# TODO: é possível que alguns artigos não possuam o self.acron
prefix = xml_sps.media_prefix
url_xml = storage.register(xml_path, prefix)
static_assets, static_additionals = get_document_assets_path(
obj_xml, list_files, folder)
registered_assets = put_static_assets_into_storage(static_assets, prefix,
storage)
for additional_path in static_additionals.values():
storage.register(os.path.join(additional_path), prefix)
if obj_xml:
renditions = get_document_renditions(folder, _renditions, prefix,
storage)
manifest_data = ManifestDomainAdapter(
manifest=manifest.get_document_manifest(
obj_xml, url_xml, registered_assets, renditions))
try:
session_db.documents.add(data=manifest_data)
session_db.changes.add({
"timestamp": utcnow(),
"entity": "Document",
"id": manifest_data.id()
})
logger.info("Document-store save: %s", manifest_data.id())
except AlreadyExists as exc:
logger.exception(exc)
return obj_xml, manifest_data.id()
def register_document(folder: str, session_db, storage) -> None:
logger.info("Processando a Pasta %s", folder)
list_files = files.list_files(folder)
obj_xml = None
prefix = ""
xml_files = files.xml_files_list(folder)
medias_files = set(list_files) - set(xml_files)
if len(xml_files) > 1:
raise exceptions.XMLError("Existe %s xmls no pacote SPS",
len(xml_files))
else:
try:
x_file = xml_files[0]
except IndexError as ex:
raise exceptions.XMLError("Não existe XML no pacote SPS: %s", ex)
xml_path = os.path.join(folder, x_file)
obj_xml = xml.loadToXML(xml_path)
xml_sps = SPS_Package(obj_xml)
prefix = xml_sps.media_prefix
url_xml = storage.register(xml_path, prefix)
assets = []
for m_file in medias_files:
assets.append({
"asset_id":
m_file,
"asset_url":
storage.register(os.path.join(folder, m_file), prefix),
})
if obj_xml:
manifest_data = ManifestDomainAdapter(
manifest=manifest.get_document_manifest(obj_xml, url_xml, assets))
try:
session_db.documents.add(data=manifest_data)
session_db.changes.add({
"timestamp": utcnow(),
"entity": "Document",
"id": manifest_data.id()
})
logger.info("Document-store save: %s", manifest_data.id())
except AlreadyExists as exc:
logger.exception(exc)
return obj_xml, manifest_data.id()
def media_prefix(self):
if not self.scielo_id:
raise exceptions.XMLError("Não existe Scielo-Id no XML: %s",
repr(self))
return f"{self.issn}/{self.scielo_id}"
def register_document(folder: str, session, storage, pid_database_engine, poison_pill=PoisonPill()) -> None:
"""Registra registra pacotes SPS em uma instância do Kernel e seus
ativos digitais em um object storage."""
if poison_pill.poisoned:
return
logger.debug("Starting the import step for '%s' package.", folder)
package_files = files.list_files(folder)
xmls = files.xml_files_list(folder)
if xmls is None or len(xmls) == 0:
raise exceptions.XMLError(
"There is no XML file into package '%s'. Please verify and try later."
% folder
) from None
xml_path = os.path.join(folder, xmls[0])
constructor.article_xml_constructor(xml_path, folder, pid_database_engine, False)
try:
obj_xml = xml.loadToXML(xml_path)
except lxml.etree.ParseError as exc:
raise exceptions.XMLError(
"Could not parse the '%s' file, please validate"
" this file before then try to import again." % xml_path,
) from None
xml_sps = SPS_Package(obj_xml)
pid_v3 = xml_sps.scielo_pid_v3
try:
session.documents.fetch(id=pid_v3)
except DoesNotExist:
pass
else:
logger.debug(
"Document '%s' already exist in kernel. Returning article result information",
pid_v3,
)
return get_article_result_dict(xml_sps)
prefix = xml_sps.media_prefix or ""
url_xml = storage.register(xml_path, prefix)
static_assets, static_additionals = get_document_assets_path(
obj_xml, package_files, folder
)
registered_assets = put_static_assets_into_storage(static_assets, prefix, storage)
for additional_path in static_additionals.values():
storage.register(os.path.join(additional_path), prefix)
renditions = get_document_renditions(folder, prefix, storage)
document = Document(
manifest=manifest.get_document_manifest(
xml_sps, url_xml, registered_assets, renditions
)
)
try:
add_document(session, document)
if renditions:
add_renditions(session, document)
except AlreadyExists as exc:
logger.error(exc)
else:
logger.debug("Document with id '%s' was imported.", document.id())
return get_article_result_dict(xml_sps)
def media_prefix(self):
if not self.scielo_pid_v3:
raise exceptions.XMLError("Não existe scielo-pid-v3")
return f"{self.issn}/{self.scielo_pid_v3}"